You are on page 1of 55

GUIDE TO ASSET MANAGEMENT

Part 5B: Roughness


Guide to Asset Management
Part 5B: Roughness
Guide to Asset Management Part 5B: Roughness

Summary
This report contains guidelines for, and background notes on, the conduct of response type and
profile-based roughness measurement for road network management purposes in Australia and
New Zealand. The guidelines define roughness as being concerned with road surface profile
wavelengths between 0.5 m and 50 m. The guidelines describe two broad measures of
roughness, viz NAASRA Roughness Meter (NRM roughness values in counts/km) and the
International Roughness Index (with a specific focus on Lane IRIqc, expressed in units of m/km),
and mark the introduction to Australasia of Lane IRIqc as a standard form of reporting road
pavement roughness. A standard reporting interval of 100 m is favoured. The guidelines discuss
the frequency and extent of network roughness surveys, and reference separate Austroads
specifications and test methods for undertaking network roughness surveys. A glossary of terms
used in measuring road roughness is also included.

Keywords
Data collection, data analysis, evenness, International Roughness Index (IRI), measurement,
measuring equipment, monitoring, pavement evaluation, profilometer, riding quality, road
management, road network, road profile, road surface properties, roughness, roughness meter

First Published 2006


2nd edition September 2007

© Austroads Inc. 2007

This work is copyright. Apart from any use as permitted under the Copyright Act 1968, no part may
be reproduced by any process without the prior written permission of Austroads.

National Library of Australia Cataloguing-in-Publication data:


ISBN 978-1-921329-21-0

Austroads Project No. AP1009


Austroads Publication No. AGAM05B/07

Project Manager
Dr Wesley Soet, MR WA

Prepared by
Michael Moffatt, ARRB Group

Published by Austroads Incorporated


Level 9, Robell House
287 Elizabeth Street
Sydney NSW 2000 Australia
Phone: +61 2 9264 7088
Fax: +61 2 9264 1657
Email: austroads@austroads.com.au
www.austroads.com.au

This Guide is produced by Austroads as a general guide. Its application is discretionary. Road
authorities may vary their practice according to local circumstances and policies.

Austroads believes this publication to be correct at the time of printing and does not accept
responsibility for any consequences arising from the use of information herein. Readers should
rely on their own skill and judgement to apply information to particular issues.
Guide to Asset Management
Part 5B: Roughness

Sydney 2007
Austroads profile
Austroads’ purpose is to contribute to improved Australian and New Zealand transport outcomes
by:
ƒ providing expert advice to SCOT and ATC on road and road transport issues
ƒ facilitating collaboration between road agencies
ƒ promoting harmonisation, consistency and uniformity in road and related operations
ƒ undertaking strategic research on behalf of road agencies and communicating outcomes
ƒ promoting improved and consistent practice by road agencies.

Austroads membership
Austroads membership comprises the six state and two territory road transport and traffic
authorities and the Australian Department of Transport and Regional Services in Australia, the
Australian Local Government Association and Transit New Zealand. It is governed by a council
consisting of the chief executive officer (or an alternative senior executive officer) of each of its
eleven member organisations:

ƒ Roads and Traffic Authority New South Wales


ƒ Roads Corporation Victoria
ƒ Department of Main Roads Queensland
ƒ Main Roads Western Australia
ƒ Department for Transport, Energy and Infrastructure South Australia
ƒ Department of Infrastructure, Energy and Resources Tasmania
ƒ Department of Planning and Infrastructure Northern Territory
ƒ Department of Territory and Municipal Services Australian Capital Territory
ƒ Australian Department of Transport and Regional Services
ƒ Australian Local Government Association
ƒ Transit New Zealand

The success of Austroads is derived from the collaboration of member organisations and others in
the road industry. It aims to be the Australasian leader in providing high quality information, advice
and fostering research in the road sector.
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

SUMMARY
These guidelines constitute Part 5B of the Guide to Asset Management, published by Austroads.

Austroads recognises the importance of road network roughness data as an input to a range of
asset management decisions as well as to a number of corporate performance indicators used by
road agencies throughout Australia and New Zealand. The purpose of these guidelines is to
promote consistency and improved quality in measuring and reporting road roughness throughout
Australia and New Zealand.

In the context of these guidelines, the term ‘roughness’ has a precise technical meaning.
However, in dealing with non-technical stakeholders, road agencies use terms such as ‘ride
comfort’, ‘rideability’, ‘smoothness’, and ‘unevenness’ as synonyms.

Whilst these guidelines are intended to be as independent as possible of the technology used for
measuring roughness, they have been prepared in the context that inertial laser profilometers are
the only devices currently conducting sealed road network level roughness surveys in Australasia.

Austroads has endorsed the International Roughness Index (IRI) as the reporting unit for road
roughness in Australasia. After a considerable transition period, continuing use of the traditional
reporting unit (NAASRA roughness meter counts, or NRM) is no longer considered appropriate and
all roughness results should now be reported exclusively in terms of IRI.

The guidelines discuss the frequency and extent of network roughness surveys, and reference
separate Austroads specifications and test methods for undertaking network roughness surveys.

The document also discusses the possible analyses that can be conducted with roughness data,
and highlights the contribution that roughness data can make to sophisticated decision-support
systems. Discussion on the use of roughness data as a performance indicator is also included.

The uses of roughness data at both the network and project level are discussed.

The guidelines are based in large part on material previously published by Austroads as Guidelines
for Road Condition Monitoring, Part 1 – Pavement Roughness (Austroads 2001a) (AP-G65.1/01).
The material has been restructured, and new content added, to form part of the Guide to Asset
Management.

Austroads 2007

— i—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

CONTENTS
1 INTRODUCTION ............................................................................................................ 1
1.1 Context within the Guide to Asset Management............................................................. 1
1.2 Overview of Roughness.................................................................................................. 1
1.3 Objective......................................................................................................................... 2
2 SPECIFICATION AND TEST METHOD ........................................................................ 4
2.1 Definition of Roughness.................................................................................................. 4
2.2 Frequency of Roughness Surveys.................................................................................. 4
2.3 Specification of Roughness Surveys .............................................................................. 5
2.4 Data Reporting................................................................................................................ 6
3 TOOLS AND TECHNIQUES .......................................................................................... 7
3.1 Classes of Measuring Devices ....................................................................................... 7
3.2 Measurement of Roughness........................................................................................... 7
3.3 Profile Analysis ............................................................................................................... 8
3.3.1 General ............................................................................................................. 8
3.3.2 Quarter-Car model (endorsed by Austroads) ................................................... 8
3.3.3 Half-Car Model (not endorsed by Austroads) ................................................... 9
4 DATA QUALITY AND VALIDATION ........................................................................... 10
4.1 Validation of Distance Measurement ............................................................................ 10
4.2 Validation of Roughness Measurement........................................................................ 10
4.3 Repeatability and Bias .................................................................................................. 10
5 ANALYSIS.................................................................................................................... 11
5.1 Assessing the Validity of Roughness Data ................................................................... 11
5.2 Distribution Analysis ..................................................................................................... 12
5.3 Performance Indicators................................................................................................. 12
5.4 Other Analysis Techniques........................................................................................... 12
6 APPLICATION ............................................................................................................. 14
6.1 General ......................................................................................................................... 14
6.2 Network Screening ....................................................................................................... 14
6.3 Work Prioritisation......................................................................................................... 14
6.4 Performance Indicators................................................................................................. 14
6.5 Intervention Levels........................................................................................................ 15
6.6 Decision-support Tools ................................................................................................. 15
6.7 Network Performance Scenario Analysis ..................................................................... 16
6.8 Budget Forecasting....................................................................................................... 16
7 PROJECT LEVEL APPLICATION............................................................................... 17
7.1 Roughness Collection at the Project Level ................................................................... 17
7.2 Auditing of Pavement Construction Ride Quality.......................................................... 18
7.3 Roughness Data as a Diagnostic Tool ......................................................................... 18
REFERENCES ...................................................................................................................... 20

Austroads 2007

— ii —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

COMMENTARY A GLOSSARY OF TERMS RELATED TO ROAD PAVEMENT


ROUGHNESS MEASUREMENT .......................................................... 22
COMMENTARY B A BRIEF HISTORY OF ROAD ROUGHNESS MEASUREMENT ........ 24
COMMENTARY C DEFINITION OF ROUGHNESS ............................................................ 25
COMMENTARY D FREQUENCY OF ROUGHNESS SURVEYS........................................ 26
COMMENTARY E SPECIFICATION AND TEST METHODS............................................. 27
COMMENTARY F DATA REPORTING .............................................................................. 31
COMMENTARY G MEASUREMENT OF ROUGHNESS .................................................... 32
COMMENTARY H PROFILE ANALYSIS............................................................................ 36
COMMENTARY I VALIDATION OF EQUIPMENT ............................................................ 40
COMMENTARY J NETWORK LEVEL DISTRIBUTION ANALYSIS EXAMPLE ............... 43
COMMENTARY K DETERMINATION OF SMOOTH TRAVEL EXPOSURE ..................... 46

Austroads 2007

— iii —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

TABLES
Table 2.1: Survey frequency.............................................................................................. 5
Table 2.2: Austroads standard specification and test methods for pavement
roughness measurement .................................................................................. 5
Table 3.1: World Bank classes of roughness measuring methods
(after Bennett et al. 2005) ................................................................................. 7
Table 7.1: Levels of roughness (after Austroads 2003)................................................... 18
Table 7.2: Pavement defects which may contribute to increased pavement
roughness ....................................................................................................... 19

FIGURES
Figure 1.1: Structure of the Guide to Asset Management with focus on Part 5 .................. 3
Figure 3.1: Inertial laser profilometer .................................................................................. 9
Figure 5.1: Roughness of a link over time ........................................................................ 11
Figure 7.1: Walking profiler ............................................................................................... 17

Austroads 2007

— iv —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

1 INTRODUCTION
1.1 Context within the Guide to Asset Management
These guidelines form a portion of the Guide to Asset Management published by Austroads. The
Guide aims to encapsulate and promote best practice and emerging issues with regard to the
management of road related assets. Preparation of the Guide is underway, and will ultimately be
comprised of six Parts, each covering specific aspects of asset management. The overall structure
of the Guide is contained in Figure 1.1.

These guidelines constitute Part 5B of the Guide to Asset Management, published by Austroads.

A considerable amount of these guidelines is based on material previously published by Austroads


as Guidelines for Road Condition Monitoring, Part 1 – Pavement Roughness (AP-G65.1/01).

1.2 Overview of Roughness


Pavement roughness is the most common objective measure of the general condition of a road.

Roughness is a condition parameter that characterises deviations from the intended longitudinal
profile of a pavement. Measurement of roughness focuses on characteristic dimensions that affect
vehicle dynamics and hence road user costs, ride quality and dynamic pavement loads.

When communicating with non-technical stakeholders, road owners often use terms such as ‘ride
comfort’, ‘rideability’, or ‘smoothness’ as synonyms for roughness.

A glossary of terms used with regard to the network level measurement of road roughness is
available together with a brief history of roughness measurement.

Roughness information can be used for a variety of purposes:


ƒ road network condition monitoring (including performance measures)
ƒ screening of candidate treatment sections
ƒ network level prioritisation and whole of life cycle cost analysis
ƒ quality assessment of new works at a project level
ƒ research (e.g. deterioration modelling, development of truck ride indicators).

Roughness information is also used to assist in:


ƒ assessing the suitability of a road for its users
ƒ assessing the relative condition of roads and networks
ƒ predicting the cost of travelling on the road.

Not all high roughness readings are the result of pavement distress. Additionally, it should be
remembered that roughness is usually determined by modelling the response of a car travelling at
80 km/h. High roughness readings at roundabouts, small local streets and other low speed
environments will suggest a poor ride comfort level which the travelling public, travelling at less
than 80 km/h, may not truly experience.

The guidelines aim to provide consistency in roughness specification and practice for road asset
management and are of relevance to contractors, condition monitoring service providers as well as
road agency personnel.

Austroads 2007

— 1—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

The advantages of a consistent approach include reduced establishment and overhead costs for
hardware and software, a stronger focus on quality, and improved opportunities for benchmarking
among road owners and service providers.

These guidelines describe currently preferred practice for the objective measurement and reporting
of road roughness, including testing of survey devices to ensure quality results.

1.3 Objective
The overall objective of these guidelines is to promote standardisation, so that:
ƒ useful road condition data of high quality is collected
ƒ cost efficiency of collecting and processing road condition data is improved
ƒ road condition data collected will be of increased value to road owners.

These guidelines are intended to assist in the collection and reporting of pavement roughness
information for monitoring road networks.

These guidelines are not intended to supersede the specific operating instructions for NAASRA
Roughness Meters, as set out in the NAASRA booklet Standard Operating Instructions for the
NAASRA Roughness Meter (NAASRA 1981). However, where there are differences of detail, then
these guidelines should generally be given precedence.

