PUBLICATION SUPERSEDED

This document was published by Austroads.

Whilst it has been superseded by later publications, it

has been made available electronically on the ARRB
Knowledge Base for historical purposes with the
permission of Austroads.

For those seeking current material and guidance,

please visit the Austroads online publications database:

http://www.onlinepublications.austroads.com.au/
So

GUIDE TO TRAFFIC MANAGEMENT

Part 3: Traffic Studies and Analysis

 Guide to Traffic Management
Part 3: Traffic Studies and Analysis
Guide to Traffic Management Part 3: Traffic Studies and Analysis

Summary
The Austroads Guide to Traffic Management has 13 parts and provides comprehensive coverage
of traffic management guidance for practitioners involved in traffic engineering, road design, town
planning and road safety.
Part 3 – Traffic Analysis is concerned with the collection and analysis of traffic data for the purpose
of traffic management and traffic control within a network. It serves as a means to ensure some
degree of consistency in conducting traffic studies and surveys. It provides guidance on the
different types of traffic studies and surveys that can be undertaken, their use and application, and
methods for traffic data collection and analysis.
Part 3 updates previous Austroads guidance on traffic studies and capacity analysis. It covers
applications of the theory presented in Part 2 of the guide, and provides guidance on traffic
analysis for uninterrupted and interrupted flow facilities and for various types of intersections. It
outlines sound methods of analysis for effective traffic management, design and control.

Keywords
Traffic management, road network, data collection, data analysis, field study, speed, origin
destination traffic, level of service, traffic capacity, traffic flow, intersection, unsignalised
intersection, roundabout, signalised intersection.

First Published August 2009

© Austroads Inc. 2009

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.

ISBN 978-1-921551-77-2

Austroads Project No. NP1380

Austroads Publication No. AGTM03/09

Project Manager
John Erceg, Main Roads WA

Prepared by
David Bennett, Rod Troutbeck, Peter Croft, David McTiernan, Agasteena Shah

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: [email protected]
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 Traffic Management
Part 3: Traffic Studies and Analysis

Sydney 2009
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, the Commonwealth Department of Infrastructure, Transport, Regional Development
and Local Government in Australia, the Australian Local Government Association and New
Zealand Transport Agency. It is governed by a council consisting of the chief executive officer (or
an alternative senior executive officer) of each of its 11 member organisations:
ƒ Roads and Traffic Authority New South Wales
ƒ Roads Corporation Victoria
ƒ Department of Transport and 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 Department of Infrastructure, Transport, Regional Development and Local Government
ƒ Australian Local Government Association
ƒ New Zealand Transport Agency.

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 TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

CONTENTS
1 INTRODUCTION ............................................................................................................ 1
1.1 Scope and Context ......................................................................................................... 1
1.2 Introduction to Traffic Studies and Traffic Analysis......................................................... 3
1.3 Structure of this Part ....................................................................................................... 5
2 TRAFFIC STUDIES AND SURVEYS............................................................................. 6
2.1 Traffic Studies and Data Needs...................................................................................... 6
2.2 Planning and Designing Traffic Studies.......................................................................... 7
2.2.1 Systems Approach to Traffic Studies................................................................ 7
2.2.2 Preliminary Planning......................................................................................... 8
2.2.3 Sample and Experimental Design .................................................................. 10
2.2.4 Survey Form Design ....................................................................................... 10
2.2.5 Pilot Survey..................................................................................................... 10
2.2.6 Conducting Survey ......................................................................................... 10
2.2.7 Data Coding and Entry ................................................................................... 11
2.2.8 Data Analysis.................................................................................................. 12
2.2.9 Presentation of Results................................................................................... 12
2.2.10 Archiving ......................................................................................................... 13
2.3 Statistical Methods and Sampling................................................................................. 13
2.3.1 Statistical Methods.......................................................................................... 13
2.3.2 Sampling Methods .......................................................................................... 13
2.3.3 Sampling Error and Bias................................................................................. 16
2.3.4 Sample Size Determination ............................................................................ 16
2.4 Data Integrity ................................................................................................................ 17
2.5 Traffic Surveys.............................................................................................................. 18
2.5.1 Traffic Volume Surveys (Appendix A)............................................................. 19
2.5.2 Speed Surveys (Appendix B).......................................................................... 20
2.5.3 Travel Time, Queuing and Delay Surveys (Appendix C) ................................ 20
2.5.4 Origin-destination Surveys (Appendix D )....................................................... 21
2.5.5 Pedestrian and Bicycle Surveys (Appendix E)................................................ 22
2.5.6 Noise, Fuel and Emission Surveys (Appendix F) ........................................... 23
2.5.7 Vehicle Mass and Dimension Surveys (Appendix G) ..................................... 24
2.5.8 Parking Surveys (Appendix H)........................................................................ 24
2.5.9 Traffic Generation Surveys (Appendix I)......................................................... 25
3 TRAFFIC ANALYSIS – CAPACITY AND LEVEL OF SERVICE................................. 26
3.1 Types of Traffic Facilities .............................................................................................. 26
3.2 Capacity, Level of Service, Degree of Saturation ......................................................... 26
3.2.1 Capacity.......................................................................................................... 26
3.2.2 Level of Service (LOS).................................................................................... 27
3.2.3 Service Flow Rate........................................................................................... 29
3.2.4 Degree of Saturation....................................................................................... 30
3.3 Factors Affecting Capacity, Level of Service, Degree of Saturation ........................ 31
3.3.1 Ideal Conditions .............................................................................................. 31
3.3.2 Roadway Conditions....................................................................................... 31
3.3.3 Terrain Conditions .......................................................................................... 31
3.3.4 Traffic Composition......................................................................................... 31
3.3.5 Pedestrians and Cyclists ................................................................................ 32
3.3.6 Driver Population ............................................................................................ 32
3.3.7 Control Conditions .......................................................................................... 32

Austroads 2009
— i—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

4 UNINTERRUPTED FLOW FACILITIES....................................................................... 33

4.1 Single Lane Flow .......................................................................................................... 33
4.1.1 Capacity.......................................................................................................... 33
4.2 Two-lane Two-way Roads ............................................................................................ 35
4.3 Multi-lane Roads........................................................................................................... 39
4.4 Freeways ...................................................................................................................... 42
4.4.1 Basic Freeway Segments ............................................................................... 43
4.4.2 Ramps and Ramp Junctions........................................................................... 46
4.4.3 Weaving sections............................................................................................ 49
5 INTERRUPTED FLOW FACILITIES ............................................................................ 55
5.1 Metered Flow ................................................................................................................ 55
5.1.1 Capacity.......................................................................................................... 55
5.1.2 Queue Lengths and Delays ............................................................................ 56
5.2 Urban Arterial Roads with Interrupted Flow.................................................................. 57
5.2.1 Capacity.......................................................................................................... 57
5.2.2 Analysis of Level of Service............................................................................ 58
5.2.3 Descriptions of Levels of Service.................................................................... 59
6 INTERSECTIONS......................................................................................................... 60
6.1 Unsignalised Intersections............................................................................................ 61
6.1.1 Unsignalised Intersections with Minor Roads ................................................. 61
6.1.2 Unsignalised Intersections with Major Roads ................................................. 61
6.2 Roundabouts ................................................................................................................ 68
6.3 Traffic Simulation and Computer Analysis Models ....................................................... 71
6.4 Signalised Intersections................................................................................................ 71
6.4.1 Concepts and Definitions................................................................................ 71
6.4.2 Capacity Analysis ........................................................................................... 74
6.4.3 Measures of Intersection Performance........................................................... 78
6.4.4 Signal Calculations ......................................................................................... 81
REFERENCES ...................................................................................................................... 82
APPENDIX A TRAFFIC VOLUME SURVEYS ........................................................ 90
APPENDIX B SPEED SURVEYS.......................................................................... 120
APPENDIX C TRAVEL TIME, QUEUING AND DELAY SURVEYS ..................... 129
APPENDIX D ORIGIN-DESTINATION SURVEYS ............................................... 138
APPENDIX E PEDESTRIAN AND BICYCLE SURVEYS ..................................... 146
APPENDIX F NOISE, FUEL AND EMISSION SURVEYS.................................... 153
APPENDIX G VEHICLE MASS AND DIMENSIONS SURVEYS .......................... 160
APPENDIX H PARKING SURVEYS ..................................................................... 168
APPENDIX I TRAFFIC GENERATION SURVEYS ............................................. 176
APPENDIX J ERROR IN SEASONAL ADJUSTMENT FACTOR........................ 179

Austroads 2009
— ii —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

TABLES
Table 1.1: Parts of the Guide to Traffic Management ........................................................ 2
Table 2.1: Relative error .................................................................................................. 17
Table 3.1: Performance measures used for defining levels of service ............................ 29
Table 4.1: Adjustment factors for lane width and lateral clearance ................................. 34
Table 4.2: Average passenger car equivalents for heavy vehicles on grades with
single lane flow ............................................................................................... 34
Table 4.3: LOS criteria for two-lane highways in Class II ................................................ 38
Table 4.4: Level of service criteria for multi-lane highways ............................................. 41
Table 4.5: LOS criteria for basic freeway segments ........................................................ 44
Table 4.6: Approximate capacity of ramp roadways in passenger cars/hour .................. 48
Table 4.7: Capacity values for merge areas in passenger cars/hour .............................. 48
Table 4.8: Capacity values for diverge areas in passenger cars/hour ............................. 48
Table 4.9: Level of service criteria for merge and diverge areas ..................................... 49
Table 4.10: Level of service criteria for weaving segments. .............................................. 52
Table 4.11: Capacity for various weaving segments ......................................................... 54
Table 5.1: Typical mid-block capacities for urban roads with interrupted flow................. 57
Table 5.2: Design categories of urban streets ................................................................. 58
Table 5.3: Average travel speeds .................................................................................... 58
Table 6.1: Intersection volumes below which capacity analysis is unnecessary ............. 61
Table 6.2: Base critical headways and follow-up times for two-way stop controlled
intersections.................................................................................................... 65
Table 6.3: Level of service criteria for two-way stop controlled intersections .................. 66
Table 6.4: Base saturation flows in through car units per hour by environment class
and lane type .................................................................................................. 75
Table 6.5: Through car equivalent (through car units per vehicle) for different types
of vehicles and movements ............................................................................ 77

Austroads 2009
— iii —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

FIGURES
Figure 2.1: Systems approach to traffic studies.................................................................. 8
Figure 2.2: An example of a 3x3 Latin square design....................................................... 15
Figure 3.1: Levels of service and service flow rates ......................................................... 30
Figure 4.1: Speed-flow and percent-time-spent-following relationships for directional
segments with base conditions....................................................................... 37
Figure 4.2: LOS criteria for two-lane highways in Class I ................................................. 38
Figure 4.3: Speed-flow curves with LOS criteria for multi-lane roads. .............................. 41
Figure 4.4: Speed-flow relationship for basic freeway segments...................................... 44
Figure 4.5: Influence area at ramps .................................................................................. 46
Figure 4.6: Influence-area density for on-ramp flows of between 5 and 15% of the
freeway flows; the acceleration lane length is 100 m ..................................... 47
Figure 4.7: Influence-area density for off-ramp flows of between 5 and 15% of the
freeway flows; the deceleration lane length is 100 m ..................................... 47
Figure 4.8: Diagrammatic representation of traffic on freeway weaving areas. ................ 49
Figure 4.9: Classification of weaving types ....................................................................... 51
Figure 4.10: Level of service for weaving sections of Type A, with 40% weaving traffic
and varying total traffic.................................................................................... 52
Figure 4.11: Level of service for weaving sections of Type A, with 4000 passenger
car units per hour total traffic and with different proportions of weaving
traffic ............................................................................................................... 53
Figure 5.1: Typical metered flow situation ........................................................................ 55
Figure 6.1: Major and minor streams at a T-intersection. ................................................. 62
Figure 6.2: Traffic stream ranks at two-way stop controlled intersections ........................ 63
Figure 6.3: Two-stage headway acceptance analysis ...................................................... 66
Figure 6.4: 95th percentile queue lengths.......................................................................... 67
Figure 6.5: Unsignalised intersection vehicle storage requirements................................. 68
Figure 6.6: Major and minor flows at a roundabout. ......................................................... 69
Figure 6.7: Upper and lower bound capacity estimates for roundabout (inscribed
diameter 40 m, entry lane width 4 m) ............................................................. 70
Figure 6.8: Upper and lower bound capacity estimates for roundabout (inscribed
diameter 60 m, entry lane width 4 m, dominant/subdominant lane flow
ratio 1.2).......................................................................................................... 71
Figure 6.9: Typical intersection plan and phasing diagram............................................... 72
Figure 6.10: The basic capacity model ............................................................................... 73

Austroads 2009
— iv —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

1 INTRODUCTION
1.1 Scope and Context
Part 3 of the Austroads Guide to Traffic Management has been given the title Traffic Studies and
Analysis to define the limitations on its scope within the contexts of:
ƒ the 13 different Parts of the Guide to Traffic Management
ƒ other guides spanning the range of Austroads publications.

The structure and content of the 13 Parts of Guide to Traffic Management is discussed in Part 1 –
Introduction to Traffic Management. The 13 parts are outlined in Table 1.1.

Austroads 2009
— 1—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table 1.1: Parts of the Guide to Traffic Management

Part Title Content
Part 1 Introduction to Traffic Introduction to the discipline of traffic management.
Management Breadth of the subject and the relationship between the various parts of the guide.
Part 2 Traffic Theory An introduction to the characteristics of traffic flow and the theories, models and
statistical distributions used to describe many traffic phenomena.
Processes that practitioners should consider.
Part 3 Traffic Studies and Analysis Traffic and transport data collection surveys and studies.
Traffic analysis for mid-block situations (including freeways/motorways).
Analysis of signalised and unsignalised intersections, including roundabouts.
Part 4 Network Management Broader issues and aspects of managing networks of roads to provide effective
traffic management for all road users.
Network needs of freight, public transport, pedestrians, cyclists and private motor
vehicles.
Network management objectives, operational objectives, network performance
measures.
Part 5 Road Management Focus on managing mid-block traffic conditions.
Good practice for access management, allocation of space to various road users,
lane management.
Application of speed limits.
Part 6 Intersections, Interchanges and Types of intersection.
Crossings Selection of intersection type and appropriate use.
Traffic considerations in traffic management for intersections, interchanges and
other crossings.
Part 7 Traffic Management in Activity Planning and traffic management of activity centres and associated transport nodes.
Centres Principles for various types of centre.
Part 8 Local Area Traffic Management Principles and processes.
Issues and resources.
Selection of schemes and treatments.
Design and implementation of schemes and devices.
Part 9 Traffic Operations Applications used in traffic operations.
Current practice for common systems including: traffic signals, congestion
management, incident management, traveller information.
Manual systems used in these application areas.
Event management.
Information management issues and principles.
Related systems integration and interoperability issues.
Part 10 Traffic Control and Signing and marking schemes.
Communication Devices Traffic signs, static and electronic.
Pavement markings and delineation.
Traffic signals and islands.
Part 11 Parking Parking policy.
Demand and supply.
Data and surveys.
On-street and off-street.
Types of parking and parking control.
Part 12 Traffic Impacts of Developments Relationship to road level of service and access management.
Development profile and trigger points for treatment.
Traffic impact assessment.
Part 13 Road Environment Safety Principles and management of the safety of road environments within a traffic
management context.
Links to relevant sections of the GRD (Guide to Road Design) and GRS (Guide to
Road Safety)

Austroads 2009
— 2—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

In the context of the guide, Part 3 Traffic Studies and Analysis outlines the importance of traffic
data and its analysis for the purpose of traffic management and traffic control within a network. It
serves as a means to ensure some degree of consistency in conducting traffic studies and
surveys. It provides guidance on the different types of traffic studies and surveys that can be
undertaken, their use and application, and methods for traffic data collection and analysis.

Part 3 provides a means to assist in implementing the various traffic management and control
measures described in:
ƒ Part 4, which is concerned with the overall operational improvement needs of the road
network as a whole (Austroads 2009e)
ƒ Part 5, which deals in detail with mid-block traffic management issues that apply to individual
lengths of road (Austroads 2008b)
ƒ Part 6, which deals with traffic management issues and solutions at particular intersections,
interchanges and crossings (Austroads 2007)
ƒ Part 9, which covers the operational management of road space for all users and describes
current practice for common systems including traffic signal systems, congestion
management systems, incident management systems and traveller information systems
(Austroads 2009e).

Part 3 covers applications of the theory presented in Part 2 (Austroads 2008a), and provides
guidance on traffic analysis for uninterrupted and interrupted flow facilities and for various types of
intersections. It outlines sound methods of analysis for effective traffic management, design and
control.

Part 3 updates previous Austroads guidance on traffic studies and capacity analysis, and makes
reference to other seminal publications such as the US Highway Capacity Manual (TRB 2000) and
the handbook from Monash University (2003).

1.2 Introduction to Traffic Studies and Traffic Analysis

In the context of Part 3, the term ‘traffic study’ involves the collection, analysis and synthesis of
data relating to traffic and its characteristics. The term ‘traffic survey’ refers generally to the
procedures and activities involved with measurement and data collection. In practice the term
‘traffic study’ is often used generically to refer to the combination of survey and analysis activities.

Traffic studies have a central place in traffic engineering and transport planning. Their importance
is expanding due to the rapid developments in data collection technologies, and the corresponding
ability to collect and analyse more data of better quality. There are also increasing demands to
improve the operational efficiency in traffic systems to meet travel and transport objectives for the
community.

Traffic studies and surveys provide essential information for much of traffic engineering planning,
design and practice. The results are used in traffic planning, traffic management, traffic and
environmental control, economic studies, road safety studies, land use-transport interaction
studies, and in establishing and revising design standards and system models.

Traffic surveys are the means of collecting traffic data, providing basic information about the
operating characteristics of the traffic. This typically includes information on vehicle flows, speeds,
travel times, delays, pedestrian and cyclist movements, journey origins and destinations, and
vehicle types and characteristics.

Austroads 2009
— 3—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The information may be sought at a number of levels of detail, from broad indications of traffic
conditions over a region to a detailed measurement of individual vehicle movements at a chosen
location. A good working knowledge of survey techniques and technology is required by all traffic
practitioners.

A systematic approach to traffic studies and surveys provides a central framework for drawing
together the wide range of needs, requirements and techniques for the various types of traffic
survey. This assists a proper understanding of traffic data and their use in analysing and
determining the characteristics of traffic systems.

Traffic studies include the synthesis and analysis of traffic information for the purposes of traffic
management and control, and for the design of roads and associated traffic facilities. They often
involve evaluations of existing conditions or projections relating to proposed changes. The
analysis of traffic characteristics for these purposes requires an understanding of the basic
concepts of capacity analysis and level of service.

Capacity analysis is fundamental to the planning, design and operation of roads and transport
services. It provides quantitative techniques for measuring the effectiveness of existing transport
facilities in moving traffic and people, and is used to determine the impacts of proposed system
improvements (e.g. additional traffic lanes, signal timing adjustments, enhanced public transport
services).

Capacity analysis provides the basis for determining the number of traffic lanes to be provided,
having regard for the volume and composition of traffic and the prevailing roadway and traffic
control conditions. For a given number of traffic lanes, capacity analysis provides a means of
determining the traffic-carrying performance of a road under the prevailing roadway, traffic and
control conditions.

Quite different analytical approaches are applicable for the basic types of traffic facility, which are
categorised as:
ƒ Uninterrupted flow facilities, on which traffic flow conditions are the result of interactions
between vehicles in the traffic stream, and between vehicles and the geometric
characteristics of the road
ƒ Interrupted flow facilities, on which traffic flow conditions are subject to the influence of fixed
elements such as traffic signals, stop signs or other controls, which cause traffic to stop
periodically.

Signalised intersections and major unsignalised intersections influence the capacity and traffic
performance of interrupted flow facilities, and specific approaches are applicable for analysing the
capacity of various intersection types.

Capacity analysis is not restricted to roads. The Highway Capacity Manual (TRB 2000) referred to
subsequently as HCM 2000, provides concepts and methods for undertaking capacity analysis for
pedestrians and cyclists on footpaths and bicycle facilities.

A primary objective in planning, designing and managing traffic facilities is to achieve a balance
between maximising flow rates and ensuring acceptable operating conditions for users of the
facilities. The notion of acceptable operating conditions is encompassed in the concept of level of
service.

Level of service is a qualitative measure for ranking operating conditions, based on factors such
as speed, travel time, freedom to manoeuvre, interruptions, comfort and convenience.

Austroads 2009
— 4—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

1.3 Structure of this Part

Section 2 describes an overall systems approach to traffic studies, including statistical and
sampling issues. Major components of traffic studies are identified, and the interactions between
these components are highlighted. Section 2 also summarises various types of traffic studies and
surveys, and the Appendices provide further information and guidance for each type of survey. In
most survey areas, new technologies are increasingly employed for data collection and analysis.
These are mentioned in context but are not considered in detail in this guide.

Section 3 presents an overview of the concepts of capacity, level of service and degree of
saturation and the factors which affect them.

Sections 4, 5 and 6 provide information and guidance on capacity analysis as applied to

uninterrupted flow facilities, interrupted flow facilities and intersections, respectively. Substantial
reference is made to detailed information available in HCM 2000. In these sections traffic analysis
methods using graphs and equations are included to give practitioners an understanding of the
fundamental principles and relationships. In current practice, most methods are pursued via
computer-based analysis, and in the context of electronic publications many of the graphical
approaches and equations could eventually be replaced by on-line spreadsheets.

Austroads 2009
— 5—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

2 TRAFFIC STUDIES AND SURVEYS

2.1 Traffic Studies and Data Needs
Traffic studies have a central place in transport planning and traffic engineering. They are
performed to:
ƒ provide a basis for planning and designing traffic facilities, including the selection of
geometric standards, economic analysis, impact assessment, and the determination of
priorities
ƒ assist traffic operation by indicating the needs for traffic control devices such as signs, traffic
signals, pavement markings, and school and pedestrian crossings
ƒ evaluate the effects of road safety measures and other changes made for traffic by
conducting ‘before and after’ studies
ƒ determine the basic characteristics and the general laws of traffic behaviour
ƒ provide heavy vehicle and freight data to improve pavement analysis and design capability,
bridge management capability, and the monitoring of road network performance.

Road usage is also linked to funding of maintenance programs, so collection and analysis of data
on road traffic performance can be a fundamental component of submissions for funding
allocations.
The importance of traffic studies is expanding in the face of three influences (Taylor, Young and
Bonsall 1996):
ƒ the rapid developments in data collection hardware, driven by microprocessor developments
and the corresponding ability to collect and analyse more data of better quality
ƒ the increased demands placed on the traffic engineering profession to obtain improved
efficiency of operations in traffic systems in order to meet increasingly complex objectives for
the community
ƒ the traffic engineer’s own demands for better quality data to permit the use of the most recent
traffic planning and design techniques and modelling procedures.

The purposes for which traffic data are required may be summarised as (Young, Taylor and Gipps
1989):
ƒ Monitoring. The collection of information about traffic conditions prevailing at any time, and
as they change over time.
ƒ Forecasting. The use of data on existing traffic systems as one of the inputs to a procedure
for estimating what the traffic would be like under different conditions, either now or in future.
ƒ Calibration. The use of traffic data to estimate the values for one or more parameters in a
theoretical or simulation model.
ƒ Validation. The verification of a theoretical or simulation model against information
independent of that used to calibrate the model.

Information typically sought in traffic surveys includes:

ƒ flows of vehicles (possibly classified by vehicle type) and pedestrians
ƒ numbers of waiting vehicles or pedestrians
ƒ flows of bicycle users

Austroads 2009
— 6—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

ƒ numbers of parked vehicles

ƒ numbers of occupants in vehicles
ƒ speeds of vehicles
ƒ travel times, delays and their components
ƒ origins and destinations of journeys
ƒ vehicle mass and dimensions
ƒ fuel consumption and emissions.

In most cases, such information is generally obtained by passive observation and measurement,
without making direct contact with road users or directly interfering with traffic flows. Technological
developments in data acquisition and analysis are facilitating the efficiency and effectiveness of
passive observation methods.

The common traffic surveys are described in Section 2.5 and details are provided in Appendix A to
Appendix F.

A comprehensive overview of traffic and transportation surveys is provided in Institute of

Transportation Engineers (1994) and survey methodology and data analyses are examined in
Taylor, Young and Bonsall (1996).

General guidance on conducting traffic studies and surveys is given in Currin (2001), Monash
University (2003), Homburger et al (2007) and Institute of Transportation Engineers (2009).

2.2 Planning and Designing Traffic Studies

All traffic studies require careful planning and administration of data collection, collation and
analysis tasks.

2.2.1 Systems Approach to Traffic Studies

The systems approach allows the identification of processes common to all traffic studies, which
form the fundamental building blocks of good study design. Figure 2.1 illustrates the process, its
components and interactions.

Figure 2.1 indicates that traffic studies should follow a series of logical, interconnected steps
leading towards the final outcome of the study. It is also to be noted that there are significant
linkages between the various activities. These linkages are of three types – forward, feedback and
backward:
ƒ Forward linkages (solid lines) indicate the basic progression through the process.
ƒ The feedback linkages (dotted lines) between component activities indicate that two or more
activities must be performed sequentially in a closed loop. For example, having performed
the pilot survey, it may be necessary to redesign the survey form and then pilot test the new
form.
ƒ The backward linkages (dashed lines) indicate where information may be transferred back
from a later activity to one which preceded it. For example, the design of the survey form
may be affected by the coding methods used to extract data from the forms, while the type of
data analysis may influence coding itself. It is important that consideration is given to
backward linkages so that decisions made early in the survey process do not preclude
options for later data analysis.

Austroads 2009
— 7—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure 2.1: Systems approach to traffic studies

The following sections outline each component in the traffic analysis system of Figure 2.1, and
highlight major considerations required for the design of various survey types.

2.2.2 Preliminary Planning

The initial activities in a study are of paramount importance. The best way to ensure the success
of a traffic study is to recognise and explore the following steps:

Study objectives
The study team, in conjunction with the client for whom the study is to be conducted, should
develop objectives specifying the basic questions to be answered in both qualitative and
quantitative terms. For example, prior agreement with the client should be made on the level of
sampling and the level of statistical significance before a survey starts.

Review of existing information

Before embarking on the collection of a new set of data, the investigator should ascertain just how
much is known about the subject in question. Existing knowledge is available through the review
of published literature, knowledge of current and completed studies and other relevant data sets
compiled by other agencies. If these sources indicate that the subject has been partially
researched, the existence of other data sets could be useful. In particular, it may provide new
ideas about the study and assist the survey design, e.g. by providing estimates of population
parameters for use in determining sample size.

Austroads 2009
— 8—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Formation of hypotheses
If new data need to be collected to satisfy the objectives, then the correct types of data must be
collected. The only way to ensure the relevance of the new data is to develop the possible
hypotheses to be tested. This is a fundamental part of the methodology of scientific inquiry on
which traffic studies should be based. The scientific approach consists of four major phases:
hypothesis formulation, observation, testing, and refinement of hypothesis. Data collection is the
basis of observation, and this must be preceded by hypothesis formulation. The recommended
practice is to establish a range of plausible hypotheses, and then determine the survey
requirements so that each hypothesis may be properly tested.

Definition of parameters
After hypothesis formulation, there is often a need to refine the meanings of several concepts and
terms. Some effort is then needed to clearly define the terminology to be used in the survey.

This is important for two reasons. Firstly, without agreement among analysts and users of survey
data on terminology, ambiguities and disagreements over concepts and parameters are likely.
Secondly, agreement allows comparisons between different data sets, survey methods and
analyses.

A clear definition of parameters in a traffic study leads to a better interpretation of the results.

Resource determination
In most studies, there will be constraints in terms of the level of resources that can be committed to
the survey. These are generally time, people and money, and a trade-off between available
resources and the accuracy obtainable from a sample survey is possible (see Survey Methods
below).

Survey content
The proper adoption of survey objectives facilitates the task of determining survey content.

Data items should only be included in the survey content if they are required for the intended
analysis. Nevertheless, even this approach may result in a long list for survey content. It might
also be noted that some data additional to basic needs, collected at little or no extra cost, might be
useful in future. The task of refining the list will be influenced by the trade-off between costs and
available resources.

Survey methods
The choice of a survey method is crucial for the efficiency of the overall survey effort. The selected
method will usually result from a compromise between the objectives of the survey, expressed in
the survey content list, and the available resources. The quality of the data obtained will depend
on the survey method selected and the amount of quality control performed. However, more
precise survey techniques with higher quality control will generally consume more survey
resources.

The use of video technologies for recording a wide variety of road and traffic data is expanding
(See Roper 2004, for example). Issues relating to data quality, storage, access and integrity are
just as pertinent for video methods as they are for other approaches using manual or computer-
based methods.

The trade-off for survey type must be made on the basis of time, money and people so that a
feasible survey method and sample size can be found for the successful completion of the survey.

Austroads 2009
— 9—
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

2.2.3 Sample and Experimental Design

All traffic surveys are sample surveys. Traffic systems contain very large numbers of individual
units and factors. It is uneconomic, if not virtually impossible, to observe all of them. What is
required is to survey a suitably chosen sample from the population, on the basis of the
experimental design, sampling methods and statistical theory in order to draw out statistically valid
inferences (e.g. see Section 2.3 and textbooks such as Miller et al. 1999 and Taylor et al. 2000).

2.2.4 Survey Form Design

For standard surveys (e.g. turning movement counts or speed surveys), standard forms may exist.
Most Austroads member authorities will have their own forms developed over many years to meet
their particular needs. However, special surveys may require new forms to be developed, e.g. lane
saturation flows at signalised junctions (see Parts 6 and 9 of the guide and Akçelik 1989).

An aspect of traffic survey design often overlooked is the design of forms for recording data.
Careful attention to the design of a form can often lead to more efficient job performance and lower
error rates by record and data entry officers. Modern information technologies have made the task
of preparing quality survey forms much easier and cheaper; for example, electronic recognition of
bar codes, or intelligent data collection forms mounted on laptop computers or portable digital
assistant devices.

2.2.5 Pilot Survey

Pilot testing is an important component of a study design, but is often neglected. The usual
reasons for ignoring pilot surveys are lack of time or money. Experience indicates that these are
false economies for all but the smallest or most standard surveys. Pilot studies provide assistance
with many aspects of the final survey design and analysis, including:
ƒ adequacy of sampling and analysis method
ƒ adequacy of survey forms
ƒ efficiency of training methods
ƒ suitability of coding and editing
ƒ cost and duration of survey and analysis
ƒ suitability of traffic control methods on-site, where applicable
ƒ effectiveness of technology or data collection devices.

2.2.6 Conducting Survey

The key to successful traffic studies lies in the attention paid by the study manager to the details of
survey administration. Frequently, a survey is undertaken by third parties according to a brief, with
the analysis and reporting undertaken ‘in-house’. The brief should include:
ƒ general outline of type of survey to be undertaken and purpose
ƒ details of field survey and survey design (if included)
ƒ competencies required and field instruction detailed
ƒ criteria for data acceptance
ƒ structure of data outcomes
ƒ provision of method for conducting survey, data quality plan and training proposed by
consultants.

Austroads 2009
— 10 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Proposals should be assessed on the basis of management capability (scheduling, contingency

planning, staff recruitment and training, quality plan, resource programming), technical capability
(project management skills of key personnel, level of technical support), documented prior
experience and price.

The following checklist may help to reduce the severity of problems arising during a survey:
ƒ Personnel must receive training in the purpose of the survey and in the methods of
measurements to be employed. On-hand support by competent administrators is especially
critical where casual personnel are employed specifically for the survey. Field staff should
be encouraged to provide feedback, particularly of unusual events that might impinge on the
quality of the data collected. The client should also be closely involved in the training
process and identify deficiencies in the process.
ƒ Replacement field staff should be available, especially on the first day of a multi-day survey.
During the survey, rosters of rest periods for observers are essential, and replacement staff
will be needed to cover these periods.
ƒ Survey forms should be prepared and distributed, as far as possible, on the eve of the
survey. The pre-study briefing is ideal for this distribution.
ƒ Procedures should be in place to ensure quality control of data and survey operations. Plans
and procedures should be in place to accommodate failure to obtain information. An
example would be where an observer either fails to collect travel time data over a route due
to collection equipment malfunction or there are systematic errors introduced due to
non-compliance with instructions.
ƒ Occupational health and safety policies must be followed. Rest and meal breaks must
comply with relevant industrial awards. They are essential in maintaining the alertness and
concentration of field personnel and reducing the errors of observation and recording.
Depending on the survey type, a maximum three-hour shift (preferably two-hour maximum)
should be considered. Often the full complement of field staff will be required for active
surveying only during peak periods. In between these times some staff can take rest periods
while others share their duties. The obvious proviso is that the times of peak traffic
movements are known before the survey schedule is set. Proper rostering of field personnel
taking account of the above factors may avoid or reduce high overtime costs and safety
risks.
ƒ Privacy legislation must be followed. Data of a confidential nature should not be disclosed
without proper clearance.

2.2.7 Data Coding and Entry

The time spent in the field is usually a small proportion of the total time required to produce the
study results. Data entry and compilation is time-consuming and new technologies often provide
the potential for significant savings.

Data coding is the process of converting data from field records into computer-readable formats.
Editing is the parallel process of data scanning to detect recording errors and logical errors in the
data. Editing may be performed manually or automatically.

Austroads 2009
— 11 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The major opportunity for time saving is the combination of coding and editing into a single data-
entry process, where the data are entered interactively by the user. Many inconsistencies can be
detected immediately by the editing program, which enables the user to quickly correct faulty data
entries. Further time savings are possible with the use of hand-held data loggers, laptop personal
computers (PCs) and voice recognition methods for field recording: the data are ready for analysis
as soon as recording finishes. In some cases, the same PC used to collect the data can perform
analyses in the field. The possibilities for using PCs and wireless Internet technologies for data
transfer are expanding rapidly.

2.2.8 Data Analysis

Data analysis is an important element of any traffic study. It is the means by which the collected
data are used to test the hypotheses set during the planning of the study. Statistical procedures
are employed to make this part of the process as effective as possible. Textbooks such as Miller
et al. (1999) and Taylor et al. (2000) provide good references for the application of statistical
analysis in traffic studies. See also Section 2.3 for guidance on methods for ensuring adequate
sample sizes for statistical testing.

2.2.9 Presentation of Results

The preparation and distribution of reports is the primary means of conveying the study methods
and results to other people. There is little point in undertaking a study if it is not adequately
reported, and proper preparation of the report is essential. The following principles should be used
when preparing a survey report:
ƒ The purpose and scope of the survey should be fully explained.
ƒ The report must cater to the types of readers for whom it is being written, the likely extent of
their knowledge and understanding, the types of problems and questions that are likely to be
of interest, and the kind of language (technical or otherwise) to which they are accustomed.
Statistical technicalities should be translated into language that will be understood by readers
who are interested in the broad and substantive results of the study.
ƒ Statistical results should be framed jointly in terms of percentage error and confidence
interval. Because a study typically deals with a sample of a broader target population, the
degree to which the sample is representative of that population must be addressed. It is
appropriate to specify a risk factor that the sample is unrepresentative. Commonly, a risk of
1 in 20 chance (sample) is taken, which translates to the ‘95% confidence interval’.
ƒ Qualitative and graphical descriptions of the data are useful in developing an understanding
of the survey results. Many readers find tables difficult to comprehend. Some verbal
summary of the salient points in a table, and graphs and diagrams displaying table contents,
make a report easier to understand and less prone to misinterpretation.
ƒ One successful course is to prepare three types of reports: a preliminary report, a summary
report and a technical report. The preliminary report contains the primary results and should
be released early to enable those who wish to use the data immediately to do so. The
summary report specifies the objectives of the traffic study, the general method used, the
major analytical results and the interpretation of those results. The technical report
documents the survey and provides complete details of the results.

Austroads 2009
— 12 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

In general, a traffic study report should include: summary, introduction, study objective and
hypothesis, experimental design, study procedure and statistical results. The results should
include descriptive statistics such as tables, graphs, histograms, pie-charts, box plots, stem and
leaf plots, which are readily available from PC software. Standard statistical test results should be
clearly stated, including the strength of relationships between variables and relevant confidence
intervals. The report should outline the statistical methods used, present and discuss the results,
and draw conclusions. Data listings can be included in appendices where appropriate.

2.2.10 Archiving
The effective archiving of data and details of the survey procedure is a means of ensuring that the
data will be available for secondary analysis at a future date.

This important task involves documentation of survey technique, storage of the data with sufficient
information to permit easy retrieval, indication of where the data are stored, and nomination of
those who have continuing responsibility for maintaining the database.

In summary, the systems approach to the design and management of traffic studies provides a
formal structure for indicating the importance of good practice in study planning, design, execution
and data analysis. There are significant linkages between these components, and the systems
approach offers valuable insights into the traffic study process. The needs for reliable and
representative traffic data are so great, and data collection is so expensive that ‘ad hoc’ unrelated
survey procedures should be avoided.

2.3 Statistical Methods and Sampling

2.3.1 Statistical Methods
The use of statistical methods is an important component of traffic analysis. Statistical procedures
can be divided into three general groups:
ƒ data reduction – provides an initial look at the data, and involves procedures for descriptive
statistics, which describe the nature of the observed data
ƒ hypothesis testing – uses the theories of statistical inference to estimate the nature of the
population from the sample data
ƒ relationship determination – examines the relationships between observed variables, checks
the quality and does preliminary exploratory analysis to confirm if survey data is sufficient.

A full description of each of these groups is beyond the scope of this Part of the Guide to Traffic
Management. Details are available in statistics textbooks. Where some of the procedures from
these groupings are required in this Part, a brief account is given in the section for the particular
type of traffic study.

2.3.2 Sampling Methods

Almost all traffic data collection is based on sample surveys. This section indicates how to select a
sample.

A prerequisite for good sample design is that there is a clearly defined the objective for the study.
Otherwise, it is only too easy to collect too much data, which is expensive and wastes resources;
too little data, which results in inconclusive, incomplete answers; or the wrong data, which provides
useless results and wastes resources.