Austroads 2007

— 2—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

AUSTROADS GUIDE GUIDE TO ASSET MANAGEMENT

ASSET MANAGEMENT PART 1 – INTRODUCTION TO ASSET MANAGEMENT


(INCLUDING SCOPE OF OTHER PARTS)
BRIDGE TECHNOLOGY

PAVEMENT TECHNOLOGY PART 2 – STAKEHOLDER / COMMUNITY


REQUIREMENTS (OUTCOMES)
PROJECT DELIVERY

PROJECT EVALUATION PART 3 – ASSET STRATEGIES

ROAD DESIGN PART 4 – PROGRAM DEVELOPMENT AND


IMPLEMENTATION
ROAD SAFETY

ROAD TRANSPORT PLANNING PART 5 – PAVEMENT PERFORMANCE

TRAFFIC MANAGEMENT PART 6 – BRIDGE PERFORMANCE

PART 7 – ROAD RELATED ASSETS PERFORMANCE

PART 8 – ASSET VALUATION AND AUDIT

PART 5 – PAVEMENT PERFORMANCE

PART 5 – PAVEMENT PERFORMANCE

PART 5A – INVENTORY
PART 5B – ROUGHNESS

PART 5C – RUTTING

PART 5D – STRENGTH

PART 5E – CRACKING

PART 5F – SKID RESISTANCE

PART 5G – TEXTURE

PART 5H – PERFORMANCE MODELLING

Figure 1.1: Structure of the Guide to Asset Management with focus on Part 5

Austroads 2007

— 3—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

2 SPECIFICATION AND TEST METHOD


2.1 Definition of Roughness
Road roughness is a condition parameter which characterises deviations from the intended
longitudinal profile of a road surface with characteristic dimensions that affect vehicle dynamics
(and hence road user costs), ride quality and dynamic pavement loading. Roughness is a
measure of surface irregularities with wavelengths between 0.5 m and 50 m in the longitudinal
profiles of either or both wheelpaths in a traffic lane.

Historically roughness values were derived from the physical response of a vehicle to a road
surface (e.g. a response type device like the NAASRA Roughness Meter). NAASRA Roughness
Meter (NRM) systems are not recommended for use at the sealed road network level, and they
have not been used for some time.

Current practice is to measure the longitudinal profile of the road and to mathematically model the
response of a hypothetical vehicle (e.g. a profilometer and quarter-car simulation). The wheelpath
profiles measured by a profilometer have been correlated to the response of an NRM, and as a
result many roughness results determined by a profilometer have been reported in units of
counts/km as if they had been measured by an NRM.

Austroads has determined that, at a road network level, roughness should be reported as Lane
IRIqc (commonly referred to simply as IRI (m/km)). This is the standard, composite IRI value
representing the roughness of a traffic lane within a section of road. It is determined by averaging
two individual Single Wheelpath IRIqc values obtained separately in each wheelpath of a lane.
After a considerable transition period, continuing use of NRM counts as a reporting unit is no
longer considered appropriate, and all roughness results should now be reported exclusively in
terms of IRI. This also provides consistency with the majority of worldwide practice.

Unless otherwise specified, all IRI values reported in this report refer to the quarter-car model
(Lane IRIqc) and not the half-car model (Lane IRIhc) (see Section 3.3).

It should be emphasised that both ‘IRI’ and ‘NRM’ are parameters for expressing roughness, and
are not units of measure themselves. By way of example, ‘length’ is a parameter, which may have
a unit of measure of ‘kilometres’. To write ‘4.3 IRI’ is incorrect, whereas ‘IRI of 4.3 m/km’ is
acceptable.

A travel speed of 80 km/h is built into the definition of IRI (and was also the standard operating
speed for a NAASRA Roughness Meter). Consideration should be given to highlighting this fact
when reporting roughness results for low speed environments and/or when communicating with
less experienced practitioners.

2.2 Frequency of Roughness Surveys


The frequency of network surveys is dependent on the performance, function and use of the road
network being considered. Available budgets, local policies, specific reporting schedules and
contractual obligations may also dictate a specific survey frequency.

Survey frequency can vary from one to five years. As a general guide, the survey frequencies
shown in Table 2.1 can be used. Some background information regarding the selection of these
frequencies is available.

Austroads 2007

— 4—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

The frequencies shown in Table 2.1 presume that the overall change in roughness trend over time
is known. In order to initially establish these trends, more frequent surveys of road roughness are
needed.

For road network roughness monitoring, as a minimum, the road is surveyed in one direction only,
in the lane with the heaviest traffic (heaviest loading as distinct from highest traffic volume).

Where a road carriageway is physically separated, the most heavily trafficked lane on each
carriageway is surveyed.

Table 2.1: Survey frequency

Road network Frequency


Heavily trafficked arterial roads 1 year
Roads with high rates of deterioration 1 year
Arterial roads with average deterioration 2 – 3 years
Low trafficked local roads with low deterioration 5 years

2.3 Specification of Roughness Surveys


Until recently, and despite a long history of data collection across the Australasian road networks,
no nationally adopted procedures have existed for the collection of data and the validation of test
equipment. Each road authority has developed its own specification for these types of survey
works. Recent Austroads work has developed a standard specification, and associated test
methods, for the collection of roughness data (Moffatt, Sharp & Ferguson 2006). The specification
and test methods are freely available from the Austroads web site:
www.austroads.com.au/asset/test.html.

The web site also contains Commentaries for each of the specification and test methods. These
commentaries provide additional notes and advice to practitioners, and have been structured to be
read in conjunction with the relevant specification or test method.

Table 2.2 lists the specification and test methods (Austroads 2007a – 2000f) that are relevant to
the collection of roughness data. A general overview of the structure of the specification and test
methods, and guidance on their use within a contract context, is provided in a separate
commentary. The specification and test methods have been prepared for the testing of pavements
at the network level. Care should be taken when applying the specification or test methods at the
project level.

Table 2.2: Austroads standard specification and test methods for pavement roughness measurement

Number Title
Specification: AG:AM/S001 Specification for Pavement Roughness Measurement with an Inertial Laser Profilometer
Test Methods: AG:AM/T001 Pavement Roughness Measurement with an Inertial Laser Profilometer
AG:AM/T002 Validation of an Inertial Laser Profilometer for Measuring Pavement Roughness (Reference
Device Method)
AG:AM/T003 Validation of an Inertial Laser Profilometer for Measuring Pavement Roughness (Loop Method)
AG:AM/T004 Pavement Roughness Repeatability and Bias Checks for an Inertial Laser Profilometer
AG:AM/T005 Distance Measurement Validation of Road Condition Monitoring Vehicles

Austroads 2007

— 5—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

2.4 Data Reporting


Roughness should be recorded for each 100 m segment as Lane roughness, IRI (or more
precisely, Lane IRIqc) (m/km), to not more than two decimal places. Any decisions to be based
upon collected roughness data (e.g. maintenance intervention levels) should be based upon
roughness results rounded to not more than one decimal place.

Data must be properly referenced in order to be meaningful for use in decision making processes.
Accordingly, roughness data must be reported using an established predefined location
referencing system such as the RTA NSW Roadloc system (RTA NSW 2002a). During the data
collection survey, the location of significant road features such as bridges, intersections,
administrative borders, etc. must be incorporated in roughness reports, to enable each 100 m
segment to be uniquely referenced in terms of the road agency’s location system.

Roughness reports must also clearly identify the lane surveyed and the direction and speed of
travel during the survey, as well as the date and weather conditions, and any impediment to the
survey, or missing or invalid results and their cause (e.g. roadworks, traffic congestion, local area
of wet surface, water over road, or other obstacle on the road, lane change manoeuvre for
overtaking, etc.).

Other methods of analysing and reporting roughness data, particularly at a network level are
discussed in Section 5.

Austroads 2007

— 6—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

3 TOOLS AND TECHNIQUES


3.1 Classes of Measuring Devices
The World Bank has defined four classes of roughness measuring methods as shown in Figure
3.1. The method or device for measuring roughness should be chosen on the basis of factors such
as the degree of accuracy required, the potential use of the data, and the cost.

Table 3.1: World Bank classes of roughness measuring methods (after Bennett et al. 2005)

Class Description Examples of equipment


Class 1 This class is the highest accuracy for roughness measuring devices. Laser profilers: non-contact lightweight
Precision profilers The profile is measured as a series of closely spaced accurate profiling devices and portable laser
elevation points in the wheelpath. Equipment in this class can be profilers
divided in two broad groups, those using laser technology and Manually operated devices: e.g. rod and
manually operated equipment. level, TRL beam, Face
Dipstick/ROMDAS z-250, ARRB Walking
Profiler.
Class 2 This class considers dynamic profile measuring methods that APL profilometer, profilographs (e.g.
Other profilometer determine profile elevations by either elevation data or summarising California, Rainhart), optical profilers, and
methods statistics calculated from elevation data. Accuracy of these devices is inertial profilers (GMR).
dependent on the technology used, being less accurate than Class 1.

Class 3 Class 3 equipment includes mechanical or electronic devices that NAASRA Roughness Meter, Roadmaster,
IRI estimates from indirectly evaluate pavement profiles. Measures obtained using these ROMDAS, ARRB Roughometer, TRL
correlation equations devices require calibration through correlations with standardised Bump Integrator, rolling straightedge.
roughness values.

Class 4 Subjective ratings are produced by either riding over the section or Key code rating systems, visual
Subjective conducting a visual inspection. inspections, ride over section.
ratings/uncalibrated
measures

3.2 Measurement of Roughness


In general, it is recommended that a Class 1 device be used for road network roughness surveys.
Devices that meet Class 1 requirements are readily available in Australasia. Class 2 devices are
not readily available, and modern laser equipment that measures road profile is almost exclusively
able to meet the Class 1 requirements. It is considered that response type devices (Class 3) would
generally not be used for the management of sealed road networks within Australasia. Such
devices have proven effective, however, in the management of rough and remote unsealed roads.
Due to their dependence on vehicle suspension characteristics, the risk of inaccurate results is
higher with response type roughness measuring devices. The significance of these inaccuracies
decreases as the roughness of the road increases.

When conducting roughness measurements using a profilometer, the following requirements


should be met:
ƒ Within the traffic lane being surveyed, the survey vehicle should follow the most common
wheelpaths used by cars. Where the most common wheelpaths used by cars are not clear,
the survey vehicle should follow the centre of the lane.
ƒ Results should be processed and reported at 100 m intervals.

Austroads 2007

— 7—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

ƒ Equipment should be used in accordance with the manufacturer’s instructions and in


accordance with the appropriate test method (e.g. Austroads (2007b), RTA NSW (2002b),
VicRoads (2001), Austroads (2007g)).
ƒ Equipment used should measure longitudinal profile simultaneously in both wheelpaths in the
wavelength range of 0.5 m to 50 m over the full range of operating speeds.
ƒ The wheelpaths are deemed to be offset 0.75 m from the centreline of the vehicle.
ƒ The profile should be sampled at longitudinal distance spacings no greater than
approximately 50 mm increments.
ƒ Height measurement accuracy should be at least 0.2 mm.

The inertial laser profilometer in Table 3.1 shows two lasers measuring the profile in each
wheelpath (the vehicle also has a third laser mounted between the wheelpaths for measuring
surface texture).

3.3 Profile Analysis


3.3.1 General
Profile analysis applies only to measurement methods that involve converting an actual longitudinal
profile to a roughness statistic (e.g. IRI or NRM).

There are two different methods of averaging the two wheelpath profiles to yield a roughness
reading for the whole lane width. Whilst both of these methods are discussed below, it must be
emphasised Austroads endorses the quarter-car model and not the half-car model.

3.3.2 Quarter-Car model (endorsed by Austroads)


Roughness values are determined by simulating the travel of a quarter-car model over the
longitudinal profile.

For the determination of the IRI of a lane, the IRI of each of the two wheelpaths is determined
using the IRI algorithm, and the resultant two wheelpath IRI values averaged to yield the IRI of the
lane. This is fully termed ‘the Lane IRI determined using the quarter-car model’, or the Lane IRIqc.

Each wheelpath profile is analysed through the ‘quarter-car’ model to determine the Single
Wheelpath IRIqc for each wheelpath for each 100 m segment. For each segment, the IRI of the
lane is then determined by averaging the two wheelpath IRI results using the following equation.

Single Wheelpath IRIqc ( inner ) + Single Wheelpath IRIqc (outer )


Lane IRIqc =
2

If required, the estimated NAASRA Roughness Meter (counts/km) for the lane can be obtained by
use of the following equation (Prem 1989).

NAASRA (counts/km ) = 26.49 ⋅ Lane IRIqc (m/km) − 1.27

Results are reported at 100 m intervals, with IRI results reported to no more than two decimal
places (m/km), and NAASRA counts/km are rounded to the nearest integer.

A conversion chart allowing for conversion of roughness data between the IRI and NRM scales is
provided.