Austroads 2009
— 13 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The following main elements of sample design may be identified:

ƒ definition of target population
ƒ definition of sampling unit
ƒ selection of sampling frame
ƒ choice of sampling method
ƒ consideration of likely sampling errors and biases
ƒ determination of sample size.

Target population
Changes to a traffic system are often aimed at a particular class of traffic or road user, and their
performance needs to be assessed with respect to that class or group. Seldom will changes have
equal effects on all road user groups. This may bring complications when trading off positive
effects for some groups (e.g. users of radial arterial roads) against negative effects for others (e.g.
cross-town travellers), when assessing the impacts of, say, signal coordination systems. The
particular subgroup of road users or sites that the survey is intended to cover is the target
population. Clear definition of the target population will subsequently reduce the level of
uncertainty associated with decision-making, and lead to optimisation of the resources needed to
complete the survey.

The population being sampled must be representative of the target population. The survey might
include the collection of supplementary information to determine the nature of the differences
between the sample and target populations.

Sampling unit
A population consists of individual elements, e.g. individual vehicles. In one study, the focus may
be on individual vehicles, whereas in another study the focus may be on types of vehicles. The
definition of a sampling unit therefore depends on the nature and purpose of the study, and is
constrained by the efficiency, effectiveness and productivity (of survey personnel) in collecting the
required data.

Examples of a sampling unit include individuals, vehicles, households, car parks, and road links.

Sampling frame
The sampling frame is the ‘register’ of the target population. It is the list or definition that identifies
the members (units) of the target population. A proper specification of the sampling frame is
required for sampling to proceed, and for any inferences to be made about the target population
from the sample results.

The choice of the sampling frame depends on the needs and constraints of the particular study.
For instance, the sampling frame might consist of the list of vehicles registered as being garaged in
a certain area, or it could be all those vehicles using a particular road or junction during a specified
period. The main requirement is that the sampling frame can be quantifiably identified so that the
size of the target population may be determined.

Sampling method
The methods for selecting samples from a target population include random sampling and
judgement sampling. In random sampling, all members of the target population have a chance of
being selected in the sample, whereas judgement sampling uses personal knowledge, expertise
and opinion to identify sample members.

Austroads 2009
— 14 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Judgement samples avoid the statistical analysis necessary for random samples (indeed it is
wrong to apply statistical analysis to them), and have a certain convenience. There is a particular
role for judgement sampling in exploratory or pilot surveys where the intention is to examine the
possible extremes of outcomes with minimal resources. However, judgement samples have no
statistical meaning. They cannot represent the target population and the results are almost
certainly biased.

A random sampling scheme should be used, to ensure that the sample taken is statistically
representative. Random samples may be taken by basic sampling methods such as:
ƒ simple random sampling
ƒ systematic sampling
ƒ stratified random sampling
ƒ cluster sampling.

Simple random sampling is the most commonly used method. A sample is selected by some
method that allows each possible sample to have an equal probability of being chosen, and each
unit in the target population has an equal probability of being included in any one sample.
Sampling may be either with replacement (i.e. any member may be selected more than once in
any sample draw) or without replacement (i.e. after selection in one sample, that unit is removed
from the sampling frame for the remainder of the draw for that sample). Selection of that sample is
by way of a number assigned to each unit in the sampling frame, with repeated random digits
being the most convenient means of drawing a random sample. Random number tables are
contained in statistical textbooks, or may be generated by computer.

Systematic sampling, stratified random sampling and cluster sampling all attempt to approximate
simple random sampling. They have been developed for their precision, economy or ease of
application. However, it is the principle of simple random sampling that lies behind the
mathematical theories of statistical inference, i.e. the process of drawing inferences about a
population from information on samples drawn from that population.

For a systematic sampling in two dimensions, the Latin square design should be used. This design
can minimise biases (Miller et al. 1999) and is a balanced experimental design that aims to control
extraneous sources of variability while investigating the impact of different treatments. It adjusts
for these sources of variability by systematically blocking in two directions, so that there are two
restrictions on the rows and columns of the square.

For example, let the treatment be three methods of counting traffic, and the two factors affecting
traffic counting accuracy be field staff and equipment type. A Latin square design requires equal
numbers of field operators and equipment types. Let there be three field operators and three
equipment types. Then the Latin Square design is a three-by-three matrix (Figure 2.2) with each
method of counting traffic being used once by each operator using each equipment type. Each
counting method appears once and only once in each row and in each column of the matrix.

Field staff A Field staff B Field staff C

Equipment type I Method 1 Method 2 Method 3
Equipment type II Method 2 Method 3 Method 1
Equipment type III Method 3 Method 1 Method 2

Figure 2.2: An example of a 3x3 Latin square design

Austroads 2009
— 15 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Applications of the Latin square design in traffic surveys were reported for example in Johnston et
al. (1982) (see also Appendix C).

2.3.3 Sampling Error and Bias

Nearly all data observations form samples from some larger population. There is a possibility that
the sample does not adequately reflect the nature of the parent population. The random
fluctuations that are inherent in the sampling process can be examined using the methods of
statistical inference. There may also be biases (systematic errors) in sample data, and these may
distort an analysis that assumes all errors are random by nature. For example, a traffic
management measure that produces unequal benefits for different road user groups may cause
the proportions of those groups to alter in the area influenced by the measure, as road users adjust
to the new system. The proper treatment of samples is therefore important if biases and variations
in the unseen section of the population are not to invalidate an analysis.

The means for overcoming systematic errors lie in proper experimental design and survey
planning, as described in Section 2.2.

All possible means of samples, drawn from a target population, have a distribution that is different
from the target population. The mean of the sample means is equal to the mean of the population.
However, the standard deviation of the sample means, which is termed the ‘standard error’, is not
equal to the standard deviation of the population. The standard error is important in the selection
of a sample size. Standard error indicates:
ƒ the size of the error in the estimation of the true mean (or proportion) due to random
variations in extracting sample statistics
ƒ the accuracy of estimation of a population parameter from the sample statistic.

2.3.4 Sample Size Determination

Traffic surveys for specific investigations attempt to provide data for the estimation of particular
population parameters, or to test statistical hypotheses about a population. In either case the size
of the sample selected will be an important element, and the reliability of the estimate will increase
as sample size increases. On the other hand, the cost of gathering data will also increase with
increased sample size. A trade-off may occur, and the additional returns from an increase in
sample size will need to be evaluated against the additional costs incurred. If the target population
is infinite (i.e. very large compared to the sample size), then the standard error SX of the mean of
the variable x is given by:
ı
SX
n

where ı is the population standard deviation and n is the sample size

This equation implies that as sample size increases, standard error decreases in proportion to the
square root of n. To double the precision of an estimate will require the collection of four times as
much data. Consequently, there is value in seeking an optimum sample size. Note that it is the
absolute size of the sample that determines sampling precision, rather than the fraction of the
target population.

Austroads 2009
— 16 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

These results provide preliminary information on how to estimate a desirable sample size. For
example, assume that the aim is to estimate a mean speed at a site, with an accuracy of 0.5 km/h
(i.e. the standard error). If it is known that the standard deviation of the speed distribution is
6.7 km/h (e.g. from a pilot survey), then by applying the above equation, or using
Table 2.1, it is found that:
2
§ 6.7 ·
n ¨ ¸ 180 (rounding up 179.6)
© 0.5 ¹

Table 2.1: Relative error

Relative error (%) Sample size n
20 25
15 44
12.5 64
10 100
7.5 178
5 400
2.5 1,608
2 2,500
1 10,000
Notes:
1. Relative error (%) = (standard error of mean) / (standard deviation of data)
2. n = 1 / (relative error)2.

Issues relating to sample sizes are of particular concern in the conduct of speed surveys, and
further discussion of sample size determination is given in Appendix B.

2.4 Data Integrity

The objective of data integrity is to provide accurate data in a user-friendly format on demand by a
wide range of users. The data should contain a statistically valid sample drawn from a population.
It should be recorded accurately in a consistent and convenient format and organised for easy
access, transfer and reporting. Limitations include equipment failure, non-representativeness of
data, and problems associated with processing and editing. Perfect data is impossible to obtain.
Every effort is still required to ensure that the quality of data is acceptable at the end of a chain of
activities at each stage of which compromises would have been made.

Data procedures and travel time survey management have been through the process of quality
certification to AS/NZS 9001-2000 on requirements in quality systems for design, development,
production, installation and servicing. Overseas guidelines such as the US Federal Highway
Administration’s Traffic Monitoring Guide (FHWA 2001) have also promoted good data collection
practice.

Daltrey (2002) and others noted the difficulty in achieving consistency in the data collected within a
state road authority, across state road authorities and across all states. This is partly due to
differences in road use data requirements and the methods for collection, processing and
reporting. The data quality issues that need to be addressed include:

Austroads 2009
— 17 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

ƒ accuracy
ƒ timeliness
ƒ accessibility
ƒ completeness and relevance
ƒ comparability and compatibility.

Other issues of concern are:

ƒ methodologies
ƒ reporting, publication and feedback
ƒ provision of analytical tools
ƒ limitations in equipment, resource and field options
ƒ collection, processing and distribution costs.

The availability of data from Intelligent Transport Systems (ITS) and other sources would require
better quality control at each source before all data sources could be shared, comprehended and
coordinated. For example, data from a freeway incident management system, digital speed
cameras or a signal control system are valuable sources of useful information. However, the
limited resources of the freeway and signal operational units are often employed for operational
improvements rather than analysing these data to become useful traffic statistics, e.g. checking
real-time data streams for zero data. Compatibility of these data with existing vehicle classification
systems is another issue that should be noted.

It is worth repeating that traffic phenomena often arise by chance from the combinations of many
factors, all of which have high degrees of variability. Environmental factors such as rain and
magnetic fields also affect equipment accuracy. It is therefore necessary to treat events in traffic
systems as random variables, and use statistical theory to investigate the impacts of the interacting
factors. Unusual variability within traffic statistics derived from one survey station should be
verified if possible from surrounding survey stations or previous data at the same station. Although
equipment failure accounts for some errors, other errors arise because of the occurrence of
abnormal traffic events in the connecting road network. The success of a statistical investigation
depends largely on the size and validity of the traffic survey used to collect the data.

2.5 Traffic Surveys

Traffic surveys have long been a major part of the traffic engineer’s work. Most traffic control and
design problems demand some knowledge of the operating characteristics of the traffic concerned.

The most common types of traffic surveys are overviewed in Sections 2.5.1 to 2.5.9, the details of
which can be found in the various appendices as indicated. These surveys are as follows:
ƒ traffic volume surveys (Appendix A)
ƒ speed surveys (Appendix B)
ƒ travel time, queuing and delay surveys (Appendix C)
ƒ origin-destination surveys (Appendix D)
ƒ pedestrian and cyclist surveys (Appendix E)
ƒ noise, fuel and emission surveys (Appendix F)

Austroads 2009
— 18 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

ƒ vehicle mass and dimension surveys (Appendix G)

ƒ parking surveys (Appendix H)
ƒ traffic generation surveys (Appendix I).

These surveys range from ‘point-based’ studies that involve the measurement and analysis of data
collected at a point in space (e.g. traffic volume and speed) to ‘link-based’ studies that involve
measurement of a parameter (e.g. travel time) over a road link, and to ‘area-based’ studies that
involve data collection over an area (e.g. for origin-destination studies, or trips generated by land
use developments).

Part 3 does not directly cover the various types of social science surveys used elsewhere in
transport and traffic planning, such as questionnaires and household interviews relating to travel
behaviour and choice.

The traffic studies described in Part 3 relate fundamentally to longitudinal movement in the
direction of travel. It is also possible to study lateral movements efficiently and accurately with new
technologies such as image processing, laser beams, and light tracking. As pointed out in Gunay
(2003), lateral movement studies are of more importance in countries where lane discipline is ‘less
tidy’, especially for understanding the capacity utilisation of a lane or an approach. The issue of
lateral movement is not addressed further in Part 3.

2.5.1 Traffic Volume Surveys (Appendix A)

Traffic volume data may be used for facility planning and design, to monitor trends in use, to assist
in highway classification and budgeting or to validate predictions from traffic forecasting models.
Traffic count data have traditionally been the most important observations for inferences about the
state of a traffic system. However, vehicle classification data are increasingly important in shaping
the information for all phases of the planning, design and operation of traffic networks.

The period over which a count is recorded is an important consideration in assessing the load
borne by a traffic facility. Consequently, traffic count information should always be expressed as a
flow rate, i.e. the number of traffic units per unit time. For road traffic, the units are vehicles per
hour (veh/h) or vehicles per day (veh/day), sometimes with the type of vehicle more precisely
defined, e.g. passenger cars per hour (pc/h), passenger car units per hour (pcu/h), axles or axle-
pairs (passenger car equivalents) per hour, trucks or heavy vehicles of different types per hour
(trucks per hour), or cyclists per hour. Traffic volumes also need to consider people movement
(passenger occupancy). Consideration needs to be given to high occupancy vehicles (HOV),
buses and transit lanes.

Traffic flow rates (commonly called volumes) are used to establish the following:
ƒ relative importance and role of a road in a traffic system
ƒ variations in the levels of traffic flow over time
ƒ extent of the use of a facility in terms of its capacity to carry traffic
ƒ distribution of travel demand in a network
ƒ coordination of traffic signals
ƒ estimating the loading on pavements and bridges from classified vehicle counts in the
absence of weigh-in-motion (WIM) equipment.

Austroads 2009
— 19 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Traffic count data are important in economic evaluation, environmental analysis (e.g. urban air
quality) and as an exposure measure in road crash studies. Further uses of traffic count data are
shown in Table A 1 in Appendix A.

2.5.2 Speed Surveys (Appendix B)

Speed is an important road design parameter. Knowledge of vehicle speeds in a traffic system
provides useful information about travel conditions, levels of service, and quality of traffic flow.
Data on vehicle speeds are necessary for both setting design standards, and as a measure of the
effects of changes to a traffic system such as installing a traffic control device or widening a curve.

Speed data can be used in the context of setting speed limits, establishing the need and
effectiveness of traffic control devices, examining road safety issues to identify crash
countermeasures, and for establishing trends in vehicular speeds. Speed data can be collected
directly or indirectly.

The most common form of direct measurement is by radar and lidar (an optical remote sensing
technology using laser light pulses rather than radio waves as in radar). Indirect speed
measurement commonly involves measuring travel time for vehicles between two detectors
separated by a known distance. The travel times may be measured manually or electronically.

Knowledge of speed data is used extensively in road safety programs. Spot speed data are useful
for traffic engineers in the study of driver behaviour as they provide estimates of the range of likely
vehicle speeds. The prevailing distribution of speeds at a site under different environmental
conditions can also be determined. Speed distributions provide indications of traffic conditions at
the observation site and are useful in assessing the need for appropriate traffic control devices,
speed limits or advisory speed signing. They are also useful aids in studies concerning overtaking
manoeuvres and the effects of lane widths and lateral clearances. ‘Before and after’ speed studies
can help in assessing driver responses to new warning signs, road markings, street lighting and
pavement surfaces.

The more useful applications of speed survey data are:

ƒ determining the need for traffic control devices, including speed zoning
ƒ evaluating the effectiveness of traffic improvements, in before and after studies
ƒ assisting in the design of roadways or intersections
ƒ finding relationships between speeds and crashes or between speeds and geometric
features
ƒ undertaking an economic analysis
ƒ determining the range and magnitude of speeds as a basis for formulating design standards.

2.5.3 Travel Time, Queuing and Delay Surveys (Appendix C)

Travel time has been identified by Austroads as an important system performance measure and
regular travel time surveys are now conducted by state road authorities (Austroads 1997). The
most common way of collecting travel time data is having observers recording data as they travel
over the road network. Some organisations have developed instrumented vehicles for conducting
travel time surveys and include global positioning systems (GPS). A system developed in NSW,
known as the Automatic Network Travel Time System (ANTTS) involves installing an electronic tag
on vehicles and monitoring their progress over the road network.

Austroads 2009
— 20 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The use of advanced technology and probe vehicles has become increasingly common in recent
years for the automatic collection of travel time data (Monash University 2003).

A travel time and delay study measures the average travel and running times along sections of a
route while at the same time collecting information on the location, duration and cause of delays. A
delay study measures stopping time delay at specific points such as intersections or railway level
crossings. From the measured times, average travel times and running speeds can be calculated.

Travel time and delay data are used to:

ƒ assess the quality of the traffic route
ƒ evaluate the ‘before and after’ effects of traffic improvements
ƒ undertake economic analysis
ƒ identify locations and causes of congestion
ƒ determine needs for traffic signals
ƒ develop optimum timing sequences at traffic signals.

Travel time and delay data are often sought as performance measures of some part of a transport
network. Several other performance indicators are available, and these may be more relevant for
some investigations. They include:
ƒ variability in travel times and delays
ƒ delays to pedestrians
ƒ total person delay
ƒ number of vehicular stops
ƒ fuel consumption
ƒ vehicle emissions
ƒ vehicle headways
ƒ gap acceptance.

Note that while most of the technical literature refers to ‘gap acceptance’, the procedure involved is
actually headway acceptance. Refer to Section 6 for further detail on headway and gap
acceptance, and to Part 2 (Austroads 2008a) of the guide for the underlying theory.

The specific objectives of any study will provide guidance as to the appropriate performance
indicators and the corresponding survey methods to collect the required data.

2.5.4 Origin-destination Surveys (Appendix D )

Origin-destination surveys provide valuable information on where motorists desire to travel, i.e. the
origins, destinations and travel times, as well as volumes in conjunction with traffic surveys. They
range from relatively simple studies that determine the amount of traffic bypassing a small country
town, to comprehensive transportation surveys that are used in the planning and design of
transport systems in large metropolitan areas. Data on trip patterns (trip distribution) are essential
in situations where diversion of traffic may result from a new traffic management scheme and it is
difficult to locate traffic count equipment at the point where diversion occurs, whether this be at the
local, regional or metropolitan level.

Austroads 2009
— 21 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

There are several techniques used for gathering origin-destination data. Some are used more
commonly than others, but each has its advantages and disadvantages. Origin-destination
surveys are liable to significant errors and biases and therefore a good understanding of the
techniques and their inherent problems is required to produce useful data. Origin-destination
surveys are usually labour-intensive and costly to undertake. Observations under conditions of
darkness can be difficult, and care is needed to avoid wasting resources.

There are now imaging technologies to automatically recognise numberplates in electronic tolling
systems that can be used under most light conditions with good accuracy and these can potentially
be useful for these surveys (Appendix D).

2.5.5 Pedestrian and Bicycle Surveys (Appendix E)

The provision of facilities for pedestrians and cyclists has been steadily increasing due to an
increased focus on user needs and safety. Data on some of the movements made by pedestrians
and cyclists can be collected using methods similar to those described in other sections on
collecting traffic data. The nature of pedestrian and bicycle movements, however, is not as
restricted to specific roadways as that of vehicles, hence the greater difficulty in collecting
information. Bicycles are defined as vehicles under road traffic regulations and therefore have a
right to use virtually all roads.

Studying pedestrian and bicycle movements may also be complicated by the spatial distribution of
routes they can choose. For example, they can easily reverse their direction of travel and exit a
system where they enter. The main similarities between vehicles and pedestrians or cyclists occur
when pedestrians or cyclists are constrained to a footpath, road lane or corridor, as this situation is
similar to vehicles on a road.

Any study of pedestrian or cyclist behaviour requires a clear statement of the problem to be
addressed and a statement of the objectives of the study. This statement should lead to a set of
parameters to be measured by the study. The Australian Bicycle Council (2000) recommends that
base data be collected in study areas that are consistent with the geographic areas used by the
Australian Bureau of Statistics, so as to ensure consistency with population characteristics.

The majority of data collected in pedestrian and bicycle surveys will come from sample surveys.
When deciding on the size of the sample, it is necessary to consider confidence limits, levels of
confidence and inherent variability (Section A.2). A trade-off exists between the required accuracy
of the sample, and therefore the size of the sample, and the cost of the study.

The sampling of cyclists is difficult because information on bicycle ownership is rarely available.
The concentration on particular groups such as school children or bicycle clubs will also not
provide information on all bicycle users. Interviewing in the field may provide an overall idea of
travel characteristics but survey locations need to be selected carefully and in a random manner to
ensure a broad spectrum of cyclists is interviewed.

An area of concern in the study of pedestrian and bicycle movements is the unit of measurement.
A study of pedestrian flow may use the pedestrian as the unit of measurement. If the pedestrian
units are not independent, as in the case of a parent with a child, it may be necessary to divide the
pedestrians into a number of categories. Another factor that needs to be taken into account occurs
in pedestrian congestion studies. When defining the amount of space occupied by each
pedestrian, disabled pedestrians and pedestrians with prams and shopping trolleys need to be
considered. These ‘pedestrian modules’ are larger than an average pedestrian and can affect the
final results. Pedestrian trip purposes may also be an important issue for the planning of
pedestrian facilities.

Austroads 2009
— 22 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Various ongoing household travel surveys exist, and useful data on bicycle and pedestrian trips
can be obtained from them. They include the Victorian Integrated Survey of Travel and Activity
(Victoria Department of Transport 2007) and the Sydney Household Travel Survey (NSW
Department of Planning 2006). The surveys recorded daily travel patterns, including bicycle and
walking trips, of household members in Melbourne and Sydney respectively. Other databases
such as the Bicycle Imports of the Bicycle Industries and Traders Association and the Serious
Injury Database of the Australian Transport Safety Bureau also provide useful bicycling and
pedestrian data.

When using existing information, it is necessary to consider the original purpose of the data, the
represented population (e.g. were children under ten included?), the treatment of ‘multi-mode trips’
and the sampling techniques used.

2.5.6 Noise, Fuel and Emission Surveys (Appendix F)

Fuel consumption and environmental degradation are of special importance in traffic studies and in
the evaluation of traffic plans and management schemes.

Road traffic systems can have significant impacts on:

ƒ noise and vibrations
ƒ fuel consumption and means of conserving liquid fuels
ƒ air pollution, including gases and particulates from vehicle emissions.

Pollution problems may be considered at two levels: immediate and prolonged exposure. In
extreme cases, the pollutant may be a danger to the physical health and wellbeing of the
individuals subjected to it. Excessive noise and concentrations of some air pollutants such as
carbon monoxide (CO) may inflict immediate damage. Prolonged exposure to large concentrations
of other pollutants, or smaller doses of noise and CO, may lead to harmful effects. The excessive
levels referred to above are rarely if ever solely or even largely attributable to road traffic.
However, individuals may become distressed or anxious in the presence of pollutants, at levels
well below those hazardous to health. Appendix F considers some of the survey management and
data collection techniques that the traffic engineer will encounter. Very often the traffic engineer
will not be directly involved in the actual data collection and analysis in these specialist areas. This
will be the province of another authority, and may require the expertise of another discipline, such
as mechanical engineering, environmental engineering or meteorology. Traffic engineers will be
contributing their expertise in traffic systems to a multi-disciplinary study team. They should have
some familiarity with the approaches adopted by other team members, so that their work can be
properly and successfully related to a traffic impact study.

Austroads 2009
— 23 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

2.5.7 Vehicle Mass and Dimension Surveys (Appendix G)

Information on vehicle mass and dimensions is necessary for effective management of road
transport networks. Relevant data have traditionally been obtained by diverting a sample or all
heavy vehicles into a specially equipped station in which each axle and/or the gross vehicle mass
is measured using a weighbridge or other special weighing equipment. The vehicle dimensions
can be measured manually. Vehicle mass is also regularly measured on roadside or commercial
weighbridges for regulatory and enforcement purposes. These methods, particularly the latter,
may involve bias because overloaded vehicles alerted via mobile phones or CB radio may avoid
routes on which a survey is being made. The development of weigh-in-motion (WIM) technology
has meant vehicle mass data is now more easily obtainable. The CULWAY systems described in
Appendix G (Section G.3.2), in particular, are expected to provide unbiased estimates of vehicle or
axle loads because they are installed in culverts and not readily detected by passing drivers.
Appendix A (Section A.5) describes vehicle classification systems by vehicle characteristics (such
as vehicle types, lengths or axle configurations) which are related to vehicle mass.

2.5.8 Parking Surveys (Appendix H)

Parking plays an important role in the provision of road transport services, allowing drivers to store
their vehicles at the origins and destinations of their journey. Parking can be of two main types,
on-street (kerbside, median) or off-street (residential, commercial). On-street parking directly
affects the flow of traffic. The utilisation of the kerbside lane is lower than adjacent lanes and the
full capacity of the roadway cannot be realised. Travel time and the variability of travel times will
increase. Parking bans and clearways on arterials have been introduced in most cities especially
during peak hours. Special parking permits are issued to residents to park on local streets where
off-street parking shortages exist.

Collecting and analysing data on parking characteristics forms the basis of the design process for
parking systems, and assists in managing potential effects upon traffic operations. The demand
for parking is generally related to the land use served. Valuable information may be needed on the
demand for parking, the supply of parking spaces, and the use of existing facilities. Typical data
collected include:
ƒ spatial distribution of parking demand
ƒ types and extent of parking requirements
ƒ numbers, type and location of parking facilities
ƒ parking restrictions and costs
ƒ duration and turnover of parking activities.

Depending on the type of information required, data collection can be undertaken by interview
surveys (including reply paid questionnaires) or observational surveys (including both cordon
counts and patrol surveys). Technologies are available for conducting observational surveys by
video with automatic data logging facilities, or by electronic tagging.

Part 11 of the Guide presents advice and guidance on the parking management process, in terms
of parking policy, demand and supply, and parking control and management systems.

Austroads 2009
— 24 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

2.5.9 Traffic Generation Surveys (Appendix I)

An important aspect of the planning and design of land use developments is the provision of
adequate traffic access. These developments include shopping centres, sports and recreation
centres, office blocks and multi-unit residential developments. Besides the immediate connections
of the development to the road system, traffic planners and engineers are concerned with the
effects of the new development on the traffic system in the surrounding area. A fundamental issue
is the amount of new traffic generated by a development. This traffic may place an additional load
on the traffic system, and could result in the need for upgrading parts of the road network or
provision of new traffic facilities.

Traffic generation is the measured level of traffic activity associated with a site, development or
land use. It is the amount of vehicular traffic arising from the number of person-trips associated
with a development. Traffic generation is usually measured in terms of the number of trip ends
(the total of trip production and trip attraction) at a site, per unit time. Data collection is by means
of observational or questionnaire surveys.

Part 12 of the guide is concerned with identifying and managing the impacts upon the road system
arising from land use developments. It provides guidance on the need and criteria for assessing
the traffic impacts of those developments, and a detailed procedure for identifying and assessing
the impacts, and mitigating their effects.

Austroads 2009
— 25 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

3 TRAFFIC ANALYSIS – CAPACITY AND LEVEL OF SERVICE

The concepts of capacity analysis and level of service are fundamental to the planning, design and
operation of traffic facilities.

The 2000 edition of the US Highway Capacity Manual (TRB 2000), referred to as HCM 2000 in this
guide, provides detailed information on capacity analysis and is the primary reference document on
this topic. Additional material, drawing on substantial Australian work, is well summarised in the
Monash University (2003) handbook, from which some of the following material is taken.

3.1 Types of Traffic Facilities

Traffic facilities may be classified into two broad categories:
ƒ Uninterrupted flow facilities, on which traffic flow conditions are the result of interactions
between vehicles in the traffic stream, and between vehicles and the geometric and
environmental characteristics of the road; there are no fixed elements external to the traffic
stream, such as traffic control signals, that cause interruptions to traffic flow; examples
include two-lane rural roads, multi-lane rural roads, and motorways.
ƒ Interrupted flow facilities, on which traffic flow conditions are subject to the influence of
fixed elements such as traffic signals, stop signs or other controls which cause traffic to stop
periodically, irrespective of the total amount of traffic; examples include urban streets,
unsignalised and signalised intersections, and roundabouts.

Uninterrupted flow and interrupted flow are terms that describe the type of road facility and not the
quality of traffic flow on it.

On uninterrupted flow facilities the causes of traffic congestion are basically due to the amount of
traffic using them, and are thus internal to the traffic stream itself. Capacity analysis for
uninterrupted flow facilities is presented in Section 4.

On interrupted flow facilities control devices external to the traffic stream require traffic to stop
periodically for their satisfactory operation. Capacity analysis for interrupted flow facilities is
presented in Section 5.

The capacity of an urban street is largely related to the type and density of intersections. At
unsignalised intersections, the major road traffic normally has priority over the minor road. From
that perspective, unsignalised intersections cause neither reduced capacity nor delay. When the
volumes of cross and/turning traffic at intersections with minor roads are small, capacity
considerations are usually not significant. A roundabout is a particular form of unsignalised
intersection and can give rise to quite different capacity considerations. For signalised
intersections the additional element of time allocation among the various conflicting movements
must also be considered. Capacity analysis for intersections is presented in Section 6.

3.2 Capacity, Level of Service, Degree of Saturation

3.2.1 Capacity
Capacity is defined in the HCM 2000 as the maximum hourly rate at which persons or vehicles can
reasonably be expected to traverse a point or uniform section of a lane or roadway during a given
time period under the prevailing roadway, traffic and control conditions. The concept applies
equally to motorised vehicular traffic and to bicycle and pedestrian traffic.

Austroads 2009
— 26 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The following points should be noted with respect to the above definition:
ƒ The time period used in capacity analyses should be one hour, but in practice analysis
typically focuses on a 15-minute period (specifically, the peak 15 minutes of the peak hour),
which is usually accepted as being the shortest interval during which stable flow exists. In
Australia a period of 30 minutes, or the full peak hour is often used.
ƒ The prevailing roadway, traffic and control conditions should be reasonably uniform for the
section of facility being analysed.
ƒ Roadway conditions refer to the geometric characteristics of the road, including the type of
facility and its development environment, the number of lanes in each direction, lane and
shoulder widths, lateral clearances, design speed and horizontal and vertical alignments.
ƒ Traffic conditions refer to the characteristics of the traffic stream using the road, including
vehicle type and the lane and directional distribution of the traffic.
ƒ Control conditions refer to the types and specific design of the control devices and traffic
regulations applicable to the particular section of road.

Capacity analysis typically focuses on vehicle capacity, in terms of vehicles (or passenger car
equivalents) per hour. In comparisons between different transport modes and systems designed to
increase vehicle occupancy (e.g. high-occupancy vehicle lanes) it is relevant to also consider the
number of persons per hour passing a point (Monash University 2003).

3.2.2 Level of Service (LOS)

Level of service is a qualitative measure describing operational conditions within a traffic stream,
and their perception by motorists and/or passengers. A level of service definition generally
describes these conditions in terms of factors such as speed and travel time, freedom to
manoeuvre, traffic interruptions, comfort and convenience, and safety.

A key issue is determining the level of service that is deemed acceptable, and whether that level
should be a projected level for future operation of a facility, or the level existing at the current
operation of the facility.

In general, there are six levels of service, designated A to F, with level of service A representing
the best operating condition (i.e. free flow) and level of service F the worst (i.e. forced or
breakdown flow).

The levels of service for uninterrupted flow facilities are described as follows in the HCM 2000.
The levels of service for interrupted flow facilities are described in Section 5.2.3.

Austroads 2009
— 27 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Level of service A A condition of free flow in which individual drivers are virtually
unaffected by the presence of others in the traffic stream.
Freedom to select desired speeds and to manoeuvre within the
traffic stream is extremely high, and the general level of
comfort and convenience provided is excellent.

Level of service B In the zone of stable flow where drivers still have reasonable
freedom to select their desired speed and to manoeuvre within
the traffic stream. The general level of comfort and
convenience is a little less than with level of service A.

Level of service C Also in the zone of stable flow, but most drivers are restricted
to some extent in their freedom to select their desired speed
and to manoeuvre within the traffic stream. The general level
of comfort and convenience declines noticeably at this level.

Level of service D Close to the limit of stable flow and approaching unstable flow.
All drivers are severely restricted in their freedom to select their
desired speed and to manoeuvre within the traffic stream. The
general level of comfort and convenience is poor, and small
increases in traffic flow will generally cause operational
problems.

Level of service E Traffic volumes are at or close to capacity, and there is virtually
no freedom to select desired speeds or to manoeuvre within
the traffic stream. Flow is unstable and minor disturbances
within the traffic stream will cause breakdown.

Level of service F In the zone of forced flow, where the amount of traffic
approaching the point under consideration exceeds that which
can pass it. Flow breakdown occurs, and queuing and delays
result.

The appropriate level of service for a particular jurisdiction will be determined in the context of the
policies indicating what are regarded as acceptable levels.

The level of service concept may be used as the basis of capacity and operational analysis for all
types of road facilities. While there is a range of parameters that could be used to define levels of
service for each type of facility, for practical purposes certain quantitative performance measures
have been developed for the different types of facility to assist in defining levels of service. These
are summarised in Table 3.1 (adapted from the HCM 2000).

Austroads 2009
— 28 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table 3.1: Performance measures used for defining levels of service

Element Level of service measure Performance measure used
(Note 1)
Vehicular
Interrupted flow
Urban street speed speed
Signalised intersection delay delay
Two-way-stop intersection delay delay
Roundabout delay delay
Interchange ramp terminal delay delay
Uninterrupted flow
Two-lane highway speed, percent time spent speed
following
Multi-lane highway density speed
Freeway
Basic segment density speed
Ramp merge or diverge density speed
Weaving speed speed
Other road users
Public transport (Note 2) speed
Pedestrians space, delay speed, delay
Cyclists event, delay speed, delay
Notes:
1. Service measure for a given facility is the primary performance measure and determines the level of service.
2. Several measures capture the multidimensional nature of transit performance when defining level of service (HCM 2000, Chapter 27) (TRB 2000).
Source: adapted from Exhibit 3-1 HCM 2000 (TRB 2000).

3.2.3 Service Flow Rate

Service flow rates indicate the maximum hourly rate at which persons or vehicles can reasonably
be expected to traverse a point under the prevailing roadway, traffic and control conditions while
maintaining a designated level of service. They indicate the vehicle or person capacity for each
level of service and are used to determine the level of service corresponding to actual traffic
volumes. As with capacity, the service flow rate is generally taken over a 15-minute time period.

Each type of facility has five service flow rates, one for each of the levels of service A to E, as
illustrated in Figure 3.1. At level of service F flow breakdown occurs, and a meaningful service
flow rate cannot be specified.

Austroads 2009
— 29 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

120
Free flow speed, FFS = 110 km/h 1450
Average passenger car speed (km/h)

100

80
LOS A B C D E

60

40

20

0
0 400 800 1200 1600 2000 2400
Flow rate (pc/h/ln)

Regime 1 (undersaturated) Regime 2 (queue discharge) Regime 3 (oversaturated)

Source: Adapted from Transportation Research Board( 2000), exhibits 13-4 and 23-3.

Figure 3.1: Levels of service and service flow rates

Figure 3.1 depicts a freeway operation curve for a free speed of 110 km/h. Other facilities and
freeways with different operating speeds will have different curves. The different levels of service
are described by the density (pc/km/ln).

At each level of service the service flow rate is defined as the maximum for that level. Service flow
rates are discrete values, whereas the level of service represents a range of conditions. Service
flow rates therefore effectively define the flow boundaries between the levels of service.

3.2.4 Degree of Saturation

The concept of degree of saturation is used in the capacity and operational analysis of
intersections.

The degree of saturation of a signalised intersection approach may be defined as the ratio of the
arrival flow (demand) to the capacity of the approach during the same period.

The degree of saturation of an intersection approach ranges from close to zero for very low traffic
flows up to one for saturated flow or capacity. A degree of saturation greater than one indicates
oversaturated conditions in which long queues of vehicles build up on the critical approaches. In
general the lower the degree of saturation the better the quality of traffic service.

Austroads 2009
— 30 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

3.3 Factors Affecting Capacity, Level of Service, Degree of Saturation

3.3.1 Ideal Conditions
In the analysis of capacity, level of service or degree of saturation, the starting point is often to
select values that are applicable to ideal conditions, and then to apply correction or adjustment
factors that reflect the actual roadway, traffic and control conditions.

In general, an ideal condition is one for which further improvements will not result in any increase
in capacity or level of service or decrease in the degree of saturation. More specific details of ideal
conditions are given in later sections.

3.3.2 Roadway Conditions

Roadway conditions that affect capacity, level of service and degree of saturation include:
ƒ type of facility and its development environment
ƒ traffic lane widths
ƒ shoulder widths and/or lateral clearances
ƒ design speed
ƒ horizontal and vertical alignment.

Correction factors for these conditions are given in later sections.

Note that these are based on US data but generally apply to Australian conditions. Judgement
may need to be exercised in their use for Australian conditions.

3.3.3 Terrain Conditions

For capacity and related analyses, the general terrain of a roadway is classified into three
categories, namely:
Level terrain – any combination of grades and horizontal and vertical alignment permitting heavy
vehicles to maintain about the same speed as passenger cars.

Rolling terrain – any combination of grades and horizontal and vertical alignment causing heavy
vehicles to reduce their speeds substantially below those of passenger cars, but not causing them
to operate at crawl speeds for any significant length of time.

Mountainous terrain – any combination of grades and horizontal and vertical alignment causing
heavy vehicles to operate at crawl speeds for significant distances and/or at frequent intervals.

It is also necessary to consider pedestrian and cyclist’s needs as their presence in the traffic mix
for the given terrain is likely to affect the capacity.

3.3.4 Traffic Composition

For capacity and related analyses, road traffic is typically classified into three categories, namely:
Passenger cars – vehicles registered as passenger cars, plus other light vehicles and light
delivery vans having no more than four single tyres.

Trucks – vehicles having more than four single tyres, and involved primarily in the transport of
goods or services.

Austroads 2009
— 31 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Buses – vehicles having more than four single tyres, and involved primarily in the transport of
people. For capacity analysis, passenger cars towing caravans, boats and other similar recreation
equipment should be included in this category.