Austroads 2007

— 8—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Suitable public domain software for calculating the IRI and analysing road profiles can be found on
the University of Michigan Transport Research Institute Road Roughness Web site:
www.umtri.umich.edu/erd/roughness (UMTRI 2005). The mathematical steps used to calculate the IRI
parameter are outlined in Sayers et al. (1986a and 1986b).

3.3.3 Half-Car Model (not endorsed by Austroads)


Another means of determining a Lane IRI value is to average the profiles of each wheelpath and to
use the IRI algorithm to then determine the IRI of this average profile. This is termed as ‘the Lane
IRI determined using the half-car model’, or Lane IRIhc. Austroads does not endorse this method
for determining the IRI of a lane.

Unless otherwise specified, all IRI values contained in this Guide refer to the quarter-car model
(Lane IRIqc) and not the half-car model (Lane IRIhc).

Figure 3.1: Inertial laser profilometer

Austroads 2007

— 9—
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

4 DATA QUALITY AND VALIDATION


4.1 Validation of Distance Measurement
Austroads Test Method AG:AM/T005 Distance Measurement Validation of Road Condition
Monitoring Vehicles defines the procedure for validating the distance measurement performance of
vehicles conducting condition monitoring surveys (such as inertial laser profilometers).

The procedure involves the comparison of the distance reported by the road condition vehicle
travelling over a 1 km length of road, with the exact length of that road as measured by precise
ground survey techniques.

4.2 Validation of Roughness Measurement


Two alternative Austroads Test methods have been developed for equipment validation.

AG:AM/T002 Validation of an Inertial Laser Profilometer for Measuring Pavement Roughness


(Reference Device Method) compares the measurement of the test profilometer to the
measurement from a static or manual reference device such as the ARRB Walking profiler, or a
surveyor’s staff and level. The test sites selected for the conduct of the validation exercise must
meet specific roughness characteristics. The Test Method contains a worked example of the
calculations required.

The second method, AG:AM/T003 Validation of an Inertial Laser Profilometer for Measuring
Pavement Roughness (Loop Method), involves the comparison of roughness data from the
profilometer requiring validation against a reference set of data collected by other similar
profilometers. The procedure was developed by RTA NSW and has been used by a number of
Australian road authorities.

4.3 Repeatability and Bias


Measurement repeatability is an indication of the variation in a series of measurements about the
mean of those measurements. Bias error indicates whether a device is systematically measuring
high or low when compared to a reference set of measurements.

AG:AM/T004 Pavement Roughness Repeatability and Bias Checks for an Inertial Laser
Profilometer outlines the process by which these checks can be conducted. The standard
Specification (see Section 2.3) requires that these checks are conducted as part of the validation
method AG:AM002, and additionally every 30 days during the collection of data. The test method
compares two sets of roughness readings, taken some time apart, along a length of road. The
Test Method contains worked examples of the calculations required.

Austroads 2007

— 10 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

5 ANALYSIS
This section discusses tools and techniques that can be used to determine the validity of the
collected roughness data. In addition, discussion is provided on ways in which value can be added
to roughness data.

5.1 Assessing the Validity of Roughness Data


Simple statistics can be used to quickly determine whether the surveyed roughness data appears
to be valid. Key statistics for roughness such as the minimum, median, maximum, inter-quartile
range, mean and standard deviation can be calculated at the link, road or network level and
compared against the corresponding roughness statistics for previous years. Graphical displays
such as histograms and cumulative distribution curves are also effective in summarising the
distribution of the roughness values collected. If the trends summarised by the above methods are
consistent with engineering judgement and knowledge of the road network, then the surveyed
roughness data might be assessed as valid.

It would be expected that the roughness of a road section that had not received recent
maintenance would demonstrate a stable or gradual increasing roughness level. Analysing the
trend of roughness for a given section of road requires an accurate referencing system to ensure
that the same length of road is being compared over time.

Consider Figure 5.1 which displays the trend in roughness values collected (at 100 m intervals)
over a road segment over three years. The series of roughness values collected during 2003 is
consistent with the series of roughness values collected during 2002. During 2004, there is a
significant reduction in the roughness values between chainage 23 and chainage 24 km.
Confirmation would be required as to whether, or not, maintenance was undertaken at this
location.

7
2004
2003
6
2002

5
Roughness (IRI m/km)

0
20 21 22 23 24 25 26 27 28 29 30
Reference chainage (km)

Figure 5.1: Roughness of a link over time

Sudden increases in roughness are unusual across large networks and such an increase may be
an indicator of a change in the measurement process.

Austroads 2007

— 11 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

When considering the possibility that a roughness data set is of suspect quality, it is important to
investigate whether there are other factors that may be affecting the road condition. It is common
for roughness surveys to be conducted concurrently with other pavement condition parameters
(usually rutting and surface texture) and in conjunction with digital image collection. Cross
checking against these other data sources can prove very helpful in determining the validity of
specific roughness results.

5.2 Distribution Analysis


Histograms and cumulative distribution graphs of roughness data can be highly effective means of
conveying network trends. Whilst generating these types of graphs is useful for examining a single
set of roughness results, the charts can provide significant insight if a time series of results is
plotted.

5.3 Performance Indicators


A key indicator for quantifying the ride quality of the network is the Smooth Travel Exposure (STE)
measure. The STE is the proportion of travel undertaken each year on roads with conditions
above the targeted conditions for those roads, and is defined as follows:

Tnf
STE = 100 ⋅
TvC
where
Tnf = the year' s travel measured in vehicle kilometres travelled (VKT)
on roads that meet the targeted condition
TvC = the year' s travel measured in VKTs for the entire network being reported

The STE measure is an Austroads National Performance Indicator (NPI) and is reported for
targeted conditions of roads with IRI roughness less than or equal to 4.2 and 5.3 m/km. As an NPI
the STE is only reported for the sealed road network. The target conditions of 4.2 and 5.3 m/km
correspond with NRM readings of 110 and 140 counts/km respectively.

The detailed method for determining the STE is included as commentary. Full details of this and
other Austroads NPIs can be accessed from www.austroads.com.au.

5.4 Other Analysis Techniques


There are other advanced analyses that can be conducted using measured profile data. The most
common is to decompose the complex shape of a longitudinal profile into a series of sinusoids via
the Power Spectral Density (PSD) function. This is a complex mathematical area, the application
of which is generally confined to the research community. The technique is effective in filtering
from a profile the different sinusoid or wave shapes that may provide insight into specific vehicular
responses.

Summaries of a range of alternative profile analysis techniques can be found in Sayers and
Karamihas (1996).

Another area that has received considerable attention in recent times has been the attempt to
quantify a roughness measure that better reflects the different ways in which light and heavy
vehicles respond to a given profile. Current research indicates that the longer wavelengths in a
longitudinal profile have a bigger influence on the ride quality of a truck than for passenger cars.
Additionally, the ‘pitch and roll’ behaviour of trucks appears to be sensitive to transverse profiles
(i.e. the profile at right angles to the road centreline).

Austroads 2007

— 12 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

The determination of a truck ride quality index comparable to the quarter-car model that is the
heart of the IRI calculation has often been discussed in international forums, but has yet to reach
an internationally accepted resolution. Recent VicRoads work (Hassan and McManus 2004) has
resulted in the development of the Heavy Articulated Truck Index (HATI), which is the first non-IRI
or non-NRM roughness based measure to have been collected for an entire Australian state road
network.

Another analysis technique used in the development of HATI has been the examination of the
differences between the wheelpath IRI results.

Austroads 2007

— 13 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

6 APPLICATION
6.1 General
Application of roughness data includes the setting and implementation of network management
policy. Discussions about policy setting are beyond the scope of these guidelines, and will be
covered elsewhere in the Guide to Asset Management. The following sub-sections, however,
provide a very brief overview of the range of applications that roughness data can have, by itself or
in combination with other parameters, on policy issues.

6.2 Network Screening


A key application of roughness data is to allow a network manager to screen an entire network and
separate those lengths of road whose condition warrants further investigation from the larger
portion of the network which does not. Roughness alone can play a role as a screening tool,
however it is usual to make some allowance for the level of traffic on the links within a network.
This ensures that proportional consideration is placed on roads which have higher traffic levels.

6.3 Work Prioritisation


Roughness data can also play a role, when coupled with traffic level data, in determining the
prioritisation that competing works projects should be given in a works program. It should be
remembered that roughness data, like most condition data, conveys a picture of the functional
condition of a pavement, but does not include any statement as to the level of use of that
pavement. Accordingly, work prioritisation usually, at least, considers the traffic level on the road
links in question as well as the condition.

Advanced decision-support tools (see Section 6.6), which make use of roughness data as input,
can also consider whole-of-life costing concepts in the prioritisation of works.

6.4 Performance Indicators


Roughness is well suited as an indicator of network performance. The uses of such performance
indicators are varied.

Roughness results are often used for internal monitoring of network behaviour, and the figures
contained in Commentary J are examples of such a use. The authority that generated these
figures uses them as a means of quickly tracking the overall behaviour of the network.

Additionally, the same authority generates the Austroads Smooth Travel Exposure index (which
considers the traffic level on the roads in addition to the roughness of the roads themselves) as a
means of benchmarking their performance against other Australian state road agencies. These
results are published by Austroads, along with the other National Performance Indicators, at
www.austroads.com.au. The authority also considers these indicators to be of interest to the
broader stakeholders of the network they manage, and accordingly also publishes the indicators in
its annual report.

These sorts of indicators can also be used as a contractual means of monitoring the performance
of maintenance service providers. These service providers may be separate departments of the
network owning authority, or, as is becoming increasingly the case, an outsourced maintenance
contractor. In some cases, current levels of roughness for separate components of the network
have been established, and the maintenance organisation has been charged with the responsibility
for maintaining the network at, or possibly improving the network from those levels.

Austroads 2007

— 14 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Some network owners use a more sophisticated approach and use the current roughness
characteristics of the network as an input into network condition projection models. The projection
models are then used to determine the worst case and target network conditions over the period of
the maintenance contract. These two cases form the upper and lower limits within which the
maintenance contractor must ensure the distribution of network performance is placed.

6.5 Intervention Levels


Roughness data is often used as an intervention trigger for treatment works. Network managers
typically establish intervention levels with specific treatment types recommended for different levels
of roughness (e.g. a maintenance response may be to undertake surface regulation for pavement
lengths with roughness in excess of IRI 4.2 m/km (110 NRM counts/km)). Maintenance responses
can range in type from establishing a need for further investigation, through to complete
reconstruction of the road.

Time series of roughness data can be useful in assessing the effectiveness of current intervention
levels, and in formulating potential future intervention levels. By way of example, consider the
network summarised in Figure J2 (Commentary J). The currently applied intervention levels do not
appear to be holding the network in a stable condition 1 . A change in the intervention levels is
probably required to assist in halting the decline of the network.

When establishing intervention levels for a network it is also important to have an appreciation of
the current condition of that network. It is important to ensure that any new intervention levels, and
maintenance responses, are affordable, practical and likely to result in the changes in the network
characteristic that are targeted. Examination of current, and preferably past, roughness data can
be of considerable assistance in addressing these issues.

6.6 Decision-support Tools


Decision-support tools differ from pure information systems in that they have the ability to process
data to yield new parameters, rather than just store and report on existing data. These tools can
provide valuable input to the network manager when making management decisions, by
processing condition parameters, such as roughness, and determining maintenance strategies,
work plans, budget estimates and the like.

Decision-support tools are wide ranging in nature and complexity, and a detailed discussion of their
nature is beyond the scope of these guidelines, and is addressed elsewhere in the Guide to Asset
Management. Some tools use simple decision rules to prepare theoretical work programs based
on a static network condition observation. Other systems, such as HDM-4, use a wide range of
pavement condition parameters as inputs into complex incremental step models to predict future
pavement, and ultimately network performance, under a variety of maintenance and rehabilitation
strategy scenarios.

Given the broad nature of this topic, a future section of this Guide is to be dedicated to the subject
of performance modelling, and it is anticipated that these issues will be discussed, and guidance
given.

1
For the sake of this discussion the possibility that there are insufficient funds available to undertake all
recommended interventions has not been considered.

Austroads 2007

— 15 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

6.7 Network Performance Scenario Analysis


A significant benefit of some decision-support tools is their ability to estimate future network
conditions under a range of alternative operating conditions. This has enabled scenario analyses
to have considered a wide range of issues including:

ƒ the effect on network performance resulting from changes in maintenance philosophy (e.g. a
change from reactive maintenance to stitch-in-time philosophies)
ƒ the effect on the network of an increase in vehicle load regulations
ƒ the effect on the network of an increase in traffic levels
ƒ the net economic benefit to society, usually determined on the basis of the net present value
(NPV) sum of the road agency and road user costs, of alternative maintenance regimes
ƒ estimation of greenhouse gas contribution from vehicle emissions resulting from different
maintenance regimes, or demand levels.

6.8 Budget Forecasting


A key role played by most established data systems, and decision-support tools, used by road
network managers is to assist in forecasting future budgetary requirements to ensure that the
network meets its performance targets. Extensive scenario analysis can also be used to assess
the impacts of policy changes on budgetary estimates, and similarly optimised management
policies can be found to match imposed budgetary levels.

Austroads 2007

— 16 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

7 PROJECT LEVEL APPLICATION


This section addresses a few key issues regarding collection and use of data at a project level.
For additional information the reader is referred to the Austroads Guide to Pavement Technology.

7.1 Roughness Collection at the Project Level


At the network level the use of non-automated collection methods is clearly impractical, but they
may have application at the project level. The accuracy of the automated systems is certainly
adequate for project level, and such systems are used in contract auditing roles at the project level;
however, for some small scale project works alternative collection means might prove to be
economical.

The determination of IRI requires a longitudinal profile as input. A standard surveyor’s staff and
level can be used, to a limited extent for research and equipment verification applications.
Alternatively Face Corporation’s DipStick™ (www.faceco.com), and more commonly in Australasia,
ARRB Group’s Walking Profiler (www.arrb.com.au) (Figure 7.1) provides automated methods of
conducting a project level roughness survey on foot.

Austroads has established a standard test method for the determination of IRI roughness using an
ARRB Walking Profiler (Austroads Pavement Test PAT: 2001).

Austroads is also preparing standard specifications and test methods for the collection of road
roughness, and it is anticipated that these methods will also be applicable for collection of data at
the project level.

Figure 7.1: Walking profiler

Austroads 2007

— 17 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

7.2 Auditing of Pavement Construction Ride Quality


A key application of roughness data at the project level is its use in contractual specification.
Some road agencies provide target roughness values for construction works, with penalties
applying for non-conformance of those levels. The measurement of roughness for this use is
undertaken in much the same manner as it is for network level surveys, using equipment calibrated
and verified to the same levels, but with a higher emphasis on repeat survey runs, and different
reporting requirements. Results are usually reported as the average of three repeat survey runs of
the test pavement

As noted above, Austroads is currently preparing standard specifications and test methods for the
collection of road roughness, and it is anticipated that these methods will also address the
collection and reporting of roughness values for auditing of pavement construction works.

7.3 Roughness Data as a Diagnostic Tool


Roughness data can provide some insight into the distress of a pavement, especially when
considered with other condition parameters and field observations.

Acceptable values of roughness are usually established by road authorities, and whilst they may
vary between jurisdictions, some indicative levels are listed in Table 7.1. When considering
roughness data for a given road the results should be compared against acceptable values, and
judgement made as to whether further investigation is warranted. Time series of roughness data
can provide insight into the likely future trend of roughness for road lengths, and allow estimations
as to when investigatory and intervention levels may be reached in the future.

Table 7.1: Levels of roughness (after Austroads 2003)

Indicative investigation levels for


Typical maximum desirable roughness for new roughness (IRI m/km)
Road function
construction or rehabilitation (length 500 m)
Isolated areas Length > 500 m

Freeways and other high-class facilities 1.6 4.2 3.5

Highways and main roads (100 km/h) 1.9 5.3* 4.2

Highways and main roads ( < 80 km/h) 1.9 6.1 5.3

Other local sealed roads No limits defined † No limits defined† No limits defined†

Notes
* Lower values may be appropriate where total traffic or heavy vehicle volumes are high.
† Roughness levels depend on local conditions and traffic calming measures.

A range of pavement distress types can lead to an increase in road roughness, including those
shown in Table 7.2. One of the most significant off-pavement contributors to increased pavement
roughness is the inability of the drainage system to prevent moisture ingress to the pavement
structure.

Austroads 2007

— 18 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Table 7.2: Pavement defects which may contribute to increased pavement roughness

Flexible pavements Concrete pavements


ƒ Localised depressions ƒ Stepping / faulting
ƒ Ruts ƒ Rocking
ƒ Potholes ƒ Pumping
ƒ Patches ƒ Spalling
ƒ Corrugations ƒ Patches
ƒ Shoving ƒ Installation of services
ƒ Delamination/debonding ƒ Temperature induced curling of slabs
ƒ Stripping
ƒ Cracking
ƒ Installation of services

It is important to consider whether there may be other factors in the road environment which could
be contributing to any localised high roughness readings. Changes in pavement profile resulting
from intersecting streets, roundabouts, rail crossings, bridge abutments and bridge decks can all
create high roughness readings. It should also be remembered that IRI roughness is determined
by modelling the response of a car travelling at 80 km/h. High roughness readings at roundabouts,
small local streets and other low speed environments will suggest a poor ride comfort level which
the travelling public, travelling at less than 80 km/h, may not truly experience. It is important to
recognise that high roughness readings resulting from these types of causes are not indicators of
pavement distress. In these circumstances considerable judgement should be exercised in
determining if engineering treatments would be efficacious or warranted.

Austroads 2007

— 19 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

REFERENCES
Aitken, T 1900, Road making and maintenance: a practical treatise for engineers, surveyors, and others, C.
Griffin and Company, London; J.B. Lippincott, Philadelphia.

Austroads 2001a, Guidelines for road condition monitoring: Part 1 – pavement roughness, AP-G65.1/01,
Austroads, Sydney.

Austroads 2003, Guide to the selection of road surfacings, 2nd edition, APG63/03, Austroads, Sydney.

Austroads 2007a, Specification for Pavement Roughness Measurement with an Inertial Laser Profilometer.
Austroads Specification AG:AM/S001, Austroads, Sydney.

Austroads 2007b, Pavement Roughness Measurement with an Inertial Laser Profilometer. Austroads Test
Method AG:AM/T001, Austroads, Sydney.

Austroads 2007c, Validation of an Inertial Laser Profilometer for Measuring Pavement Roughness
(Reference Device Method). Austroads Test Method AG:AM/T002, Austroads, Sydney.

Austroads 2007d, Validation of an Inertial Laser Profilometer for Measuring Pavement Roughness (Loop
Method). Austroads Test Method AG:AM/T003, Austroads, Sydney.

Austroads 2007e, Pavement Roughness Repeatability and Bias Checks for an Inertial Laser Profilometer.
Austroads Test Method AG:AM/T004, Austroads, Sydney.

Austroads 2007f, Distance Measurement Validation of Road Condition Monitoring Vehicles. Austroads Test
Method AG:AM/T005, Austroads, Sydney.

Austroads 2007g, Determination of the International Roughness Index (IRI) using ARRB Walking Profiler.
Austroads Test Method AG:PT/T450, Austroads, Sydney.

Bennett, CR, Chamorro, A, Chen, C, de Solminihac, H & Flintsch, GW 2005, Data Collection Technologies
for Road Management, Version 1.0-6 April 2005, East Asia Pacific Transport Unit, The World Bank,
Washington, DC.

Country Roads Board of Victoria 1930, 17th Annual Report, Kew, Vic., p 41.

Country Roads Board of Victoria 1931, 18th Annual Report, Kew, Vic., pp 33-35.

Descornet, G 1990, ‘Reference road surfaces for vehicle testing’, Routes Roads, no. 272, pp.57-71.

Fong, G & Brown DN 1997, Transfer function based performance specifications for inertial profilometer
systems, Central Laboratories Report 97-529351, Opus International Consultants, Lower Hutt, New
Zealand.

Hassan, R & McManus K 2004, Providing better ride quality for freight industry, a report prepared for
VicRoads, consultancy agreement RD888.

Hveem, FN 1960, 'Devices for recording and evaluating pavement roughness', Road Roughness and
Skidding Measurements: 1960, bulletin 264, Highway Research Board, Washington DC, pp.1-26.

Moffatt, MA, Sharp, KG & Ferguson, RA 2006, Austroads standardised measurement of road condition.
Proceedings 22nd ARRB Conference, Canberra, ARRB Group, Vermont South, Victoria.

Austroads 2007

— 20 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

National Association of Australian State Road Authorities 1981, Standard operating instructions for the
NAASRA Roughness Meter and guide for the serviceability rating of road pavements, MEC-4, National
Association of Australian State Road Authorities, Sydney.

Prem, H 1989, NAASRA Roughness Meter calibration via the road-profile based International Roughness
Index (IRI), ARR 164, Australian Road Research Board, Vermont South, Vic.

Prem, H 1998, Development and evaluation of a method for validation of pavement roughness
measurements, contract report RE7135, ARRB Transport Research, Vermont South, Vic.

Prem, H 1999, ‘Measurement of reference profiles for high-speed profiler validation’, Road and Transport
Research, vol. 8, no. 1, pp 84-91.

Roads and Traffic Authority, New South Wales 2002a, Reference guide to linear referencing, Version 1,
Revision 1. August 2002, RTA NSW, Sydney.

Roads and Traffic Authority, New South Wales 2002b, Measurement of ride quality of road pavements by
laser profiler, test method T187, February 2002, RTA NSW, Sydney.

Sayers, MW 1995, ‘On the calculation of International Roughness Index from longitudinal road profile’,
Transportation Research Record 1501, pp 1-12.

Sayers, MW, Gillespie, TD & Paterson, WDO 1986a, Guidelines for conducting and calibrating road
roughness measurements, technical paper 46, The World Bank, Washington DC.

Sayers, MW, Gillespie, TD & Queiroz, CA 1986b, The International Road Roughness experiment:
establishing correlation and a calibration standard for measurements, technical paper 45, The World
Bank, Washington DC.

Sayers, MW & Karamihas, SM 1996, The little book of profiling: basic information about measuring and
interpreting road profiles, Revision 10, October 1996, University of Michigan Transportation Institute
(available from UMTRI web site).

Standards Australia 2002, Road and traffic engineering - glossary of terms, AS 1348-2002, Standards
Australia, Sydney.

Transit New Zealand 1998, State Highway distance marking manual, SM051, Transit New Zealand,
Wellington.

University of Michigan Transportation Research Institute 2005, RoadRuf: software for analyzing road profiles,
public domain software available from www.umtri.umich.edu/erd/roughness. (Accessed April 2005).

VicRoads 2001, Pavement roughness (inertial laser profiler method), manual of testing, RC422.03,
GeoPave, VicRoads, Melbourne.

Austroads 2007

— 21 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

COMMENTARY A GLOSSARY OF TERMS RELATED TO ROAD


PAVEMENT ROUGHNESS MEASUREMENT
Term Description
Accelerometer A sensor that measures acceleration.
ARRB Walking Profiler A device for static measuring of road surface profile. The ARRB Walking Profiler consists of a slope
sensitive beam mounted on two feet spaced 241.3 mm apart, with an on-board computer which records
distance and elevation, and calculates IRI as it progresses. The measuring mechanism positions the
beam in a continuing cycle such that for each new beam placement, the rear foot of the beam accurately
overlies the position formerly occupied by the front foot. Though the beam mechanism is supported from
the body of the device, it is disengaged for a short period each time it is placed on the pavement. The
device travels on two wheels and is pushed manually at about 0.8 km/h. The ARRB Walking Profiler is
accepted as a Class 1 profile measuring device.
Autospectral density A mathematical function that is usually displayed in graphical form showing the frequency composition of
a signal. The frequency composition of a road profile when presented in this form can be used to
indicate the distribution of roughness amplitude with spatial frequency or wavenumber (the inverse of
wavelength), and the roughness components that are most likely to affect the IRI. For example,
corrugations in the road profile will appear as a peak in the autospectral density function at a spatial
frequency that corresponds to the distance between consecutive high (or low) points of the corrugations.
Bias A statistical term to indicate whether a device is systematically measuring high or low when compared to
a reference set of measures (also see Commentary I).
Class 1 profile measuring One of two accuracy classes (Class 1 and Class 2) for devices that measure road surface profile, as
device defined by the World Bank (Sayers et al. (1986a), Bennett et al. (2005)). For acceptance as a Class 1
device, accuracy of within ±2% over a 320 m length is necessary, both in terms of repeatability and
agreement with the reference or estimated true profile, as measured by the ‘rod and level’ method or
using the Face Corporation’s ‘DipstickTM’. ARRB’s Walking Profiler is accepted as a Class 1 profile
measuring device. (Note that the World Bank also defined Class 3 for response-type roughness devices
such as the NAASRA roughness meter which can be correlated with IRI, and Class 4 for subjective
ratings and uncalibrated measures) (also see Section 3.1).
Condition parameter A quantifiable expression of a specific defect in an asset. For example, roughness, surface texture, skid
resistance, edge break, deflection, etc., are pavement condition parameters. Reporting can be either by
bins (e.g. good, fair, poor, bad, etc.) or on a continuous numerical scale (IRI, NRM, rut depth (mm), crack
width (mm), % area patched, etc.).
DipstickTM A proprietary device for static measuring of road surface profile. DipstickTM (Face Corporation) consists
of a precision inclinometer on two supports, normally spaced 300 mm apart, and adjustable between
50 mm and 300 mm. The inclinometer measures and records the difference in height between the two
supports, and the device ‘walks’ along the line being profiled, by pivoting through 1800 after each
measurement. DipstickTM is accepted as a Class 1 profile measuring device.
Filtering A calculation procedure to change a series of numbers (a signal) into another series of numbers (another
signal). The main use of signal filtering with profilometer data is to replace each profile point with the
average of several adjacent points (a ‘moving average filter’).
Half-car model A term used to indicate a measure of roughness based on the simulated response of a hypothetical
vehicle to the measured profiles of two wheelpaths in a traffic lane, similar to one axle of a car (also see
‘quarter-car model’).
Inertial profilometer A vehicle-based road profile measuring system that includes an accelerometer to provide the reference
datum, a height sensor for datum-to-ground measurement, a longitudinal distance sensor, a computer,
and various electronics to power the sensors and connect them to the computer. Early inertial
profilometers (e.g. the original General Motors Research (GMR) type) sensed height above the road
surface using a follower wheel instrumented with a potentiometer. This required testing at low speeds to
avoid bouncing, was subject to mechanical wear, and has been replaced with a non-contact laser-based
sensor in the modern inertial profilometers now used in Australia and New Zealand.
(also see ‘profilometer’)