This classification of vehicles into three categories is generally consistent with current classification
counting, although it is noted that modern loop detectors classify vehicles into four categories. The
Austroads vehicle classification categories are described in Appendix A.

3.3.5 Pedestrians and Cyclists

Pedestrians form the largest single road-user group. Most individual trips, whatever the primary
mode used, begin and/or finish with a walk section. Walking is a fundamental component of all
travel. Pedestrian safety and mobility must be explicit factors in the planning, design,
implementation and maintenance of road facilities.

Initiatives are being implemented at the planning and policy levels, which give explicit considera-
tion to the walking mode and to pedestrian safety and amenity.

The framework for pedestrian planning and design lies within the practice of traffic engineering.
Pedestrians may be considered as being similar to vehicles, operating on a transport network
consisting of footways, stairs, travelators etc. Network capacity can be defined, demand measured
or predicted, operational levels calculated, and areas of congestion and hazard identified. To
achieve maximum safety, the pedestrian network itself should be separate from, but integrated
with, the main road and public transport system. This will necessitate regular crossings in order to
sustain the coverage and continuity of the network for walking.

Cyclists also form a significant road-user group with specific needs. Solutions addressing their
capacity needs can significantly affect road capacity, level of service and degree of saturation.

Because of increasing environmental, physical and financial constraints, attention must be given to
issues of traffic calming and demand management in order to encourage pedestrian, cycling and
higher vehicle occupancy modes of transport. Planning, design and management of traffic facilities
must cater for pedestrian or cyclist traffic along or across roads as this can affect road capacity.

Parts 5 and 6 of the Guide provide information on requirements and provisions for pedestrians and
cyclists on road sections and at intersections.

3.3.6 Driver Population

The traffic stream characteristics associated with the information provided in certain sections of this
Part, are representative of regular weekday commuter drivers and other regular users of a facility.
In situations where weekend or recreation, or even mid-day (off-peak) drivers are a significant
proportion of the traffic stream, the capacity and/or level of service may be reduced.

An adjustment factor is provided where relevant, but the range of values given is relatively wide
and engineering judgement and/or local data should be used in selecting the appropriate value.

3.3.7 Control Conditions

The control of the time available for individual traffic movements is a critical factor affecting the
capacity, level of service and/or the degree of saturation of interrupted flow facilities. Typical forms
of control include traffic signals, stop and give-way signs, turn restrictions, lane-use controls and
parking restrictions. The effects of various forms of control are described in later sections.

Austroads 2009
— 32 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

4 UNINTERRUPTED FLOW FACILITIES

4.1 Single Lane Flow
In certain situations traffic flow may be constrained to a single traffic lane without overtaking,
typical examples being a single lane provided in one direction on an undivided urban road with
reversible lane flow, in a tunnel or at a construction or maintenance site. In these types of
situations, depending on the length of the single lane restriction and on the volume of traffic, the
speed of all vehicles will tend to the speed of the slowest vehicle, and traffic will operate in
accordance with car-following mechanisms.

Several of the earliest studies of traffic capacity examined single lane traffic flow. A linear
relationship between the average speed and the density of vehicles was assumed, and the
capacity was related to the free speed (space mean speed at low flows) and the jam density (the
maximum density for stopped traffic) as follows:

C = k j Vf / 4

where

C = capacity (vehicles per hour)

Vf = free speed (km per hour)

kj = jam density (vehicles per km)

In some cases, the speed-spacing relationship was based on limited field observations, but in most
cases it was determined by using factors such as driver reaction time, coefficients of friction and
braking distances. The capacity of a single lane so determined varied from 1000 to 4800 vehicles
per hour, but with most results in the range 1500 to 2400 vehicles per hour.

4.1.1 Capacity
If there is a small bunch of vehicles in a single traffic lane all moving at relatively high speed, the
average headway between them may be as small as about 1.5 seconds, giving a corresponding
potential capacity of 2400 vehicles per hour. However, on a long length of road without overtaking,
these small headways usually cannot be sustained over a long period of time. Also, it will be noted
that lane volumes of up to 2200 vehicles per hour have been recorded on urban freeways in peak
period conditions during which there is virtually no lane changing.

If single lane conditions without overtaking are retained over a significant length of road, then as
traffic volumes increase, a long unbroken line of vehicles (or a long bunch) develops and the
speeds of all vehicles tend to that of the slowest vehicle, and stop-start conditions may develop.
Once this occurs, the capacity of a single lane is reduced to that equivalent to a headway of about
two seconds, i.e. to a capacity of about 1800 passenger cars per hour. In general, this figure can
be regarded as the capacity of a single lane without overtaking.

Factors affecting capacity

The capacity of a single traffic lane will be affected by factors such as the pavement width and
restricted lateral clearances, the presence of heavy vehicles and grades.

Austroads 2009
— 33 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The capacity of a significant length of a single traffic lane for the prevailing roadway and traffic
conditions can be calculated by using the following equation:
C = 1800 fW fHV

where

C = capacity in vehicles per hour under prevailing roadway and traffic conditions

fW = adjustment factor for narrow lanes and lateral clearances, obtained from Table 4.1

fHV = adjustment factor for heavy vehicles

= 1/[1+ PHV (EHV – 1)]

PHV = the proportion of heavy vehicles in the traffic stream, expressed as a decimal

EHV = the average passenger car equivalents for heavy vehicles obtained from Table 4.2.

Factual data on adjustment factors that should be applied to the capacity of 1800 passenger cars
per hour for a single lane is very limited. In the absence of specific measurements at comparable
sites, the adjustment factors given in Table 4.1 and Table 4.2 may be used as a reasonable
approximation, although care needs to be exercised in their use.

A worked example is given in section C1.1.1 in Commentary 1.

Where a roadway with two or more lanes in one direction is restricted over a short length (say less
than about 100 m) to a single lane flow and provided that the upstream merge is adequately
designed, the capacity may be calculated using a base flow rate of 2400 veh/h rather than
1800 veh/h as in the above equation.

Table 4.1: Adjustment factors for lane width and lateral clearance
Lateral clearances on each Lane Width
side (m)
3.7 m 3.2 m 2.7 m
2 1.00 0.90 0.70
1 0.90 0.80 0.63
0 0.65 0.60 0.50

Table 4.2: Average passenger car equivalents for heavy vehicles on grades with single lane flow
Grade Passenger car equivalents
Level 2.00
Moderate 4.0
Long sustained 8.0

Austroads 2009
— 34 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

4.2 Two-lane Two-way Roads

Two-lane rural roads have one lane for use by traffic travelling in each direction. Overtaking of
slower vehicles requires the use of the opposing traffic lane when sight distance and gaps in the
opposing traffic stream permit. At low traffic volumes and under ideal conditions, drivers are able
to travel at their desired speed without interference. As traffic volumes increase, and as roadway,
terrain and traffic conditions become less than ideal, drivers are increasingly affected by the
presence of other vehicles on the road, and bunches form in the traffic stream. Vehicles in these
bunches are subjected to delay because of the inability to overtake slower moving vehicles.

The analysis process outlined in Part 3 is based on the HCM 2000 (TRB 2000). That manual
should be consulted for further explanation of the concepts and a detailed description of the
process. Software that provides solutions in agreement with the HCM processes is available, from
McTrans (2009) for example, for all HCM procedures.

The HCM 2000 distinguishes between two categories of two-lane highways, as follows:
ƒ Class I –These are two-lane highways on which motorists expect to travel at relatively high
speeds. Two-lane highways that are major intercity routes, primary arterials connecting
major traffic generators, daily commuter routes, or primary links in state or national highway
networks generally are assigned to Class I. Class I facilities most often serve long-distance
trips or provide connecting links between facilities that serve long-distance trips.
ƒ Class II –These are two-lane highways on which motorists do not necessarily expect to travel
at high speeds. Two-lane highways that function as access routes to Class I facilities, serve
as scenic or recreational routes that are not primary arterials, or pass through rugged terrain
generally are assigned to Class II. Class II facilities most often serve relatively short trips,
the beginning and ending portions of longer trips, or trips for which sightseeing plays a
significant role.

The two classes of road perform markedly different functions and are not totally dependent on the
highway’s role in the hierarchy. A highway between major urban centres through mountainous
terrain might be classified as Class II instead of Class I if it is established that drivers appreciate
that higher speeds are inappropriate.

The HCM 2000 provides a method to analyse the performance of a two-way segment: as
directional segments, with each direction of travel considered separately; for sections with steep
grades and for segments with passing lanes. The two-way analysis assumes roughly similar flows
in each direction.

The performance of a particular road section is calculated from the predicted performance of base
or ideal conditions, which include:
ƒ lane widths greater than or equal to 3.6 m
ƒ clear shoulders wider than or equal to 1.8 m
ƒ no no-overtaking zones
ƒ all passenger cars
ƒ no impediments to through traffic, such as traffic control or turning vehicles
ƒ level terrain
ƒ a 50/50 directional split of traffic for the analysis of two-way flow (i.e. both directions at once).

Austroads 2009
— 35 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The process estimates the free flow speed (FFS) based on the base free flow speed (from the
base conditions) and termed the BFFS. The FFS is based on the lane widths, shoulder widths and
the average number of access points per kilometre. Reduce the shoulder width or the lane width
and free speed is reduced. Increase the frequency of access points and the free speed is reduced.
The actual travel speed is a function of the free flow speed and the passenger car equivalent flow
rate which is based on the proportion of heavy vehicles, the grade (or terrain type) and broad flow
rates. The effect of an additional vehicle at low flows differs from the effect of an additional vehicle
at higher flows.

The percent-time-spent-following is a measure of the level of opportunities to overtake. The

per cent-time-spent-following approaches 100 asymptotically:
ƒ as the traffic demand increases
ƒ as the percentage of no-overtaking zones increases
ƒ as the directional split moves closer to 90/10.

The HCM 2000 considers a no-overtaking zone to exist when the sight distance is below 300 m.
For the analysis of an existing roadway, a no-overtaking zone could be defined by a barrier line.
This would make it easier for practitioners to measure the zone in the field, or from plans, than to
determine where the sight distance is below 300 m.

Figure 4.1 shows the expected average travel speed and the percent-time-spent-following for ideal
or base conditions.

Austroads 2009
— 36 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Source: From Exhibit 12-7 in the HCM 2000 (TRB 2000).

Figure 4.1: Speed-flow and percent-time-spent-following relationships for directional segments with base
conditions.

The level of service (LOS) criteria used for the different highway classes differ. The HCM 2000
notes that:
ƒ The LOS for Class I highways on which efficient mobility is paramount is defined in terms of
both percent-time-spent-following and average travel speed
ƒ On Class II highways, mobility is less critical, and LOS is defined only in terms of percent-
time-spent-following
ƒ Drivers generally tolerate higher levels of percent-time-spent-following on a Class II facility
than on a Class I facility, because Class II highways usually serve shorter trips and different
trip purposes.

Figure 4.2 shows the LOS criteria for Class I highways. The criteria for Class II highways are
shown in Table 4.3. The level of service can be evaluated from field measurements on an existing
road. The percent of vehicles within three seconds, collected at fixed regular intervals and
averaged for the section in question, is a good substitute for the percent-time-spent-following.

Austroads 2009
— 37 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Source: Exhibit 20-3 in the HCM 2000 (TRB 2000).

Figure 4.2: LOS criteria for two-lane highways in Class I

Table 4.3: LOS criteria for two-lane highways in Class II

LOS Percent time-spent-following
A ” 40
B > 40–55
C > 55–70
D > 70–85
E > 85
Note: LOS F applies whenever the flow rate exceeds the segment capacity.
Source: Exhibit 20-4 in the HCM 2000 (TRB 2000).

The HCM 2000 indicates that the capacity of a two-lane highway is 1700 pc/h for each direction of
travel and is nearly independent of the directional distribution of traffic. For extended lengths of
two-lane highway, the capacity will not exceed 3200 pc/h for both directions of travel combined.

The HCM 2000 describes the conditions for the different levels of performance of two-lane
highways in the following terms:
ƒ LOS A describes the highest quality of traffic service, when motorists are able to travel at
their desired speed. Without strict enforcement, this highest quality would result in average
speeds of 90 km/h or more on two-lane highways in Class I. The overtaking frequency
required to maintain these speeds has not reached a demanding level, so that overtaking
demand is well below overtaking capacity, and bunches of three or more vehicles are rare.
Drivers are delayed no more than 35% of their travel time by slow-moving vehicles. A
maximum flow rate of 490 pc/h total in both directions may be achieved with base conditions.
On Class II highways, speeds may fall below 90 km/h, but motorists will not be delayed in
bunches for more than 40% of their travel time.

Austroads 2009
— 38 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

ƒ LOS B characterises traffic flow with speeds of 80 km/h or slightly higher on level terrain
Class I highways. The demand for overtaking to maintain desired speeds becomes
significant and approximates the overtaking capacity at the lower boundary of LOS B.
Drivers are delayed in bunches up to 50% of the time. Service flow rates of 780 pc/h total in
both directions can be achieved under base conditions. Above this flow rate, the number of
bunches increases dramatically. On Class II highways, speeds may fall below 80 km/h, but
motorists will not be delayed in bunches for more than 55% of their travel time.
ƒ LOS C describes further increases in flow, resulting in noticeable increases in bunch
formation, bunch size, and frequency of overtaking impediments. The average speed still
exceeds 70 km/h on level-terrain Class I highways, even though unrestricted overtaking
demand exceeds overtaking capacity. At higher volumes the chaining of bunches and
significant reductions in overtaking capacity occur. Although traffic flow is stable, it is
susceptible to congestion due to turning traffic and slow-moving vehicles. Percent-time-
spent- following may reach 65%. A service flow rate of up to 1190 pc/h total in both
directions can be accommodated under base conditions. On Class II highways, speeds may
fall below 70 km/h, but motorists will not be delayed in bunches for more than 70% of their
travel time.
ƒ LOS D describes unstable traffic flow. The two opposing traffic streams begin to operate
separately at higher volume levels, as overtaking becomes extremely difficult. Overtaking
demand is high, but overtaking capacity approaches zero. Mean bunch sizes of five to
10 vehicles are common, although speeds of 60 km/h still can be maintained under base
conditions on Class I highways. The proportion of no-overtaking zones along the roadway
section usually has little influence on overtaking. Turning vehicles and roadside distractions
cause major shock waves in the traffic stream. Motorists are delayed in bunches for up to
80% of their travel time. Maximum service flow rates of 1830 pc/h total in both directions can
be maintained under base conditions. On Class II highways, speeds may fall below 60 km/h,
but in no case will motorists be delayed in bunches for more than 85% of their travel time.
ƒ At LOS E, traffic flow conditions have a percent-time-spent-following up to 100% on Class I
and Class II highways. Even under base conditions, speeds may drop below 60 km/h.
Average travel speeds on highways with less than base conditions will be slower, even down
to 40 km/h on sustained upgrades. Overtaking is virtually impossible at LOS E, and
bunching becomes intense, as slower vehicles or other interruptions are encountered. The
highest volume attainable under LOS E defines the capacity of the highway, generally
3,200 pc/h total in both directions. Operating conditions at capacity are unstable and difficult
to predict. Traffic operations seldom reach near capacity on rural highways, primarily
because of a lack of demand.
ƒ LOS F represents heavily congested flow with traffic demand exceeding capacity. Volumes
are lower than capacity and speeds are highly variable.

4.3 Multi-lane Roads

Multi-lane roads have two or more lanes for use by traffic in each direction. They may be classified
as:
ƒ Divided – when opposing directions of traffic are physically separated by a median.
ƒ Undivided – when opposing directions of traffic are not physically separated.

Multi-lane roads have at-grade intersections including signalised intersections; this attribute
distinguishes them from freeways. Multi-lane roads and urban roads have different traffic signal
densities. Both HCM 2000 procedures should be reviewed when analysing urban roads in
suburban areas.

Austroads 2009
— 39 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The recommended analysis procedure is based on the HCM 2000 Chapters 12 and 21. The
analysis is extended for road sections with higher grades extending over longer distances. The
analysis involves a comparison with base conditions. The base represents the highest operating
level of a multi-lane highway with the following characteristics. Software, from McTRANS for
example, provides solutions in agreement with these HCM 2000 procedures. The characteristics
are:
ƒ 3.6 m minimum lane widths
ƒ 3.6 m minimum total lateral clearance in the direction of travel—this represents the total
lateral clearances from the edge of the travel lanes to obstructions along the edge of the road
and in the median (in computations, lateral clearances greater than 1.8 m are considered in
computations to be equal to 1.8 m)
ƒ only passenger cars in the traffic stream
ƒ no direct access points along the roadway
ƒ a divided highway
ƒ free flow speed equal to 100 km/h.

The free flow speed (FFS) is estimated from the free flow speed for the base conditions. A value
between 8 and 10 km/h above the speed limit could be used if there is no other evidence from
similar roads. The free flow speed is reduced if lane width is reduced, if the total lateral is reduced,
and as the frequency of access points is increased. The free speed on an undivided road is
2.6 km/h lower than on a divided road with the same characteristics.

The design flow rate is calculated for a peak 15 minutes and accounts for the percentage and type
of heavy vehicles, whether drivers are commuters or not, and on the variation of the traffic within a
peak hour. The HCM 2000 uses a peak hour factor (PHF) term in almost all procedures. The PHF
is the ratio between the hourly flow and the peak 15-minute flow rate.

When using the HCM 2000 procedure, there may be a need to consider the vehicle equivalency of
Australian trucks. There is little evidence to indicate appropriate values to be used, but a
sensitivity analysis could be used to establish if higher vehicle equivalency values are appropriate.

The HCM 2000 reports that ‘capacities for recreational traffic can be up to 20% lower than for
commuter traffic on the same highway; however, the (free flow speed) FFS does not seem to be
similarly affected. If this possible effect of driver population is taken into account, locally derived
data should be obtained and used carefully, according to the methodology for multilane highways.’

The average trip speed is then estimated using the free flow speed and the design flow rate.
Figure 4.3 shows the relationship between these parameters although the HCM 2000 provides
equations for a computerised solution. A vehicle density, given by the design flow rate divided by
the average trip speed, is the prime measure used in the evaluation of the LOS. Other
characteristics for different LOS are shown in Table 4.4.

Austroads 2009
— 40 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Note: parts of the curves are obscured by grid lines.

Source: Exhibit 21-3 in the HCM 2000 (TRB 2000).

Figure 4.3: Speed-flow curves with LOS criteria for multi-lane roads.

Table 4.4: Level of service criteria for multi-lane highways

Free flow speed Criteria A B C D E
100 km/h Maximum density (pc/km/ln) 7 11 16 22 25
Average speed (km/h) 100.0 100.0 98.4 91.5 88.0
Maximum volume to capacity ratio (v/c) 0.32 0.50 0.72 0.92 1.00
Maximum service flow rate (pc/h/ln) 700 1100 1575 2015 2200
90 km/h Maximum density (pc/km/ln) 7 11 16 22 26
Average speed (km/h) 90.0 90.0 89.8 84.7 80.8
Maximum volume to capacity ratio (v/c) 0.30 0.47 0.68 0.89 1.00
Maximum service flow rate (pc/h/ln) 630 990 1435 1860 2100
80 km/h Maximum density (pc/km/ln) 7 11 16 22 27
Average speed (km/h) 80.0 80.0 80.0 77.6 74.1
Maximum volume to capacity ratio (v/c) 0.28 0.44 0.64 0.85 1.00
Maximum service flow rate (pc/h/ln) 560 880 1280 1705 2000
70 km/h Maximum density (pc/km/ln) 7 11 16 22 28
Average speed (km/h) 70.0 70.0 70.0 69.6 67.9
Maximum volume to capacity ratio (v/c) 0.26 0.41 0.59 0.81 1.00
Maximum service flow rate (pc/h/ln) 490 770 1120 1530 1900
Source: Exhibit 21-2 in the HCM 2000 (TRB 2000).

Austroads 2009
— 41 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The HCM 2000 provides a description of the different LOS as follows:

ƒ LOS A describes completely free flow conditions. The operation of vehicles is virtually
unaffected by the presence of other vehicles, and operations are constrained only by the
geometric features of the highway and by driver preferences. The ability to manoeuvre
within the traffic stream is good. Minor disruptions to flow are easily absorbed without a
change in travel speed.
ƒ LOS B also indicates free flow, although the presence of other vehicles becomes noticeable.
Average travel speeds are the same as in LOS A, but drivers have slightly less freedom to
manoeuvre. Minor disruptions are still easily absorbed, although local deterioration in LOS
will be more obvious.
ƒ In LOS C, the influence of traffic density on operations becomes marked. The ability to
manoeuvre within the traffic stream is clearly affected by other vehicles. On multi-lane
highways with an FFS above 80 km/h, the travel speeds reduce somewhat. Minor
disruptions can cause serious local deterioration in service, and queues will form behind any
significant traffic disruption.
ƒ At LOS D, the ability to manoeuvre is severely restricted due to traffic congestion. Travel
speed is reduced by the increasing volume. Only minor disruptions can be absorbed without
extensive queues forming and the service deteriorating.
ƒ LOS E represents operations at or near capacity, an unstable level. The densities vary,
depending on the FFS. Vehicles are operating with the minimum spacing for maintaining
uniform flow. Disruptions cannot be dissipated readily, often causing queues to form and
service to deteriorate to LOS F. For the majority of multi-lane highways with FFS between
70 and 100 km/h, passenger-car mean speeds at capacity range from 68 to 88 km/h but are
highly variable and unpredictable.
ƒ LOS F represents forced or breakdown flow. It occurs either when vehicles arrive at a rate
greater than the rate at which they are discharged or when the forecast demand exceeds the
computed capacity of a planned facility. Although operations at these points and on sections
immediately downstream appear to be at capacity, queues form behind these breakdowns.
Operations within queues are highly unstable, with vehicles experiencing brief periods of
movement followed by stoppages. Travel speeds within queues are generally less than
48 km/h. Note that the term LOS F may be used to characterise both the point of the
breakdown and the operating condition within the queue. Although the point of breakdown
causes the queue to form, operations within the queue generally are not related to
deficiencies along the highway segment.

4.4 Freeways
A freeway is a divided road with two or more lanes for traffic travelling in each direction, with no at-
grade intersections and with full control of access from abutting property.

Freeways have three types of elements, namely:

ƒ Basic freeway segments – which are sections of freeway that are outside the influence of any
ramp or weaving area.
ƒ Ramps and ramp terminals – which provide access to and from the freeway.
ƒ Weaving areas – which are sections of freeway on which two or more vehicle flows must
cross, such as when a merge area is closely followed by a diverge area.

Austroads 2009
— 42 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

These elements can be combined to evaluate a freeway facility. The HCM 2000 provides a
method to evaluate the performance of a freeway that has congested elements for more than one
15-minute time period.

The recommended analysis procedure is the current procedure in the HCM 2000, and the default
values documented in that manual should be used. Research has shown that the speed-flow
relationship for freeways in the USA, Canada, Germany, the UK and Australia has similar values
for comparable freeway geometries and free speeds.

Guidance on road space allocation and lane management for freeways is provided in Part 5 of the
guide (Austroads 2008b).

4.4.1 Basic Freeway Segments

As for other uninterrupted facilities, the evaluation of a basic freeway segment relies on an
evaluation of the average travel speed. The first consideration is the free flow speed, which could
be measured in the field, or estimated using relationships in the HCM 2000 (Chapter 23).

Basic freeway segments can be analysed for sections with a specific grade. This process is
typically applied to sections with steeper grades over extended lengths. The process is not
discussed here and the HCM 2000 should be consulted for further information.

The free flow speed is a function of the base free flow speed and a number of adjustment
parameters. The base conditions are:
ƒ minimum lane widths of 3.6 m
ƒ minimum left-shoulder lateral clearance between the edge of the travel lane and the nearest
obstacle or object that influences traffic behaviour of 1.8 m
ƒ minimum median lateral clearance of 0.6 m
ƒ traffic stream composed entirely of passenger cars
ƒ five or more lanes for one direction (in urban areas only)
ƒ interchange spacing at 3 km or greater
ƒ level terrain, with grades no greater than 2%
ƒ a driver population composed principally of regular users of the facility
ƒ under these base conditions the free speed is 120 km/h in rural areas and 110 km/h in urban
areas.

The free flow speed is reduced when the lane width is reduced, when the left shoulder width is
reduced, when the number of lanes is reduced and when the interchange density is increased.

The design flow rate is a function of the type and proportion of heavy vehicles and whether or not
drivers are commuters. The effect of long combination vehicles such as road trains may need to
be considered in this approach, but appropriate values for the vehicle equivalencies are not readily
available.

The level of service is calculated from the vehicle density, being the design flow rate divided by the
average passenger-car speed. Figure 4.4 shows the speed-flow curve for basic freeway
segments. Results can be measured or estimated and plotted directly onto this curve to predict the
level of service. Table 4.5 is a list of the level of service criteria with density being the prime term.

Austroads 2009
— 43 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Source: Exhibit 23-3 in the HCM 2000 (TRB 2000).

Figure 4.4: Speed-flow relationship for basic freeway segments

Table 4.5: LOS criteria for basic freeway segments

Criteria LOS
A B C D E
FFS= 120 km/h
Maximum density (pc/km/ln) 7 11 16 22 28
Minimum speed (km/h) 120.0 120.0 114.6 99.6 85.7
Maximum (v/c) 0.35 0.55 0.77 0.92 1.00
Maximum service flow rate (pc/h/ln) 840 1320 1840 2200 2400
FFS= 110 km/h
Maximum density (pc/km/ln) 7 11 16 22 28
Minimum speed (km/h) 110.0 110.0 108.5 97.2 83.9
Maximum (v/c) 0.33 0.51 0.74 0.91 1.00
Maximum service flow rate (pc/h/ln) 770 1210 1740 2135 2350
FFS= 100 km/h
Maximum density (pc/km/ln) 7 11 16 22 28
Minimum speed (km/h) 100.0 100.0 100.0 93.8 82.1
Maximum (v/c) 0.30 0.48 0.70 0.90 1.00
Maximum service flow rate (pc/h/ln) 700 1100 1600 2065 2300
FFS= 90 km/h
Maximum density (pc/km/ln) 7 11 16 22 28
Minimum speed (km/h) 90.0 90.0 90.0 89.1 80.4
Maximum (v/c) 0.28 0.44 0.64 0.87 1.00
Maximum service flow rate (pc/h/ln) 630 990 1440 1955 2250
Source: Exhibit 23-2 in the HCM 2000 (TRB 2000).

Austroads 2009
— 44 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

On basic freeway segments, the level of service definitions from the HCM 2000 are:
ƒ LOS A describes free flow operations. Free flow speeds prevail. Vehicles are almost
completely unimpeded in their ability to manoeuvre within the traffic stream. The effects of
incidents or point breakdowns are easily absorbed at this level.
ƒ LOS B represents reasonably free flow, and free flow speeds are maintained. The ability to
manoeuvre within the traffic stream is only slightly restricted, and the general level of
physical and psychological comfort provided to drivers is still high. The effects of minor
incidents and point breakdowns are still easily absorbed.
ƒ LOS C provides for flow with speeds at or near the FFS of the freeway. Freedom to
manoeuvre within the traffic stream is noticeably restricted, and lane changes require more
care and vigilance on the part of the driver. Minor incidents may still be absorbed, but the
local deterioration in service will be substantial. Queues may be expected to form behind
any significant blockage.
ƒ LOS D is the level at which speeds begin to decline slightly with increasing flows and density
begins to increase somewhat more quickly. Freedom to manoeuvre within the traffic stream
is more noticeably limited, and the driver experiences reduced physical and psychological
comfort levels. Even minor incidents can be expected to create queuing, because the traffic
stream has little space to absorb disruptions.
ƒ At its highest density value, LOS E describes operation at capacity. Operations at this level
are volatile, because there are virtually no usable gaps in the traffic stream. Vehicles are
closely spaced leaving little room to manoeuvre within the traffic stream at speeds that still
exceed 80 km/h. Any disruption of the traffic stream, such as vehicles entering from a ramp
or a vehicle changing lanes, can establish a disruption wave that propagates throughout the
upstream traffic flow. At capacity, the traffic stream has no ability to dissipate even the most
minor disruption, and any incident can be expected to produce a serious breakdown with
extensive queuing. The ability to manoeuvre within the traffic stream is extremely limited,
and the level of physical and psychological comfort afforded the driver is poor.
ƒ LOS F describes breakdowns in vehicular flow. Such conditions generally exist within
queues forming behind breakdown points. Breakdowns occur for a number of reasons:
— Traffic incidents can cause a temporary reduction in the capacity of a short segment,
so that the number of vehicles arriving at the point is greater than the number of
vehicles that can move through it.
— Points of recurring congestion, such as merge or weaving segments and lane drops,
experience very high demand in which the number of vehicles arriving is greater than
the number of vehicles discharged.
— In forecasting situations, the projected peak-hour (or other) flow rate can exceed the
estimated capacity of the location.

Note that in all cases, breakdown occurs when the ratio of existing demand to actual capacity, or of
forecast demand to estimated capacity, exceeds 1.00. Operations immediately downstream of
such a point, however, are generally at or near capacity, and downstream operations improve
(assuming that there are no additional downstream bottlenecks) as discharging vehicles move
away from the bottleneck.

LOS F operations within a queue are the result of a breakdown or bottleneck at a downstream
point. LOS F is also used to describe conditions at the point of the breakdown or bottleneck and
the queue discharge flow that occurs at speeds lower than the lowest speed for LOS E, as well as
the operations within the queue that forms upstream. Whenever LOS F conditions exist, they have
the potential to extend upstream for significant distances.

Austroads 2009
— 45 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

4.4.2 Ramps and Ramp Junctions

The recommended procedure for the evaluation of ramps is given in the HCM 2000 (Chapter 25).
The evaluation of ramps considers vehicle interactions that occur within 450 m of the ramp, shown
as the influence area in Figure 4.5. This figure also shows important traffic flow variables used in
the analysis. The traffic flows are recorded in passenger car units per hour and are a function of
the type and proportion of heavy vehicles and the attributes of the driving population.

The terminology for the peak 15-minute traffic flows is as follows:

VF is the total traffic across all lanes of the carriageway entering the ramp area
VR is the ramp traffic
V12 is the traffic in the two kerb-side lanes of the carriageway entering the ramp area
VR12 is the traffic in the two kerb lanes and on the ramp
VFO is the total traffic across all lanes of the carriageway exiting the ramp area.

VFO VF
Influence area
V12
VR12

LA VR

450 m

VFO VF
Influence area
V12

LD
VR
450 m

Figure 4.5: Influence area at ramps

The process to evaluate ramps is to predict the traffic in the two kerb-side lanes. (If there are only
two lanes on the carriageway, then this flow is equal to the total flow.) The proportion of traffic in
the two kerb-side lanes depends on the proximity and type on the previous upstream ramp and the
proximity and type on the next downstream ramp.

The density of the merge area is then calculated using a linear relationship with the peak 15-
minute ramp flow, VR, the flow in the two kerb-side lanes, V12, and the acceleration lane length LA.
The HCM 2000 defines the acceleration lane length measured from ‘the point at which the right
edge of the ramp lane or lanes and the left edge of the freeway lanes converge to the end of the
taper segment connecting the ramp to the freeway. The point of convergence can be defined by
painted markings or physical barriers or by some combination of the two. Note that both taper area
and parallel ramps are measured in the same way.

Austroads 2009
— 46 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The calculated densities for on-ramp traffic onto a three-lane freeway carriageway are shown in
Figure 4.6. The on-ramp traffic flow is between 5 to 15% of the freeway carriageway traffic VF.
The densities are reasonable independent of the ramp flows when plotted against the total
downstream traffic (freeway and ramp flows). This plot is for a 100 m acceleration lane. Increase
the acceleration lane by 100 m and the density will decrease by 1.3 pc/km/ln.

25
LoS E

20 LoS D
Density (pc/km/ln)

15 LoS C

10 LoS B

5
LoS A

0
0 1000 2000 3000 4000 5000 6000 7000
Freeway and ramp traffic flow (veh/h)

Figure 4.6: Influence-area density for on-ramp flows of between 5 and 15% of the freeway flows; the acceleration
lane length is 100 m

For an off-ramp the calculated densities in the influence area of a three-lane freeway are shown in
Figure 4.7. Again changing the off-ramp traffic flow has only a marginal affect on the traffic
densities. Increase the deceleration lane length by 100 m and the density in the influence area will
decrease by 1.8 pc/km/ln.

25
LoS E

20
LoS D
Density (pc/km/ln)

15
LoS C
10
LoS B
5

LoS A
0
0 1000 2000 3000 4000 5000 6000 7000
Freeway traffic flow (veh/h)

Figure 4.7: Influence-area density for off-ramp flows of between 5 and 15% of the freeway flows; the deceleration
lane length is 100 m

Austroads 2009
— 47 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Capacities of the freeway approaching the diverge area, departing from the merge or diverge area,
and of the ramp and the influence areas, are given in HCM 2000 (Chapter 25). The HCM 2000
lists the capacity of ramp roadways, merge areas and diverge areas, as reproduced in Table 4.6,
Table 4.7and Table 4.8.

The level of service is estimated from the densities of vehicles in the influence area. Appropriate
densities are given in Table 4.9.

Table 4.6: Approximate capacity of ramp roadways in passenger cars/hour

Free flow speed of ramp, SFR (km/h) Capacity (pc/h)
Single-lane ramps Two-lane ramps
> 80 2200 4400
> 65–80 2100 4100
> 50–65 2000 3800
• 30–50 1900 3500
< 30 1800 3200
Source: Exhibit 25-3 in the HCM 2000 (TRB 2000).

Table 4.7: Capacity values for merge areas in passenger cars/hour

Freeway free flow Maximum downstream freeway flow, v (pc/h) Max desirable flow entering
speed (km/h) Number of lanes in one direction influence area, VR12 (pc/h)

2 3 4 >4
120 4800 7200 9600 2400/ln 4600
110 4700 7050 9400 2350/ln 4600
100 4600 6900 9200 2300/ln 4600
90 4500 6750 9000 2250/ln 4600
Source: Exhibit 25-7 in the HCM 2000 (TRB 2000).

Table 4.8: Capacity values for diverge areas in passenger cars/hour

Freeway free flow Maximum upstream, vFI or downstream freeway flow, v (pc/h) Max desirable flow entering
speed (km/h) Number of lanes in one direction influence area, VR12 (pc/h)

2 3 4 >4
120 4800 7200 9600 2400/ln 4400
110 4700 7050 9400 2350/ln 4400
100 4600 6900 9200 2300/ln 4400
90 4500 6750 9000 2250/ln 4400
Source: Exhibit 25-14 in the HCM 2000 (TRB 2000).

Austroads 2009
— 48 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table 4.9: Level of service criteria for merge and diverge areas
LOS Density (pc/km/ln)
A ”6
B > 6–12
C > 12–17
D > 17–22
E > 22
F Demand exceeds capacity
Source: Exhibit 25-4 in the HCM 2000 (TRB 2000).

The HCM 2000 notes that levels of service in merge and diverge influence areas are defined in
terms of density for all cases of stable operation, from LOS A to E. LOS F exists when the demand
exceeds the capacity of upstream or downstream freeway sections or the capacity of an off-ramp.
The different levels of service are as follows:
ƒ LOS A represents unrestricted operations. Density is low enough to permit smooth merging
and diverging, with virtually no turbulence in the traffic stream.
ƒ At LOS B, merging and diverging manoeuvres become noticeable to through drivers, and
minimal turbulence occurs. Merging drivers must adjust speeds to accomplish smooth
transitions from the acceleration lane to the freeway.
ƒ At LOS C, speed within the influence area begins to decline as turbulence levels become
noticeable. Both ramp and freeway vehicles begin to adjust their speeds to accomplish
smooth transitions.
ƒ At LOS D, turbulence levels in the influence area become intrusive and virtually all vehicles
slow to accommodate merging and diverging. Some ramp queues may form at heavily used
on-ramps, but freeway operation remains stable.
ƒ LOS E represents conditions approaching capacity. Speeds reduce significantly, and
turbulence is felt by virtually all drivers. Flow levels approach capacity, and small changes in
demand or disruptions within the traffic stream can cause both ramp and freeway queues to
form.

The HCM 2000 process allows for closely spaced ramps, and ramps with more than one lane. The
manual should be consulted for details about the analysis process.

4.4.3 Weaving sections

The analysis of weaving sections is less developed than other freeway analysis procedures. The
analysis process requires the weaving and the non-weaving traffic to be estimated. This is shown
diagrammatically in Figure 4.8:

Vo1
A C

V w2 Vw1

Vo2
B D

Figure 4.8: Diagrammatic representation of traffic on freeway weaving areas.

Austroads 2009
— 49 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

In this figure, Vo1 and Vo2 are the outer, non-weaving flow rates and Vw1 and Vw2 are the two
weaving flow rates.

The number of lanes crossed by both weaving traffic flows is used to classify weaving sections.
The HCM 2000 (Chapter 13) defines the different types of weaving sections as follows, and these
are illustrated in Figure 4.9:
ƒ Type A weaving segments – all weaving vehicles must make one lane change to complete
their manoeuvre successfully. All of these lane changes occur across a lane line that
connects from the entrance gore area directly to the exit gore area. The most common form
of Type A weaving segment is formed by a one-lane on-ramp followed by a one-lane off-
ramp, with the two connected by a continuous auxiliary lane. All on-ramp vehicles entering
the freeway must make a lane change from the auxiliary lane to the shoulder lane of the
freeway. All freeway vehicles exiting at the off-ramp must make a lane change from the
shoulder lane of the freeway to the auxiliary lane. This type of configuration is also referred
to as a ramp-weave.
ƒ Type B weaving segments – one weaving movement can be made without making any lane
changes, and the other weaving movement requires at most one lane change.
ƒ Type C weaving segments – similar to those of Type B in that one or more through lanes are
provided for one of the weaving movements. One weaving movement requires a minimum of
two lane changes for successful completion of a weaving manoeuvre while the other
movement can be made without making a lane change.