Austroads 2007

— 22 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Term Description
IRI International Roughness Index (IRI) is a widely accepted measure of roughness developed by the World
Bank in the 1980s and adopted by the World Road Association (PIARC). IRI is a numerical
representation of a road profile, designed to replicate the traditional roughness measures obtained from
response-type road roughness measuring systems such as the NAASRA roughness meter. IRI
represents the vertical response to the measured longitudinal road profile of a hypothetical quarter-car
(as replicated by a mathematical model) travelling at 80 km/h. IRI has been found to be highly portable,
that is, different roughness or profile measuring devices are capable of producing outputs expressed in
IRI. A small number of slightly different IRIs have been defined as shown in Commentary G, Table G.1,
e.g. Single Wheelpath IRIqc, Lane IRIqc, and Lane IRIhc. IRI is a dimensionless measure expressed in
Australasia in m/km (and in US as inches per mile, conversion factor 0.0158).
IRI validation A process to check that the roughness outputs (IRI m/km, or NRM counts/km) from an operational
roughness survey device are within specified tolerances.
Lane IRI Road surface profiles are measured along one or more wheelpaths. A ‘Lane IRI’ is therefore a
combination of the surface profiles of two wheelpaths. Austroads has endorsed the determination of
‘Lane IRI’ based upon the ‘quarter-car’ model, and the resultant ‘Lane IRIqc’, commonly expressed simply
as IRI.
Longitudinal profile The shape of a pavement surface measured as vertical distances from the same datum parallel to the
traffic flow (AS 1348:2002).
NAASRA roughness meter A standard mechanical device used extensively in Australia and New Zealand since the 1970s for
measuring road roughness by recording the upward vertical movement of the rear axle of a standard
station sedan relative to the vehicle’s body as the vehicle travels at a standard speed along the road
being tested. A cumulative upward vertical movement of 15.2 mm corresponds to one NAASRA
Roughness Count.
Network level A type of road condition survey or data analysis where the main purpose is to monitor network
performance or assist with network asset management decisions, as distinct from project decisions.
Non-contact A term to describe a system of measuring where there is no physical contact between the instrument and
the object being measured, e.g. accelerometers and laser sensors are capable of measuring a road
surface profile without physical contact with the road surface.
Profile A two-dimensional slice of the road surface, taken along an imaginary line (Sayers and Karamihas
1996). The shape of a pavement surface measured in a vertical plane parallel to the traffic flow
(AS 1348:2002).
Profile index A summary number calculated from the many numbers that make up a profile.
Profile filtering see filtering.
Profiler Some literature, especially from the United States, refers to profilometers as profilers. In Australasia, the
term ‘profiler’ mostly means a self propelled machine which removes a controlled depth of pavement
material. An exception is the ‘ARRB Walking Profiler’ (see above) which is a commercial name for
proprietary technology with international patents.
Profile validation A process to check that the profile obtained from an operational roughness survey device is within
specified tolerances, with respect to the reference profile (also see Commentary I.)
Profilometer A device for producing a series of numbers related in a well-defined way to a reference profile.
Roughness measuring devices, other than mechanical response-type devices and most static devices,
are commonly referred to as profilometers (also see ‘inertial profilometer’).
Project level A type of road condition survey or data analysis where the main purpose is to assist with decisions about
proposals for a specific project on a short length of road, as distinct from network decisions.
Quarter-car model A term used to indicate a measure of roughness based on data from one wheelpath in a lane, similar to
one wheel of a car (also see ‘half-car model’).
Reference profile That part of the true profile which is relevant to road management, and can be measured (also see ‘true
profile’).
Repeatability A statistical term to indicate the extent of variation in outputs about the mean.
Response-type roughness A device consisting of a mechanical system which produces vertical relative movement in response to
measuring device the road surface profile, and a measuring system which aggregates the relative motion. A NAASRA
roughness meter is a response-type roughness measuring device.

Austroads 2007

— 23 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Term Description
Roughness A condition parameter that characterises deviations from the intended longitudinal profile of a road
surface, with characteristic dimensions that affect vehicle dynamics (and hence road user costs), ride
quality and dynamic pavement loading. A measure of surface irregularities with wavelengths between
0.5 m and 50 m in the longitudinal profile of one or two wheelpaths in a traffic lane, reported in
dimensionless units as either International Roughness Index (IRI, m/km) or as NAASRA Roughness
Meter counts (NRM, counts/km) for the lane.
The consequence of irregularities in the longitudinal profile of a road with respect to the intended profile
(AS 1348:2002).
Signal A term used in electrical engineering to describe a fluctuating voltage that contains information. This
term was introduced to road profile measuring because the raw outputs of early inertial profilometers
were in the form of fluctuating voltages. The term ‘signal’ is now used in the measurement of road
profiles to mean a series or sequence of numbers, e.g. the numbers that form the road surface profile.
Signal processing Conditioning or analysis of signals, principally to improve the quality of measurements, and to convert
the data into useful information.
Single wheelpath IRI Road surface profiles are measured along one or more wheelpaths. A ‘Single Wheelpath IRI’ is a
statistic to describe the surface profile of one wheelpath within a traffic lane. (also see ‘IRI’)
Spatial frequency see wavenumber.
Static measurement method A method of measuring surface profile where the device is not moving while the measurement is taken.
Examples include the ‘rod and level’ method, the Face DipstickTM, and the ARRB Walking Profiler.
Transfer function In general terms, a transfer function is a mathematical function describing the dynamic input-output
characteristics of a system to sinusoidal frequency excitations. As specifically applied to the
measurement of road profile, a transfer function is the ratio of the measured road surface profile to the
reference profile, over the frequency or wavelength range of interest. Transfer functions are usually
presented in graphical form. Transfer functions can be used to characterise profilometers to identify both
satisfactory performance and deficiencies over the frequency range that has most influence on pavement
roughness measurements. The transfer function for a profilometer that is performing satisfactorily has a
magnitude equal to one across the range of wavelengths or frequency range of interest.
True profile A highly accurate but unmeasurable and theoretical concept (also see ‘reference profile’).
Validation A standardised process, covering either ‘IRI valdiation and ‘profile valdiation, or only ‘IRI validation, to
test the validity of roughness results from a measuring device. Testing is preferably carried out against
the results from a static Class 1 measuring device (see Section 4 and Commentary I).
Wavenumber The inverse of wavelength, also known as spatial frequency. For example, a wavelength of 0.5 m
corresponds to a spatial frequency of 2 cycles/m.

COMMENTARY B A BRIEF HISTORY OF ROAD ROUGHNESS


MEASUREMENT
An 1833 report describes a ‘Road Indicator’ invented by John MacNeill in England as an
instrument to measure the force required to pull a standard wagon over a road. Calibration for
grade and seasonal conditions enabled a standard condition measure to be reported for each
discrete section, and averages for routes. These numerical reports were useful in tracking the rate
of change of road condition with time, comparing sections within routes, and comparing the overall
condition of different routes.

In his book titled Road Making and Maintenance published in 1900, Thomas Aitken (1900)
describes a ‘Viagraph’ invented by J Brown, an engineer practising in Belfast, Ireland as:

a straight edge, 12 foot long and 9 inches wide, applied continuously to the road surface,
along which it is drawn, containing an apparatus for recording on paper a profile of the road
surface being tested, and the sum of unevenness is indicated by a numerical index.

Austroads 2007

— 24 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Hveem (1960) describes a profilometer developed by the State of Illinois in 1922 comprising a
frame supported by 32 bicycle wheels mounted in pairs. Like many other mechanical profilometers
mounted on straight edges, it measured the profile in terms of depth below the peaks or high points
on the road surface within the length of the straight edge.

In Australia and New Zealand, response type mechanical systems were used to measure
roughness as early as 1930 (CRB 1930, 1931). For approximately 20 years from about 1970,
NAASRA roughness meters were used for nearly all measurement of road pavement roughness in
Australia and New Zealand. In 1981, based on the combined experience of the Australian state
road authorities, NAASRA issued Standard Operating Instructions for the NAASRA Roughness
Meter (NAASRA, 1981). This NAASRA booklet effectively became the national standard for
measuring road roughness.

Advances in technology in accelerometers, height sensors and distance measurement, combined


with the rapid development of information technology during the 1980s, enabled digital recording of
pavement surface profile, and led to the introduction of non-contact automated profilometers to
measure road roughness. Use of inertial profilometers with ultrasonic or laser sensors to measure
pavement roughness has grown significantly, and continues to increase.

The World Bank developed the International Roughness Index (IRI) in the 1980s (Sayers et al.
1986b), based on an international calibration exercise performed in Brazil in 1982 involving many
types of roughness measuring equipment, including response type devices and non-contact
profilometers. IRI is now the most widely used measure of road roughness in the world.

COMMENTARY C DEFINITION OF ROUGHNESS


Roughness is a measure of the rideability of the road surface, and indicates the relative comfort
offered to the road user. From an economic perspective roughness also provides a link to vehicle
operating costs.

A response type road roughness measuring system records the movement of the rear axle relative
to the vehicle’s sprung mass (body) as the vehicle travels along the road at constant speed. Since
the early 1970s road pavement roughness has been measured in Australia using the NAASRA
Roughness Meter (NRM). This device measured the cumulative total relative upward
displacement between axle and body of a standard vehicle, registered in units of counts per
kilometre of distance travelled at either of two principal standard speeds, 80 km/h or 50 km/h. One
NAASRA roughness count corresponds to a measured axle-to-body separation of 15.2 mm (that is,
displacement in one direction).

The drawbacks of NRM, and other response type systems, have been concerns about maintaining
their calibration and about repeatability and reproducibility of the results. Outputs are very
dependent on vehicle suspension characteristics (e.g. shock absorbers, springs and tyres) and the
speed of travel.

Over the last 20 years, advances in technology have made it possible to cost-effectively and
accurately measure longitudinal profile at highway speed. Computer modelling can be used to
predict any aspect of vehicle response to the measured profile.

The drawback of the NRM systems, and the improved reliability and affordability of road profile
measuring systems has resulted in the current use of profilometers for almost all network-level
roughness surveys in Australia and New Zealand.

Austroads 2007

— 25 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Roughness for a lane is reported in terms of the International Roughness Index (IRI) - the average
of the results of the application of a computer model of a standard ‘quarter-car’ to the measured
longitudinal road profile of each wheelpath. The simulated travel speed of the IRI quarter-car
model is 80 km/h.

Extensive research has shown very high and consistent correlations between mathematical
expressions of the shape of road surface profiles and traditional measures of roughness from
response type devices. In particular, there is very good correlation between IRI and NAASRA
Roughness Counts (r2 > 0.99, (Prem 1989)), allowing the derivation of NAASRA values from
profile-based IRI measurement, and vice versa (refer to Section 3.3 of these guidelines).

Reporting roughness as IRI provides consistency with the majority worldwide trend. IRI is
therefore the preferred reporting unit for roughness. Austroads has endorsed the use of IRI for
reporting of all roughness results and no longer considers reporting roughness in terms of
NAASRA Roughness Meter counts/km as appropriate.

COMMENTARY D FREQUENCY OF ROUGHNESS SURVEYS


On typical two-lane two-way roads, it is usually appropriate to record roughness in one lane in one
direction only, and assume similar roughness in the other lane. The cost of undertaking roughness
surveys is generally the main influence on the number of traffic lanes where roughness is
measured.

On multi-lane roads, the usual practice is to survey either the left, slow or most heavily trafficked
lane. While surveys of the lane with the heaviest loading (as distinct from the highest traffic
volume) may bias reported network roughness, the results are far more useful at the project level
in monitoring deterioration and defining needs. For more than 95% of the road network, the most
heavily trafficked lane and the lane most subject to deterioration is the left or slow lane. For this
reason the network bias due to selection of the most heavily trafficked lane is not expected to be
significant. In urban conditions, it is usually not practical to survey the kerbside lane.