The configuration of the weaving segment has a marked effect on operations because of its
influence on lane-changing behaviour. A weaving segment with 1000 veh/h weaving across
1000 veh/h in the other direction requires at least 2000 lane changes per hour in a Type A
segment, since each vehicle makes one lane change. In a Type B segment, only one movement
must change lanes, reducing the number of required lane changes per hour to 1000. In a Type C
segment, one weaving flow would not have to change lanes, while the other would have to make at
least two lane changes, for a total of (at least) 2000 lane changes per hour.

Austroads 2009
— 50 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Type A
A C

B D

Type B
C
A

D
B

Type C
A C

B D

Type C A
Note: Two Type Cs

C
B

Figure 4.9: Classification of weaving types

Configuration has a further effect on the proportional use of lanes by weaving and non-weaving
vehicles. Since weaving vehicles must occupy specific lanes to efficiently complete their
manoeuvres, the configuration can limit the ability of weaving vehicles to use outer lanes of the
segment. This effect is most pronounced for Type A segments, because weaving vehicles must
primarily occupy the two lanes adjacent to the crown line. It is least severe for Type B segments,
since these segments require the fewest lane changes for weaving vehicles, thus allowing more
flexibility in lane use.

The length and width of the weaving segment are two geometric parameters that describe the area
used by weaving vehicles. The weaving length is defined in the HCM 2000 as ‘measured from a
point at the merge gore where the right edge of the freeway shoulder lane and the left edge of the
merging lane(s) are 0.6 m apart to a point at the diverge gore where the two edges are 3.7 m
apart’. All weaving vehicles must make their lane changing within the length of the weaving
segment. If the length of a weaving segment decreases, the intensity of lane changing and
resulting turbulence increases. The weaving width is the number of lanes affected or influenced by
the weaving traffic.

Austroads 2009
— 51 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The analysis process estimates the average speed of the weaving and non-weaving traffic
assuming that the weave is unconstrained. This assumption is later tested and adjusted if
necessary. Given the average speeds and flows, the traffic density is calculated and compared
with values listed in Table 4.10.

Table 4.10: Level of service criteria for weaving segments.

Density (pc/km/ln)
Level of Service Freeway weaving segment Multilane and collector-distributor weaving
segments
A ” 6.0 ” 8.0
B > 6.0–12.0 > 8.0–15.0
C > 12.0–17.0 > 15.0–20.0
D > 17.0–22.0 > 20.0–23.0
E > 22.0–27.0 > 23.0–25.0
F > 27.0 > 25.0
Source: Exhibit 24-2 in the HCM 2000 (TRB 2000).

Figure 4.10 illustrates the predicted densities when the weaving traffic is 40% of the total traffic.
The different curves relate to different total traffic. The weaving section has 4 lanes and is Type A.
The densities and the LOS are largely dependent on the total traffic and to a lesser extent on the
length of the weaving section. Figure 4.11 demonstrates that the effect of changing the proportion
of weaving traffic is marginal.

40

35 Total flow
LoS F 8000 pcu/h
30
Density(pc/km/ln)
(pc/km/ln)

25 LoS E 6000
20 LoS D
Density

15 LoS C 4000
10 LoS B 2000
5
LoS A
0
0 200 400 600 800 1000
Weave Length (m)

Figure 4.10: Level of service for weaving sections of Type A, with 40% weaving traffic and varying total traffic

Austroads 2009
— 52 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

40

35 Per cent
LoS F weaving traffic
30
(pc/km/ln)
Density(pc/km/ln)

25 LoS E
20 LoS D
Density

15 LoS C 40 per cent

10 LoS B 10 per cent
5
LoS A
0
0 200 400 600 800 1000
Weave Length (m)

Figure 4.11: Level of service for weaving sections of Type A, with 4000 passenger car units per hour total traffic
and with different proportions of weaving traffic

The HCM 2000 provides several tables for the capacity of a weaving segment. Only one table is
reproduced here as.in Table 4.11.

Austroads 2009
— 53 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table 4.11: Capacity for various weaving segments

Volume ratio, VR (C) Type A weaving segments—100km/h free flow speed
Length of weaving segment (m)
150 300 450 600 7501
Three-lane segments
0.10 5,470 6,110 6,480 6,730 6,910
0.20 5,000 5,640 6,020 6,290 6,490
0.30 4,610 5,240 5,620 5,900 6,110
0.40 4,290 4,900 4,9903 5,2503 5,4603
0.454 4,000 4,5203 4,7903 5,0403 5,2003
Four-lane segments

0.10 7,300 8,150 8,630 8,970 9,220

0.20 6,660 7,520 8,030 8,380 8,650
0.30 6,0803 6,8303 7,3103 7,6503 7,9203
0.355 5,7803 6,5203 6,9903 7,3303 7,6003
Five-lane segments
0.10 9,120 10,180 10,790 11,210 11,5002
0.206 8,330 9,5003 10,0803 10,5103 10,83033
Source: Exhibit 24-8 in the HCM 2000 (TRB 2000).
Notes:
1. Weaving segments longer than 750 m are treated as isolated merge and diverge areas using the procedures of the HCM 2000 Chapter 25.
2. Capacity constrained by basic freeway capacity.
3. Capacity occurs under constrained operating conditions.
4. Three-lane Type A segments do not operate well at volume ratios greater than 0.45. Poor operations and some local queuing are expected in such cases.
5. Four-lane Type A segments do not operate well at volume ratios greater than 0.35. Poor operations and some local queuing are expected in such cases.
6. Five-lane Type A segments do not operate well at volume ratios greater than 0.20. Poor operations and some local queuing are expected in such cases.

Austroads 2009
— 54 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

5 INTERRUPTED FLOW FACILITIES

5.1 Metered Flow
Metered flow occurs on interrupted flow facilities when vehicles arriving at a given point on the road
are required to pass through it in accordance with certain rules. Typical examples of metered flow
are at toll booths, at various drive-in facilities (such as bottle shops, bank teller points, take-away
food outlets and parking stations), at points where a reduction in the number of traffic lanes
requires traffic to merge, and at freeway entrance ramps where a traffic control signal may be used
to restrict the volume of traffic entering the freeway. Basically, metered flow is a queuing situation,
and queuing theory may be applied to it.

Details of queuing theory applicable to this section are given in Guide to Traffic Management
Part 2 (Austroads 2008a).

A typical metered flow situation is shown in Figure 5.1. Vehicles arrive at the metering point at an
average arrival rate of r vehicles per second. They may or may not queue before being serviced
depending on whether or not other vehicles are being serviced, and they then pass through the
metering point at an average service rate of s vehicles per second.

Figure 5.1: Typical metered flow situation

The main characteristics of the process are as follows:

ƒ The arrival pattern, namely the average arrival rate and the distribution of the arrivals.
ƒ The number of service channels. There may be one service channel in the case of two lanes
merging into one lane or at a metered freeway entrance ramp, and there may be one or
several at a toll booth or at a drive-in facility.
ƒ The queue discipline. First-come first-served is usual, although in special cases other
disciplines may be used. For example, buses or high occupancy vehicles may be given
priority.
ƒ The service pattern, namely the average service rate and the distribution of the service
times.

5.1.1 Capacity
Under stable flow conditions, the capacity of a metering point is given by:
C = 3,600 s

where

C = capacity in vehicles per hour

s = average service rate in vehicles per second

Austroads 2009
— 55 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

At a point where two lanes merge into one lane, the average service time per vehicle may vary from
1.5 seconds if vehicles are not required to queue to about 2.0 seconds if a queue develops and
stop-start conditions result. Corresponding capacities are 2400 vehicles per hour and
1800 vehicles per hour respectively.

At toll booths, the average service time per vehicle may vary from 4.5 to 15 seconds depending on
the collection arrangements. Corresponding capacities per booth are 800 vehicles per hour and
250 vehicles per hour. AS 2890.1 (Parking facilities – Off-street car parking) gives the following
capacities:
ƒ automatic ticket issuing and boom gate 400 veh/h/lane
ƒ manually operated 250 veh/h/lane.

In other metering situations, the average service time per vehicle and hence channel capacity will
depend on the particular case. At existing facilities, service times may be measured directly, and
for proposed facilities service times may be estimated or preferably obtained by direct
measurement at comparable existing facilities.

Factors Affecting Capacity

In general, the actual servicing operation (such as, for example, toll collection) usually has by far
the most significant effect on the service time in a metering situation. However, in some cases,
and particularly when service times are relatively small (say Iess than three or four seconds), the
departure lane width and grade, and the presence of heavy vehicles may adversely affect the
service time.

Each case should be considered on its merits, and if service times are likely to be small, some
small allowance should be made for the effect of departure lanes less than 3.0 metres wide, and of
upgrades steeper than 3%. As a guide, the allowances for these factors at signalised
intersections, given in Section 6, may be used in the absence of actual measured data from other
comparable sites. Also, as an approximation, each heavy vehicle may be considered equivalent to
two passenger cars. However, it should be noted that such adjustments are necessary only for
small service times. Except in unusual cases, these types of adjustments should not be necessary
for service times over about 4 seconds, and certainly not when they exceed six seconds.

5.1.2 Queue Lengths and Delays

For many metering situations, it is satisfactory to treat the arrival of vehicles at the metering point
as Poisson distributed. In most situations, vehicles are usually serviced on a first-come first-served
basis, and it can be assumed that service times follow a negative exponential distribution.

On this basis, then for a single service channel the following relationships apply:
ƒ average arrival rate: r vehicles per second
ƒ average service rate: s vehicles per second
ƒ utilisation factor: ȡ = r / s
ƒ probability of n vehicles in the system, including the one being serviced
P(n) = (1 – ȡ) ȡ n

ƒ probability of more than n vehicles in the system, including the one being serviced
P(>n) = ȡ n+1

ƒ the mean queue length, including the vehicle being serviced

Austroads 2009
— 56 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

nq = ȡ / (1- ȡ)

ƒ the variance of the mean queue length, including the vehicle being serviced
ı2 (n) = ȡ / (1 – ȡ)2

ƒ the mean waiting time (delay) in the system, including the time being serviced
Wm = nq / r = 1 / s (1 – ȡ) seconds.

As the flow through a metering point approaches capacity, the utilisation factor ȡ approaches 1 and
this situation is associated with long queues and long waiting times. As far as practicable,
metering points should be designed so that the utilisation factor does not exceed about 0.8. A
worked example is given in section C1.2.1 in Commentary 1.

5.2 Urban Arterial Roads with Interrupted Flow

5.2.1 Capacity
Table 5.1 sets out typical mid-block capacities for various types of urban roads with interrupted
flow, with unflared major intersections and with interruptions from cross and turning traffic at minor
intersections.

When improvements to isolated intersections are being considered, without any change to
upstream conditions, the figures in Table 5.1 can be taken as limiting values.

Table 5.1: Typical mid-block capacities for urban roads with interrupted flow
Type of lane One-way mid-block capacity (veh/h)
Median or inner lane
Divided road 1000
Undivided road 900
Middle lane (of a 3 lane carriageway)
Divided road 900
Undivided road 1000
Kerb lane
Adjacent to parking lane 900
Occasional parked vehicles 600
Clearway conditions 900

Peak period mid-block traffic volumes may increase to 1200 to 1400 vehicles per lane per hour on
any approach road when the following conditions exist or can be implemented:
ƒ adequate flaring at major upstream intersections
ƒ uninterrupted flow from a wider carriageway upstream of an intersection approach and
flowing at capacity
ƒ control or absence of crossing or entering traffic at minor intersections by major road priority
controls
ƒ control or absence of parking
ƒ control or absence of right turns by banning turning at difficult intersections
ƒ high volume flows of traffic from upstream intersections during more than one phase of a
signal cycle

Austroads 2009
— 57 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

ƒ good co-ordination of traffic signals along the route.

5.2.2 Analysis of Level of Service

Level of service on arterial roads may be evaluated using the methods described in the HCM 2000
(TRB 2000). The procedure involves classification of urban and suburban arterial roads into four
design categories, which are shown in Table 5.2.

Table 5.2: Design categories of urban streets

Design category
Criterion
High speed Suburban Intermediate Urban
Driveway/access density Very low density Low density Moderate density High density
Arterial type Multi-lane divided; Multi-lane divided; Multi-lane divided or Undivided one-
undivided or two- undivided or two- undivided; one- way, way, two-way, two
lane with shoulders lane with shoulders two-lane or more lanes
Parking No No Some Significant
Separate right-turn lanes Yes Yes Usually Some
Signals/km 0.3–1.2 0.6–3.0 2–6 4–8
Speed limit 75–90 km/h 65–75 km/h 50–65 km/h 40–55 km/h
Pedestrian activity Very little Little Some Usually
Roadside development Low density Low to medium Medium to moderate High density
density density
Urban street class if a major I II II III or IV
arterial
Urban street class if a minor II III or IV IV
arterial
Source: Adapted from TRB (2000).

Level of service (LOS) for these arterials is defined by the average travel speed of all through
vehicles. The average travel speed is a function of the spacing of traffic signals and the average
intersection delay. Table 5.3 shows the average travel speeds that are either measured or
computed using the HCM 2000 established procedures, depending on whether it is an existing or a
proposed road. Speed measurements can be obtained from a travel time or delay survey.

Table 5.3: Average travel speeds

Urban street class I II III IV
Range of free flow speeds 90 to 70 km/h 70 to 55 km/h 55 to 50 km/h 55 to 40 km/h
(FFS)
Typical FFS 80 km/h 65 km/h 55 km/h 45 km/h
LOS Average travel speed (km/h)
A > 72 > 59 > 50 > 41
B > 56–72 > 46–59 > 39–50 > 32–41
C > 40–56 > 33–46 > 28–39 > 23–32
D > 32–40 > 26–33 > 22–28 > 18–23
E > 26–32 > 21–26 > 17–22 > 14–18
F ” 26 ” 21 ” 17 ” 14
Source: HCM 2000 (TRB 2000).

Austroads 2009
— 58 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

5.2.3 Descriptions of Levels of Service

The traffic conditions associated with the various levels of service for urban and suburban arterial
roads with interrupted flow conditions are described in HCM 2000 as follows:
LOS A Primarily free flow operations at average travel speeds, usually about 90% of the FFS
(free flow speed) for the given street class. Vehicles are completely unimpeded in their
ability to manoeuvre within the traffic stream. Control delay at signalised intersections is
minimal.
LOS B Reasonably unimpeded operations at average travel speeds, usually about 70% of the
FFS for the street class. The ability to manoeuvre within the traffic stream is only slightly
restricted, and control delays at signalised intersections are not significant.
LOS C Stable operations; however, ability to manoeuvre and change lanes in mid-block locations
may be more restricted than at LOS B, and longer queues, adverse signal coordination,
or both may contribute to lower average travel speeds of about 50% of the FFS for the
street class.
LOS D A range in which small increases in flow may cause substantial increases in delay and
decreases in travel speed. LOS D may be due to adverse signal progression,
inappropriate signal timing, high volumes, or a combination of these factors. Average
travel speeds are about 40% of FFS.
LOS E Characterised by significant delays and average travel speeds of 33% of the FFS or less.
Such operations are caused by a combination of adverse progression, high signal
density, high volumes, extensive delays at critical intersections, and inappropriate signal
timing.
LOS F Characterised by urban street flow at extremely low speeds, typically 25% to 33% of the
FFS. Intersection congestion is likely at critical signalised locations, with high delays,
high volumes, and extensive queuing.

Austroads 2009
— 59 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

6 INTERSECTIONS
In general, the more important unsignalised and signalised intersections determine the overall
capacity and traffic performance of interrupted flow facilities. Analysis of these types of
intersections usually involves consideration of their capacity and of the mid-block or route capacity
of the approach roads. In some cases a balance may need to be achieved between the
sometimes conflicting requirements of capacity and traffic safety. On the other hand, at
unsignalised intersections carrying relatively low volumes of cross and turning traffic, capacity is
usually not a major factor, and the emphasis in design is generally aimed at traffic safety
objectives.

The capacity analysis of an intersection may be based on existing traffic volumes or on estimated
future traffic volumes depending on the particular problem being addressed. For example, if the
objective is to correct an existing deficiency or to review existing traffic signal timings, present-
day peak-period volumes would be appropriate. However, if the objective is to prepare an ultimate
design in order to decide possible future land requirements, estimated future traffic volumes
should be used.

If future traffic volumes are to be used, they may be obtained as follows:

ƒ use traffic modelling techniques to produce estimates of future traffic volumes and turning
movements
ƒ alternatively, if sufficient information is not available for modelling, or if it is thought to be
inappropriate:
 determine present-day peak-period traffic volumes and turning movements
 determine the critical approach road
 estimate the potential future capacity of the critical approach road when fully developed
 factor up the existing traffic volumes and turning movements at the intersection by a
factor equal to the potential future capacity of the critical approach road divided by the
existing volume on it.

Even if the first method is used, it should be checked by the second to ensure that estimated future
approach road volumes at the intersection are realistic.

In certain circumstances, the installation of traffic signals may improve capacity, traffic operation
and/or safety at an intersection.

Although each case should be considered on its merits, typical guidelines for the installation of
traffic signals are discussed in Section 2 Part 6 (Austroads 2007).

Austroads 2009
— 60 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Information on the design and traffic management of both unsignalised and signalised intersections
is given in the Guide to Road Design (GRD) Part 4 (Austroads 2009a) and Guide to Traffic
Management Part 6 (Austroads 2007) respectively. Guide to Traffic Management Part 6 is
restricted to traffic management at locations where different traffic streams intersect each other
and provides guidance in relation to all types of road intersections, including grade-separated
interchanges, as well as rail crossings and pedestrian and cyclist crossings of roads, bicycle paths
and shared paths. Guidance on road space allocation is provided in Guide to Traffic Management
Part 5 (Austroads 2008b). It provides a summary of the road space requirements for general traffic
use. Pedestrians are covered as a generic group but should be taken to include all road users that
are deemed to be pedestrians under legislation that applies to the particular jurisdiction (e.g.
people using wheelchairs). Guide to Traffic Management Part 5 summarises the road space
requirements for different road user groups and provides more detail on providing for people with a
disability. Guide to Traffic Management Part 6 provides information relating to pedestrian crossing
facilities.

6.1 Unsignalised Intersections

This section considers the capacity of unsignalised intersections. The capacity of signalised
intersections is dealt with in Section 6.4. Refer also to Section 5.5 in Guide to Traffic Management
Part 2 (Austroads 2008a) for example applications of gap acceptance analysis to unsignalised
intersections.

6.1.1 Unsignalised Intersections with Minor Roads

At unsignalised intersections with minor roads where there are relatively low volumes of cross and
turning traffic, capacity considerations are usually not significant, and capacity analysis is
unnecessary. Table 6.1 sets out details of intersection volumes below which capacity analysis is
unnecessary.

While flaring for capacity is not required at intersections with traffic volumes below those shown in
Table 6.1, separate lanes for left or right-turning vehicles may be provided on the major road for
improved safety.

Table 6.1: Intersection volumes below which capacity analysis is unnecessary

Type of road Light cross and turning volumes
maximum design hour volumes
vehicles per hour (two way)
Two-lane major road 400 500 650
Cross road 250 200 100
Four-lane major road 1000 1500 2000
Cross road 100 50 25

6.1.2 Unsignalised Intersections with Major Roads

The operation of unsignalised intersections was extensively studied in the US in the 1990s and the
results are included in the HCM 2000. Unsignalised intersections are one of the most complicated
intersections to analyse and there have been a number of approaches developed over the years.
All approaches have their advantages and disadvantages, and in most cases can only effectively
be evaluated with a computer program.

Austroads 2009
— 61 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

In Australia, the analysis of unsignalised intersections should be evaluated using SIDRA

INTERSECTION (SIDRA Solutions 2009) or an equivalent program. This section documents the
concepts used in the analysis of unsignalised intersections.

Determining the Conflicting Streams

Unsignalised intersections have a number of conflicting streams that have an order of priority. For
instance, at a T-intersection with three vehicular streams (Figure 6.1), stream 1 has the highest
priority, being on the major road. Stream 2 has the second order priority as drivers in this stream
need to give way to stream 1 vehicles. Stream 3 has a lowest order of priority again, as stream 3
drivers need to give way to stream 2 vehicles, which in turn have to give way to stream 1 vehicles.

Stream 3

Stream 1

Stream 2

Figure 6.1: Major and minor streams at a T-intersection.

When there are only vehicular movements, then the priority of each stream can be determined by
the driving code. If there are pedestrian movements then the order of priority is more difficult. The
HCM 2000 refers to the different levels of priority as the ‘rank’ of the traffic stream. Figure 6.2
illustrates the rank of each stream at a cross intersection.

In this figure streams 3 (left turn from the major road) and 15 (pedestrian movement) have equal
priority. However, it might be questioned whether the left turners have, or take priority over the
pedestrian movement, or whether is it the other way around and pedestrian movement takes
priority over left turners. In fact driver and pedestrian behaviour would suggest that at different
times each stream may have priority over the other. Consequently, the order of the priority of
streams in not clear and if the analysis is to be as realistic as possible, then there may a need to
accommodate a range of behaviours. Brilon and Wu (2002) have developed a probabilistic
method that allows for this changing priority phenomenon.

Austroads 2009
— 62 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

10 11 12

16
6
5 14
4
1
13 2
3
15
Rank Traffic streams
1 2, 3, 5, 6, 15, 16
2 1, 4, 9, 12, 13, 14
3 8, 11
4 7, 10
987

5 14
4

13 2
3
15
Rank Traffic streams
1 2, 3, 5, 15
2 4, 9, 13, 14
3 7

9 7

Source: Exhibit 17-3 in the HCM 2000 (TRB 2000).

Figure 6.2: Traffic stream ranks at two-way stop controlled intersections

Critical headway and follow-up headway

The critical headway and the follow-up headway are two basic terms that describe how a driver will
decide whether to depart or not. The critical acceptance headway describes the assumed
minimum headway between conflicting vehicles that is acceptable to an entering minor stream
driver (in a stream of rank 2, 3 or 4). It is assumed that all drivers are consistent and behave in the
same way. The follow-up headway is the assumed headway between minor stream drivers in the
one stream that depart in a longer acceptable headway between vehicles in higher-ranking
streams.

Austroads 2009
— 63 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Values of the critical acceptance headway are difficult to measure in an unbiased fashion and the
best method is a maximum likelihood method (Brilon et al. 1999; Troutbeck 1992). The critical
acceptance headway and follow-up time values used in the HCM 2000 represent the most well
evaluated values available at this time (Kyte et al. 1997). SIDRA INTERSECTION incorporates
these values, but they may be adjusted with local knowledge.

The HCM 2000 lists (Equations 17-1 and 17-2) the critical headway and follow-up times for
movement x as
tc,x = tc,base + tc,HV PHV + tc,G G – tc,T – t3,LT

and tf,x = tf,base + tf,HV PHV

where

tc,x = critical headway for movement x (s)

tc,b = base critical headway from Exhibit 17-5 (s)

ase

tc,HV = adjustment factor for heavy vehicles (1.0 for two-lane major streets and 2.0 for
four-lane major streets) (s)

PHV = proportion of heavy vehicles for minor movement

tc,G = adjustment factor for grade (0.1 for Movements 9 and 12 and 0.2 for Movements
7, 8, 10, and 11) (s)

G = percent grade divided by 100

tc,T = adjustment factor for each part of a two-stage gap acceptance process (1.0 for
first or second stage; 0.0 if only one stage) (s)

t3,L = adjustment factor for intersection geometry (0.7 for minor-street left-turn
T movement at three-leg intersection; 0.0 otherwise) (s)

tf,x = follow-up time for minor movement x (s)

tf,base = base follow-up time from Exhibit 17-5 (s)

tf,HV = adjustment factor for heavy vehicles (0.9 for two-lane major streets and 1.0 for
four-lane major streets)

PHV = proportion of heavy vehicles for minor movement

Base values and of tc and tf, for passenger cars are given in Table 6.2.

Austroads 2009
— 64 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table 6.2: Base critical headways and follow-up times for two-way stop controlled intersections
Base critical gap, tc,base (s)
Vehicle movement Four-lane major street Follow-up time, tf,base (s)
Two-lane major street
base
Right turn from major 4.1 4.1 2.2
Left turn from minor 6.2 6.9 3.3
Through traffic on minor 6.5 6.5 4.0
Right turn from minor 7.1 7.5 3.5
Source: Exhibit 17-5 in the HCM 2000 (TRB 2000).

Impedance Factors
Rank 3 and 4 streams are impeded by the queuing in the streams of higher rank (1 and 2).
A rank 3 vehicle cannot depart unless the queue of rank 2 vehicles has cleared. A process to
account for the impedance factors was developed by Bennett in the 1980s and a description is
given in Bennett (2003).

Essentially, the capacity is first calculated based on the total conflicting flow. This flow is adjusted
to account for queuing in the higher-ranking streams (which will have priority). The capacity
adjustment factors for rank 3 streams (Figure 6.2) are the product of the probabilities that the
conflicting rank 2 streams will not have a queue. The capacity adjustment factor for the rank
4 streams, namely the right turns from the minor street at a cross intersection, is also given by the
product of the probabilities that the conflicting rank 2 and 3 streams will not have a queue, and by
an additional adjustment factor. This additional factor, for the impedance to rank 4 streams, is not
a simple one and was first found by a review of output from a large number of simulation runs (it
has later been established theoretically). There has been some discussion that the impedance
factors are not required. Kyte et al. (1997) found that the use of impedance factors improved the
estimates of capacity recorded in field observations. It is recommended that they be used in an
analysis.

Two-stage Crossing
The HCM 2000 provides for a two-stage headway acceptance procedure. The two-stage process
is illustrated in Figure 6.3. Right turners from the minor road and through and right turners from the
minor road can cross one carriageway at a time. At each stage, the number of conflicting vehicles
is less than it would be if the median was not available for storage. However, the capacity is
restricted to some extent, as the median has limited storage. The addition of the first space
provides the greatest advantage, with each subsequent space in the median providing less
additional increase in capacity.

Austroads 2009
— 65 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure 6.3: Two-stage headway acceptance analysis

The HCM 2000 Levels of Service Criteria

The HCM 2000 judges the level of service on the average control delay as shown in Table 6.3.
The HCM states that:
‘the LOS criteria for TWSC (two-way stop-controlled) intersections are somewhat
different from the criteria used in Chapter 16 for signalised intersections primarily
because different transportation facilities create different driver perceptions. The
expectation is that a signalised intersection is designed to carry higher traffic
volumes and experience greater delay than an unsignalised intersection’.

Table 6.3: Level of service criteria for two-way stop controlled intersections

Level of service Average control delay

(s/veh)
A 0–10
B > 10–15
C > 15–25
D > 25–35
E > 35–50
F > 50

Source: Exhibit 17-2 in the HCM 2000 (TRB 2000).

Austroads 2009
— 66 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Recommended Analysis Procedures

The predicted average control delay is generally used to describe the performance of the
intersection and can be used to compare the intersection’s performance with other forms of control.
The degree of saturation, the flow divided by the capacity, is another measure to quantify the
intersection’s performance.

The control delay is increased when the conflicting flows are increased and as the critical headway
and follow-up times are increased. The critical headway and the follow-up headway tend to be
longer as the manoeuvre becomes more difficult and as the subject stream has to give way to
more streams.

SIDRA INTERSECTION provides a convenient analysis tool. It can account for the effect of short
lanes, for environmental effects and for extensive queues. It is recommended that the SIDRA
INTERSECTION or the HCM 2000 software be used to analyse unsignalised intersections.

The HCM 2000 does provide a useful graph, and procedure, to estimate the 95th percentile queue
length to estimate storage requirements. This figure is reproduced below as Figure 6.4 and uses
two basic terms, the volume to capacity ratio and the maximum number of departures, in the
analysis period, which is assumed to be 15 minutes in the HCM 2000 procedures.

Source: Exhibit 17-19 in the HCM 2000 (TRB 2000).

Figure 6.4: 95th percentile queue lengths

Austroads 2009
— 67 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Storage Requirements
The storage requirements such as, for example, the storage length required for an unsignalised
right-turn movement may be calculated as follows:
ƒ Determine minor stream movement lane volume rm.
ƒ Determine maximum service rate s for the minor stream movement. Usually this is taken as
the absorption capacity C for the movement.
ƒ Calculate utilisation rate ȡ (ratio of the arrival rate to the service rate) i.e. ȡ = rm / s.
ƒ Decide on a probability that the mean queue length will not be exceeded. As a minimum
95% should be selected.
ƒ Use Figure 6.5 to determine the maximum queue length appropriate to the selected
probability level.
ƒ Allow 6 m for each passenger car and 12 m for each heavy vehicle in the maximum queue
length.

Figure 6.5: Unsignalised intersection vehicle storage requirements

Part 6 Guide to Traffic Management provides guidance on the capacity and flow at unsignalised
intersections from a traffic management perspective.

6.2 Roundabouts
Roundabouts are a particular form of unsignalised intersection. Details of their design and
operational analysis and traffic management are given in Part 6 Guide to Traffic Management
(Austroads 2007) and Part 4 Guide to Traffic Management (Austroads 2009e) to Road Design.
Australian practice for determining the capacity of roundabouts is based on headway acceptance
theory. Traffic entering the roundabout gives way to, and accepts opportunities (headways) in the
circulating traffic stream, as depicted in Figure 6.6.

Austroads 2009
— 68 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

For the analysis of roundabouts, each leg of the roundabout is analysed separately, although traffic
characteristics may be affected by traffic movement characteristics at other legs.

The current Australian analysis procedure for roundabouts is largely based on field data collected
and analysis procedures developed at ARRB (Troutbeck 1989; Akçelik 1998).

This section describes the basic concepts used in the analysis procedures. It is recommended that
the SIDRA INTERSECTION program be used.

Circulating flow

Entry flow Exiting

(lane by lane) flow

Figure 6.6: Major and minor flows at a roundabout.

Previous Austroads guidance on roundabouts was based on the research by Troutbeck (1989) and
recommended that the performance of each lane be analysed separately.

The behaviour of drivers in each entry lane is defined by a critical acceptance headway (previously
termed a critical acceptance gap) and the follow-up time. These two terms are correlated and
together they are the critical acceptance parameters. The critical acceptance parameters are
different for drivers in different entry lanes and are dependant on traffic conditions. The lane with
the larger flows has shorter critical acceptance parameters and a higher capacity. Equal flows in
different entry lanes do not produce equal lane degrees of saturation.

All entering drivers tend to give way to all circulating vehicles. The circulating roadway is modelled
by a dichotomised headway model that has a proportion of the circulating vehicles travelling in
headways platoons with short headway of between 1 and 2 seconds. This model is more able to
represent the likely platooning in streams.

The capacity and delay estimation procedures are based on traditional gap acceptance equations,
which account for a dichotomised headway model for the circulating traffic. The capacity equation
is based in the assumption that the major stream, the circulating traffic, does not adjust its speed or
relative position when a vehicle enters the roundabout. This approach is still the most common
one used in guides around the world (e.g. HCM 2000).

The use of the headway acceptance values recorded in previous Austroads guidelines with the
absolute priority capacity theory equations, results in an overestimate of capacity for some
roundabouts particularly those with higher circulating flows.

Akçelik (1998) included a process to account for increased usage of one circulatory lane. This
tendency is related to the increased bunching of traffic from a dominant approach and explains
how traffic from one approach affects the performance of other approaches.

Austroads 2009
— 69 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The critical acceptance parameters would be reduced if the entry flow is much larger than the
circulating flow. This effect increases capacity for circulating flows less than 900 pcu/h. Refer to
Commentary 2 for a detailed description of the development of the analysis procedure.

SIDRA INTERSECTION is recommended for the analysis of roundabouts as it produces a range of

entry capacities based on the traffic flow characteristics at other approaches to the roundabout.
This range is illustrated in Figure 6.7 (case 1 – single lane roundabout with an inscribed diameter
of 40 m and entry lane width of 4 m) and Figure 6.8 (case 2 – single lane roundabout with an
inscribed diameter of 60 m, entry lane width of 4 m and ratio of the dominant to the subdominant
lane flows of 1.2). Although SIDRA INTERSECTION has not been calibrated with field based
capacity data, it is based on extensive field data of critical acceptance headways. The model is,
however, well regarded by the profession and its use promotes consistency among analyses. It is
not recommended that the HCM 2000 procedure for roundabout analysis be used at this time.

Note: output from SIDRA INTERSECTION.

Figure 6.7: Upper and lower bound capacity estimates for roundabout (inscribed diameter 40 m, entry lane width 4 m)

Austroads 2009
— 70 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Note: output from SIDRA INTERSECTION.

Figure 6.8: Upper and lower bound capacity estimates for roundabout
(inscribed diameter 60 m, entry lane width 4 m, dominant/subdominant lane flow ratio 1.2)

6.3 Traffic Simulation and Computer Analysis Models

In addition to the analytical procedures outlined in Sections 6.1 and 6.2, unsignalised intersections
and roundabout operation may be analysed by computer simulation models such as INSECT,
INTANAL and SIDRA. Simulation should be used for complex intersections that are outside
SIDRA capabilities because of its current calibration.

Part 6 of the guide covers the use, performance, signalisation, road space allocation, lane
management and functional design of a roundabout from a traffic management perspective.
Design details are covered in Parts 4, 4A and 4B of the Guide to Road Design (Austroads 2009a,
Austroads 2009b and Austroads 2009c).

6.4 Signalised Intersections

In certain circumstances, the installation of traffic signals may improve capacity, traffic operation
and/or safety at an intersection. Although each case should be considered on its merits, typical
guidelines for the selection and installation of traffic signals include consideration of traffic volume,
continuous traffic, crashes, pedestrian safety, intersection configuration and traffic management
plans. These factors, together with functional layout, road space allocation, lane management,
signal phasing and timings, co-ordination and other information associated with signalised
intersections are presented in Part 6 of the guide (Austroads 2007). Further information on the
warrants and standards associated with traffic signals is given in AS 1742.2.

6.4.1 Concepts and Definitions

At signalised intersections, capacity calculations tend to focus on individual traffic movements (or
lane groups) and individual approaches, rather than on the intersection as a whole. A traffic
movement (or lane group) can be defined as one or more lanes on an intersection approach
having common right-of-way provision and similar lane utilisation characteristics.

Austroads 2009
— 71 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Separate traffic movements (or lane groups) may be as follows:

ƒ An exclusive right-turn lane (or lanes).
ƒ An exclusive left-turn lane (or lanes).
ƒ All straight-through lanes on the approach. All shared lanes (i.e. a combined left turn and
straight-through lane, or a combined right-turn and straight-through lane) may also be
included in this group provided they have a similar degree of utilisation – otherwise they can
be considered separately.

The allocation of right-of-way to individual movements at an intersection is determined by the traffic

signal phasing system. A signal phase is the state of the signals during which one or more
movements receive right-of-way. At least one movement receives right-of-way at the start of the
phase, and at least one movement loses right-of-way at the end of it. A movement that receives
right-of-way in more than one phase is called an overlap movement. One complete sequence of
phases is called a signal cycle.

Figure 6.9 shows an intersection and phasing diagram for a three-phase traffic signal installation at
a typical T-intersection.

Source: Akçelik (1981).

Figure 6.9: Typical intersection plan and phasing diagram

Austroads 2009
— 72 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The basic traffic signal capacity model, which is illustrated in Figure 6.10, assumes that when the
signal changes to green, the flow across the stop line increases rapidly to a rate called the
saturation flow, which remains constant until either the queue is exhausted or the green period
ends.

Source: Akçelik (1981).

Figure 6.10: The basic capacity model

As indicated in Figure 6.10, the basic model replaces the actual departure flow curve by a
rectangle of equal area, the height of which is equal to the saturation flow and the width of which is
the effective green time. Thus, the area under each curve (i.e. the actual departure flow curve, and
the rectangle of equal area) is the maximum number of departures in an average cycle.

As also indicated in Figure 6.10, the inter-green time is the time from the end of the green period
for one phase to the start of the green period for the next phase. It is equal to the sum of the
amber time plus any all-red time. The cycle time is the sum of all inter-green and green times
summed over all phases.

Start and end lag times are also shown in Figure 6.10. The movement lost time is equal to the
difference between the start and end lag times, and is equal to the inter-green time plus the
difference between the start loss and the end gain. The difference between movement lost time
and movement inter-green time will vary from site to site. However, in general, movement lost time
can be assumed to be equal to the inter-green time or taken as the inter-green time plus one
second.

The movements that determine the capacity and timing requirements of an intersection are called
critical movements. If all movements are non-overlap movements, there is one critical movement
per phase, i.e. the movement requiring the longest time in that phase. When there are overlap
movements, it is necessary to identify the critical movements or those that lie on the critical (or
longest) path. Intersection lost time is the total lost time summed over the critical movements.

Austroads 2009
— 73 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

6.4.2 Capacity Analysis

Most analyses are performed with computer programs, which can analyse more complex phasings
and intersection layouts. The process presented here is a description of the basic theory and
analysis process.

Capacity of a Movement
Capacity at signalised intersections is based on the concept of saturation flow. The saturation flow
rate may be defined as the maximum rate of flow that can pass through a given traffic movement
(or intersection approach) under the prevailing roadway and traffic conditions, expressed in
vehicles per unit time – usually vehicles per hour.

The capacity of a movement can be expressed as:

C = Sg/c

where

C = the capacity, in vehicles per hour

S = the saturation flow rate, in vehicles per hour

g = the effective green time per cycle that is available for the particular
movement, in seconds

c = the cycle time, in seconds.

The ratio of effective green time to cycle time is called the green time ratio, and is denoted by u
u = g/c

The ratio of the arrival flow, Q vehicles per hour, to the saturation flow S is called the flow ratio, y.
Thus,
y = Q/s

The movement degree of saturation, x, is the ratio of the arrival flow to capacity (i.e. the volume to
capacity ratio) and is expressed as:
x = Q/C = Qc/Sg = y/u

For a movement to have adequate capacity, it is necessary that:

C>Q or x<1

Thus:
Sg>Qc or u>y

Base Saturation Flows

Base saturation flows at signalised intersections are normally expressed in through car units. A
through car unit is a straight-ahead passenger car. The actual numbers of through and turning
trucks and turning cars must be multiplied by adjustment factors (which are given later) to convert
them to equivalent through car units.

Austroads 2009
— 74 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Base saturation flow values in through car units per hour by environment class and lane type are
set out in Table 6.4.

Table 6.4: Base saturation flows in through car units per hour by environment class and lane type
Environment class Lane type
1 2 3
A 1850 1810 1700
B 1700 1670 1570
C 1580 1550 1270
Source: Akçelik (1981) Table 5.1.

The environment classes A, B and C listed in Table 6.4 are defined as follows:
ƒ Class A – ideal or nearly ideal conditions for the free movement of vehicles on both
approach and exit sides, including good visibility, very few pedestrians, and almost no
interference due to loading and unloading of goods vehicles or parking turn over (typically,
but not necessarily, on a suburban residential or parkland area).
ƒ Class B – average conditions, including adequate intersection geometry, small to moderate
numbers of pedestrians, some interference by loading and unloading of goods vehicles or
parking turn over and vehicles entering and leaving premises (typically, but not necessarily,
in an industrial or shopping area).
ƒ Class C – poor conditions, including large numbers of pedestrians, poor visibility,
interference from standing vehicles, loading and unloading of goods vehicles, taxis and
buses, and high parking turn over (typically, but not necessarily, in a central city area).

The lane types 1, 2 and 3 listed in Table 6.4 are defined as follows:
ƒ Type 1 – through lane – a lane containing through vehicles only.
ƒ Type 2 – turning lane – a lane that contains any type of turning traffic, such as an exclusive
left-turn lane, an exclusive right-turn lane, or a shared lane from which vehicles may turn left
or right or continue straight through. There should be an adequate turning radius, and
negligible pedestrian interference to turning vehicles.
ƒ Type 3 – restricted turning lane – a lane similar to a type 2 lane, but with turning vehicles
subject to a small turning radius and some pedestrian interference.

Adjustment of Base Saturation Flows

The base saturation flows in Table 6.4 must be adjusted to allow for various factors, namely lane
width, gradient, and traffic composition (which takes into account vehicle type and the mix of
turning traffic). The adjustment can be applied to each lane separately, or to the movement as a
whole after adding the lane saturation flows if the factors are equally applicable to all lanes.

Austroads 2009
— 75 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The adjustment is made as follows:

S = fw fg Sb / fc

where

S = the estimated saturation flow in vehicles per hour

Sb = the relevant base saturation flow in through car units per lane, from Table 6.4

fw = lane width factor

fg = gradient factor

fc = traffic composition factor

The lane width factor, w, is as follows:

0.55+0.14w for lane widths between 2.4 and 3.0 m

1.00 for lane widths between 3.0m and 3.7 m

0.83+0.05w for lane widths between 3.7 and 4.6 m.

For a varying lane width, use the width at the narrowest point within 30 m of the stop line. The exit
lane must be at least as wide as the approach lane – if it is narrower use its width.

The gradient factor is given by:

fg = 1 ± 0.5 (percent gradient) / 100

+ being for a downgrade, and – for an up grade.

The traffic composition factor is given by:

fc = Ȉ ei Qi / Q

where

Qi = flow in vehicles per hour per vehicle type and movement i

Q = total movement flow in vehicles per hour

ei = through-car equivalent of vehicle traffic and movement, from Table 6.5

As indicated in Table 6.5, vehicles are classified as cars or heavy vehicles, a heavy vehicle being
defined as any vehicle with more than two axles or with dual tyres on the rear axle. All other
vehicles are regarded as cars.

Austroads 2009
— 76 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Also, Table 6.5 classifies turns as unopposed or opposed. An opposed turning vehicle is one that
has to give way to, and seek gaps in a higher priority opposing movement, whereas an unopposed
turning vehicle can turn without restriction as on a green arrow signal. There are two types of
unopposed turns, namely:
ƒ Normal unopposed turns. These may be either left or right turns, where the radius of the turn
is reasonably large, i.e. at least 15 m, and where there is little or no interference by
pedestrians.
ƒ Restricted unopposed turns. These may be either left or right turns, where the radius of the
turn is less than 15 m, and where turning vehicles are subject to some interference by
pedestrians. Turns in which vehicles are subject to interference by heavy pedestrian flows
may be treated as opposed turns.

Table 6.5: Through car equivalent (through car units per vehicle) for different types of vehicles and movements

Vehicle Through Unopposed turn Opposed

Normal Restricted turn

Car 1 1 1.25 e0
Heavy vehicles 2 2 2.5 e0 = 1
Source; Akçelik (1981) Table 5.2.

For the case of long combination vehicles, the figures for heavy vehicles in Table 6.5 would need
to be extended. As a broad guide, NRTC (1998) has determined general passenger car
equivalents for Austroads vehicle classes (see Section A.5 in Appendix A) as follows:

ƒ vehicle classes 2 to 5: two passenger car equivalents

ƒ vehicle classes 6 to 9: three passenger car equivalents
ƒ vehicle classes 10 to 11: four passenger car equivalents
ƒ vehicle class 12: five passenger car equivalents

With opposed turns, the value of the opposed turn equivalent eo in Table 6.5 depends on signal
timings and opposing movement characteristics. Opposed turns may include:
ƒ filter right turns giving way to vehicles in an opposing stream
ƒ left turns or right turns from one-way streets giving way to pedestrians
ƒ filter left turns under ‘left turn at any time with care’ rules giving way to pedestrians or to
vehicles in the opposing through and right-turn streams
ƒ filter left turns under ‘left turn on red’ rules.

A detailed procedure for determining eo for both exclusive lanes and shared lanes is given in
Akçelik (1981). The parameter, eo is estimated in an iterative fashion, as eo is affected by signal
phasings and timings and eo affects the signal timings. A good first approximation is to use a value
of eo equal to 3.

If relatively precise calculations are desired in any particular case, adjustments can be made for
lane underutilisation, for short lanes and for lane blockages. This involves relatively complex
analysis and is best done through the use of computer programs such as SIDRA. For details of
these adjustments, see Akçelik (1981).

Austroads 2009
— 77 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The process outlined by Akçelik (1981) includes the use of the practical degree of saturation. This
value is chosen by the user and is the target degree of saturation for the critical movements. The
practical degree of saturation is usually within the range of 0.8 to 0.95 and is used to find the
practical cycle time.

Once the saturation flow rate for a particular traffic movement has been calculated, the capacity
and degree of saturation can be determined using the equations given earlier. Also, by using the
results for each of the separate traffic movements on any given approach, the approach capacity
and its degree of saturation can be calculated.

After evaluating the signal timings, a degree of saturation should range from zero when the arrival
flow rate is zero to the practical degree of saturation for the critical movements. If a cycle time is
too short or there are insufficient lanes to handle all traffic (that is the saturation flows are
insufficient for the demand) then the degree of saturation can be greater than the practical degree
of saturation. A worked example is given in section C1.3.1 in Commentary 1.

6.4.3 Measures of Intersection Performance

Capacity calculations at signalised intersections have most meaning when applied to a traffic
movement or to an intersection approach. However, it is desirable to express the operational
efficiency of a signalised intersection in terms of various measures of performance.

A basic measure of signalised intersection performance is the intersection degree of saturation,

which is defined as the largest of the individual movement degrees of saturation. The minimum
intersection degree of saturation for a given cycle time is obtained when the critical movement
green times are proportional to the corresponding flow ratios. This in fact is an equal degree of
saturation situation in which the degrees of saturation for all critical movements are equal. In this
situation:

Yc
X
c L

where

X = intersection degree of saturation

Y = intersection flow ratio

= Ȉy for the critical movements

c = cycle time

L = intersection lost time

= sum of the critical movement lost times.

Because the intersection degree of saturation is defined as being equal to the largest movement
degree of saturation, if the condition X < 1 is satisfied, then x < 1 for all movements.

It should be noted that equal saturation of all critical movements is not always achieved or
desirable and, hence, intersection degree of saturation may not accurately represent the capacity
or saturation of the intersection.

Austroads 2009
— 78 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

When X for a signal controlled intersection is less than one, the intersection is said to be under-
saturated. When X = 1, the intersection is saturated, or operating at capacity. When X > 1 the
intersection operation is described as over-saturated. As both queue length and delay increase
rapidly as X approaches 1, it is usual in design to attempt to keep X to less than 0.9.

Other measures of intersection performance include delay, number of stops and queue length, and
some equations to enable these measures to be approximately estimated for a given vehicle
movement at an isolated fixed time signal are given below.

The seminal report by Akçelik (1989) comments on these equations as follows. Each of these
equations, which are taken from Akçelik (1981), can be considered as having a uniform and an
overflow component. The uniform component is based on the assumption of regular arrivals and is
expressed mainly in relation to the red time. The overflow component is expressed as a function of
the average overflow queue, i.e. the average number of vehicles left in the queue at the end of the
green period. Overflow queues are due to over-saturation which may last only for a few signal
cycles, (i.e. low to moderate degrees of saturation), or which may persist for a long period of time
(i.e. degrees of saturation > 1).

The average overflow queue length for a movement at isolated fixed time signals for a degree of
saturation x > xo is given by:
ª 12( x  x 0 ) º
n0 0.25CTf «Z  Z 2  »
¬« CTf ¼»

where

n0 = average overflow queue in vehicles (where there are several lanes of

vehicles, this is the total number of vehicles queued in all lanes)

C = movement capacity in vehicles per hour

Tf = the flow period, i.e. the time interval in hours during which an average
arrival demand Q persists

Z = x-1

x = movement degree of saturation = Q / C

x0 = movement degree of saturation below which the average overflow queue

is approximately zero

= 0.67+sg/600

s = saturation flow in vehicles per second

g = effective green time per cycle in seconds that is available for the
particular movement.

Austroads 2009
— 79 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The approximate value of total delay for a movement at isolated fixed time signals is given by:
qc(1  u) 2
D  n0 x
2(1  y )

where

D = total delay (vehicles hours per hour)

q = flow (vehicles per second)

c = cycle time (seconds)

u = green time ratio = g / c

y = flow ratio=q/s

n0 = average overflow queue (vehicles).

The average delay per vehicle is given by:

d = D/q

where

d = average delay per vehicle (seconds)

D = total delay (vehicle hours per hour)

q = flow (vehicles per second).

The average number of complete stops per vehicle h at isolated fixed time signals is called the
stop rate:
ª1  u n 0 º
h 0 .9 «  »
¬1  y qc ¼

where 0.9 is a reduction factor to allow for vehicles which are delayed but which do not come to a
complete stop. The number of complete stops per hour H experienced by a movement with a flow
rate of q vehicles per second is given by:
H = 3,600qh

An estimate of the maximum back of queue length nm at an isolated signal for an average cycle is
given by:
q(c  g)
nm  n0
1 y

Delay per vehicle at an isolated fixed time signal may be estimated using the formulae above.
However, delay is a complex variable that is sensitive to a variety of factors, and when evaluating
existing conditions at an intersection it is preferable to measure actual delays wherever possible.

Austroads 2009
— 80 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

6.4.4 Signal Calculations

Various computer models have been developed for analysing the operation of a traffic signal, and
to assist calculating signal timings, capacity, delays and queue lengths. One of the most
commonly used models is SIDRA. For details of this model and for information on its use in
practice see Akçelik (1987, 1990)

Other computer based traffic simulation models (e.g. SIMSET, Paramics, VISSIM) may also be
used. The use of these models in network traffic operations is outlined in Part 9 of the Guide
(Austroads forthcoming 2009f). Part 9 also includes an extensive worked example of signalised
intersection design utilising the aaSIDRA package (Akçelik and Associates 2002).

Austroads 2009
— 81 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

REFERENCES
Akçelik, R 1981, Traffic signals: capacity and timing analysis, research report ARR 123, Australian Road
Research Board, Vermont South, Vic.

Akçelik, R 1987, Introduction to SIDRA-2 for signal design, workshop papers and discussions, ARR 148,
Australian Road Research Board, Vermont South, Vic.

Akçelik, R 1989, Traffic signals: capacity and timing analysis, research report ARR 123, Australian Road
Research Board, Vermont South, Vic.

Akçelik, R 1990, Calibrating SIDRA, research report ARR180, Australian Road Research Board, Vermont
South, Vic.

Akçelik, R 1998, Roundabouts capacity and performance analysis, research report ARR 321, ARRB
Transport Research Ltd, Vermont South, Vic.

Akçelik and Associates 2002, aaSIDRA user guide, Akçelik and Associates Pty Ltd, Melbourne, Vic.

ARRB Transport Research 2003a, Incident detection station, ARRB Transport Research, Vermont South,
Vic.

Aultman-Hall, L & Hall, F 1998, ‘Research design insights from a survey of urban bicycle commuters’,
Transportation Research Record, no. 1636, pp.21-8.

Australian Bicycle Council 2000, Cycling data and indicator guidelines, Australian Bicycle Council, Sydney,
NSW, viewed 30 June 2009,
<http://www.austroads.com.au/documents/GuidelinesAndIndicators[1].pdf >

Australian Bureau of Statistics 1955 – 2008, Motor vehicle census, Australia, catalogue no. 9309.0, ABS,
Sydney, NSW.

Australian Bureau of Statistics 1963 – 2007, Survey of motor vehicle use, Australia, catalogue no. 9202.0,
ABS, Sydney, NSW.

Australian Bureau of Statistics 1994 – 2001, Freight movements, Australia, catalogue no. 9220.0, ABS,
Sydney, NSW.

Australian Bureau of Statistics 2007, Survey of motor vehicle use, Australia, 12 months ended 31 October
2006, catalogue no. 9208.0, ABS, Sydney, NSW.

Austroads 1997, Value of travel time savings, by D Rainey (ed), report no. AP-119/97, Austroads, Sydney,
NSW.

Austroads 1999, Australia cycling, the national strategy 1999-2004, AP-61/99, Austroads, Sydney, NSW.

Austroads 2000, Weigh-in-motion technology, AP-R168/00, Austroads, Sydney, NSW.

Austroads 2001, The Australian and New Zealand road system and road authorities: national performance
indicators 2000, AP-43/01, Austroads, Sydney, NSW.

Austroads 2002, An approach to the validation of road traffic noise models, AP-T14, Austroads, Sydney,
NSW.

Austroads 2009
— 82 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Austroads 2005, The Australian national cycling strategy 2005-2010, AP-C85/05, Austroads, Sydney, NSW.

Austroads 2006, The use and application of microsimulation traffic models, AP-R286/06, Austroads, Sydney,
NSW.

Austroads 2007, Guide to traffic management: part 6: intersections, interchanges and crossings,
AGTM06/07, Austroads, Sydney, NSW.

Austroads 2008a, Guide to traffic management: part 2: traffic theory, AGTM02/08, Austroads, Sydney, NSW.

Austroads 2008b, Guide to traffic management: part 5: road management, AGTM05/08, Austroads, Sydney,
NSW.

Austroads 2008c, Guide to traffic management: part 8: local area traffic management, AGTM08/08,
Austroads, Sydney, NSW.

Austroads 2009a, Guide to road design: part 4: intersections and crossings: general, Austroads, Sydney,
NSW

Austroads 2009b, Guide to road design: part 4a: unsignalised and signalised intersections, Austroads,
Sydney, NSW.

Austroads 2009c, Guide to road design: part 4b: roundabouts, Austroads, Sydney, NSW.

Austroads 2009d, Guide to road design: part 6a: pedestrian and cyclists paths, Austroads, Sydney, NSW.

Austroads 2009e, Guide to traffic management: part 4: network management, Austroads, Sydney, NSW.

Austroads 2009f, Guide to traffic management: part 9: traffic operations, Austroads, Sydney, NSW.

Aylward, D & Valentine, T 1995, ‘Bicycle and pedestrian detection and counting’, Australian pedestrian and
bicyclist safety and travel workshop, Melbourne, Victoria, ARRB Transport Research Ltd, Vermont
South, Vic., pp.1-10.

Beard, CL & McLean, JR 1974, ‘Number plate surveys on urban roads’, Australian Road Research Board
(ARRB) conference, 7th, 1974, Adelaide, South Australia, Australian Road Research Board, Vermont
South, Vic., vol.7, no.4, pp.193-204.

Bennett, DW 2003, ‘Elementary traffic flow theory and applications’ in Traffic engineering and management,
Monash University. Institute of Transport Studies, Clayton, Vic., chapter 8.2.

Bonsall, PW 1991, ‘The changing face of parking-related data collection and analysis: the role of new
technologies’, Transportation, vol.18, no.1, pp.83-106.

Brilon, W 1977, ‘Queuing model of two-lane rural traffic prediction’, Transportation Research, vol. 11, no. 2,
pp.95-107.

Brilon, W & Wu, N 2002, ‘Unsignalised intersections: a third method for analysis’, in MAP Taylor (ed.),
Transportation and traffic theory in the 21st century, Pergamon, Amsterdam, Netherlands, pp.157-178.

Brilon, W, Troutbeck, R & Konig, R 1999, ‘Useful estimation procedures for critical gaps’, Transportation
Research A, vol.33A, no.3/4, pp.161-186.

Commonwealth Scientific and Industrial Research Organisation 2000, A review of papers on creating corrupt
and missing traffic counts, CMIS 99/111, Division of Mathematical and Information Sciences, CSIRO,
North Ryde, NSW.

Austroads 2009
— 83 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Cowan, RJ & Eriksson, SCB 1972, ‘On the floating vehicle problem’, Australian Road Research Board
(ARRB) conference, 6th, 1972, Canberra, ACT, Australian Road Research Board, Vermont South,
Vic., vol.6, no.3, pp.137-143.

Currin, TR 2001, Introduction to traffic engineering: a manual for data collection and analysis, Brooks/Cole,
Pacific Grove, CA, USA.

Daff, M & Cramphorn, B 1995, ‘Bicycle and pedestrian detection and counting’, Australian pedestrian and
bicyclist safety and travel workshop, Melbourne, Victoria, ARRB Transport Research Ltd, Vermont
South, pp.223-229.

Daltrey, R 2002, ‘Data integrity’, Roads and Traffic Authority, NSW, private communication 2002.

Federal Highway Administration 2001, Traffic monitoring guide, FHWA, Washington DC, USA, viewed 30
June 2009, <www.fhwa.dot.gov/ohim/tmguide/index.htm>

Foldvary, LA 1962, Establishing an organised system of road traffic and transportation studies, Australian
Road Research Board (ARRB) conference, 1st, 1962, Canberra, ACT, Australian Road Research
Board, Vermont South, Vic., vol.1, no.1, pp.268-307.

Foley, SP 1981, ‘Traffic generated by shopping centres in Adelaide’, Australian Road Research, vol.13, no.1,
pp.12-16.

Gunay, B 2003, ‘Methods to quantify the discipline of lane-based-driving’, Traffic Engineering and Control,
vol. 44, no.1, pp.22-27.

Homburger, WS, Hall, JW, Sullivan, EC & Reilly, WR 2007, Fundamentals of traffic engineering, 16th edn,
University of California, Berkeley, CA, USA.

Horman, CB & Turnbull, HH 1974, ‘Design and analysis of roundabouts’, Australian Road Research Board
(ARRB) conference, 7th, 1974, Adelaide, South Australia, Australian Road Research Board, Vermont
South, Vic., vol.7, no.4, pp.58-82.

Institute of Transportation Engineers 1994, Manual of transportation engineering studies, Prentice Hall,
Englewood Cliffs, NJ, USA.

Institute of Transportation Engineers 2006, Transportation impact analyses for site development: an ITE
proposed recommended practice, ITE, Washington, DC, USA.

Institute of Transportation Engineers 2008, Trip generation: an ITE informational report, 8th edn, ITE,
Washington DC, USA.

Institute of Transportation Engineers 2009, Traffic engineering handbook, 6th edn, ITE, Washington, DC.

Johnston, RRM, Trayford, RS & Van Der Touw, JW 1982, ‘Fuel consumption in urban traffic: a twenty car
designed experiment’, Transportation Research, vol.16A, no.3, pp.173-184.

Jordan, P 1995, ‘Bicycle and pedestrian detection and counting’, Australian pedestrian and cyclist safety and
travel workshop, 1994, Melbourne, Victoria, ARRB Transport Research Ltd, Vermont South, Vic.,
pp.213-222.

Khan, SI & Raksuntorn, W 2001, ‘Accuracy of numerical rectification of video images to analyze bicycle
traffic scenes’, Transportation Research Record, no. 1773, pp.32-8.

Khoo, VHS & Luk, JYK 2002, ‘GPS technology and enhancement positioning developments: application to
road transport in Singapore’, Roads & Transport Research, vol.11, no.3, pp.34-45.

Austroads 2009
— 84 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Koniditsiotis, C 1990, ‘The Golden River capacitive pad weigh-in-motion system: its life and times’, Australian
Road Research Board (ARRB) conference, 15th, 1990, Darwin, Northern Territory, Australian Road
Research Board, vol.15, no.6, pp.171-188.

Kyte, M, Tian, Z , Mir, Z, Hameedmansoor, Z, Kittelson, Vandehey, M, Robinson, B, Brilon, W, Bondzio, L,

Wu, N & Troutbeck, R 1996, Capacity and level of service at unsignalized intersections: final report,
vol.1: two-way stop-controlled intersections, NCHRP web document 5, National Cooperative Highway
Research Program, TRB, Washington DC, viewed 29 June 2009,
<http://www.nap.edu/books/nch005/html>

Lay, MG 1998, Handbook of road technology, 3rd edn, Gordon and Breach, Sydney, NSW.

Lay, MG & Daley, KF 2002, ‘The Melbourne CityLink project’, Transport Policy, vol.9, no.3, pp.261-267.

Leschinski, R 1994, Bicycle detection at signalised intersections, ARR 245, Australian Road Research
Board, Vermont South, Vic.

Luk, JYK 1989, Modelling and monitoring the performance of urban traffic control systems, special report
SR 43, Australian Road Research Board, Vermont South, Vic.

Luk, JYK & Brown, JI 1987, ‘The use of piezoelectric cable as an axle sensor’, New Zealand Roading
symposium, Wellington, National Roads Board, Wellington, New Zealand, vol.2, pp.297-304.

Luk, JYK & Yang, C 2001, ‘Impact of ITS measures on public transport: a case study’, Journal of Advanced
Transportation, vol.35 no. 3, pp.303-320.

Luk, JYK, Sims, AG & Lowrie, PR 1983, The Parramatta experiment: evaluating four methods of area traffic
control, ARR 132, Australian Road Research Board, Vermont South, Vic.

McLean, JR 1978, ‘Observed speed distributions and rural road traffic operations’, Australian Road Research
Board (ARRB) conference, 9th, 1978, Brisbane, Queensland, Australian Road Research Board,
Vermont South, Vic., vol. 9, no. 5, pp.235-244.

McLean, JR 1989, Two lane highway traffic operations: theory and practice, Gordon and Breach, London,
UK.

McPherson, C 1995, ‘An improved matching algorithm for OD numberplate surveys’, Transport Engineering
in Australia, vol.1, no.1, pp.27-37.

McTrans 2009, An international technology resource, Center for Microcomputers in Transportation

(McTrans), University of Florida, USA, viewed 30 June 2009,
<http://mctrans.ce.ufl.edu/aboutMctrans/>

Menge, CW, Rossano, CF, Anderson, GS & Bajdek, CJ 1998, FHWA traffic noise model, version 1.0, US
Department of Transportation, Cambridge, USA.

Michael, R 1976, ‘Short count annual average daily traffic estimation’, Australian Road Research Board
(ARRB) conference, 8th, 1976, Perth, Western Australia, Australian Road Research Board, Vermont
South, Vic., vol. 8, no.5, pp.9-14.

Miller, I, Johnson, RA & Freund, JE 1999, Miller and Freund’s probability and statistics for engineers, 6th edn,
Prentice Hall, NJ, USA.

Mitchell, G, Mamdeo, A, May, AD & Milne, D 2003, ‘The air quality implications of urban road user charging’,
Traffic Engineering and Control, vol. 44, no.2, pp.57-62.

Austroads 2009
— 85 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Monash University 2003, Traffic engineering and management, 2 vols, Monash University, Institute of
Transport Studies, Clayton, Vic.

Moses, F 1979, ‘Weigh-in-motion system using instrumented bridges’, Transportation Engineering Journal,
vol.105, no.3, pp.233-49.

National Association of Australian State Road Authorities 1980, Guide policy for traffic noise measurement
procedures, NAASRA, Sydney, NSW.

National Association of Australian State Road Authorities 1982, Guide to traffic counting in state road
authorities, report no. TEC-13, NAASRA, Sydney, NSW.

NRTC 1998, Updating heavy vehicle charges: technical report, National Road Transport Commission,
Melbourne, Australia.

NSW Department of Planning 2006, The 2004 household travel survey summary report, Transport and
Population Data Centre, Department of Planning, NSW, viewed 30 June 2009,
<www.transport.nsw.gov.au/tdc/documents/hts-2004-2006-release.pdf >

NZ Ministry of Transport 2005, Getting there: on foot, by bicycle: a strategy to advance cycling and walking in
New Zealand transport, Ministry of Transport, Wellington, New Zealand.

Nielsen, HL & Nordic Council of Ministers 1996, Road traffic noise: Nordic prediction method, TemaNord
1996:525, Nordic Council of Ministers, Copenhagen, Denmark.

Oppenlander, JC, Bunte, WF & Kadakia, PL 1961, ‘Sample size requirements for vehicular speed studies’,
Highway Research Bulletin, no.281, pp.68-86.

Pearson, RA & Foley, GD 2001, Emerging issues in the Australian transport industry 2000-2015, research
report ARR 350, ARRB Transport Research Ltd., Vermont South, Vic.

Peters, RJ 1984, ‘AXWAY: a system to obtain vehicle axle weights’, Australian Road Research Board
conference, 12th, 1984, Hobart, Tasmania, Australian Road Research Board, Vermont South, Vic.,
vol.12, no.1, pp.19-29.

Quail, DJ & West, RP 1992, ‘The Sydney Harbour Bridge bus lane’, Australian Road Research Board
conference, 16th, 1992, Perth, Western Australia, Australian Road Research Board, vol.16, no.5,
pp.263-85.

Reilly, WR & Gardner, CC 1976, ‘Techniques for measuring delay at intersections’, Transportation Research
Record, no.644, pp.1-7.

Richardson, A & Battellino, H 1997, ‘Similar, yet different: some emerging trends in travel surveys in
Australia’, Australasian Transport Research Forum (ATRF), 21st, 1997, Adelaide, South Australia,
University of South Australia, Transport Systems Centre, Adelaide, SA, vol. 21, no.1b, pp.101-24.

Richardson, AJ 1974, ‘An improved parking duration study method’, Australian Road Research Board
conference, 7th, 1974, Adelaide, South Australia, Australian Road Research Board, Vermont South,
Vic., vol.7, no.2, pp.397-413.

Richardson, AJ & Graham, NR 1982, ‘QDELAY: a signalized intersection survey method’, Australian Road
Research Board conference, 11th, 1982, Melbourne, Victoria, Australian Road Research Board,
Vermont South, Vic., vol.11, no.4, pp.34-52.

Richardson, AJ, Ampt, ES & Meyburg, AH 1995, Survey methods for transport planning, Eucalyptus Press,
Melbourne,Vic.

Austroads 2009
— 86 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Roads and Traffic Authority 1995, Guide to traffic generating developments, 2nd edn, RTA, Sydney, NSW.

Roads and Traffic Authority 2001, ‘SH10-Pacific Highway travel time and delay monitoring: study report’,
Transport Planning Section, RTA, Sydney, NSW.

Roads and Traffic Authority 2002, ‘Sydney transit lane and bus lane survey’, Transport Planning Section,
RTA, Sydney, NSW.

Robertson, HD 1994, ‘Travel-time and delay studies’, in HD Robertson, JE Hummer & DC Nelson (eds),
Manual of transportation engineering studies, Prentice Hall (for Institute of Transportation Engineers
(ITE)), Englewood Cliffs, NJ, USA, pp.52-68.

Roper, PJ 2004, ‘Data collection technology at ARRB Transport Research’, Road and Transport Research,
vol.13, no.1, pp.88-91.

Rouke, A & Bell, MGH 1994, ‘An image processing system for pedestrian data collection’, International
conference on road traffic monitoring and control, 7th, 1994, London, United Kingdom, IEEE, Los
Alamitos, CA, USA., publication no. 391, pp.123-6.

Samuels, SE 1988, Development and evaluation of highway speed weigh-in-motion systems in Australia,
special report SR 40, Australian Road Research Board, Vermont South, Vic.

Samuels, SE & Saunders, RE 1982, ‘The Australian performance of the UK DOE traffic noise prediction
method’, Australian Road Research Board conference, 11th, 1982, Melbourne, Victoria, Australian
Road Research Board, Vermont South, Vic., vol.11, no.6, pp.30-44.

SIDRA Solutions 2009, SIDRA INTERSECTION: micro-analytical software for intersection evaluation and
design, Akcelik and Associates, Greythorn, Vic., viewed 30 June 2009,
<www.sidrasolutions.com/traffic_software_intersection.htm>

Stewart, PM, Luk, JYK, Brown, JI & Gunstone, E 1986, ‘Laboratory and in-ground tests of triboelectric and
piezoelectric cables as vehicle detectors’, Australian Road Research Board conference, 13th and
Road Engineering Association of Asia and Australasia (REAAA) conference, 5th, 1986, Adelaide,
South Australia, Australian Road Research Board, Vermont South, Vic., vol.13, no.7, pp.107-22.

Stopher, P 2001, ‘Using passive GPS as a means to improve spatial travel data’, World Congress on
Intelligent Transport Systems, 8th, 2001, Sydney, New South Wales, Australia, ITS Australia,
Melbourne, Vic., 11pp.

Taylor, MAP 1982, ‘Travel time variability: the case of two public modes’, Transportation Science, vol.16,
no.4, pp.507-21.

Taylor, MAP 2000, ‘Network modelling of the traffic, environmental and energy effects of lower urban speed
limits’, Road & Transport Research, vol.9, no.4, pp.48-57.

Taylor, MAP, Bonsall, PW & Young, W 2000, Understanding traffic systems: data, analysis and presentation,
2nd edn, Ashgate, Aldershot, UK.

Taylor, MAP, Young, W & Bonsall, PW 1996, Understanding traffic systems: data, analysis and presentation,
Avebury, Aldershot, UK.

Transfund New Zealand 2001, Guide to estimation and monitoring of traffic counting and traffic growth,
research report 205, Transfund, Wellington, NZ.

Transport WA 2000, Perth walking: the metropolitan region pedestrian strategy, Department of Transport,
Perth, WA.

Austroads 2009
— 87 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Transportation Research Board 2000, Highway capacity manual: HCM 2000, TRB, Washington, DC.

Troutbeck R 1989, Evaluating the performance of a roundabout, special report 45, Australian Road Research
Board, Vermont South, Vic.

Troutbeck, RJ 1992, Estimating the critical acceptance gap from traffic movements, research report 92-5,
Queensland University of Technology, Brisbane, Qld.

Troutbeck, R 2002, ‘The performance of uncontrolled merges using a limited priority process’, in
Transportation and traffic theory in the 21st century: proceedings of the 15th international symposium
on transportation and traffic theory, 2002, Adelaide, South Australia, MAP Taylor (ed.), Pergamon,
Amsterdam, Netherlands, pp.463-82.

Troutbeck, RJ & Dods, JS 1986, ‘The ARRB video analysis data acquisition system’, Australian Road
Research Board conference, 13th and Road Engineering Association of Asia and Australasia (REAAA)
conference, 5th, 1986, Adelaide, South Australia, Australian Road Research Board, Vermont South,
Vic., vol.13, no.7, pp.183-95.

Troutbeck, RJ & Kako S 1999, ‘Limited priority merge at unsignalised intersections’, Transportation Research
Part A, vol.33, no.3-4, pp.291-304.

UK Department of the Environment 1975, Calculation of road traffic noise, HMSO, London, UK.

UK Department of Transport 1988, Calculation of road traffic noise, HMSO, London, UK.

Vaughan, RJ 1968, ‘The estimation of highest hourly volumes’, Australian Road Research Board conference,
4th, 1968, Melbourne, Victoria, Australian Road Research Board, Vermont South, Vic., vol. 4, no.1,
pp.633-44.

VicRoads 2008, 'Road network priorities and plans – illustrative examples', Internal working papers,
VicRoads, Kew, Vic.

Victoria Department of Transport 2009, Victorian integrated survey of travel and activity 2007, Department of
Transport, Melbourne, Vic.

Vincent, EN 1986, ‘Vehicle classification and mass data collection systems’, Australian Road Research
Board conference, 13th and Road Engineering Association of Asia and Australasia (REAAA)
conference, 5th, 1986, Adelaide, South Australia, Australian Road Research Board, Vermont South,
Vic., vol.13, no.8, pp.96-108.

Virkler, MR 1998, ‘Prediction and measurement of travel time along pedestrian routes’, Transportation
Research Record, no.1636, pp.37-42.

Young, W 1996, ‘Management and design of parking’, in KW Ogden & SY Taylor (eds.), Traffic engineering
and management, Department of Civil Engineering, Monash University, Clayton, Vic., 30pp.

Young, W, Taylor, MAP & Gipps, PG 1989, Microcomputers in traffic engineering, Research Studies Press,
Taunton, UK.

Australian/New Zealand Standards

AS 1055.1-1997, Acoustics: description and measurement of environmental noise: general procedures.

AS 1055.2-1997, Acoustics: description and measurement of environmental noise: application to specific

situations.

Austroads 2009
— 88 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

AS 1055.3-1997, Acoustics: description and measurement of environmental noise: acquisition of data

pertinent to land use.

AS 1742.2-2009, Manual of uniform traffic control devices part 2: traffic control devices for general use.

AS 2702-1984, Acoustics: methods for the measurement of road traffic noise.

AS/NZS 2890.1-2004, Parking facilities: off-street car parking.

AS/NZS 9001-2000, Quality management system: requirements.

AS/NZS ISO 9001-2008, Quality management systems: requirements.

AS IEC 61672.1-2004, Electroacoustics: sound level meters: specifications.

DR 01043, Methods of test for fuel consumption of motor vehicles designed to comply with Australian Design
Rule 37.

Austroads 2009
— 89 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX A TRAFFIC VOLUME SURVEYS

This appendix provides an introduction to traffic volume counting and the estimation of associated
traffic parameters. Road authorities and other private service providers carry out most traffic
counting in Australia. It is essential to the many users that such data are presented on a
consistent basis, and uniformity of methods and procedures is desirable. This section describes a
basis for the design of traffic counting systems that provide traffic data at a consistent and
specified level of accuracy and with a consistent coverage of the road system.

Table A 1: Types of traffic volume data required for different purposes

Purpose Type of traffic data required
1. Road financing and budgeting Total vehicle kilometres of travel (VKT) for different classes of road
2. Classification of roads and road network planning AADT, traffic composition and trends
3. Classification of traffic for noise and environment studies AADT, traffic composition and speed
4. Validation and adjustment of trip data collected in transportation AADT, AAWT, 24 hour or peak hour across cordon or screen lines
studies or at major intersections
5. Development and maintenance and improvement programs AADT, VKT, DHV; traffic composition and growth trends for
and economic evaluation of alternatives important corridors
6. Selection of design standards for road geometry and pavement AADT, DHV, traffic composition and growth trends for a particular
thickness road; possibly bicycle and pedestrian data
7. Interchange and intersection deign Turning movements, traffic composition, peak hour volume and
trends; possibly bicycle and pedestrian data
8. Evaluation of level of service DHV, hourly flow variation patterns; AADT maybe adequate for low
volume rural roads
9. Supply of exposure data for determining accident rates AADT, VKT, traffic composition
10. Implementation and appraisal of safety programs VKT on different road sections and areas, AADT and peak hour
volumes on particular roads, movements at intersections;
sometimes, traffic composition and pedestrian movements
11. Establishment of warrants for traffic control devices AADT, peak hour volumes; sometimes traffic composition and
pedestrian movements
12. Estimating the loading on pavement and bridges* (in the Classified vehicle counts and axle configurations
absence of weigh-in-motion equipment)
* the selection of design standards for pavement thickness is now largely determined from weigh-in-motion equipment that provides a direct measure of pavement
loading by vehicle types.
Note: AADT = annual average daily traffic; AAWT = annual average weekday traffic; DHV = design hour volume; VKT = vehicle kilometres of travel. See Austroads
Glossary for full definitions.

A.1 Methods of Counting

Systematic counting programs have been undertaken for many years by Austroads member
authorities.

The results of these counting programs are still published at regular intervals. Models for
estimating design traffic volumes exist from these data sources particularly for rural areas where
there are often long historical records. The science of systematic traffic counting is well advanced,
but still an active area of research. Road authorities have also developed software for the
correction of corrupt or missing counts recorded by permanent traffic counting stations (e.g. CSIRO
2000; Section 2.3).

Austroads 2009
— 90 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Computer-based traffic information systems are being developed. These systems provide a
common platform to retrieve and store traffic flow and other data from freeway, signal and other
emergency management systems. An example is the Transport Management Integration System
(TMIS) of the NSW Roads and Traffic Authority. Figure A 1 shows the Transport Operations Room
of the Transport Management Centre at Eveleigh, Sydney, where the TMIS is located. The
technology for measuring traffic counts consists of vehicle sensors and data loggers. When a
vehicle appears at an observation station, the sensor generates a signal that is received and
stored inside the logger. The data within the logger can be downloaded on-site to a portable
computer at regular intervals, or uploaded through telemetry to a central computer.

Traffic volume counts can be carried out manually or by automatic traffic counters.

Source: RTA NSW.

Figure A 1: The operations room of the RTA Transport Management Centre

A.1.1 Manual Traffic Counting

Manual counts are usually carried out at intersections where turning movement volumes are
required or at sites where detailed classification data are needed, e.g. number of passengers or
vehicle body types. Mechanical hand counter and PC-based data loggers are available to facilitate
data logging. Manual counting is much more expensive than automatic counting and is generally
used for short period studies only. The number of persons required to count a particular site will be
governed by the amount of data required and the experience of the personnel involved. For
example, an experienced person can record up to 12 items, say, three traffic movements of four
vehicle classes under light to moderate traffic flow.