Given the relatively slow rate of change of roughness generally within Australasia, and the degree
of accuracy that can be achieved, annual surveys are not expected to show measurable changes
in network roughness. Intervals of two to three years would appear to be a more practical and
cost-effective approach for arterial roads where the rate of deterioration is not high. More frequent
surveys may be required in areas of rapid deterioration, or to increase the sample size to provide
more confidence in derived deterioration rates. Policy or contractual requirements may also define
a specific collection frequency (e.g. roughness may be an annual performance indicator in a
maintenance contract).

Austroads 2007

— 26 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

COMMENTARY E SPECIFICATION AND TEST METHODS


E.1 Background
Most road authorities conduct regular pavement condition surveys to collect data for pavement
attributes including roughness, rutting, strength and texture. This data is used for a range of
purposes including monitoring and predicting the performance of the network, determining the
timing and selection of appropriate intervention treatments, and monitoring the performance of
maintenance contractors. As this data can be used in developing investment programs and to
reward or penalise maintenance contractors, it is essential that it is collected to an accepted
standard and that this is reflected in the specification for the condition survey work. Asset
managers also need to be confident that the data from successive surveys is comparable,
particularly for the development of performance prediction models.

Despite a long history of data collection across the Australasian road networks, no nationally
adopted procedures exist for the collection of data and the validation of test equipment. Each road
authority has developed its own specification for these types of survey works. Recognising the
value of high quality condition data, and in line with its strategic goal of promoting national
uniformity of practice, Austroads sponsored Project AT1006 – Standardised Methods of Road
Condition Monitoring – as part of the technical research program conducted by ARRB.

The project developed standard specifications and test methods for the following condition
parameters:
ƒ pavement roughness
ƒ pavement rutting
ƒ pavement surface macro-texture
ƒ pavement deflection measured with a Falling Weight Deflectometer (FWD)
ƒ pavement deflection measured with a deflectograph.

Table E.1 lists the specification and test methods (Austroads 2007a – 2000f) that are relevant to
the collection of roughness data.

Table E.1: Austroads standard specification and test methods for pavement roughness measurement

Number Title
Specification: AG:AM/S001 Specification for Pavement Roughness Measurement with an Inertial Laser Profilometer
Test Methods: AG:AM/T001 Pavement Roughness Measurement with an Inertial Laser Profilometer
AG:AM/T002 Validation of an Inertial Laser Profilometer for Measuring Pavement Roughness (Reference
Device Method)
AG:AM/T003 Validation of an Inertial Laser Profilometer for Measuring Pavement Roughness (Loop Method)
AG:AM/T004 Pavement Roughness Repeatability and Bias Checks for an Inertial Laser Profilometer
AG:AM/T005 Distance Measurement Validation of Road Condition Monitoring Vehicles

E.2 Development
The following key principles were established early in the drafting stages of the project, and all
drafts and comments were reviewed with respect to their effect on the following principles:
ƒ the specifications and test methods must be relevant to the national, state and local
government networks, and personnel managing those networks

Austroads 2007

— 27 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

ƒ standardisation of practice is desirable


ƒ equipment validation exercises do not need to be tied to specific jurisdictions, and inter-
jurisdictional acceptance of equipment validation is desirable
ƒ specification requirements that cannot readily be measured or assessed should be avoided.

Additionally, there were two, often conflicting, imperatives that had to be balanced when drafting
material:
ƒ a specification could not be a vague guideline, but must provide clear statements of
requirements
ƒ for a specification to obtain wide acceptance, it must be flexible enough to cater for the range
of different requirements that occur between jurisdictions.

In the preparation of the draft specifications and test methods, considerable effort was made to
balance these issues. For example, in order to ensure wide applicability, no attempt was made to
specify requirements for the locations of testing because of the wide range of location referencing
systems, testing spacings and number of test lanes used throughout Australasia.

A full description of the development of the specifications and test methods can be found in
Moffatt, Sharp and Ferguson (2006)

E.3 Availability of Specifications and Test Methods


Preparation of the specifications and test methods is now complete, and the final documents are
freely available from the Austroads web site: www.austroads.com.au/asset/test.html.

The web site also contains Commentaries each of the specification and test methods. These
commentaries provide additional notes and advice to practitioners, and have been structured to be
read in conjunction with the relevant specification or test method.

E.4 Role of the Specifications and Test Methods


The test methods contain specific actions to be undertaken, however for activities other than
calibration, the test methods provide little information regarding when each test needs to be
conducted. For example, the test method for conducting on-going repeatability testing of a
roughness measuring profilometer does not contain any statements as to when such testing must
occur. This information is provided in the specification.

Each specification contains the following three annexes that must be completed by the client:
ƒ Annex 1 – list of roads to be surveyed
ƒ Annex 2 – location referencing system
ƒ Annex 3 – data file format.

The specification cannot be used whilst these annexes remain empty. It is anticipated that clients
with established procedures for procuring data collection contracts will have little difficulty in
populating these annexes. Some guidance is included in each Annex to assist less experienced
clients in the completion of the document.

The specifications consider the interaction between three distinct parties, as shown in Table E.2.

Austroads 2007

— 28 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Table E.2: Role of different parties considered in the specification

Role Description
Client The organisation for whom the data is to be collected
Contract Supervisor The representative of the client organisation
Supplier The operators of the survey equipment and suppliers
of the resultant data

The test methods do not make reference to the specifications, and can be used independently of
them. Accordingly, the test methods can be used in-house by organisations collecting data, where
there may not be a need for a formal specification. Alternatively, the test methods could be used
for project level data collection, for which the contract model envisaged by the specifications may
be inappropriate.

In the majority of cases, it is anticipated that the specification, and in turn the test methods, will be
framed within broader contract documentation specific to a given application. Such overarching
contract documentation provides an opportunity for the client to specify additional requirements, or
make specific exclusions, from the Austroads specification.

Whilst the specifications could be amended on a contract-by-contract basis to reflect the


requirements for the specific project, this is not recommended as it would dilute the standardised
nature of the Austroads specification. It is preferred that the contract documentation requires
adherence to the Austroads specification, but separately lists additional amendments,
requirements or exclusions.

Table E.3 presents the intended roles of the specifications, test methods and broader contract
documentation within a condition data collection contract.

Table E.3: Document roles within a condition data collection contract

Document type Description


Test methods Instructions for carrying out a specific activity
ƒ statements about which activities to undertake and when to undertake them
ƒ list of roads to be surveyed
Specifications
ƒ details of the client’s referencing system (i.e. the system to be used to report the data)
ƒ details of the data format for the presentation of the final data.
ƒ legal terms of contract
ƒ quality plan and procedures
Additional contract ƒ provision for risk (including types of insurances)
documentation ƒ definitions of responsibilities (e.g. traffic control)
ƒ additional requirements
ƒ specific exclusions from Austroads specification.

When considering additional requirements or exclusions to the specifications and test methods it is
important to remember that, during their development, these documents have been closely
examined by asset managers, tender documentation writers and equipment users. Accordingly,
the specification and test methods should be considered as both technically sound and practical to
use. Care should be taken when adding requirements to the specification to ensure that there is a
need to do so, and that the additions are achievable and relevant.

Austroads 2007

— 29 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

E.5 Generic Structure of the Standard Specifications and Test Methods


E.5.1 Survey Method
The survey method defines how a data collection survey is to be undertaken. It requires that the
equipment be validated once per annum in accordance with the referenced validation procedure.
The method usually also contains minimum equipment requirements (i.e. an equipment
specification).

E.5.2 Validation Method


A validation method defines how equipment validation is to be undertaken. It does not specify how
often the validation needs to occur. It is required that, as part of the validation exercise, a
repeatability check be undertaken in accordance with a referenced test method. In some cases
alternative validation methods are presented. It is explicitly stated that there is no requirement that
validations be undertaken on roads within the client’s jurisdiction. Accordingly, under the terms of
the specifications, the data collection service provider need not conduct separate validations for
each client, but must only ensure that the minimum timetable of validations is adhered to.

E.5.3 Repeatability Method


The repeatability method defines how checks for repeatability of measurements are to be
undertaken. The method does not specify how often these checks need to occur. In most cases
these repeatability checks also include an assessment of the bias error of measurements.

E.5.4 Specification
The specification requires that the survey method be followed, and provides a timetable for
conducting equipment validations in accordance with the validation method(s). In addition, a
timetable for conducting ongoing checks of the repeatability of measurement following the
repeatability method is specified. For large scale data collection exercises it is additionally
required that a 100 km trial survey be undertaken and the results presented in final reporting
format for client checking and approval. This ‘sanity check’ aims to ensure that the entire data
collection system, and not just the equipment, is able to meet the needs of the client, and also
allows the client to compare collected data with historical records and other data sources.

E.6 Overview of Specification and Test Methods


E.6.1 Calibration, Measurement and Reporting
Whilst it is required that equipment be calibrated, no specific instructions are provided as to how, or
when, this calibration must be undertaken. Reference is made to the appropriate documentation
provided by the manufacturer of the equipment. The test methods do, however, identify triggers for
calibration, including the replacement of key equipment components (e.g. laser transducers).

As a minimum it is required that roughness be measured using the International Roughness Index
(IRI, m/km) in both the outer and inner wheelpaths of the test lane. Both of these roughness
values are to be reported every 100 metres along the test lane. Additionally, the lane IRI is to be
determined (the average of the two wheelpath IRI values) and similarly reported for every 100 m
length.

Conversion of IRI recorded roughness to the older NAASRA Roughness Meter scale is considered
by the test method, but is not mandatory.

The survey test method also includes daily equipment checks that must be conducted before
survey activity can begin.

Austroads 2007

— 30 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

E.6.2 Equipment Validation and Repeatability of Measurement Checks


The specification requires that a validation exercise must have been passed within the last year.
Two alternative test methods are provided for equipment validation.

The first method compares the measurement of the tested profilometer to the measurement from a
static or manual reference device such as the ARRB Walking Profiler, or a surveyor’s staff and
level. The test sites selected for the conduct of the validation exercise must meet specified
roughness characteristics. The method requires the results of multiple survey passes at a variety
of speeds to be compared, using linear regression analyses, against the data collected by the
reference device. The tests are to be carried out at a variety of test speeds in order to
demonstrate that the profilometer’s results are not affected by survey speed. The method requires
that the repeatability checks described below be conducted as part of the validation exercise.

The second method involves the comparison of roughness data from the profilometer requiring
validation against a reference set of roughness data collected by other profilometers. The
procedure was developed by RTA NSW (2002) and has been used by a number of Australian road
authorities. As the method incorporates its own repeatability checks it does not require that the
separate repeatability method be conducted as part of the validation exercise.

It is required that checks for repeatability and bias of measurement be undertaken as part of the
first validation method above. Additionally it is required that these checks be conducted at a
minimum of every 30 days during the collection of data. The test method compares two sets of
100 m average lane roughness readings along a 10 km length of road. A variety of statistical
comparisons are made between these data sets, based upon coefficients of variation, and pass/fail
limits are specified.

COMMENTARY F DATA REPORTING


Measurement of road roughness should generally be accurate to within 3%. It is therefore
misleading to record roughness measures more precisely than the following:
ƒ IRI (m/km): 2 decimal places
ƒ NAASRA Roughness Counts (counts/km): the nearest whole number.
Austroads has endorsed the International Roughness Index (IRI) as the reporting unit for road
roughness in Australasia. After a considerable transition period, continuing use of the traditional
reporting unit (NAASRA roughness meter counts, or NRM) is no longer considered appropriate and
all roughness results should now be reported exclusively in terms of IRI.
The interval at which roughness is calculated has an impact on the ‘apparent’ severity of
roughness. Long intervals such as 1 km tend to average out the higher roughness and make it
more difficult to detect changes in roughness and identify specific discrete features in the profile.
In Australia and New Zealand an interval of 100 m is accepted as the appropriate standard
reporting interval to detect changes over time (provided longitudinal referencing remains accurate
and consistent), and is long enough to produce a repeatable result.
Some road agencies use ‘rubber-banding’ to match surveyed lengths to lengths in the road
reference database. As a result, survey segments could vary, from say 99 metres to 101 metres,
depending on the tolerance allowed in the specification. This is solely a product of the road
agency’s road location system, and is unlikely to have a discernible impact on data consistency at
the network level or between road agencies.

Austroads 2007

— 31 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Inaccuracies in profile measurement occur in inertial profilometers whenever the speed of the
survey vehicle drops below about 25 km/h. This can occur at intersections where survey speed
may fluctuate to avoid stopping at traffic lights and therefore has a particular effect on metropolitan
surveys. The main reason is that accelerometers are unable to detect longer wavelength profile
shape at low travel speeds. The effect of this is usually that the roughness result is grossly
inflated. Profile points taken at less than this speed should be discarded prior to the calculation of
IRI. Where this leaves less than a 50 m continuous length of data, no result should be reported for
that interval (see Figure F.1).