A.1.2 Automatic Traffic Counting

Automatic counters are normally used for recording 24-hour counts and the hourly, daily or
seasonal variations in traffic volume. The equipment normally consists of a data logger and an
axle or vehicle sensor, or both.

Austroads 2009
— 91 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Automatic Axle Counts

Axles are generally detected by a pneumatic rubber tube stretched across the road surface. The
pulse generated in the tube when an axle crosses it closes the contact at the connected air-switch,
and so the axle is registered. A typical response from a tube axle detector is shown in Figure A 2
for the case of a two-axled vehicle. Pneumatic tubes are susceptible to damage and have
relatively short life spans. They need frequent checking for splits or breakages and therefore are
mainly suitable for short-term traffic studies.

Special electrical cables such as ‘triboelectric’ (friction sensitive) and ‘piezoelectric’ (pressure
sensitive) cables may also be used as axle detectors (Stewart et al. 1986; Luk and Brown 1987).
They work by generating a detectable electric charge as a vehicle passes over the cable.
Piezoelectric cables are normally set in grooves cut in the surface, making them permanent. They
are used in the CULWAY weigh-in-motion systems described in Appendix G. In-ground piezo-
electric cables are durable, but more expensive than aboveground pneumatic tubes. One type of
piezo-resistive material changes the resistance of the sensor under pressure and is suitable in
detecting pedestrians at zero or low speeds (Aylward and Valentine 1995). Section G.3 in
Appendix G also describes the use of quartz piezo-cable for weigh-in-motion applications.

When tube or cable detectors are used, it is important that they are laid perpendicular to the flow of
traffic to reduce the chance of multiple counts. A notable exception is in the case where detectors
installed in a ‘zed’ pattern are used to measure lateral position also.

Automatic counting equipment that uses pneumatic tubes and other axle detectors either provides
counts of ‘axle pairs’ from one detector, or classifies the vehicles according to axle spacings using
two detectors. If it provides counts of axle pairs only, an appropriate correction factor needs to be
used. This factor is usually derived from vehicle classification counts – generally taken manually at
representative sites. Alternatively, a correction factor may be obtained from a direct correlation
between an axle detector count recording at sites also equipped with inductive loop detectors that
register the passage of a complete vehicle unit.

Figure A 2: Typical response from an axle detector when crossed by a two-axled vehicle

Austroads 2009
— 92 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure A 3: Four positions of a two-axled vehicle passing over two-axle sensors

The principle of using a pair of axle sensors for vehicle classification is illustrated in Figure A 3.
Two axle sensors (e.g. pneumatic tubes) are spaced at a known distance L metres apart. Four
vehicle actuation pulses (T1, T2, T3, T4) are recorded when a two-axled vehicle passes over the
sensors. The speed (v) and wheelbase (w) are determined using the following equations.

L
Speed v (m/s)
T34

Wheelbase w = v T13

where

T13 = time difference between T1 and T3

T34 = time difference between T3 and T4.

Austroads 2009
— 93 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The correction factor (CF) for converting axle-pairs to vehicle numbers is defined as:
CF = (Total (manual or automatic) vehicle count) / (‘axle pair’ count).

Figure A 4 shows typical correction factor graphs for highways and main roads in Victoria (similar
graphs would be available in most other states).

Figure A 4: Axle count correction factors

Automatic Vehicle Counts

An alternative form of traffic detection is to register the passage and/or presence of a vehicle. The
most widely used vehicle presence detector is the inductive loop. Other technologies include
microwave or radar scanning, infrared, acoustic, magnetic and video imaging devices.

A microwave sensor can be mounted on a pole by the roadside and offers the benefit of remote
sensing without road works to install loop sensors. Disadvantages include lower resolution than
loop sensors (lane-by-lane or at specified spots) and susceptibility for error in adverse weather
conditions.

Video imaging technology has advanced rapidly in recent years, especially as an enforcement and
monitoring device for tolled roads such as the CityLink in Melbourne. There, vehicle registrations
can be identified from an image of the numberplate as the vehicle passes through a gantry where
the video camera is mounted. Video cameras have long been used for monitoring a freeway or
other traffic facilities. Some of these cameras are dedicated for collecting data such as volume,
occupancy and speed, which are suitable for automatic incident detection and as speed and red
light cameras for law enforcement.

Video imaging is used in some jurisdictions to capture registration numbers of speeding vehicles.
In NSW, it is also used in the Safe-T-Cam system to monitor heavy vehicle travel times across the
statewide road network.

The inductive loop sensor is by far the most accepted vehicle count technology by Australian road
authorities. It consists of several loops of wire embedded in the pavement, or attached to the road
surface, as a temporary detector. Loop detectors embedded in the road pavement are more
durable with a life of about five years under normal traffic conditions, but they are more expensive
than surface-mounted rubber tube detectors.

Austroads 2009
— 94 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

An alternating current is passed through the inductive loop. When a mass of metal (such as a
vehicle chassis or an engine) passes through the electromagnetic field of the loop, the inductance
of the loop changes. These changes are used to indicate the passage or presence of a vehicle.
Figure A 5 shows the typical response of a loop detector when crossed by a passenger car. The
reduction in loop inductance depends on the size and metallic content of the vehicle. For example,
a passenger car generates a change of about five per cent for the loops typically used in traffic
counting. The inductive loop detector is used extensively for automatic traffic counting, traffic
surveillance and traffic signal control. The analysis of inductive profiles is also used for vehicle
classification.

Figure A 5: Typical Response of a loop detector crossed by a passenger car

For accurate counting, care needs to be taken in selecting the size, shape and positioning of the
loop within the traffic lanes. Over-counting can occur when loops in adjacent lanes are too close
and the same vehicle is detected in both lanes. On the other hand, motorcycles or small vehicles
straddling both lanes may be missed by loops placed too far apart.

Dual loop configurations can capture three or four classes of vehicle. Some recently developed
systems can capture up to eight classes by detecting the induced magnetic profiles of vehicles
travelling over the loops. This enables a distinction between trucks and buses, which have the
same axle configuration but differ in mass. There are also available specialised loop
configurations to capture bicycle movement by direction.

Two types of loop designs are commonly used by road authorities for vehicle counting on a lane of
traffic flow. A loop for counting at mid-block is often a square loop of 2 m x2 m, and a pair of these
loops at a known distance (about 4 m) apart is usually used for speed measurement and vehicle
classification. A second loop design is the SCATS loop for signal operation and traffic counting.
The SCATS loop has a width of 2 m and an overall length of about 4.5 m and is located near a
stop line of an approach. It provides reasonably accurate counts over a time period of, say,
15 minutes but the counting accuracy is less over the length of a signal cycle due to the loop
location and length.

Austroads 2009
— 95 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The SCATS loop does have some shortcomings as a vehicle counter. Because of the size of the
loop and its location just before the stop line, at the start of the green signal it is possible for
vehicles to ‘bridge’ the loop: the rear of one vehicle having not left the loop’s field of influence
before the front of the following vehicle enters it. Under those circumstances the two vehicles will
be counted as only one. The impact that this effect would have on the accuracy of a count would
vary from site to site. It might also be more significant on approaches with short greens and at
sites with short cycle lengths.

In general, inductive loop detectors are not suitable for unsealed roads. Pneumatic tubes may be
used as axle sensors on unsealed roads provided the pavement is hard and smooth. Their life is
short where pavement deformation or surface erosion occurs under traffic.

Microprocessor data recorders log traffic volumes at pre-determined intervals (e.g. daily, hourly,
15 minutes, 5 minutes, etc.) or alternatively they simply log individual axle event data for later
downstream processing. They can store large amounts of data that can be retrieved also from a
remote site using a data modem for subsequent analysis.

A.2 Sources of Counting Error

Errors in traffic counts may arise from the collection of raw count data, the estimation of a mean
count from a series of observed (sample) counts, and the period over which continuous counts are
taken. Counting errors would also depend on the type of equipment and method of counting (the
pneumatic method would yield an error type different from that of the inductive loop method). It is
always essential to follow the equipment manufacturer’s instructions to minimise the counting error.

A fundamental observation about all traffic count surveys is that errors and biases are inevitable
and must be allowed for. Neither manual nor automatic counts are free of possible errors.
Although the magnitudes of specific errors at any one site are difficult to ascertain, some
indications of likely errors are available if the site can be compared to other surrounding sites. The
following rules are known:
ƒ Completely accurate counts do not exist due to human errors, mechanical failures,
environmental factors such as rain and magnetic fields, vehicles straddling two lanes,
interference from opposing traffic, misinterpreted multi-axled vehicles (e.g. vehicles following
others too closely or changing speed), vehicles turning across axle-detectors, and other
factors which affect the reliability of raw traffic counts.
ƒ Manual counts tend to underestimate traffic volumes, due to missed observations. These
errors of omission are likely to occur more frequently on heavily trafficked roads or when
inexperienced personnel are used. A rule-of-thumb is that undercounting of about five per
cent or higher may be expected from manual traffic count surveys.
ƒ Automatic counters using single-axle detectors may be expected to overestimate traffic
volumes, unless an axle count correction is applied to the raw data. The magnitude of this
error will depend on the proportion of multi-axled or heavy vehicles in the traffic (Figure A 4).
ƒ If the axle sensors of a classifying counter are placed closer together than the specification,
vehicle speeds are over-stated and vehicles could be wrongly classified.
ƒ Time clock errors in automatic counters may result in apparent ‘time shifts’ of peak demand
periods. Counts should be taken over intervals of at least 24 hours to permit corrections to
be applied. Long-term records must also be corrected for daylight saving time.

Austroads 2009
— 96 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The counting personnel should be well prepared in terms of counting equipment, suitable seats,
appropriate clothing including contingency provisions for adverse weather, amenities (food, toilet),
and carry authority certificates. Conspicuous signage is not recommended as it may affect count
results.

A common and useful assumption is that traffic count estimates have a normal probability
distribution about the true traffic volume. The assumption of statistical normality means that the
methods of statistical inference can be applied to generate confidence intervals for traffic volume
estimates i.e. a likely range of values of the true traffic flow.

A.3 Estimation and Accuracy Requirements

Frequently used traffic characteristics include AADT, DHV, VKT and peak-hour volume. DHV and
VKT can be derived from AADT and therefore the measurement and estimation of AADT is
important. Global estimates of VKT over the total road network are also estimated by other
methods, e.g. the Survey of Motor Vehicle Usage of the Australian Bureau of Statistics (ABS from
1963 onward). In NSW, vehicle odometer readings are recorded at annual inspections and the
information if transferred could provide useful VKT data in that jurisdiction.

The use of AADT assumes that weekend traffic remains fairly constant over time, although less so
than weekday traffic. However, a much larger component of weekend traffic is discretionary and
factors such as rising fuel prices might well impact on the underlying assumptions. The AAWT
might well become a better indicator for planning purposes.

The accuracy of an AADT estimate depends mainly on the period of time during which traffic was
counted. It also depends on the accuracy of the seasonal adjustment factor used to convert the
average daily traffic (ADT) to AADT as described in the following section.

Similarly, the accuracy of the estimation of VKT depends primarily on the number of road links on
which AADT is measured and the accuracy of the estimates of AADT. If every link with relatively
uniform traffic cannot be sampled, it also depends on the dispersion or ‘scatter’ of the estimates.
Accuracy can be improved by grouping together road segments with similar traffic characteristics.
These groups may be geographic regions, functional categories, or volume strata, depending on
the data sought. For example, seasonal variations have the greatest impact on AADT estimates
and regional groupings are usually preferable. On the other hand, volume stratification is used to
improve the accuracy in estimating a VKT.

Thus, there is a clear distinction between the estimation of AADT and VKT:
ƒ AADT is estimated by making short-term counts, usually on a rotational basis, at selected
sites, calculating ADT and applying an adjustment factor. This factor is derived from a
seasonal (12 month) count on the road in question, or on a road exhibiting similar seasonal
traffic variations.
ƒ VKT on a network is estimated by measuring or estimating AADT on selected links, and
making an estimate of the total network travel on the basis of travel on each link. The
accuracy of the result will depend on the number of links selected and the dispersion of
AADT on the network. It can be improved by adopting statistical sampling procedures.

Accuracy requirements for both AADT and VKT are set out in Table A 2 and Table A 3.

Austroads 2009
— 97 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table A 2: Traffic characteristics accuracy requirements – individual sites or road sections1, 2

Traffic characteristic Maximum error requirements
AADT < 100 50%
101-300 35%
301-1100 25%
>1100 15%
Trend in AADT 10%
Traffic composition 20%
Notes:
1. Accuracies for individual sites at the 68% confidence limits.
2. Traffic composition errors are expressed as percentage errors of the vehicle class proportions.
Source: NAASRA (1982).

Table A 3: Traffic characteristics accuracy requirements – groups of roads1

Traffic characteristic Australia State City3 Rural area3 AADT
range2
VKT – Annual total 3% 5-7% 10% 10% 25%
VKT – Annual change 1% 3% 5% 5% 10%
VKT – Annual composition 3% 5% 10% 10% 15%
Notes:
1. Accuracies or groups of roads are expressed as errors at the 95% confidence limits.
2. Recommended AADT ranges are given in Table A.5.
3. If these accuracies are achieved, those for States and Australia will also be obtained.
Source: NAASRA (1982).

A.4 Measurement and Estimation of AADT

Some member authorities carry out comprehensive area traffic volume surveys in which traffic
volumes are usually expressed in terms of AADT. These are measured on all important roads in a
state or region comprising several local government areas. By definition AADT is measured at
sites where counts can be taken continuously for a full year. A sample count method, however,
may be used to estimate AADT based on the assumption that the seasonal variation of daily
volumes remains stable over a long period of time and over a long length of road.

Other approaches that are being used amongst Austroads member authorities include those
reported in Transfund New Zealand (2001) and the US Federal Highway Administration (FHWA
2001).

A.4.1 Daily and Seasonal Variations in Traffic Volume

Traffic volumes vary between hours of the day, days of the week, and between weeks or months of
the year.

Austroads 2009
— 98 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The variations are cyclical and repetitive year after year under normal conditions. The traffic flows
at a site can be regarded as a time series, with variations occurring as a result of the combination
of random, cyclic and trend effects. Some of the variations are due to:
ƒ Daily variations – stem from hour by hour changes in levels of traffic demand. Figure A 6
shows typical examples for urban and rural arterial roads. Distinct peaks may be observed,
corresponding to peak levels of travel-related activities by people. In rural areas a single
peak in mid-afternoon is common, while urban roads tend to have two peaks (morning and
evening). Urban roads also tend to show directional differences in flows.
ƒ Weekly variations – especially differences between weekday and weekend, may be
observed as in Figure A 7. Typical weekday flows on arterial roads in urban areas are higher
than weekend flows, while the rural variations are less well defined and may reflect the
influences of recreational traffic.
ƒ Seasonal variations – tend to be fairly consistent for a given type of route and location. This
observation forms the basis of the methods for estimating AADT and DHV from weekday
counts (Section A.4.3). Figure A 8 indicates typical patterns of variations over the months of
the year. Urban roads generally show small variations, whereas rural roads may show
significant changes, usually as a function of the level of recreational and vacation traffic.
ƒ Trend effects – arise from the changes in the general levels of traffic activity at a site over an
extended period, usually as a reflection of changes in land use, population and economic
activity in a region.

Figure A 6: Typical hour-by-hour variations in two-way flow on arterial roads

Austroads 2009
— 99 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure A 7: Typical variations in traffic flow over days of the week on arterial roads

Figure A 8: Typical variations in traffic flow on arterial roads over months of the year

A.4.2 Counting Stations

Roads having similar traffic patterns may be grouped together, and different types of counting
stations used to monitor traffic volumes in each group. These groups can remain substantially
intact for several years, even though the AADT values on different road sections within the pattern
group may vary over wide limits.

Counting stations are categorised as Pattern Stations or Short-Term Stations according to whether
counting proceeds long enough to establish seasonal patterns or not.

Pattern Stations
Pattern Stations may be either:
ƒ Permanent Stations, which are monitored continuously, or
ƒ Seasonal Stations, which may be continuously counted or frequently sampled, but usually in
a particular locality for twelve months.

Austroads 2009
— 100 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

At permanent or continuously counted Seasonal Stations, the AADT may be calculated directly.
This represents a measured value. Where less than a full year’s data is available, techniques for
synthesising missing data are used (CSIRO 2000). However, at Seasonal Stations where counting
is not continuous, AADT is estimated by using appropriate adjustment factors. The adjustment
factor for a particular Seasonal Station is obtained by comparing its output with that from a
Permanent Station, which exhibits a similar pattern.

It should be noted that seasonal volume variations are much more closely related to climatic and
geographic characteristics than to AADT. Consequently, the application of a seasonal variation
factor derived at a Permanent Station in the same AADT stratum, but in a different geographic
area, is less likely to yield an accurate result than if the Permanent Station were similarly located
geographically.

Short Term (Coverage Count) Stations

At Short Term Stations, traffic volumes are measured for only a brief period (usually less than a
week), once during a survey period. As at a Seasonal Station, AADT may be estimated by
multiplying the ADT obtained at the Short Term Station by a Seasonal Adjustment Factor derived
at a suitable Permanent Station (Section A.4.3).

The adequacy of this approach depends upon the assumption that each Short-Term Station is well
represented by the Permanent Station chosen. The selection of Pattern Station locations is
therefore very important (Transfund New Zealand 2001). If the choice is strictly random, it is
unlikely that adequate coverage will be obtained. There is also a danger that permanent sites,
placed at positions of known congestion or other vital interest, may not be indicative of the traffic
pattern in the vicinity.

To increase the confidence in the pattern information gained, it is recommended that the collection
of data at all Short-Term Stations in a pattern group be made at the same time as the related
Seasonal Station is being sampled. This should be possible with careful deployment of available
short-term count equipment and staff, provided the number of stations in each pattern group is not
excessive. An example is given in Section A.4.3 to illustrate the error in estimating AADT using
short-term counts.

Location and Density of Counting Stations

The accuracy of an AADT estimate depends partly on the reliability of grouping Short-Term
Stations with Pattern Stations, and the method of grouping is important. Areas with similar
economy, culture and development tend to show similar traffic patterns. The most effective
grouping is therefore generally by geographical region as already mentioned.

Pattern Stations are best located by randomly selecting the number of sites on roads from each of
a number of AADT strata, within each geographical region. As a general rule, the density of
stations in each stratum should be proportional to the product of the total length of road and the
square root of the mean AADT in that stratum. This will provide a weighting factor in favour of low-
volume roads, which tend to exhibit more daily variation in traffic volumes.

Usually, the number of Short-Term Stations is determined by the extent of the road system and the
availability of funds. It is usual to count on each leg of all major intersections, and at identifiable
sites remote from major intersections. The variations in AADT at these sites are expected to differ
from those measured at major intersections.

Austroads 2009
— 101 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The number of Pattern Stations, and the ratio of Pattern Stations to Short-Term Stations, cannot be
rationally assessed in advance, because the assessment will depend on the regional grouping and
the reliability of matched patterns. The achievement of consistent and specified accuracy levels in
AADT estimation is essentially an iterative process. Table A 4 can be a guide on the densities of
Pattern Stations (NAASRA 1982).

Table A 4: Densities of pattern stations

Counting station density (km/station)
Type of road Pattern stations
Short-term stations
Permanent* Seasonal
Rural
Freeways, arterials 100-200 200-300 Up to 25 (usually 13-17)
Local Roads 5000-9000 1000-2000 Short term stations may
be characterised by one
Pattern Station
Urban
Freeways, arterials 20-50 30-50
Collectors, distributors 100-150 50-100
Local roads >1000 4000-6000
* lower figures (or higher densities) applicable if relatively few Seasonal Stations used.

A.4.3 Estimation of AADT

To measure the AADT on a particular road segment, it is necessary to locate a counting station on
that segment for a year, then divide the total counted traffic by the number of days in the year.

If traffic were counted on a random day in the year, the result would only approximate the AADT.
The accuracy of the approximation would depend on the weekly and seasonal pattern for the road
segment, the day of the year the count was made, and on the ADT counted because daily
variations tend to decrease with increasing traffic volumes.

The longer the period during which traffic is counted, the closer the ADT obtained will approximate
the AADT. For example, the ADT derived from a seven-day count will tend to be a more accurate
estimate of AADT than a one-day count. Similarly, the ADT derived from several short-term counts
throughout the year will more closely approximate the AADT, especially if the particular road
segment exhibits a high seasonal variation. A typical relationship between count duration, ADT
and estimate accuracy is illustrated in Figure A 9 (NAASRA 1982; Michael 1976). Where a broad
counting program has been established and seasonal patterns identified, the AADT at a particular
location may be estimated by multiplying a sample count (say, two to six days duration) by the
seasonal adjustment factor derived from a Pattern Station representative of the required location
as follows:

(AADT)j = ADTij ˜(SAF)i,k

where

(AADT)j = the AADT at the required location j

(ADT)ij = the sample count in the season (month, week, etc.) i at the location j

(SAF)i,k = = the seasonal adjustment factor for the season (month, week, etc.) i at a
Pattern Station, k, representative of the required location j.

Austroads 2009
— 102 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure A 9: Expected percentage error for AADT from ADT and count durations (n-day counts)
for a 75% confidence level

An example below illustrates the estimation of AADT and the error due to count duration using
Figure A 9.

Let the ADT obtained from six-day counts at a Short Term Station X be 6335 vehicles. Let the
ADT obtained from six-day counts at a Permanent Station Y be 13,221 vehicles. Let the AADT
obtained from one-year counts at a Permanent Station Y be 14,176 vehicles.

Then the seasonal adjustment factor = AADT/ADT for Permanent Station Y

= 14,716/13,221 = 1.113

AADT for Short Term Station X = ADT x seasonal adjustment factor

= 6,335 x 1.113 = 7,051 vehicles

Error due to the count duration using 6-day counts (Figure A 9) is given by:
E6 = (11.5 x 6,335 + 707)/6,335

= 11.6% at 75% confidence level

Absolute error from count duration

= ± 11.6% x 7051

= ± 818 vehicles.

The accuracy of an AADT estimate therefore depends on the accuracies of both the short-term
count and the seasonal adjustment factor. The latter depends on the accuracy with which the
Pattern Station reflects the fluctuations at the Short Term Stations. The estimation of error in
seasonal adjustment factor and its combination with the error due to count duration are described
in Appendix J.

Austroads 2009
— 103 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Transfund New Zealand (2001) provides the factors to convert short-term counts to estimates of
AADT. These factors are categorised by road types, day-of-week, time periods of day (part-days)
and vehicle types. The day and part-day conversion factors are shown in Table A 5 for Auckland
and non-Auckland roads in New Zealand. The validity of these factors for other places has yet to
be established, and the error for AADT estimates using part-day factors can be greater than 30%.

Some examples of using Table A 5 are as follows:

ƒ if the short-term day count on a Wednesday on a rural urban fringe site is 10,000 vehicles
per 24 hour, then the estimated AADT is 1.10 x 10,000 = 11,000 vehicles
ƒ if the short-term part-day (7 am – 9 am) count on a Wednesday at the same rural urban
fringe site is 1300 vehicles per two hours, then the estimated AADT is 8.61 x 1,300 = 11,193
vehicles.

Table A 5: Some conversions factors from short term counts to AADT estimates
Road types Day factors
Mon Tues Wed Thu Fri Sat Sun
Urban arterial 1 (Auckland) 0.98 0.95 0.93 0.92 0.88 1.17 1.37
Urban arterial 1 (non-Auckland) 1.01 0.98 0.95 0.94 0.89 1.09 1.24
Urban arterial 2 (Auckland) 0.99 0.96 0.95 0.93 0.89 1.13 1.31
Urban arterial 2 (non-Auckland) 1.00 0.97 0.94 0.93 0.88 1.11 1.30
Urban CBD (Auckland) 1.02 1.00 0.97 0.94 0.86 1.07 1.34
Urban industrial (Auckland) 0.85 0.85 0.84 0.84 0.84 1.84 2.66
Rural urban fringe 1.14 1.14 1.10 1.07 1.07 0.94 0.86
Rural strategic 1 1.05 1.01 0.99 0.97 0.97 1.09 1.10
Rural strategic 2 1.11 1.14 1.10 1.05 1.05 1.03 0.91
Rural recreation summer 1.07 1.17 1.13 1.05 1.05 1.11 0.88
Rural recreation winter 1.15 1.25 1.21 1.12 1.12 1.03 0.82
Road types Part-day factors for typical Monday – Thursday
7-9am 9am-12pm 1-4pm 4-6pm
Urban arterial (Auckland) 5.63 5.66 5.10 5.78
Urban arterial (non-Auckland) 7.31 5.17 4.60 6.00
Urban CBD (Auckland) 7.86 5.77 5.22 6.36
Urban industrial (Auckland) 5.57 4.13 3.72 5.97
Rural urban fringe 8.61 5.59 5.49 6.64
Rural strategic 1 8.40 4.74 4.41 5.99
Rural strategic 2 10.2 5.17 4.89 6.88
Rural recreation summer 16.9 4.84 4.08 7.77
Rural recreation winter 13.8 5.24 4.76 7.91
Source: Transfund NZ (2001).

Austroads 2009
— 104 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

A.4.4 Estimation of Design Hourly Volume from AADT

Choice of Design Hourly Volume (DHV)
When designing a road a balance must be made between the construction cost and the level of
service. The objective of the designer is generally to achieve the desired level of service at
acceptable costs. Traffic demand usually varies widely in a day and throughout a year, and it
would be uneconomic to design a road for the maximum hourly volume that could be expected.
Instead, a lower volume is chosen which will be exceeded for a particular number of hours during
the year. During this time congestion may be noticeable, and a lower level of service experienced
by motorists.

The 30th highest hourly volume (denoted as 30 HV) is often used in designing rural roads, taking
into consideration the traffic growth over a design period. Figure A 10 illustrates the relationship
between hourly volume and annual traffic for various types of roads. If a DHV higher than the
30 HV is chosen, a considerable increase in road construction costs will be incurred for a relatively
small decrease in the number of congested hours of operation. On the other hand, choosing a
DHV less than a 50 HV saves relatively little in expenditure per extra congested hour.
Nevertheless, the 30 HV may be inappropriately high for predominantly recreational routes, and an
80 HV or 120 HV may be chosen. In these cases the choice of DHV often needs special economic
consideration.

Figure A 10: Typical relationships between hourly volumes and AADT

Determination of Design Hourly Volume

If hourly volumes are available for the whole year, the n HV can be found directly by sorting the
data. Similarly, if intermittent counting has been adopted (say, one week in four), a value can be
readily obtained, although with less accuracy.

To determine the n HV for a Short Term (Coverage) Count Station, where little or no hourly data is
available for the whole of a year, it is necessary to obtain a relationship between n HV and AADT.
The simplest method is to select a representative Pattern Station (Permanent or Seasonal) where
full hourly data is available, and use the ratio of n HV/AADT from that location for the Short Term
Station. In practice, a number of Short-Term Stations are often allocated to a particular
representative Pattern (Permanent or Seasonal) Station. Where a number of Pattern Stations are
available in the one pattern group, regression equations may be used to get n HV from AADT.

Austroads 2009
— 105 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

On urban roads that are subject to pronounced peaks, it may not be appropriate to establish
capacity analysis or design on a full peak-hour flow. This is because higher flows for shorter
periods (e.g. 15 minutes) may result in unacceptable congestion for even a short period. The peak
15 minutes flow rate, converted to an equivalent hourly rate, may be used in such situations.
Alternatively, a peak-hour factor (PHF) can be specified for the site and the full hour flow is divided
by PHF to give a higher design volume to cater for possible heavier congestion.

Further details of the method of estimating DHV can be found in Vaughan (1968).

A.4.5 Estimation of Vehicle Kilometres of Travel (VKT)

The total daily travel (the daily VKT) on all segments in a road network is the sum of the product of
AADT on each segment and the segment length. The yearly VKT or total yearly travel is therefore
the total daily travel multiplied by the number of days in that year (365 or 366 days), and care must
be exercised in comparing yearly VKT. Measurements of VKT and trends in VKT are required for
Australia as a whole, for states and territories, or for regions within states, e.g. capital cities, major
provincial cities and the remaining rural areas. Within these regions, VKT may be required by the
class of road, individual road or road section.

VKT may be estimated by determining the mean AADT of a sample of road segments representing
the system under consideration, and then multiplying this by the total length of roads in the system.
The accuracy of this estimate depends on the sample size and the dispersion of AADT in the
network. If the distribution is skewed, the median AADT is more appropriate for estimating VKT.

Stratification
The AADT dispersion may be reduced by restricting the group to road segments carrying a limited
range of AADT. These groups, or ‘strata’, will each include a relatively large number of segments,
which can be regarded as a ‘population’ for statistical sampling. The relative proportions of roads
within various traffic volume strata will differ in various regions throughout a state or an area.
Stratification between urban and rural areas is of prime importance, and is also desirable within
urban and rural regions because of differences in economy and development.

Most road segments carry relatively low volumes. Strata limits should be adjusted to provide
similar amounts of travel in each stratum, and the strata limits therefore show a bias towards low
volume roads. Table A 6 provides some stratification for urban and rural roads.

Stratification by AADT volume requires a preliminary estimate of the AADT on each segment so
that it can be allocated to a group. Existing data may be used for this purpose.

Stratification can also be by traffic patterns, which are dependent on economic-geographical

factors. Foldvary (1962) noted that the areas of similar traffic patterns agreed reasonably well with
Statistical Divisions in NSW, and gave an explanation on using both AADT and traffic patterns for
two-dimensional random sampling.

Austroads 2009
— 106 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table A 6: Suggested volume stratification of road segments

Stratum Suggested range for urban Suggested range for rural
roads (AADT) roads (AADT)
1 0-300 0-100
2 301-1,100 101-300
3 1,101-2,000 301-700
4 2,001-4,000 701-1,100
5 4,001-7,000 1,101-2,000
6 7,001-16,000 2,001-4,000
7 16,001-20,000 4,001-7,000
8 20,001-35,000 7,000 plus
9 35,001-50,000
10 50,000 plus

Number of AADT Observations

The recommended method of estimating the number of AADT observations required to calculate
VKTij in any region i and in stratum j is as follows:
Calculate the following estimates from the most recent available data:
VKTi = amount of travel in region i

ni
= ¦ VKT
j 1
ij

SDij = standard deviation of the AADT observations in region i and stratum j

Lij = total length of road in region i and stratum j

Ni = the number of strata in region i.

Calculate the required number of AADT observations in region i and stratum j using the following
expression (NAASRA 1982):
ni
L ij SD ij
Nij
(0.05VKTi ) 2
¦ L SD
j 1
ij ij

where
Nij = the required number of AADT observations in region i and stratum j

An example calculation to illustrate the procedure is shown in

Austroads 2009
— 107 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table A 7 with eight strata (ni = 8). The site is on a rural road. The estimates of VKTij, Lij and SDij
are from existing inventory data in region i.

Austroads 2009
— 108 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table A 7: Calculation of the Number of VKT Stations

Inventory data in region i
AADT stratum Lij x SD ijI No, of stations
VKT ij L ij SD ij
j = 1 to 8 (veh-km) N ij
(x103 veh-km) (km) (veh)_
0-100 2586 94821 29 2,749,809 14.07 (14)
101-300 2750 15275 60 916,500 4.7 (5)
301-700 3277 6931 111 769,341 3.94 (4)
701-1100 2409 2699 110 296,890 1.52 (2)
1101-2000 3496 2310 279 644,490 3.30 (4)
2001-4000 4013 1428 543 775,404 3.97 (4)
4001-7000 2574 496 848 420,608 2.15 (3)
7000 plus 2703 275 2459 676,225 3.46 (4)
Sum 23808 - - 7,249,267 40

The total VKTi for this region i using eight strata

8
= ¦ VKT
j 1
ij = 23,808 x 103 veh-km

The summation
8
= ¦ L SD
j 1
ij ij = 7,249,267 veh-km

The number of counting stations for each stratum (Nij) is calculated using for example, Li1= 94,821
km and SDi1 = 29 veh for j=1 and
94,821 x 29
Nij u 7,249,267 = 14.07 (or 14) stations
(0.05 u 23,808 x 10 3 ) 2

Location and Density of AADT Observations

The methods of estimating AADT are as discussed in Section A.4.3. The two main sources of
error in estimating AADT, instrument error and seasonal factors, do not have a significant effect on
the accuracy of VKT for a region provided an adequate sample of AADT observations is obtained.

Some member authorities have counting systems designed to estimate AADT. Where these AADT
estimates are sufficient in number, and meet the stipulated accuracy levels of Table A 2, it may be
possible to use these as AADT observations for VKT estimation. This can be checked by
comparing the mean and standard deviation of the AADT estimates for all roads in the stratum,
making use of existing traffic volume data in a member authority. If there is a significant difference,
it will be necessary to include additional ADT observation points.

Testing of this has shown that problems may be experienced in the lower volume categories,
particularly in urban areas. Some extra counting may have to be arranged in this area. Short-term
counts on local government roads adjusted appropriately could be used, and estimation techniques
based on housing density or other parameters might also be useful.

Austroads 2009
— 109 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

There is a need for further research in this area. Where no traffic counting is currently being
undertaken to the required degree of accuracy, observation points should be randomly selected.
The recommended procedure would be to divide the length of road in each stratum into equal
intervals, the number of such intervals equalling the number of AADT observations required. The
observation points could then be located at the same relative position within each interval.
Software tools such as the geographical information systems (GIS) are available to facilitate this
task.

Trends in VKT
Trends relate to extended periods, and are accordingly less vulnerable to inaccuracies than the
values of the characteristics to which they refer. Consequently, the number of measurements
necessary to establish values of VKT, with the required accuracy, will also be adequate to provide
the accuracy for trends, as prescribed in Table A 2. NAASRA (1982) includes the details of the
method of calculating the accuracy of trends in VKT.

A.5 Vehicle Classification

Vehicle classification data are important for all transport engineering applications. Comprehensive
classification data are now possible using new technologies. Vehicle classification surveys are
concerned with the distribution of traffic data into different types of vehicles. Traffic engineers are
interested in the composition of a traffic stream, for example, in terms of flows of cars, heavy
vehicles, motorcycles and bicycles for the planning, operation and assessment of a traffic system.

A.5.1 Needs for Vehicle Classification data

Six broad study areas may be identified in which vehicle classification data are essential. These
study areas include:
ƒ design and performance of road pavements and bridges
ƒ traffic capacity and operations analysis
ƒ vehicle categories for traffic legislation and regulation
ƒ road safety studies
ƒ parking system design
ƒ economic studies.

The first two items above are of immediate concern to the traffic engineer, and brief descriptions of
the needs for classification data in these two areas are given below. Figure A 11 also provides a
framework on the various uses of classified counts (Vincent 1986).

Austroads 2009
— 110 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Source: Vincent (1986).

Figure A 11: Usage of vehicle classification data in transport engineering

Road Pavements and Bridges

The total loads borne by a road element, and the distribution and frequency of individual loads are
important in predicting and assessing the performance of that element. The structural design of
pavements and bridges is a primary use for vehicle classification data. Structural design requires
data on the distribution of axle loads, the frequency of axle loads, and the configuration of axles
(hence vehicle types). Note that road pavement design and thereby costs are sensitive to the
presence of outliers in the patterns of axle groups taken as a traffic stream, i.e. not necessarily
individual vehicles but groupings of vehicles. Traditionally, axle load data have been collected as a
by-product of legal load enforcement procedures. Present needs require direct attention to the
actual loads using a particular facility such as a bridge, and weigh-in-motion technology is available
to efficiently obtain this data (Appendix G).

Traffic Capacity and Operations

The ability of a road system to cope with the demands imposed on it depends to a large degree on
the dimensions and performance characteristics of the vehicles using it. Low powered and/or large
vehicles may find difficulty in climbing grades, thereby reducing the level of service for a given
traffic flow, unless special provision is made to allow other vehicles to overtake. Standards for
intersection design and channelisation, and road geometry, are largely defined in terms of the
ability of a design vehicle to negotiate curves and corners, and comply with traffic regulations
without demolishing roadside furniture. The design vehicle is usually taken as one having a swept
path envelope representing at least 85% of the population of that type of vehicle. Estimates of
travel and operating costs for different vehicle classes are necessary in economic studies of road
transport needs.

Austroads 2009
— 111 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

A.5.2 Vehicle Classification Systems

A number of classification systems are currently in use. A system that groups vehicles into three
categories is illustrated in Figure A 11. This involves the automatic classification of vehicle lengths
using inductive loop sensors. The system identifies distinct break points in vehicle length
distributions for cars, rigid trucks and articulated vehicles. Figure A 12 shows the distribution of
vehicle lengths in these three broad classes, indicating the break between cars and rigid trucks
occurring at a length of 5.5 m, and between rigid and articulated trucks at 12.0 m. Overlaps or
uncertainties always exist at each break point.

Source: Vincent (1986).

Figure A 12: Length distributions of vehicles on Victorian highways in 1985

Table A 8 shows the current Austroads vehicle classification system that was updated in 1994.
This system provides for three levels of vehicle classification – Level 1 by overall vehicle length,
Level 2 by number of axles, and Level 3 by axle configuration and vehicle type. It is illustrated in
Figure A 13. The system allows flexibility for any user to aggregate vehicle classes in various
ways to suit particular purposes and economies of data collection.

The classification data can be used in conjunction with axle load data from strategically located
weigh-in-motion sites to develop load factors. Heavy vehicle counts can be converted to
equivalent standard axles (ESA) for road pavement design (Appendix G). Three-bin and four-bin
systems are being used for classification of vehicles by length. At present, the RTA Incident
Management System employs a three-bin system and VicRoads employs a four-bin system, both
of which are different from the Austroads five-bin system classified by vehicle length. The
Australian Bureau of Statistics provides data related to heavy vehicle activity using different
classification categories (Appendix G).