Figure F.1: Example graphical reporting of roughness results (NRM counts/km) showing gaps in data at low survey speeds

COMMENTARY G MEASUREMENT OF ROUGHNESS


G.1 Panel Ratings
Panel rating methods to assess roughness condition by subjective means is not accepted practice
for Australasian conditions where automated methods are readily accessible to road agencies.
However, an example of a rating system is shown in Figure G.1.

Austroads 2007

— 32 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

DETAILS ACCEPTABLE RATING

SECTION # YES 5
RATER NO VERY GOOD
4
VEHICLE UNDECIDED
GOOD
DATE 3
TIME FAIR
2
AUSTROADS GUIDE POOR
1
VERY POOR
0

Figure G.1: Example of a panel rating form

The rating that a person gives to a road, however, can be influenced by a range of factors besides
the actual condition of the road pavement, including their own standards (which can be
considerably different for a community member compared to a maintenance engineer), visual
appearance of the pavement and roadside, and general mood at time of assessment. Safety
issues for raters are also a concern.

As noted by Sayers and Karamihas (1996) there are two major problems when using panel ratings
for network use:

The rating scale is not a measure of road condition that is stable with time. For example, roads
considered ‘good’ by a panel today might be considered something else by a panel in 50 years.
[In addition] it is expensive to obtain panel ratings due to the number of people required, and the
need to transport them to the roads being rated.

G.2 Response Type Roughness Measurement


Response type devices provide a much simpler form of road roughness measurement than profile-
based devices. However, the major disadvantages of measuring roughness with response type
devices are:

ƒ Inability to analyse the data for any type of vehicle other than that used as the measurement
survey vehicle.
ƒ The dependence of the data on the nature and condition of the suspension characteristics of
the survey vehicle. Response type devices have to be calibrated regularly to correct for
changes in vehicle response (e.g. suspension, tyres, shock absorbers).
ƒ The need for the survey vehicle to travel at a constant travel speed (50 km/h or 80 km/h).

The 1981 Standard Operating Instructions for the NAASRA Roughness Meter specify calibration
when the vehicle is replaced and at intervals of not more than two years, and maintenance checks
immediately after repairs to the vehicle (e.g. tyres changed) and when any part or the whole
instrument is replaced.

Austroads 2007

— 33 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

G.3 Measuring Longitudinal Profile


Systems which involve measuring a longitudinal profile and converting it to a numerical indicator of
road roughness depend on accurate measurement of the profile. Profile measurement is affected
by sampling interval, surface texture, measurement accuracy, signalling, and filtering, as discussed
below.

G.3.1 Sample Interval and Wavelength Range


The true longitudinal profile is continuous. However, nearly all profile measuring systems take
measurements at discrete distance intervals. Ideally, the sample interval should be short enough
to capture the profile characteristics of interest. Road profile can be considered as being made up
of a large number of sinusoids of varying wavelengths and amplitudes. PIARC has adopted a
wavelength range for roughness of 0.5 m to 50 m, as shown in Table G.1. A range of 1.2 m to
30 m has been used in some places (Sayers and Karamihas 1986). To capture the features of
these sinusoids over the wavelengths of interest, the sampling interval must be no greater than half
the shortest wavelength, or 250 mm. The wider wavelength range 0.5 m to 50 m is recommended
in these guidelines to improve IRI sensitivity to specific pavement conditions, and to provide a
profile range that could be used for future calculation of truck ride quality indicators (Prem 1998).

Table G 1: Wavelength ranges for road surface characteristics (Descornet 1990)

Surface characteristics Frequency range

Wavelength Wavenumber

Microtexture < 0.5 mm > 2,000 cycles/m

Macrotexture 0.5 - 50 mm 20 - 2,000 cycles/m

Megatexture 50 - 500 mm 2 - 20 cycles/m

Roughness 0.5 - 50 m 0.02 - 2 cycles/m

G.3.2 Surface Texture


Textured surfaces can cause errors in profile measurement, for example where the sample interval
is larger than the ‘footprint’ of the measuring device (Prem 1999). These errors depend on the
type of instrument used but can be minimised by increasing the number of samples taken per unit
distance. This problem particularly applies to chip seal surfaces, which are predominant in
Australia and New Zealand. Therefore, while the minimum theoretical sample interval is 250 mm,
the preferred sample interval for Australian and New Zealand conditions is 50 mm.

G.3.3 Performance of Profile Measuring System


The performance of a profile measuring system depends on the design and proper functioning of
many pieces of hardware and software. Each element of a system has limitations on its
performance (electronic, mechanical etc.). Overall, the performance of each profile measuring
system is effectively the combined performance of all its measuring devices (height sensors,
accelerometers, filters, etc.), data acquisition system and its operational settings, processing
algorithms and profile computation methods. Profile computation methods alone can lead to
significant differences between profile measuring systems. Filtering can also have a significant
effect on the resulting reported profile.

Austroads 2007

— 34 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Class 1 measuring device


To obtain a valid measure of the IRI there are minimum requirements of measurement accuracy
that must be satisfied. To comply as a Class 1 method requires sampling of the longitudinal road
surface profile along the travelled wheelpath at least every 250 mm with a precision in elevation of
at least 0.5 mm (0.2 mm desirable). In addition, profile information within the IRI waveband (0.5 m
to 50 m) must also be recovered.

Inertial profilometers
Accelerometers are used in inertial profilometers to measure the very long wavelength profile
components as well as vehicle body movement relative to an inertial (essentially fixed) frame of
reference. When the accelerometer signal is processed and combined with the height sensor
recordings, the long, medium and short wavelength components in the profile are combined to
produce an overall profile free from unwanted vehicle bounce motion that appears in both the
accelerometer and height sensor signals. At very low speeds the long wavelength features only
produce very low acceleration levels for the accelerometers to measure. The accelerometers are
set up to measure the higher acceleration levels that occur when the profilometer is operated at
highway speeds. At lower speeds this leads to poor recovery of long wavelength information.
Generally, a profilometer speed of 25 km/h should be viewed as a lower limit for obtaining profiles
for the calculation of IRI. Results obtained when the survey speed is less than 25 km/h should be
rejected.

Laser and ultrasonic sensors


Profilometers which rely on reflection of various types of wave energy to detect the road surface
are prone to reduced accuracy in wet conditions because of reflection from the surface of water,
rather than from the surface of the road.

Ultrasonic height sensors (which are not currently used in Australia) are inadequate for accurately
measuring longitudinal profile. At highway speeds, ultrasonic sensors can take up to four height
readings per metre and rely on the speed of sound through the air to measure height. They are
more sensitive to wind and changes in temperature than laser and optical sensors which rely on
the speed of light. Typically the ‘footprint’ of an ultrasonic height sensor is about 50 mm in
diameter, compared to a fine point for laser and optical sensors. For a given speed of travel of a
profilometer along a road, an ultrasonic sensor cannot sample the profile at the same spatial
frequency as a laser or optical sensor.

Note that a profilometer may have numerous height sensors. However, normally only two sensors,
one in each wheelpath, are used to measure longitudinal profile and derive roughness. The other
sensors may be used to measure transverse profile to derive rutting. Height sensors in
profilometers that operate continuously along a road at some forward velocity (i.e. other than in
static devices) also require accelerometers to measure longitudinal profile.

Austroads 2007

— 35 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

COMMENTARY H PROFILE ANALYSIS


H.1 Lane Roughness (correlating NRM and IRI)
Two methods can be used to calculate lane roughness from profile data, ‘profile averaging’ (half-
car simulation, giving Lane IRIhc) and ‘IRI averaging’ (quarter-car simulation, giving Lane IRIqc).
Details of these methods are shown in Table H.1. Profile averaging takes more account of height
differences between the left and right wheelpaths, and so is considered a better simulation of
mechanical response type devices such as the NAASRA roughness meter. These height
differences produce a roll response in vehicles that may affect the comfort of occupants, but is not
registered in roughness values. In analysis of Australian data (Prem 1989), ARRB found that the
profile averaging method provided a slightly better correlation between IRI and NAASRA
roughness, compared with the IRI averaging method which reports the average of the two
wheelpath IRIs.

Analysis to determine Lane IRIhc (profile averaging) is more complex than for Lane IRIqc (IRI
averaging) because the inner and outer wheelpath profiles have to be synchronised by sampling
the profile at precisely the same longitudinal location in both wheelpaths. For IRI averaging (Lane
IRIqc), this synchronisation is not necessary. For single track profilers, such as the Face
Corporation’s DipStickTM and the ARRB Walking Profiler, the sampled points in the inner and outer
wheelpaths do not always line up precisely, particularly if there is some horizontal curvature and
one wheelpath is longer than the other, or if one wheelpath is rougher and therefore slightly longer.

The World Bank has adopted IRI averaging (i.e. Lane IRIqc) for use in HDM products. Careful
reference to Table H.1 will reduce the risk of confusion between these two forms of Lane IRI.
There is only a marginal difference in accuracy between the two methods, and both have strong
correlations to NAASRA roughness.

It is important to note that the numerical value of Lane IRIqc from a quarter-car model is generally
greater than Lane IRIhc from profile averaging (half-car simulation). The difference depends on the
roughness characteristics of the road. Research has shown that on typical Australian roads the
half-car Lane IRIhc is about 0.79 Lane IRIqc (Prem 1989).

Because of the different regression formulae (see Table H.1) there is little difference in the
resulting NAASRA roughness values using either method. While IRI is not necessarily a universal
standard, as most of Europe and parts of the USA do not use IRI, it is the most widely adopted
form of roughness. The IRI averaging method also makes the calculation of NAASRA roughness
from IRI more transparent and is consistent with the methods used to determine IRI elsewhere in
the world. These guidelines support use of the World Bank form of Lane IRI in the form of Lane
IRIqc.

The high correlation between NAASRA roughness and IRI means that conversion of historic data
is a reasonable option. IRI can be calculated from NRM using the regression equations in
Table H.1. A conversion chart allowing for conversion of roughness data between the IRI and
NRM scale is shown as Table H.2.

Austroads 2007

— 36 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Table H 1: Roughness measures available in Australasia

Profile-based profilometers
Measuring device Mechanical response-type
Profile readings are taken at approximately 50 mm centres along the profile being surveyed
Name of measure NAASRA Roughness Counts Single Wheelpath IRIqc IRI averaging Profile averaging
Reporting units Counts/km m/km m/km m/km
Symbol (complete) NRM Single Wheelpath IRIqc Lane IRIqc Lane IRIhc
Common symbol NRM Single Wheelpath IRIqc IRI Lane IRIhc
The cumulative total relative upward The point-to-point average of the two
physical displacement between axle and International Roughness Index calculated for The Single Wheelpath IRIqc for each wheelpath profiles is determined and the
body of a standard vehicle operating at a each 100 m segment by applying a wheelpath for each 100 m segment is quarter-car mathematical model is
Description
standard speed (1 NRM corresponds to mathematical model of a quarter-car to a computed independently and then applied to this average profile to
a relative upward physical displacement single longitudinal profile averaged to produce Lane IRIqc simulate a ‘half-car’ model, i.e.
of 15.2 mm) Lane IRIhc
Favoured by Austroads Traditional measure, being phased out No Yes No
Used in HDM-4 products No No Yes No
Used by ARRB in calibrating
Not applicable No Yes No (was used until mid 1990s)
NAASRA roughness meters
NRM = - 3.47
Conversion to NRM NRM = NRM =
Not applicable - 0.557 (Single Wheelpath IRIqc)2 + 27.50 x
(Prem 1989) – 1.27 + 26.49 x Lane IRIqc – 1.95 + 33.67 x Lane IRIhc
(Single Wheelpath IRIqc)
Correlation (r2) (Prem 1989) Not applicable 0.955 0.990 0.994
Typical relative value in
Not applicable Not applicable Lane IRIqc = 1.27 Lane IRIhc Lane IRIhc = 0.79 Lane IRIqc
Australia (Prem 1989)

Austroads 2007

— 37 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

Table H 2: Conversion tables for IRI and NRM roughness values

NRM IRI IRI NRM


(counts/km) (m/km) (m/km) (counts/km)
20 0.8 1.0 25
30 1.2 1.5 38
40 1.6 2.0 52
50 1.9 2.5 65
60 2.3 3.0 78
70 2.7 3.5 91
80 3.1 4.0 105
90 3.4 4.5 118
100 3.8 5.0 131
110 4.2 5.5 144
120 4.6 6.0 158
130 5.0 6.5 171
140 5.3 7.0 184
150 5.7 7.5 197
160 6.1 8.0 211
170 6.5 8.5 224
180 6.8 9.0 237
190 7.2 9.5 250
200 7.6 10.0 264
210 8.0 12.0 317
220 8.4 14.0 370
230 8.7 16.0 423
240 9.1 18.0 476
250 9.5 20.0 529

Note: IRI values are quarter car, i.e. IRIqc.