Austroads 2009
— 112 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table A 8: Austroads vehicle classification systems (updated in 1994)

Level 1 Level 2 Level 3
Austroads
Length Axles and classification
Vehicle type
(indicative) axle groups
Type Axles Groups Description Class Parameters
LIGHT VEHICLES
Short Short
Up to 5.5 m Sedan, wagon, 4WD, d1 d 3.2 m
2 1 or 2 1
utility, light van, bicycle, and axles = 2
motorcycle, etc.
groups = 3,
Short – towing 2.1 m d d1 d 3.2 m
3, 4 or 5 3 2
trailer, caravan, boat, etc. d2 t 2.1 m,
and axles = 3, 4 or 5
HEAVY VEHICLES
Medium
5.5 m to d1 > 3.2 m
2 2 Two axle truck or bus 3
14.5 m and axles = 2
axles = 3
3 2 Three axle truck or bus 4
and groups = 2
axles > 3
>3 2 Four axle truck 5
and groups = 2
Three axle articulated or d1 > 3.2 m,
3 3 6
rigid vehicle and trailer axles = 3 and groups = 3
d2 < 2.1 m,
Four axle articulated or
4 >2 7 or d1 < 2.1 or d1 > 3.2 m
Long rigid vehicle and trailer
axles = 4 and groups > 2
11.5 m to
19.0 m d2 < 2.1 m,
Five axle articulated or
5 >2 8 or d1 < 2.1 or d1 > 3.2 m
rigid vehicle and trailer
axles = 5 and groups > 2
6 >2 Six axle (or more) articulated or axles = 6 and groups > 2; or
9
>6 3 rigid vehicle and trailer axles > 6 and groups = 3
Medium B Double or axles > 6
>6 4 10
combination heavy truck and trailer and groups = 4
17.5 m to axles > 6
Double road train or
36.5 m >6 5 or 6 11
heavy truck and two trailers and groups = 5 or 6
Long axles > 6
Triple road train or
combination >6 >6 12
heavy truck and three trailers and groups > 6
over 33 m

Definitions:
Group: (axle group) – where adjacent axles are less than 2.1 m apart
Groups: number of axle groups
Axles: number of axles (maximum axle spacing of 10 m)
d1: distance between first and second axle
d2: distance between second and third axle.

Austroads 2009
— 113 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure A 13: Representative vehicles in Austroads 12-bin classification system

A.5.3 Methods of Collecting Vehicle Classification Data

Manual vehicle classification methods, based on either vehicle body type (e.g. surveys by the
Australian Bureau of Statistics) or axle configurations (e.g. Austroads), have been used for many
years. Manual methods are now largely confined to intersection turning movement counts. As
these surveys require considerable human resources, they are costly and generally limited to short
period counts – generally up to 12 hours duration.

Austroads 2009
— 114 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Automatic vehicle classifiers are now readily available. They usually use a pair of axle sensors for
classification by axle configurations or a pair of inductive loops for classification by vehicle lengths
(Section A.1). More accurate classified counts are possible by using an extra sensor, e.g. two
loops plus an axle sensor. However, uncertainties are present in vehicle counting and
classification, especially under congested flow conditions. Modern vehicle classifiers are
programmable to suit different classification systems and are available at low cost. They should
replace simple vehicle or axle-pair counters when these are due for replacement.

In recent years, electronic tolling systems have employed video imaging for law enforcement and
charging tolls. Numberplates and vehicle profiles are used for vehicle identification and
classification. Video imaging is therefore a potentially useful vehicle classification technology with
a high accuracy (up to 99.5%). It is expected to be more commonly employed as the cost of
implementing the technology decreases.

A new non-intrusive technology based on intersecting infra-red beams crossing a carriageway is

gaining extensive use.

Research on corrective software to improve the accuracy of vehicle classification counts (using
loops or axles) has progressed well. The NSW Roads and Traffic Authority has employed such
software, which needs to be calibrated against sample video clips of traffic flow for vehicle
classification surveys.

A.6 Data Analysis and Presentation

Traffic data should be presented in ways that make them easily and quickly understood, by people
who are unfamiliar with the methods used to collect data. Large volumes of precise data must
generally be presented as numerical tabulations. Graphical presentations permit rapid visual
interpretation, particularly for variations in time and space, and are sufficiently accurate for most
purposes. Modern computer graphics systems can readily generate time profiles and flow maps.

It is important to precisely record and identify the location of counting sites. This may be done
either on the basis of a permanent reference system devised for general inventory purposes, by
verbal description, or by making a locality plan. The availability of GPS devices and GIS software
has greatly facilitated data presentation. Unusual variations from such causes as extremes of
weather, public holidays, crashes or the like should be recorded.

A.6.1 Estimated AADT

Estimated AADT profiles for roads and highways are normally obtained from short-term counts,
using the methods outlined in Section A.4.3. They represent the bulk of published data,
particularly for rural areas. A common method of presentation is to show the observed AADT at
each counting station on the map. Figure A 14 shows a typical AADT map. Alternatively, flow
bands proportional to the AADT volumes are drawn along the surveyed routes.

Austroads 2009
— 115 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Source: DTEI, South Australia.

Figure A 14: Typical map presentation of traffic count data

Austroads 2009
— 116 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Another method is to plot AADT profiles for only selected routes, and is particularly useful in urban
areas where there may be many routes close together so that flow bands overlap. Here, the route
is represented schematically by a straight line (the abscissa), and the AADT value for each route
segment is shown as an ordinate. Figure A 15 shows an example.

Source: VicRoads.

Figure A 15: Typical AADT profile for a route

The duration of the counts and the adjustment factors used in the estimation of AADT may be of
interest to those familiar with traffic counting methods. These data are usually presented in tabular
form.

A.6.2 Seasonal Patterns and Trends

Pattern Stations provide information on daily and seasonal variations, while Permanent Stations
also provide indications of trends.

Graphical and tabular presentations may be used to show monthly, weekly and daily variation
patterns, often illustrated as proportions of average values. Graphs showing such variations
arranged in decreasing order are also useful, as they show information on variations about the
mean.

A typical Pattern Station report is shown in Figure A 16.

Austroads 2009
— 117 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Source: Queensland Department of Main Roads.

Figure A 16: Typical Pattern Station report

A.6.3 Turning Movements at Intersections

Information on turning movements is usually obtained from manual counts over relatively short
periods, typically the peak hour or over 12 hours from 7.00 am to 7.00 pm. These data are
essential for intersection and traffic signal design. Turning movement counts may be segregated
into two vehicle classes – cars and heavy vehicles – and these may be displayed, each as a
number of cars and heavy vehicle component or as a percentage of the particular total traffic
movement. Figure A 17 shows a typical example.

Austroads 2009
— 118 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Source: Queensland Department of Main Roads.

Figure A 17: Typical intersection count report

Austroads 2009
— 119 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX B SPEED SURVEYS

B.1 Definitions
There are three speed classifications that are of interest in traffic engineering:
ƒ spot speed – the instantaneous speed of a vehicle at a specified point on a road
ƒ journey or space speed – the effective speed of a vehicle on a trip between two points
obtained from the ratio of distance travelled to the time taken
ƒ running speed – the average speed over a trip while the vehicle is moving, given by the
distance travelled divided by the time the vehicle is in motion.

A comparison of journey speed and running speed for a road provides a measure of the
congestion on that road.

The average speed of a traffic stream can be defined in two ways:

ƒ time mean speed – the arithmetic mean of the measured spot speeds of all vehicles passing
a fixed roadside point during a given time interval
ƒ space mean speed – the harmonic mean of the measured speeds of all vehicles in the
stream which are within a specified length of roadway at a given instant of time.

For a given time period, a roadside observer tends to see more cars that are at higher speeds and
hence spot speeds are biased in favour of the faster vehicles on the road. The time mean speed is
therefore slightly greater (typically two to three per cent higher) than the space mean speed. The
relationship between time mean speed and space mean speed is described in Part 2 Guide to
Traffic Management (Austroads 2008a) and in other texts such as Taylor et al. (2000).

Space or journey speed belongs more properly to travel time investigations (journey speed is the
inverse of travel time per unit distance) and is discussed in detail in Appendix C. This section
concentrates on spot speeds, their measurement, analysis and uses. For both spot speeds and
space speeds, further disaggregation of the concept of speed is possible on the basis of driver
behaviour. A common system is:
ƒ actual speeds – the speeds that would be observed for all vehicles on a road
ƒ desired speeds – the hypothesised speeds which drivers will adopt when free from all other
influences
ƒ free speeds – the observable part of the total population of desired speeds.

B.1.1 Speeds on Curves

Advisory speeds are the desirable speeds for horizontal curves, intersections or other locations
where design standards or physical conditions restrict safe operating speeds to a value below the
posted regulatory speed limit on a roadway.

The speed at which a driver can negotiate a horizontal curve is related to the superelevation of the
pavement, and the friction developed between the tyres and the road. An equation based on the
acceleration of a body travelling a circular path applies. The acceleration is measured away from
the centre of the path and is given by:

Austroads 2009
— 120 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

v2
(e  f )g
R
where

v = speed

R = radius

e = superelevation

f = friction coefficient

g = acceleration due to gravity

For convenience this equation can be expressed as:

v 127R( e  f )

where
v = speed (km/h)
R = radius (m)

If a driver attempts to negotiate the corner faster than the speed calculated there is a greater risk of
a crash. Drivers would typically prefer to use a friction coefficient, which is significantly less than
the maximum possible value at a particular site.

Consider a rural road with R = 166 m, e = 0.04 and f = 0.16. The advisory speed is therefore
65 km/h.

AS 1742.4 describes the use of the ball-bank indicator to establish appropriate advisory speeds on
curves.

Variable speed limits and warning displays have been used by member authorities to match traffic
and environmental conditions, including fog and icy pavement conditions. It is known from
experience in countries with foggy weather conditions that displaying lower speeds on motorways
when warranted can reduce road crashes, especially those that occur after the initial or primary
crashes.

B.2 Methods of Speed Data Collection

Measurements of spot speeds are generally made from a specific location on the road. Various
approaches may be used to collect spot speed data:
ƒ methods involving timing
ƒ microwave radar using the Doppler effect
ƒ direct measurement using laser gun
ƒ methods involving video
ƒ global positioning system.

Austroads 2009
— 121 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

B.2.1 Methods Involving Timing

The increasing availability of electronic time and data recorders has meant that manual timing of
vehicles using a stopwatch is now used only as a last resort. The passage time of a vehicle
between two detectors, a measured distance apart, can easily be recorded. Detectors can be
placed close together (a few metres apart) as high levels of accuracy are obtainable. Thus a more
representative spot speed can be measured over that short distance and acceleration and
deceleration effects will be minimal. Data are recorded and can be uploaded at regular intervals
either on site or remotely. Additional data, such as headway and volume data can also be
collected simultaneously.

Various detectors can be used in this type of system, as discussed in Section A.1.2. These include
pairs of pneumatic tubes, tribo and piezoelectric cables, switch tapes, inductive loops and photo-
electric or electromagnetic beams.

An example of such a measuring system is the Incident Detection Station (ARRB Group Ltd
2003a) (Figure B 1). This data logger will collect data from up to 32 lanes. It can easily provide
accurate information from 2400 vehicles per lane per hour, and replaces the earlier generations of
the Vehicle Detection Data Acquisition System (VDDAS). 

Figure B 1: The Incident Detection Station

B.2.2 Microwave Radar Gun Using Doppler Effect

The earlier generations of microwave radar guns for speed measurements make use of the
Doppler effect. A microwave beam is sent to the target vehicle, which reflects a signal back to the
receiver in the radar gun.

The moving vehicle affects the frequency of the returned signal. The shift in the frequencies
between the emitted and received microwave signals is called the Doppler effect. By measuring
the amount of frequency shift and the duration of the time interval, the speed of the targeted
vehicle can be determined. A microwave radar gun has a wide cone of detection, which is about
70 m at a range of 300 m. It cannot be relied upon to give an accurate speed measurement of a
particular car, and has been replaced by laser devices for law enforcement.

B.2.3 Direct Measurement Using Laser Guns

The laser infra-red gun uses the higher optical frequency and has a small detection cone of about
1 m in diameter at a distance of 300 m between the laser gun and the targeted vehicle. An
example of a hand-held laser gun is shown in Figure B 2.

Austroads 2009
— 122 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure B 2: An example of a laser speed measurement device

The equipment employs a direct method because it relies on the measurement of the round-trip
time of the infra-red light beam to reach a vehicle and be reflected back. The gun can accurately
count the number of nanoseconds the light takes for the round trip, and making use of the speed of
light at 300,000 km/s, several samples of the distance are obtained in a fraction of a second. The
changes in distance and hence the velocity as the targeted vehicle moves can therefore be
monitored accurately.

Two types of error may be found when radar or laser speed meters are used to measure the
speeds of isolated vehicles:
ƒ equipment error, i.e. the rounding-off in the displayed speed; typically this is (-1, +0) km/h,
e.g. for an actual speed of 100 km/h, the meter reading would be between 99 and 100 km/h
ƒ angle error, which is related to the actual angle of incidence of the radar beam to the
direction of travel of the vehicle.

Figure B 3 shows a typical configuration. Unless the meter can be placed in the path of the
vehicle, there will be a finite angle Dbetween the beam and the vehicle path.

Figure B 3: Angle of incidence (D) of a radar beam to traffic

The speed reading from the meter is thus v. cos(D) where v is the actual vehicle speed. Typically
Dwill be quite small, so this error will also be small and is given by [100(1 – cos(D)) per cent].

Austroads 2009
— 123 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Note that this is a consistent error, and that the observed speed reading will always be less than v.
Some units have a built-in correction for angle error, usually for a predefined value of the angle D.
As the mass market for the units is law enforcement, the correction is often ignored. The rationale
for speed limit enforcement is that the observed speed will never exceed the actual speed of the
vehicle.

Both types of error mean that observed readings will always be slightly less than actual speeds (a
margin of leniency in speed limit enforcement). Because the laser equipment measures speed
directly (with a set absolute error in reading speed, depending on the set-up position of the unit),
the relative error will decrease as speed increases. This is the opposite result to that noted for
indirect speed measurements based on observed travel times.

B.2.4 Methods Involving Video

Video can be used to determine vehicle speeds and is becoming increasingly cheaper to use and
operate. The general method involves recording the distance moved by a vehicle in a short period
(perhaps a couple of frames), then computing the speed. An example is the Video Analysis Data
Acquisition System (VADAS) reported in Troutbeck and Dods (1986), which is now replaced by
new imaging equipment at ARRB (e.g. Roper, 2004). The advantages of using video include:
ƒ the provision of a complete, permanent record of the traffic flow, which can always be re-
analysed and re-examined at a later stage, and
ƒ additional information (e.g. vehicle classification, flows, headways, special phenomena,
overtaking, etc.) can be obtained.

Manual data extraction from a video recording is time-consuming, tedious and expensive, making
the technique not particularly useful for routine surveys. However, the continuing development of
automatic data extraction procedures should make vehicle speed data collection from video a cost-
effective alternative.

Video imaging techniques are used on freeways in various cities to extract speed data from fixed
position cameras. The operator defines virtual loops on a carriageway and superimposes them on
the video frames captured by the camera. The luminance of the marked area changes as one or
more vehicles travel over the area. These changes are detected and provide information such as
speed, volume, headway and occupancy. The accuracy of speed measurement using video is
affected by shadows of adjacent lane traffic and weather conditions, and is expected to be less
than that obtained using inductive loop sensors. Video techniques offer the significant advantage
of remote sensing and do not require cutting loops on road pavements – an important issue for
privately operated tolled roads.

B.2.5 Global Positioning System

Vehicles can be fitted with receiver units that pick up signals from the global positioning system
(GPS) satellite network (Figure B 4). GPS was the first satellite navigation system launched in
1978. It is technically known as NAVSTAR GPS (NAVigation Satellite with Timing And Ranging
Global Positioning System). The system is designed, maintained and operated by the US
Department of Defence. Among all the extraterrestrial positioning systems available today, GPS is
the most popular and widely used positioning system. It is a worldwide navigation tool that
provides three-dimensional position, velocity and time information under all weather conditions for
24 hours a day (Khoo and Luk 2002).

Austroads 2009
— 124 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure B 4: An example of a GPS receiver unit

The complete constellation of GPS consists of 21 satellites with three operational spares
circulating the earth in six orbital planes. The GPS was essentially set up for military purposes.
Civilians are allowed to use certain facilities of the full system.

Two techniques are used to determine distances or ranges. They are the code-based and the
phase-based measurements. The code-based measurement makes use of the Coarse Acquisition
code (C/A) and can achieve a maximum positioning accuracy of about 3 – 5 m. The phase-based
measurement attempts to improve this accuracy by considering further the information contained in
the phase of the carrier and is an active area of research.

Satellite positioning using one receiver is known as single-point positioning or simply point
positioning. This is the most basic usage of satellite systems for positioning and navigation. It is
widely used in recreational navigation like hiking, sailing and hunting. A point positioning accuracy
of about 10 m with a probability of 95% is now possible with code-based GPS. The accuracy of
single-point positioning is too low for serious navigation, surveying and other vehicle location
applications.

A relative or differential positioning technique is necessary to improve the accuracy. It involves the
use of two receivers, one stationary at a reference or base station and the other (called a rover)
moving or simply located at a place usually in the vicinity of the reference station. Both receivers
simultaneously track the same set of satellites. The errors common to both receivers can be
mitigated and, hence, the accuracy obtainable will be higher. Note that four satellites are needed
to fix a position with certainty because two circles representing the two ranges from two satellites
intersect at two points, and a third satellite is needed to give a third range to determine which of the
two intersecting points is the correct distance. Four satellites are therefore necessary to give a
correct three-dimensional position fix.

The accuracy of code-based differential GPS (DGPS) is about 2 – 3 m with a baseline distance
(i.e. range of coverage) of 100 – 200 km. A phase-measuring DGPS is expensive with a much
higher accuracy in the range 20 – 50 mm, but the baseline distance has been found to be less than
10 km.

The GPS receiver, whether in point processing or differential mode, keeps track of its position and
time. From time and position, it can also calculate its speed when in motion. Modern receivers
now make use of a combination of time, distance and frequency shift signals to calculate speeds
and other parameters. The speed values can be useful as an indicator of the road congestion
condition. In Singapore, such data are transmitted from GPS-equipped taxis to a traffic control
centre. The data are processed and then made available in real-time on the Internet as congestion
indicators (Luk and Yang 2001).

Austroads 2009
— 125 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

B.3 Sample Selection

Section 2 describes the processes for defining study objectives, selecting survey measures, and
selecting target populations. This section describes the procedures and methods for determining
an appropriate sample size for a proposed speed survey.

B.3.1 Selecting a Target Population

The vehicle speeds to be recorded, and hence the vehicles to be observed, must depend on the
specific objectives behind the study. This is of particular importance given the discussion of
different concepts of speeds (e.g. actual, desired or free speeds).

If actual speeds are of interest, measurements can be taken from all of the vehicles on a road.
With modern data loggers, large volumes of vehicles can be recorded with ease in comparison to
older manual recording methods.

Traditionally, free speed data have been collected by measuring the speeds of isolated vehicles
and bunch leaders. This will yield results biased in favour of the slower vehicles. McLean (1978)
introduced a method for correcting this bias on the assumption that free speeds are normally
distributed. Brilon (1977) suggested that the speeds of isolated vehicles will be biased in favour of
the faster vehicles, so that the desired speed distribution would lie somewhere between that for the
isolated vehicles alone and that for the free-moving vehicles (i.e. both isolated vehicles and bunch
leaders). Hence, Brilon’s method is to measure both the distributions of speeds of isolated
vehicles and bunch leaders, and to use these as upper and lower bounds on the distribution of free
speeds.

B.3.2 Behavioural Factors

Because of the enforcement of speed limits, drivers are naturally wary of any apparent speed
measuring system on or near the road. Detectors on the road surface may cause drivers to slow
down. Some drivers may have radar detection apparatus mounted in their vehicles. These
devices are now illegal in many places. Speed studies for traffic engineering purposes are
intended to provide information about the normal behaviour of drivers. Speed surveys should
therefore be conducted in as unobtrusive a manner as possible.

B.3.3 Sample Size for Speed Studies

Most investigations of vehicle speed relate to the range of speeds adopted by vehicles in the traffic
stream, rather than a single observation such as the mean speed. Thus, the overall distribution of
speeds is of interest, and samples must be selected so that useful data on the speed distribution
can be obtained.

Austroads 2009
— 126 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

For most purposes, the minimum sample size n required for a speed survey may be determined as
follows (Oppenlander et al. 1961):

V 2 s 2 (2  U 2 )
n
2d 2
where

V = normal deviate corresponding to the desired confidence level

s = standard deviation of the sample

U = normal deviate corresponding to the percentile being estimated

d = permitted error in the estimate.

A normal deviate is a random variable, which follows a normal distribution. For convenience,
values of V2(2+U2) are given in Table B 1 for common values of V and U.

Table B 1: Values of V2 (2 + U2) for use in sample size determination for speed surveys
Percentile being Desired Confidence Level (%)
estimated
90 95 99
15 8.3 11.8 20.5
50 5.4 7.7 13.3
85 8.3 11.8 20.5

For example, if d is defined as ±3 km/h, and the mean speed needs to be determined (i.e. 50th
percentile speed, assuming speed distribution is normal) at a 95% confidence level, then from
Table B 1, V2(2+U2) would be equal to 7.7 and therefore the minimum sample size n required to
determine the mean speed, is given by:
s 2 (7.7)
n
2(3 2 )

and therefore,
n = 0.428 s2

where the standard deviation of speeds s is in km/h.

According to McLean (1989), the standard deviation of free speeds on rural highways in Australia
is about 14% of the mean speed on the highway, for mean speeds of 90 km/h or more. For mean
free speeds less than 90 km/h, this percentage (the ‘coefficient of variation’ of the speed
distribution) decreases by about one per cent for every 10 km/h decrement in the mean. Thus, if
the mean free speed is 80 km/ h, the standard deviation s will be about 13% of 80 km/h or
10.4 km/h. This result is for all vehicles in a traffic stream. The standard deviation for trucks would
be somewhat less, perhaps 11% of the mean truck speed (in level terrain).

Austroads 2009
— 127 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

B.4 Data Analysis and Presentation of Results

Speed data typically need analysis of their distributions. Measures only of their central tendencies
(e.g. the mean speed) are seldom sufficient indicators of the overall speed situation. The extremes
of observed speeds set a range of conditions, which is useful in traffic design. For this reason the
results of a speed distribution are usually expressed in graphical terms either as a histogram or
(more commonly) as a cumulative distribution. Measures such as the mean, median and 85th
percentile speeds, and the standard deviations are then extracted from the distribution. A measure
such as the 15th percentile speed might be used to indicate difficulties for some vehicles in
maintaining travel speeds (e.g. for trucks climbing grades).

Figure B 5 shows a typical cumulative speed distribution curve. Median and 85th percentile
speeds are illustrated, and compared to the mean speed. This figure is the typical output from a
spot speed survey.

Figure B 5: Typical cumulative speed distribution and histogram from a spot speed survey

In before and after analysis, statistical tests are used to compare the observed distributions and
their parameters, in a search for significant differences that might indicate changes in driver
behaviour. Other variables (e.g. flow) must be held constant or multivariate analysis carried out.

B.5 Conclusions on Speed Surveys

Speed surveys are an important part of the traffic engineer’s investigations. Although they often
appear to be routine exercises, care is always needed in defining the types of speeds to be
observed and in selecting the experimental techniques to be used. There are definitional problems
to be resolved in deciding just what speed data are appropriate to a particular study objective.
Speed surveys are, therefore, prime examples for the application of the systems planning
approach to traffic studies outlined in Section 2. Speed data can be collected using both indirect
and direct methods, many utilising new technologies. The most suitable survey method for a
particular study will depend on many factors, including the study objectives, the characteristics of
the observation site, the data collection and analysis resources available, and the expected range
of speeds to be observed. Other data collection methods that focus on the variability of travel
times are available, and should also be considered. These methods are described in Appendix C.

Austroads 2009
— 128 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX C TRAVEL TIME, QUEUING AND DELAY

SURVEYS
All studies of travel times in traffic systems need to be approached carefully. There are many
parameters and definitions to be considered, and the nature of the underlying variable – the travel
times of individual flow units in the traffic – is complex. In particular, travel time variability is
recognised as an important parameter in travellers’ route choice decisions, and also provides
information on the congestion levels of a traffic network (Taylor 1982; Austroads 2001,
Section 6.5).

There are a large number of definitions and components in travel time investigations, and the
following definitions are commonly used:
ƒ free flow time – the time required by an unimpeded vehicle to traverse the survey section
ƒ free flow speed – the length of the survey section divided by the free flow time
ƒ travel time – the actual (observed) time taken to traverse the test section
ƒ delay – the difference between the travel time and the free flow travel time
ƒ stopped time – the period for which a vehicle is stationary while in the survey section
ƒ running time – the period of time for which a vehicle is in motion while in the survey section;
(total) travel time is then the sum of stopped time and running time
ƒ running speed – the total sectional distance divided by the running time; sometimes running
speed is used as an estimate of free flow speed.

C.1 Travel Time Surveys

Travel times may be observed in two different ways. The first way involves ‘moving observer’
methods where the observer drives a test car in the traffic stream and records its passage time.
The second way involves ‘stationary observer’ methods, where observers placed at fixed points in
the network record the times that a sample of vehicles take to travel the length of the survey
section. Table C 1 lists the various travel time survey methods described in this section, and also
indicates the data items that may be found using each method.

Austroads 2009
— 129 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Table C 1: Travel time survey methods

Basic approach Survey method Application Data obtained
Moving observer Floating car Long road sections Mean travel time, stopped time
and running time; number of
stops; flows
Chase car Long road sections Mean and variance of travel
times, stopped time and running
time; number of stops
Stationary observer Number plate Road sections, junctions, traffic Mean, variance and distribution of
control devices travel times and flows
Input-output Road sections, junctions, traffic Mean travel time and flow
control devices
Path trace Short road sections, junctions, Mean and variance of travel time,
traffic control devices free flow time, running time;
number of stops

C.1.1 Moving Observer Methods

Moving observer methods enable estimates of parameters such as mean travel times to be made
fairly easily and quickly. The basic resources required are a test vehicle, two people (driver and
observer), and a data recording system. Data recording may be by paper, pencil and stopwatch,
by hand-held or laptop computer, or by an instrumented vehicle. An instrumented vehicle is one
equipped with an ‘in-built’ computer data recording system. Two moving observer techniques are
the floating car and chase car methods. The development of more sophisticated technology has
facilitated the use of these methods, e.g. the GPS described in Section B.2.

Floating Car Method

The floating car method measures the mean travel time along a test section. A test vehicle
attempts to simulate an ‘average’ vehicle in the traffic stream, by noting the number of vehicles that
overtake the test car, and the number of vehicles that the test car overtakes. The driver tries to
equate these two numbers of overtakings, and the test vehicle is said to be ‘floating’ in the traffic if
the two numbers are the same. If it is not possible to equate the numbers of overtakings by the
end of the test section, a correction to the observed travel time can be applied (Robertson 1994).

The floating car method is appropriate for travel time surveys on long and perhaps complex routes.
The main advantage of the method is that data can be collected easily and quickly over a number
of routes. The main disadvantage is that it is difficult to collect large samples of data, as the test
vehicle has to traverse the route and then return to the starting point for the next run. This all takes
time, and in a dynamic environment where travel conditions (e.g. flow rates and traffic control
systems) are changing rapidly, repeated runs may form biased samples of travel times from any
one set of conditions. Further problems exist for the technique on arterial roads with significant
levels of platooning, in which the test vehicle may have great difficulty in floating in the stream.

Johnston et al. (1982) and Luk et al. (1983) reported the experience of using the floating car
technique in balanced experimental designs to determine the performance of different signal
coordination methods in reducing travel time, fuel consumption and other performance indices. A
fully balanced, Latin Square design is not always possible because of resource constraints but
meaningful results are still possible by considering most of the factors involved.

Austroads 2009
— 130 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Cowan and Eriksson (1972) evaluated the floating car method and demonstrated that it could
provide an accurate estimate of the mean section travel time with only a slight bias. They also
showed that the bias decreases as the flow rate increases. Section C.1.3 describes recent travel
time studies on bus and transit lanes in Sydney with the floating car method using vehicles
equipped with GPS receivers.

Chase Car Method

Here the survey vehicle follows a randomly selected vehicle in the traffic stream, copying as
closely as possible the manoeuvres of the chased vehicle. For travel time studies it is necessary
to keep track of the time performance of the vehicle (e.g. section travel times, time stopped). Care
is needed to ensure that vehicles are selected at random, or strictly according to a predetermined
strategy (e.g. by tracking a certain number of particular vehicle types) – otherwise significant
biases may result.

One obvious problem with the method is that the chased vehicle may leave the survey route at any
time. The decision then has to be made as to whether the survey vehicle aborts the run and
returns to pick up a new vehicle; continues and picks up a nearby vehicle to follow; or ‘floats’ in the
traffic until the end of the test section. The decision will largely depend on the sampling unit for the
survey. If the unit is the individual vehicle, another vehicle would have to be found and the
collected data would still be useable. If the route is the sampling unit, the run would have to be
aborted.

C.1.2 Stationary Observer Methods

Travel time observations by teams of observers, situated at a series of fixed points on the road
network, may be used to collect data on the travel times of individual vehicles in the traffic stream.
Three methods – numberplate surveys, input-output surveys, and path trace surveys – are
described below.

Some technologies available for managing traffic operations may be adapted for traffic survey
monitoring. For example, the installation of electronic toll tags in an increasing proportion of the
vehicle fleet provides an opportunity to use probe vehicles as sensors to measure speeds and
travel times. Further discussion is provided in Part 9 of the Guide (Austroads forthcoming 2009f).

Numberplate Survey Method

The most common use of numberplate surveys is in the collection of origin-destination data (see
also Appendix D). These surveys may also be used to collect data on the distribution of travel
times in a section of the road network. Typically, the section will be a length of road or a single
route in the network, but travel times across an area might also be collected. The survey operates
by having observers positioned at selected points on the road: they record the numberplates (or
partial numberplates) and arrival times of a sample of the vehicles that pass them. Setting the
survey area, time and sampling rate is discussed in detail in Section D.2. When the data are
consolidated in the office, numberplates can be matched between upstream and downstream
locations, and travel times for individual vehicles computed from the difference in arrival times.
The principles of the survey method are quite straightforward, but certain complications can arise
in practice.

The first consideration is to ensure that the workload placed on any one observer is not too great.
Beard and McLean (1974) suggested that the practical maximum recording rate for conventional
data recording methods was 600 recordings per hour. The selection of appropriate recording
equipment and proper training of observers are essential.

Austroads 2009
— 131 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The second problem is one of natural attrition of the observed traffic stream. Vehicles entering the
test section may turn off the road, or finish their journeys upstream from the exit station. Similarly,
other traffic will enter the road from side streets and not pass through the entry station. The longer
the section of road, the lower is the through traffic proportion. As a result of this second problem,
there is the need for a trade-off between section length and accuracy of estimating travel time and
speed.

Another problem relates to start-up and shutdown errors. Vehicles will be within the survey section
at the time the survey commences, and will only be observed on departure. Similarly, vehicles
entering the survey section just prior to the end of the survey may not be observed departing. This
bias is less significant as the duration of the survey increases (Section D.2.3).

Several different recording techniques are available. They include watch, paper and pencil,
cassette tape recorder, laptop or palm computers. Hand-held computers and data loggers have
simplified and improved the accuracy of the data collection process. In-built clocks can be
synchronised, reducing discrepancies between observation stations, and data does not have to be
transcribed as with a cassette recording which can introduce transcription errors.

The NSW Roads and Traffic Authority (2001) has conducted a series of numberplate surveys to
determine travel times along the Pacific Highway in NSW since 1995. These surveys have been
conducted between Hexham, North of Sydney and Tweed Heads on the Queensland border. Each
survey involved a series of leap-frogging sites along the survey route and two survey teams. The
progressive matching of numberplates between successive stations gives a distribution of travel
times over each subject length. The survey samples consisted of light vehicles in white or cream
colours and all heavy vehicles (defined as those with twin tyres on any axle and those with more
than two axles). The sample size was about 40% of all vehicles and the travel times should be
representative of the route travel times. Delays at work sites along the route were also recorded
from 1997, using stopwatches as each team travelled from a recording station to a subsequent
designated location.

In the analysis of the RTA survey data, outliers representing very low travel speeds were deleted
and median travel time values were used instead of the mean values to minimise the impact of
outlying data. These surveys therefore provide valuable time-series data to indicate the
performance of the route over time. One finding from the latest survey is that, despite steady
growth in traffic volume, the increasing capacity of the route has improved travel times of both the
light and heavy vehicles in both directions of travel, when work site delays are excluded.

Input-Output Survey
This survey method bears some similarities to the numberplate method, but is simpler in operation.
It does not try to match individual vehicle observations at different stations, and thus can only
provide estimates of mean travel times. The technique is based on the idea that the difference
between the means of two sets of observations is equal to the mean of the differences of the two
sets. Input-output surveys find the mean arrival time and the mean departure time of the traffic
stream in the test section, and calculate the mean travel time by subtracting mean departure time
from mean arrival time. The data collected at each station are the number of arrivals in successive
time intervals. Either automatic data loggers or human recorders can collect the data. The shorter
the time intervals, the more accurate is the estimate, but also the higher the workloads imposed on
field staff and consequently, the less accurate the observations. A compromise for human
recorders seems to be an interval of about 10 seconds, with longer intervals giving comparable
results – as long as the mean travel time is longer than the interval and at least 30 intervals or 5
minutes are observed.

Austroads 2009
— 132 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

It is important to check that the same general group of vehicles will be recorded at each station.
Allowances need to be made for start-up and shutdown errors as already mentioned. Input-output
analysis is best suited to closed systems, in which vehicles entering the survey zone can only
leave it via the observation points. Freeway traffic is one such system. Another application for the
technique is in recording the duration of stay of vehicles in an off-street car park.

Luk (1989) reported the use of the input-output technique to measure the travel time on a 350 m
section of an arterial road. The upstream and downstream measurement stations employed a pair
of treadle axle detectors at each station to monitor vehicle counts and wheelbase lengths. Data
loggers were used to log and process the data at a resolution of 1 second because of the relatively
short travel time. The results suggest that it is viable to monitor travel times by correlating input
and output traffic count time series, and that wheelbase lengths do not add significant extra
information for travel time estimation.

Path Trace Survey

This method is capable of providing accurate information about traffic movements and travel times
in a restricted area. Examples are direct observation, video recordings or even vehicle tagging
using GPS and other devices. The travel time and detailed path of a vehicle can be determined.
The sampling of vehicles may be a problem. A suitable vantage point is also needed for direct
observation or video recording.

C.1.3 Transit and Bus Lane Surveys Using GPS

Transit and bus lanes have been important as a congestion management tool in Australian cities.
They can provide shorter travel times for buses or high-occupancy vehicles especially during peak
hours, and have the potential to attract commuters to carpool or travel by bus. Sydney has
implemented more transit and bus lanes than other Australian cities. The transit lanes are known
as T2 lanes for two-person carpools and T3 lanes for three-person carpools. The person-
throughput on a transit or bus lane can be significantly higher than non-transit lanes (Quail and
West 1992).

The NSW Roads and Traffic Authority (2002) undertakes regular travel time and occupancy
surveys of all transit and bus lanes. The surveys set out to determine if the time saving benefit of
these lanes was being maintained. Clearly marked survey vehicles equipped with GPS receivers
are used to collect data. The police are informed about the survey. The travel time on each transit
lane is measured 12 times (i.e. six times per day over two weekdays). Each vehicle begins a
survey at the start, in the middle and at the end of a peak period. Both a.m. and p.m. peak
periods are included. One of the vehicles stays wholly within the transit lane, and the other is free
to change lanes if travel conditions are more favourable. The travel times of the two vehicles are
then averaged. This methodology reduces any bias due to variations in traffic volumes throughout
a peak period. Bus lane travel times are determined by observers riding in buses, although a
similar technique to that used on the transit lanes could be employed. Three trips are taken in
each period for two weekdays.

In addition to the travel times information, vehicle occupancy data are also collected to allow other
performance measures such as person-throughput and illegal usage to be collected.

C.2 Queuing Survey

A vehicle is in a queue when it is controlled in its actions by the vehicle in front of it or has been
stopped by a component of the traffic system. Queues can, therefore, occur in traffic moving along
the open road as well as at constrictions in the traffic system. Bunching of vehicles is, by the
above definition, a queue.

Austroads 2009
— 133 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

To illustrate the various measurements associated with a queue of stopped vehicles, consider a
signalised intersection with vehicles arriving and departing uniformly. Figure C 1 presents a time-
space diagram for such a case. Vehicles arriving during the red phase are halted and are not able
to leave during the red phase. The vehicles stopped at the intersection can depart when the green
phase starts. The vehicles stopped at the intersection when the lights turn green represent the
maximum stationary queue. During the green time, vehicles leave the intersection at a faster rate
than they arrive. Hence, the queue decreases in total size, but since it takes time for the leading
vehicles to start moving, the later vehicles remain stationary for some time after the traffic lights
turn green. The point in time when the last stationary vehicle in the queue moves, determines the
maximum back of queue. This represents the physical end of the queue as perceived by the
driver.

Figure C 1: Time space diagram for vehicles at a signalised intersection

Another queue that is of interest is termed the overflow queue. This is the number of vehicles that
are still present in the queue at the end of the green plus yellow period. The above discussion
relates to traffic signals and the event that determines the critical time of queue measurement is
the change in phase. Queues forming at uncontrolled intersections or at other constrictions in the
traffic system have the same queue formation characteristics as those present at a signalised
intersection. The main difference lies in the critical time of queue measurement. In these cases it
is likely to be the departure of a vehicle from the head of the queue that determines the critical
queue lengths.