H.2 Profile Filtering for IRI Calculation


The measured profile is filtered with a moving average that has a 250 mm base length to simulate
the effect of the tyre-to-road contact area and the way in which a tyre envelops the small sharp
unevenness features. This filter ‘smooths’ the profile by averaging individual profile points over a
250 mm length. While sampling should be undertaken at smaller intervals (these guidelines
require a sample every 50 mm), the software smooths the profile by averaging over a 250 mm
base length, before calculating IRI.

The smoothed profile is then further filtered by application of the quarter-car model (Figure H.1),
with specific parameter values that define the Single Wheelpath IRIqc, at a simulated speed of
80 km/h.

Austroads 2007

— 38 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

The IRIqc is an accumulation of the simulated motion between the sprung and unsprung masses in
the quarter-car model, divided by the length of the profile.

The resulting IRI statistic has dimensionless units of metres/kilometre (m/km). Much of the
literature from the United States refers to IRI with dimensionless units of in/mi (inches per mile).
1 IRI (m/km) is equal to 63.36 IRI (in/mi).

Figure H.1: Hypothetical IRI quarter-car model (after Sayers et al. 1986b)

Austroads 2007

— 39 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

COMMENTARY I VALIDATION OF EQUIPMENT


I.1 Validation of Distance Measurement (Test Method AG:AM/T005)
Accurate correlation of road condition data with physical location on the road network is pivotal to
the success of almost all uses of road condition data. Reliable distance measurement is therefore
an essential aspect of all road condition surveys.

Road agency databases should generally be supported with robust and unambiguous physical
reference points and permanent features such as roadside markers, bridge abutments, side road
intersections, etc. Examples of the establishment and use of referencing systems can be found in
Transit New Zealand (1998) and RTA NSW (2002a). Road agency databases should also contain
the distances between the various reference points and permanent features. A limited number of
these distances should be measured to a high degree of accuracy (0.005% or better) using
geomatic (precise ground survey) techniques, and clearly marked. The precisely measured
sections should be sealed, relatively straight, without sags or crests, and between 1 km and 15 km
long. There are a number of ongoing uses for measured sections, including calibration and checks
of a range of road condition survey host vehicles and equipment.

Preparation for a road condition survey should include inspection and any necessary maintenance
of roadside markers at reference points.

The validity of checks of distance outputs from a road condition measuring device against precisely
measured test lengths depends on an accurate method of detecting and recording the start and
end of each measured length. An automated means of detecting and recording the ends of
measured lengths is desirable because distance outputs can be checked at speed, without
additional traffic control, and without interruption to survey operations. A light sensor mounted in
the host vehicle and triggered by reflectors at right angles to the road has proved suitable for
recording the ends of measured lengths.

I.2 Validation of Roughness Measurement


I.2.1 Profile-based Measuring Devices
The major problem facing road agencies is to be assured that the roughness measuring device is
accurately measuring the road profile over the wavelength range of interest (0.5 m to 50 m). In
addition to the technical guidelines in Section 3.2, a robust method of assessing the validity of
reported roughness is needed.

The mathematical model used to run the quarter-car over the longitudinal profile can readily be
checked using the RoadRuf public domain software (UMTRI 1997). However, the measurement of
profile requires an ultimate check against the actual physical profile, which is taken to be the profile
as reported by a Class 1 profile measuring method. Class 1 instruments establish a ‘reference
profile’ which is a practical method of estimating the true profile. Class 1 instruments are therefore
suitable for validation checks on profilometers for network level surveys.

Austroads 2007

— 40 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

I.2.2 Establishing a Reference Profile


Reference device method (Test Method AG:AM/T002)
Various methods and devices are available to establish the reference profile, including rod and
level, Face Corporation’s DipStickTM, and ARRB’s Walking Profiler. An independent evaluation
(Fong and Brown 1997) of the WP in New Zealand showed that it meets the ± 2% accuracy
requirement for a Class 1 instrument defined in Sayers et al. (1986a) and Sayers (1995). The WP
can measure 0.9 km of profile per hour, compared to the DipStickTM which operates at a much
lower speed of 0.3 km of profile per hour.

Austroads Test Method AG:PT/T450 (Austroads 2007g) describes use of the WP for measuring
the reference profile on test sections. However, due to the difference in the methods of
measurement between the WP and non-contact profilometers, it is important that validation be
undertaken on smooth textured surfaces (Prem 1999).

Loop method (Test Method AG:AM/T003)


The alternative validation procedure establishes a reference profile from repeated survey runs
using a reference laser profilometer. This is less time-consuming and less disruptive to traffic than
measuring profile with a static Class 1 device which is a slow process. The alternative procedure
relies on the accuracy of the reference inertial profilometer, and an ability to distinguish changes in
the roughness of the test sites over time from changes in the performance of the reference
profilometer.

I.2.3 Profile Validation


The preferred method used by these guidelines to verify roughness measuring equipment makes
use of the comparison of the IRI results determined by the measuring equipment with the IRI
results provided by a reference device. An alternative approach would be to focus on validation of
the actual profile measurements that are used as inputs to the IRI calculation and not just the
results of that calculation – this is termed ‘profile validation.’

It is impractical to compare elevation profiles directly because they are dominated by large
amplitude, long wavelength components that have very little, if any, influence on IRI or NAASRA
roughness. Filtering out the longer wavelengths or analysing the individual wavelength ranges can
overcome this.

Prem (1999) developed a profile validation method using frequency analysis techniques. Some
problems have been encountered in applying the technique as a simple pass/fail test under
contract conditions across pavements with a wide range of surface textures. Of particular issue is
the effect that coarse surface texture can have on the comparison of profiles collected using laser
technology with those obtained using either geomatic surveys or the ARRB Walking Profiler.

Accordingly, until further research has been conducted on the sensitivity of profilers to coarse
texture, it is recommended that, in the interim, profile validation not be used as an essential
criterion for acceptance of roughness data in data collection contracts. However, if the profile
validation method is used, and agreement between the measuring equipment and a reference
device is clearly demonstrated then the data should be accepted.

The interested reader is referred to Prem (1999) for a full discussion of the profile validation
approach.

Austroads 2007

— 41 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

I.2.4 IRI Validation


Profile-based measuring devices
Given the correct longitudinal profile of a road, the calculation of IRI is a mathematical expression
of the profile using a quarter-car model. Validation of IRI is a test to ensure that the mathematical
process is correct.

Response type measuring devices


IRI can be directly calculated from a conversion of the NAASRA roughness values measured using
the NAASRA Roughness Meter. Validation of IRI is the key test to ensure the device is correctly
measuring roughness.

I.2.5 Selection of Validation Sites


The characteristics of the validation sites listed in AG:AM/T002 have been selected to ensure a
reasonable spread of roughness condition, over which the performance of the profilometer can be
assessed. The validation limits listed in the test method cannot be used if the selected validation
sites do not match the characteristics listed. The validation analysis makes use of linear
regression techniques and uses the r2 statistic as a validation limit. This statistic is heavily
influenced by the range of data over which is it determined.

I.3 Repeatability and Bias (Test Method AG:AM/T004)


Measurement repeatability and bias are important factors in ensuring that changes in roughness
can be identified over relatively short time spans. Measurement repeatability is an indication of
variation in measures about the mean, whereas bias error indicates whether a device is
systematically measuring high or low when compared to a reference set of measures. Bias error is
potentially a more serious problem that can distort perceptions of the condition of the road network
and indicate that pavement condition is better or worse than it really is. Bias error can be caused
by normal wear and tear, errors in equipment calibration, or physical damage to the measuring
device sustained during the survey.

Changes in roughness at the network level are generally of the order of 1% to 3% per annum.
Laser profilometers typically achieve repeatability of individual 100 m measurements that are no
greater than 3% of the mean and a bias error that is less than 1%. The levels of repeatability and
bias are critical in determining changes in roughness, and at current accuracy at least three years
would be required between roughness measurements to obtain statistically significant changes in
roughness.

The path followed by survey vehicles is the major source of variation affecting repeatability.
Similarly, variation in the travelled path between different drivers is the major source of variation
affecting reproducibility of results.

Austroads 2007

— 42 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

COMMENTARY J NETWORK LEVEL DISTRIBUTION


ANALYSIS EXAMPLE
This brief example demonstrates the utility of histograms and cumulative distributions in examining
network trends.

The distributions shown in Figure J.1 are based on real data for a metropolitan main road network.
It can be seen that there has been a significant change in the overall condition of the network
between 1994 and 1999. In the histograms of Figure J.1(a) the leftwards shift in the distribution
peaks and the drop in the height of the right end distribution tail are showing that the proportion of
the network length with high roughness has decreased. The right hand leading edge of the
distributions has remained relatively uniform indicating that whilst the network has shown an
overall improvement in roughness, the proportion of the network with very low roughness has not
changed. This same improvement in the network can be seen in the leftwards shift of the
cumulative distribution in Figure J.1(b).

Between 1999 and 2003 the network has changed less dramatically, possibly marking the end of a
concentrated burst of maintenance/rehabilitation activity in the network. The tail end of the 2003
distribution shows a slight increase in the proportion of high roughness lengths, and whilst it is only
a slight change it is fairly uniform, probably indicating an unchecked deterioration of part of the
network. This deterioration may well be the result of conscious management policy, or may be
unintentional. In either case, the distribution analysis has enabled the effect to become readily
identifiable.

As a second example consider the distributions plotted in Figure J.2 which summarise the
roughness of a rural road network. A clear and steadily increasing roughness level across the
entire network is evident as the distributions progressively shift to the right. Such a shift in the
roughness distributions, unless the result of deliberate policy, is usually of concern to the network
manager as it indicates that the current maintenance and rehabilitation program is not holding the
network condition in a stable state.

Another factor that might also shed some light on the differences in the two distributions is the
different nature of the maintenance and rehabilitation activities in metropolitan and rural networks.
In metropolitan networks like the one plotted above the treatment of the majority of pavement
length involves the placement of asphalt, and therefore benefits from asphalt’s general roughness
improving characteristics. Conversely, the primary treatment technique in the rural network would
be the application of reseals which have little effect on road roughness. In a rural environment only
deliberate selection of specific treatments will have an effect on roughness condition, whereas in
the metropolitan environment reduction in roughness can be obtained from treatment types that
are actually targeting other issues entirely.

Austroads 2007

— 43 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

20

18 2003

16 1999

Percentage of Network 14 1994

12

10

0
0 1 2 3 4 5 6
Roughness - IRI m/km

(a) – Distribution (histogram)

100

90 2003

80 1999
Cummulative Percentage of Network

70 1994

60

50

40

30

20

10

0
0 1 2 3 4 5 6
Roughness - IRI m/km

(b) – Cumulative distribution

Figure J.1: Roughness distribution of the main roads in a metropolitan network

Austroads 2007

— 44 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

20

18 2003

16 1999

14 1994
Percentage of Network

12

10

0
0 1 2 3 4 5 6
Roughness - IRI m/km

100

90 2003

80 1999
Cummulative Percentage of Network

70 1994

60

50

40

30

20

10

0
0 1 2 3 4 5 6
Roughness - IRI m/km

Figure J.2: Roughness distribution of the main roads in a rural network

Austroads 2007

— 45 —
GUIDE TO ASSET MANAGEMENT PART 5B: ROUGHNESS

COMMENTARY K DETERMINATION OF SMOOTH TRAVEL


EXPOSURE
The Smooth Travel Exposure indicator forms one of a suite of Austroads National Performance
Indicators (NPIs). Full details of the determination of the NPIs, including the STE, and the latest
reported values, can be found via www.austroads.com.au.

The STE aims to monitor whether roads are providing acceptable travel conditions. The STE can
be briefly described as the proportion of travel that is undertaken on roads with conditions above
the targeted conditions for those roads. The STE is reported for two different target conditions: IRI
of 4.2 and 5.3 m/km (corresponding with NRM roughness values of 110 and 140 counts/km
respectively).

The following methodology outlines how the STE is determined for a road network:
1. Determine the length of the network (i.e. the number of kilometres), which meet the targeted
condition for smooth roads as determined by the mean reading for the segment length. This
is to be undertaken for IRI roughness < 4.2 m/km and roughness < 5.3 m/km and by
summing segment lengths which have an:
Average roughness < 4.2 m/km
Average roughness < 5.3 m/km

The segment lengths used should be the finest level of data reasonably available and
normally used by the organisation, and be consistent each year.

2. Determine the annual average daily traffic for each segment length of road, and compute the
annual kilometres travelled:
Annual km travelled = annual average daily traffic x segment length x 365

3. Sum the total annual kilometres travelled on smooth road segments with IRI < 4.2 m/km.
4. Sum the total annual kilometres travelled on smooth road segments with IRI < 5.3 m/km.
5. Sum the total annual kilometres travelled on all road segments.
6. Calculate the STE values for each target condition as follows:

total annual km travelled on road segments with IRI roughness < 4.2 m/km
STE 4.2 = 100 ⋅
total annual km travelled on all road segments
total annual km travelled on road segments with IRI roughness < 5.3 m/km
STE 5.3 = 100 ⋅
total annual km travelled on all road segments

Austroads 2007

— 46 —

You might also like