Measurement of queue lengths involves an observer recording the number of stationary vehicles at
a particular point in time. This can be done by physically counting the vehicles, or by placing
marks along the road length to indicate the number of vehicles that would be in a queue of a given
physical length. Video cameras can be used to record the queue lengths for subsequent analysis
manually, or automatically employing digital imaging technologies.

Queue lengths and delay are related. The triangle in Figure C 1 represents the delay experienced
in one cycle length (in vehicle-seconds) due to uniform arrivals at a signalised intersection. For
non-uniform arrivals, the delay triangle becomes a delay polygon. By sampling the queue lengths
at different time points of a signal cycle, the average delay can be estimated. This method is
described in the following section and is known as a point-sample method.

Austroads 2009
— 134 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

C.3 Traffic Delay Surveys

C.3.1 Definition of Delay
There are many definitions of delay. The HCM 2000 defines four basic delay terms as:
Geometric delay – the component of delay that results when geometric features cause vehicles to
reduce their speed in negotiating a facility.

Control delay – the component of delay that results when a control signal causes a lane group to
reduce speed or to stop; it is measured by comparison with the uncontrolled condition.

Incident delay – the component of delay that results from an incident, compared with the no-
incident condition.

Traffic delay – the component of delay that results when the interaction of vehicles causes drivers
to reduce speed below the free flow speed.

If on a straight section of road and there is no traffic there will be no delay. Add a curve to the road
but still with no other vehicles on the road then a geometric delay is introduced. Add a stop sign or
install traffic signals, but still with no other vehicles on the road then a control delay is introduced.
Increase the traffic and a traffic delay is introduced. The additional delay above the control delay is
the traffic delay and this delay is also dependent on the type of control. Finally, if there is an
incident then the additional delay is an incident delay.

Stopped delay is the delay experienced by vehicles that have actually stopped. It is a part of both
the control delay and the traffic delay.

In general, delay (D) on a road section is given by:

D = T – Tb

where T is the journey time and Tb is a base travel time represented usually by the free flow speed.
Other choices of the base speed include the posted speed limit or average speed (Reilly and
Gardner 1976).

The travel time survey methods in Table C 1 are directly applicable for the measurement of delay
on road sections and are not repeated here. The use of the point-sample method for measuring
intersection delay requires further discussion.

C.3.2 Measurement of Intersection Delay

The measurement of intersection delay using the point-sample method requires the identification of
two locations, which could be the stop line and a point further back than the tail end of any
expected queue. The sampling can be at small, regular intervals for stopped delay or at specific
time points of a signal cycle for overall delay.

Austroads 2009
— 135 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Stopped Delay

Stopped time delay can be collected by a manual method. The manual method is based on the
calculation of stopped delay, Ds, from the formula

N
DS ¦ [ET
i 1
i  STi ]

where

N = total number of vehicles stopped

STi = time when vehicle i stopped

ETi = time when vehicle i started.

This method separately records stopped vehicles in small intervals, as well as all previously
stopped and departed vehicles. A matrix is drawn up as shown in Table C 2, and the vehicles
remaining in each time slice calculated. The stopped delay DS is thus determined by multiplying
the sum of the remaining vehicles in each period by the time intervals. That is:
N
DS ¦ RM
j 1
j

where d is the time interval and RMj is the number of vehicles remaining in the queue at the end of
the jth interval and n is the number of time intervals.

This equation is strictly correct only if all stoppages and departures occur only at the start of each
time interval, whereas, in fact they will be spread over each interval. Therefore, delay will tend to
be overestimated in an interval in which stoppages exceed departures, and underestimated in the
opposite case. Over a sufficient period, the errors tend to cancel each other.

For the data in Table C 2 this yields:

DS = 10 x(6 + 10 + 12 + 8 + 2) = 380 seconds.

Table C 2: Matrix of vehicle movements at a sample intersection

Vehicles Number of vehicles in 10 second time intervals
0-10 11-20 21-30 31-40 41-50 51-60
Stopped 6 4 3 2 2 0
Leaving 0 0 1 6 8 2
Remaining in period (RMj) 6 10 12 8 2 0

Typically, intervals of 10 to 15 seconds are used. At traffic signals, the time interval should not be
a direct proportion of the cycle time.

A variation of this method can also estimate the time travelling across the approach area. All the
vehicles in the approach are measured at each time slice, whether stopped or moving (Taylor et al.
2000).

Austroads 2009
— 136 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Overall Delay Using QDELAY

The QDELAY technique requires the recording of queue lengths at different points in time during
the signal cycle (Richardson and Graham 1982). Figure C 2 shows the form used to record the
data in the field. At the start of the green phase, the time is recorded in column A, and the number
of vehicles stopped in the queue is recorded in column B. At this point it is necessary to make a
mental note of the last vehicle in the queue at the start of the green time. If the last vehicle crosses
the stop-line before the next red phase starts, the time at which it crosses the stop line is recorded
in column C. After that vehicle crosses the stop line, the number of vehicles crossing the stop line
before the start of the red phase is recorded in column F. The time at which the signal changes
back to red is then recorded in column D. Column E is left blank in this situation.

Figure C 2: Survey form for QDELAY

If, however, the end-of-queue vehicle does not cross the stop line before the signal changes back
to red, the time at which the light changes to red is recorded in column D. The number of vehicles
in the queue at this point of time (including the last vehicle) is recorded in column E. Columns C
and F are in this case left blank. This process is repeated for every cycle of the survey period.

These data can then be analysed to provide information on: cycle settings, total flow, averages and
standard deviations of the approach delay and stopped delay per vehicle, average numbers and
standard deviations of effective stops and complete stops, and average and standard deviation of
maximum stationary queue length per cycle. It cannot, however, distinguish between delays to
vehicles making particular manoeuvres if the lanes are carrying mixed turns and through vehicles.

C.4 Data Analysis and Presentation of Results

The previous discussion has included descriptions of the methods of analysis used in travel time
and delay studies. Results of travel time and delay studies may be presented in tabular, graphical,
and diagrammatic form to show variations along the route being studied, together with the location,
cause and length of delays, if available. Where large quantities of travel time or delay data are
collected, electronic data processing methods for recording and subsequent analysis of data
should be used.

Austroads 2009
— 137 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX D ORIGIN-DESTINATION SURVEYS

D.1 Types of Origin-Destination Surveys
Some commonly used techniques for gathering origin-destination data are:
ƒ numberplate survey
ƒ roadside interview technique
ƒ postcard survey
ƒ headlight survey
ƒ windscreen sticker technique
ƒ registration address technique.

The use of GPS technology for speed measurement has been mentioned in Appendix B. The
same technology is applicable for identifying the travel path of a vehicle and hence the origin and
destination of that trip. The technology at present is still expensive and its applications are mainly
for commercial operations such as taxi dispatching and heavy vehicle or fleet monitoring. Its
usage will increase as the cost of using a full GPS system decreases in the future (Stopher 2001).

Electronic toll collection technology may also be adapted for origin-destination surveys. The
increasing prevalence of electronic toll tags in the vehicle fleet provides opportunities for identifying
the entry and exit of vehicles at points that could be used on survey cordons.

D.1.1 Numberplate Technique

This is the most common origin-destination survey technique in use. It involves stationing
observers at selected points, in and around an area, to record the number plates of vehicles and
the time at which each vehicle passes the observer. The route taken by a vehicle can then be
found by matching the observations. This technique is best suited to traffic patterns characterised
by a large number of origins and destinations, linked by a complex road network. This is the
situation commonly found in an urban area. One disadvantage of this technique is the large
amount of effort required in data coding and analysis. The use of laptop and palm PCs in traffic
data collection and analysis has helped to mitigate this problem. The technique is discussed in
further detail in Section D.2.

A recent development is the automatic recognition of number plates for electronic tolling and law
enforcement. This technology is used on CityLink in Melbourne and various toll roads in other
cities. Cameras at a toll point in CityLink take a video image of the number plate of any vehicle
without a valid electronic tag. These vehicles include account holders without a tag on that vehicle,
day-pass holders and other vehicles that could receive a fine (Lay and Daley 2002). The
technology of automatically processing video images to obtain number plates can potentially
provide valuable origin-destination data. The technology is still expensive for general data
collection and privacy laws can prevent agencies such as CityLink from making such data available
to the public.

D.1.2 Roadside Interview Technique

This type of survey presents the best opportunity to gather comprehensive data on traffic
movements, but is not often carried out due to the lack of authority; generally only Police have the
authority to require drivers to stop (other than at road works or a traffic incident). Vehicles are
directed into a roadside station, and drivers are asked to respond to a preset questionnaire about
the details of their trip.

Austroads 2009
— 138 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The technique allows the collection of information not available from simple observation. For
example, the exact origin and the ultimate destination of a trip can be found, together with
information on trip purpose, frequency and details of the route taken. The interview should be
short, simple and unambiguous to both the interviewer and interviewee. It should only seek
information relevant to the purpose of the study.

If the interviews are carried out on traffic entering and leaving a cordon line around a study area,
the flow making a through movement is actually the mean of the flows determined at the point of
entry and the point of exit. The roadside interview technique is generally not suitable for use in
urban areas where high traffic flows and restricted right-of-way makes it difficult to stop the
sampled vehicles without serious disruption to other traffic.

It is necessary also to check the legality of asking drivers to stop even with the cooperation of the
police for reasons other than the non-compliance with road rules.

D.1.3 Postcard Survey

This technique requests that drivers take and complete a self-administered questionnaire, and then
post it back to the survey organisation. Reply rates of 30-50% are usually achieved with reply-paid
post. A return of at least 20% is considered essential to a recommended maximum of five
questions. The postcard survey technique does not provide as much information as a roadside
interview, but is useful in situations where it is unsafe to delay drivers more than a minute and
provides more information than direct observation.

D.1.4 Headlight Survey

The headlight survey technique works by asking drivers at a particular site to switch their
headlights on, and keep them on for a set period or until asked to extinguish them. Observers at
downstream sites record the distribution of vehicles with headlights lit. The technique can be
applied in rural areas where there are limited trip origins. It traces the destinations of vehicles
entering an area along a given road. The technique can only be used in daylight and requires high
levels of public cooperation. Advantages of the technique include little traffic disruption and quick
and easy analysis of results. The technique can only be applied in relatively simple networks. Due
to the lack of authority in asking drivers to switch on their headlights, this technique is less
frequently being used.

D.1.5 Windscreen Sticker Survey

This method allows for the differentiation between separate origins by using different coloured
stickers. Traffic may have to be stopped briefly to issue stickers to drivers, or stickers can be
placed on parked vehicles or handed out to patrons before they leave a particular site. The
stickers are then collected at the survey cordon stations. The survey may also be a postcard
questionnaire survey, where the ‘sticker’ is a small questionnaire card. The technique may also be
used for studies of pedestrian origin-destination movements.

The method is easy to apply and data analysis is relatively simple. However, it requires advance
publicity to obtain co-operation from the public, and appropriate signs and police assistance are
necessary at the approaches to the survey points.

Austroads 2009
— 139 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

D.1.6 Registration Address Survey

This technique exploits information on car owners’ home addresses. Vehicle registration numbers
are recorded, and then the registered address of each vehicle is checked from official records. A
number of problems exist with this technique. It assumes that the registered address of each
vehicle is the actual trip origin and/or destination. Vehicles registered at commercial addresses will
be recorded as such regardless of whether they are usually garaged in residential areas.

D.2 Conducting a Numberplate Survey

The survey process should begin by defining the purposes of the survey within the context of the
traffic study to which it belongs, then setting the objectives to be met, and selecting a survey
technique for the population to be sampled. The following considerations are needed once the
number plate technique is selected for an origin-destination survey.

D.2.1 Defining a cordon

The cordon is the boundary around the study area for which trip movements are required. It may
consist of physical boundaries (e.g. rivers, hills, parkland); major transport facilities (e.g. railways,
freeways); jurisdictional boundaries (e.g. municipal boundaries); or arbitrary boundaries. The study
purposes should point to the extent of the survey area. The survey objectives and constraints
should be used to set the exact details of the cordon. Survey stations are then located on the
cordon and inside the area to permit the observation of movements into and out of the area as well
as route information. Figure D 1 differentiates internal stations from cordon stations.

Figure D 1: Stations on the cordon and inside the study area

Stations need to be located with care, both to minimise the number of stations while maintaining
the required level of detail, and to remove possible ambiguities in the data (Figure D 2). For those
stations located at junctions, observations should be made of each separate turning movement, as
indicated in Figure D 3. Normally, it is no more difficult or costly to record the flows by movement,
and valuable additional data are then available.

Austroads 2009
— 140 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure D 2: Setting cordon station locations

Figure D 3: Recording of separate turning movements at a station

D.2.2 Setting the Survey Time

The time of day for the survey should be chosen to cover the period when ‘divertible’ traffic will be
observed (e.g. peak-period flows in urban areas, or busy shopping times at a shopping centre).
The duration of the survey is also of concern. The observation must allow for ‘start-up’ and ‘shut-
down’ periods that are discussed again in Section D.3.2.

D.2.3 Setting the Sample Rate

Sampling of number plates is usually performed on the basis of a specified digit, such as the last
digit of the number plate. Sampling on the basis of the last digit ending in, say, a ‘4’ will result in an
a priori sampling rate of one in ten, while sampling all vehicles with plates ending in ‘1’ or ‘2’ would
give a sampling rate of one in five. More accurate results will be achieved if the digits ‘1’ or ‘4’ are
selected, as opposed to ‘6’, ‘8’ or ‘9’ which are easily confused. Practical experience suggests that
sampling on a single value of the last digit yields an approximate 10% sample, once traffic
observations at the station exceed 2000 vehicles.

Care needs to be taken in the application of this rule. Some states now have specialty number
plates that may not contain any numeric characters at all, and some general release number plates
actually end up with an alphabetic character, e.g. ABC123D.

A useful alternative to characters on a number plate is the colour of vehicles, which can be
assessed quickly. The sample rate is estimated by placing an automatic traffic counter at one or
more survey locations. The proportion of different vehicle colours in the traffic stream seems to be
relatively stable. For example, the white and cream colours were selected for the travel time
survey described in Section C.1.2.

Austroads 2009
— 141 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

D.2.4 Selecting the Characters to be Recorded

It is common to record four characters rather than all the characters on a number plate. The
characters can be three digits and one letter, two digits and two letters, etc. Recording more
characters requires more effort but results in less duplication of number plates. A compromise
exists between the effort required and the level of duplication. Recording less than the actual
number of characters also addresses any privacy concerns which may arise.

D.2.5 Field Data Collection

The following data are usually recorded for each observation:
ƒ number plate (to a specified number of characters)
ƒ location (the observation station)
ƒ direction of travel
ƒ time of observation
ƒ the type of vehicle.

Recording may be done with paper and a pencil, a cassette tape recorder or laptop/palm PC. The
PC has a number of advantages such as in-built clocks to log arrival times automatically, and
immediate computer-readable data, but shortcomings such as battery life, impact and water
resistance, and initial cost should also be taken into account. Other techniques have their own
favourable features. When all else fails, the humble survey form provides an intelligible and re-
constructible storage medium. Cassette recordings are useful for heavy traffic flows, but are
tedious to transcribe after the survey and may have their own special types of errors. Voice
recognition technologies are now more readily available and can be used for automatically
transcribing numberplate data.

D.3 Sources of Error in Origin-Destination Surveys

Errors in an origin-destination survey can arise from two separate sources:
ƒ errors in field data collection, including mis-recording of data observations and start-up and
shut-down errors
ƒ spurious matchings due to combinational effects in the analysis of partial numberplates.

D.3.1 Errors in Data Recording

The following errors are commonly made in the collection of numberplate data, and the order given
below is an indication of the relative frequencies of occurrence:
ƒ the number is missed completely
ƒ one or more characters are mis-recorded, or two characters are transposed
ƒ the direction of movement is recorded incorrectly
ƒ the type of vehicle is mis-recorded.

The consequence of any of these errors will usually exclude the possibility of a successful match
with another observation. Attention to methods that can reduce recording errors is, therefore,
essential in good survey practice. A primary concern is the selection of observers to be used, and
the stations they are allocated. Only experienced observers should be sent to sites where heavy
traffic volumes are expected.

Austroads 2009
— 142 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

D.3.2 Start-Up and Shut-Down Errors

As every trip in the study area will involve a finite travel time, the first vehicles leaving the cordon at
the start of the survey will be interpreted as local trips, even if they are in fact through trips. As
they entered the study area before the start of the survey, there are no records of their entries. A
similar problem occurs at the end of the survey. Two adjustment techniques are as follows:
ƒ Discard all observations leaving the cordon a time period after the start of the survey and
those entering the cordon a similar time period before the end of the survey. This time
period is approximately the time taken for a vehicle to traverse the study area at a prevailing
speed and is necessarily a function of the size and type of study area.
ƒ Use manual matching of numberplates at the start of the survey to determine the time at
which the first ‘wave’ of vehicles, entering the cordon at the start, passed each station.
Observations before this wave can be discarded. A similar technique may be applied at the
end of the survey.

D.3.3 Confused Characters

Depending on the recording technique, there will be pairs of characters that can be easily
confused. For example, in written recordings ‘S’ and ‘5’ and ‘2’ and ‘Z’ and in audio recordings ‘B’
and ‘D’ and ‘M’ and ‘N’ can be confused along with others. The potential for this type of error is
increased where one observer is calling numberplates to another observer who is writing the
information on field sheets. Handwritten data can also be misread. During the analysis of data, an
increase of matches could be achieved by replacing one of each of the problem characters with its
partner.

D.3.4 Spurious Matchings

Errors due to spurious matchings may occur in the analysis of partial numberplates. For instance,
three different numberplates ABC123, BAC123 and CAB123 could be observed. If analysis were
performed on the basis of the last four characters, the partial plates would then be C123, C123 and
B123. The first two observations would now match. This is termed a spurious match. If only the
last three characters were used, the observations could all be interpreted as the same vehicle.
With the tightening of privacy laws, it is more common for only partial numberplate data to be
recorded, so some account needs to be taken of possible errors. The compounding errors in
origin-destination numberplate surveys are likely to lead to an underestimation of the number of
through trips across the study area. Experience suggests that errors in observing and recording
numberplate data are likely to outweigh errors due to spurious matching, certainly if at least four
characters of the numberplates are observed. Errors can be further reduced if the selected
characters contain more alphabetic than numeric characters, i.e. where numberplates are
predominantly of the form ABD123, collection of the first four characters will yield less spurious
matches than if the last four characters are recorded.

D.4 Data Analysis and Presentation of Results

D.4.1 Analysis of Numberplate Data
There are many computer programs available for analysing numberplate data. Each program will
have its own internal logic, and it is possible that two different programs will generate different
results. A two-pass procedure is useful for the analysis of numberplate data, with the ability to
account for partial mis-recordings of the data. The first pass looks for exact matchings between
recording stations, creating an initial origin-destination matrix. Unmatched plates are then
searched for ‘almost matched pairs’ (e.g. one character different, or two characters transposed).
The estimated matches can then be added to the exact matches to create a total origin-destination
matrix.

Austroads 2009
— 143 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

ƒ McPherson (1995) describes one such program that takes into consideration the following
factors:
ƒ vehicle speed
ƒ vehicle route
ƒ vehicle type
ƒ time of travel
ƒ size of study area
ƒ average distance between observation stations.

The Roads and Traffic Authority also has a software package that produces origin destination trip
matrices from numberplate data. The software produces ‘nominated’ paths between any two pairs
of points, and origin destination paths between one entry and any internal point or from where last
sighted.

D.4.2 Presentation of Results

The main result from an origin-destination survey is usually a matrix showing the trips from each
origin to every destination in the study. A convenient representation of the origin-destination matrix
is the desire line diagram. This diagram shows the travel desire between origin and destination
zones represented by straight lines connecting the trip origins and destinations. The relative width
of the lines indicates the relative amount of travel between the zones.

The detail with a desire line diagram can be lost if there are many origins and destinations. It is
therefore useful if a computer program has the option to produce specific desire line diagrams for
trips from a particular origin, to a particular destination or between an origin-destination pair, as
well as a full desire line diagram. The generation of desire line diagrams can benefit from a spatial
database technology such as the geographical information system.

An example of a desire line diagram is shown in Figure D 4. For clarity, different colours are used
to illustrate desire lines from different origins.

Source: VicRoads (2008).

Figure D 4: A desire line diagram showing trips in the vicinity of Kew Junction, Melbourne

Austroads 2009
— 144 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

D.5 Summary
Origin-destination surveys provide useful information on travel patterns that can be used for traffic
planning. They are important surveys for traffic engineers, but there are some significant
difficulties in using them. These difficulties can only be overcome by the use of the best survey
practices. This type of survey is both expensive and difficult to manage, and thus deserves careful
planning, administration and data analysis, if its results are to be truly useful and valid.

New technologies such as GPS, GIS and automatic numberplate recognition are now being used
for fleet management, law enforcement and possibly electronic tolling. Origin-destination data can
be a by-product of these operations. Privacy issues will need to be addressed before such data
are readily available.

Austroads 2009
— 145 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX E PEDESTRIAN AND BICYCLE SURVEYS

Initiatives in several jurisdictions (e.g. Austroads 2005, Transport WA 2000, NZ Ministry of
Transport 2005) are increasing the profile of cycling and walking as well as striving to improve the
provision of safe facilities. The planning and design of pedestrian and bicycle facilities is covered
in Parts 4, 5 and 6 of Guide to Traffic Management (Austroads forthcoming 2009e , Austroads
2008b and Austroads 2007) and in Part 6A of the Guide to Road Design (Austroads forthcoming
2009d). This Section outlines methods for collecting data on pedestrians and bicycles.

E.1 Pedestrian Survey Methods

E.1.1 Pedestrian Counts
The number of pedestrians passing a specific location per unit of time is the flow rate or volume.
Information on volumes is required in almost all surveys (Appendix A). Volume data can be used
to establish the importance of a route, the variations in flow, the distribution of traffic, the trends in
road use and in the determination of other parameters. Traditionally, volume counts have been
conducted manually. However, a number of emerging technologies can now be used to collect
pedestrian data.

Manual Pedestrian Counts

Manual counts consist of an observer recording the flow of pedestrians past a certain point for the
required time period. The most common method of collecting volume data is by manual counts of
the flow of pedestrians at a particular point in the traffic system. In the simplest form, the observer
manually records the number of pedestrians for the time period. The demands of data collection
can be reduced by using mechanical, electrical or computerised tally counters.

Manual pedestrian counts rely on good planning and skilled observers to ensure accurate and
useful results. The number of observers will depend on the general level of traffic activity and the
data-recording task. For example, if classification of pedestrians (by demographic and/or direction
of travel) is necessary, more observers will be required. Observation sites need to be chosen so
that they provide a good view of the area but also provide protection from the weather and
inquisitive people.

Pedestrian Classification Counts

The use of a pedestrian facility is partly related to the types of pedestrians using it. To determine
the proportions of different users, classification counts are necessary. A classification count is an
extended volume count in which different types of pedestrians are identified. A clear definition of
pedestrian types is essential. For example, there is little point in recording the volume of
pedestrians by gender if the objective of the study is to determine the volume of school children
using a crossing. The adults could be put into a single group.

The greater precision required in manual classification counts means that more people are
required to collect, decode and analyse the data. A manual classification count appears simple.
Passing pedestrians are recorded for predetermined time periods using different parts of a survey
form or by hand-held counters. In practice, this procedure gives rise to considerable scope for
error. Pedestrians can be missed, double counted, incorrectly identified or entered in the wrong
place. Even if counts are properly conducted and well supervised there is still possibility for error.

Austroads 2009
— 146 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Distribution of Arrivals and Service-time Measurement

The average pedestrian arrival rate can be determined from a known volume. The analyst may,
however, be interested in the distribution of pedestrian arrivals at a particular point. This can be
obtained by recording arrival times and forming a histogram from the data. Times can be obtained
using stopwatches and coding forms or via a computer data logger. Video could also be used.

The rate at which pedestrians are serviced is also important. Queuing studies can be undertaken
using the method described above for arrival distributions. It is important to note that the service
time is strictly the time taken to service a customer. It is not the total time of the study divided by
the number of customers, but rather the total time that the server is busy divided by the number of
customers.

Automated Pedestrian Counts

Various pedestrian detection systems have been developed and they include the following
(Aylward and Valentine 1995):
ƒ Tactile sensors – two types exist. The first consists of a tile with printed circuit board
inductors mounted in a flexible pad underneath. The pressure of a pedestrian standing on
the tile distorts the shape of the inductors slightly, which is detected by a conventional loop
detector. The second consists of a tile with a piezo-cable moulded into the non-slip covering.
The tile can detect pedestrians, even when they are standing still because a pedestrian’s
muscle groups that are constantly active can be detected. Tactile sensors (Figure E 1) are
suitable for pedestrian detection at traffic signals (e.g. puffin crossings).
ƒ Above-ground sensors utilising microwave, infra-red, ultrasonic and laser detection – these
systems can be used to count pedestrians passing a fixed point. A major difficulty is in
distinguishing between closely positioned pedestrians.

Figure E 1: Tactile sensors

Video Detection
The cost of video equipment has decreased substantially, making it an affordable option for traffic
studies. Video can be used to collect many types of pedestrian parameters such as flow, volume,
classification, distribution, and route and origin-destination data. Video has a number of
advantages, including:
ƒ observations that do not disrupt the pattern of activities
ƒ the ability to record many events simultaneously or in rapid succession
ƒ a reduction in field staff requirements
ƒ permanent and complete set of records
ƒ both time and space dimensions can be recorded.

Austroads 2009
— 147 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

In the past, the major disadvantage of using video has been the long and tedious task of
transcribing data following recording. Advances have now been made in the areas of digital
recognition and image processing to allow automatic recording of data during traffic studies.

For example, Rouke and Bell (1994) describe software that can obtain information on pedestrian
movements automatically using image processing. Of the two separate algorithms developed, one
was designed to give a measure of the pedestrian density within a crowd scene and the second to
count and determine the walking direction of individual pedestrians. The reported results suggest
that the automatic measurements of crowd density and individual pedestrians are feasible, after
some detailed calibration.

E.1.2 Pedestrian Audits

Pedestrian audits provide valuable information to engineers, urban planners and local councils on
how public spaces or development plans rate from a pedestrian viewpoint (Daff and Cramphorn
1995). They are used to improve road safety, convenience and mobility, and personal safety of
pedestrians. Pedestrian audits are ideally carried out at all stages of a major road scheme, in
particular the feasibility and draft design stages to identify potential safety problems before the
development is completed (Jordan 1995).

A road safety audit is carried out by an auditor who walks the study area in a number of different
conditions - (e.g. day, night, wet) - searching for potential safety deficiencies and reviewing existing
facilities. The auditor needs to address issues relating to the most vulnerable pedestrians
including elderly, young, disabled and intoxicated pedestrians. A safety audit checklist should
include the consideration of visibility, vehicle speeds, presence of steps, paving types, levels of
passive surveillance and length of walk signals. Additional information and identification of
problems in the study area could be determined by conducting a questionnaire survey of
pedestrians in the area (Section E.2.2).

E.2 Bicycle Survey Methods

E.2.1 Manual Bicycle Counts
Manual bicycle counts to determine bicycle flows, classifications of riders, distributions of arrivals
and service times are similar to the processes described for manual pedestrian counts in
Section E.1.

E.2.2 Questionnaire Surveys

Questionnaire surveys can provide useful information on route choice, origin-destination
information, characteristics of cyclists, crash history and the adequacy of bicycling facilities.
Questionnaires can be mail-back, self-administered or interviewer-administered. The mail-back
questionnaire is useful when the respondent has little time to answer the questions. The response
to such surveys, however, can be quite low (30%) and unless information on the characteristics of
the non-respondents is known, the results could be misleading. Simple, readily understandable
questions will provide the highest response rate. Self-administered questionnaires are completed
by cyclists (or pedestrians) at the location where they are handed out. For this technique to be
successful the respondent must be captive and not pushed for time. As with the mail-back survey,
the questions must be simple and easily understood. The on-site interview involves an interviewer
asking the cyclist (or pedestrian) a series of questions, and recording the answers. Again, the
respondent must not be pressed for time and the questions asked should be kept to a minimum.
The advantages of this technique are an increased response rate and the ability to further explain
difficult questions.

Austroads 2009
— 148 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Another form of questionnaire is the household or workplace survey. These types of surveys can
be used to collect considerable information on trip purpose, route and origin-destination and socio-
economic characteristics. Household surveys are expensive to undertake, particularly on a
random sample of the population because cycling is a relatively rare activity.

An important point to keep in mind when preparing a questionnaire is the use of appropriate
definitions. One problem area is the definition of a trip. A trip can be defined as a one-way
movement of a person or vehicle between two points for a specific purpose, sometimes called a
stop (Richardson et al. 1995).

An example of a successful mail-back questionnaire survey on cyclists was one conducted by

Aultman-Hall and Hall (1998) in Ottawa and Toronto. The objective of the survey was to collect
quantitative information on crash history and total distances travelled, in order to calculate relative
crash rates on different types of facilities (roads, off-road bike paths, sidewalks). Information on
bicycle route choice was also collected, helping identify the types of facilities chosen by different
cyclists. Questionnaires plus reply-paid envelopes in clear plastic bags were wrapped around the
crossbars of bicycles and held in place with a sticker. The questionnaires were distributed to over
6,000 cyclists in the central city areas and received a response rate of 50%.

E.2.3 Bicycle Detection

Inductive loop detectors are commonly used to detect vehicles but can also be used to detect
bicycles. The loops, which are buried just below the surface of the road or cycle way record
metallic objects passing over due to a change in the inductance. Bicycles have a lower metal
content than vehicles. Bicycle inductive loop detectors therefore need to be more sensitive to
produce acceptable results. Various designs for bicycle inductive loop detectors have been
proposed to try to optimise bicycle detection (Leschinski 1994).

Commercially available bicycle detectors are also available and were installed on the Sydney
Harbour Bridge walkways using asymmetric loop detectors to detect cyclists by direction and
ignore the passage of shopping trolleys.

Piezo-detectors can also be used to detect bicycles (Figure E 2). Piezo materials change electrical
characteristics when subjected to mechanical deformation caused by pressure as discussed in
Sections A.1 and G.3. The deformation can cause a change in resistance (piezo-resistive) or the
generation of a charge (piezoelectric). The piezo-resistive sensor can detect a bicycle at low to
zero speeds, whilst the piezoelectric sensor is not effective at very low speeds.

Figure E 2: A typical piezo-resistive cable for bicycle detection

As with the detection of pedestrians, microwave, infra-red, ultrasonic and laser detection methods
can also be used to detect bicycles. Again, these types of sensors may not provide the required
accuracy due to difficulties in distinguishing between closely spaced bicycles. Figure E 3 shows
the downloading of bicycle counts to a PC, using a tape-switch as a bicycle axle sensor.

Austroads 2009
— 149 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure E 3: Downloading data from a bicycle counter

E.2.4 Video
Video recordings can be analysed to determine bicycle flow rates, speeds and headways. The
time stamp of the video including the frame number allows an accurate time recording. A
technique by Khan and Raksuntorn (2001) automatically determines bicycle-flow data and could
greatly simplify the study of bicycle flow characteristics. The technique estimates bicycle location
data by transforming screen coordinates of video frames to ground or roadway coordinates. The
process is called rectification and enables automatic recognition of location and hence speed and
acceleration data.

Further research is needed in this important area and imaging technologies are expected to
continue to improve. See for example ARRB Transport Research (2003b).

E.3 Travel Time and Delay Studies

Travel time and delay information can be collected via a number of methods, including
questionnaires, video with time-stamp facilities and tagging.

In a tagging survey, pedestrians or cyclists entering a study area are given a card showing the time
of arrival, classification and entry point. This information is updated as they pass other tagging
points. The cards are collected and stamped with appropriate time and location information at the
exit points. The method can yield a large amount of data on trip patterns and travel times. For
detailed route information the cards need to be marked at a number of locations. The delay
involved in this marking could well influence the travel time measures. Cards may also be lost or
discarded.

Virkler (1998) described a technique for determining pedestrian travel time. The technique is an
adaptation of the floating car method (Appendix C). Observers are instructed to walk at the
prevailing speed of nearby pedestrian traffic and, where practical, to pass roughly the same
number of pedestrians as those that pass them. The observers record arrival times at each
intersection queuing point and the times they step off and back on the kerb while crossing.
Walking time, queuing time and average travel speed can be collected using this method.

Austroads 2009
— 150 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Delay information is also of interest to traffic professionals. The delay to pedestrians or cyclists
can either be a delay determined at a point in the traffic system or the delay over a route. The
route delay can be determined by subtracting the unimpeded travel time along a route from the
observed travel time measured in the study. The unimpeded travel time is the average travel time
for a sample of unimpeded pedestrians or cyclists. Point delay can be determined by observing
the number of pedestrians or cyclists stopping and measuring the length of time they are stopped.
In situations where there is a large queue and the delay at different points in time is required, it
may be necessary to employ a large number of observers or use techniques that do not involve
detailed observation of individual movements. Appendix C provides further details on travel time
and delay surveys, some of which can apply in pedestrian and bicycle studies.

E.4 Behaviour and Conflict Studies

The illegal use of traffic signals and other facilities and the incidence of queue jumping could
influence the effectiveness of a facility. Information on the proportion of pedestrians and cyclists
not observing regulations and making illegal manoeuvres can be collected using manual
observation of the system. Video recording is also of help. The unusual nature of the movements
recorded, however, precludes most automatic methods of data collection.

E.5 Data Analysis and Results

The Australian Bicycle Council (2000) released a set of guidelines on the reporting of cycling data,
which is an initiative under the national cycling strategy (Austroads 2005). The guidelines provide
a framework for the reporting of comparable state and territory data, so that relevant comparisons
can be made. The development of the guidelines will allow the measurement of progress towards
a goal of increased cycling participation. The guidelines cover the reporting of bicycle ownership,
infrastructure, usage, demographics and safety.

The Council recommended that data be obtained from existing data sources such as those from
ABS population census and household travel surveys of capital cities. Surveys should also be
undertaken in stages to collect the appropriate data, which include the following in the first stage:
ƒ bicycle ownership per capita
ƒ bicycle network coverage (urban)
ƒ cycling mode share
ƒ cycling trip purposes
ƒ proportion of population cycling
ƒ cyclists age and gender
ƒ cyclist injury rates – hospital reported
ƒ cyclist crash rates – police reported.

E.6 Future Development

The development of new technologies has already automated a number of tasks in pedestrian and
bicycle studies, as was described above. In the near future, advancements to existing
technologies and the development of new techniques should see a further simplification of study
techniques by decreasing the manual component even more. These include:
ƒ Global positioning system (GPS) – GPS receivers and loggers can currently be installed in
vehicles to record route, speed and travel time information (Section 5). Hand-held units
could provide similar information about pedestrians and cyclists (Stopher 2001).

Austroads 2009
— 151 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

ƒ Geographic information system (GIS) – in all the questionnaire and interview techniques
discussed above there is an opportunity to ask respondents to mark their routes on a map of
the locality. These routes can be entered into a GIS and the frequency of trips along a
particular route can then be easily determined (Richardson et al. 1995).
ƒ Video – further development of video capture and data processing technology should enable
accurate automatic recording of pedestrian and bicycle flow, speed, congestion, route and
origin-destination data.
ƒ Smart card – non-contact smart card technology already exists for payment of public
transport fares (Luk and Yang 2001). This technology could be adapted to provide data on
pedestrian and bicycle flows, in a study similar to the tagging survey described in
Section E.3.

Austroads 2009
— 152 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX F NOISE, FUEL AND EMISSION SURVEYS

F.1 Traffic Noise
F.1.1 The Nature of Noise
Noise may be defined as unwanted sound. The biggest contributor to community noise levels in
terms of number of people affected is road traffic. Besides degrading amenity, noise affects sleep
and concentration. Exposure over the longer term to higher levels associated with heavily
trafficked roads is increasingly recognised as affecting health. Very high levels of traffic noise can
also be injurious, but such high levels are well above the normal levels of road traffic noise.

Sound is measured in decibels (dB). Decibel is a unit for measuring acoustic or electromagnetic
energy (P1) relative to a reference level (P2), with the number of decibels calculated as 10 x
log10(P1/P2). A reduction of 10 dB, i.e. a reduction of P1 (or P1/P2 ) to one-tenth of its former value,
is usually and subjectively considered equivalent to halving the sound levels. Human ears respond
to sound levels differently according to the frequency or pitch, variability and loudness of a noise.
An A-weighting scale is built into sound measuring equipment to give a better representation of
how ears respond to sound. The scale is known as the dB(A) scale. Figure F 1 shows typical
environments corresponding to noise levels measured on the dB(A) scale of measurement.

Figure F 1: Typical environments for given noise levels on the db(A) scale

Austroads 2009
— 153 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

F.1.2 Noise Indices

Noise levels vary continuously at any given site and are usually expressed in terms of maximum
noise levels, averages and statistical measures such as percentiles. The L10(18h) noise level has
been used extensively throughout Australia to measure noise pollution. It represents the sound
level exceeded 10% of the time, measured over the 18 hour period 06:00 to 24:00 hours. L10(18h)
is also sometimes referred to as the average maximum noise level (Austroads 2002). The
permitted value is commonly taken as 68 dB(A).

Increasingly, the L10(18h) is being replaced by the equivalent continuous noise level LAeq. This is
an energy average over a period of time, with typical periods being 1 h, 9 h (night) and 15 h (day).
EEC member countries, USA, Canada and several other countries use the LAeq unit to measure
noise.

Austroads (2002) reviews a number of models used for predicting traffic noise levels. They include
those from the Nielsen & Nordic Council of Ministers (Nielsen et al. 1996), the UK Department of
Transport (1988) and the US Federal Highway Administration (Menge et al. 1998). The most
widely used model in Australia is the Calculation of Road Traffic Noise, or CORTN, developed by
the UK Department of the Environment (1975) and updated by the UK Department of Transport
(1988). The procedure was adapted for Australian conditions by Samuels and Saunders (1982).

F.1.3 Surveys of Noise Levels

Traffic and environmental regulatory bodies commonly perform measurements of noise emissions
from individual vehicles. It is the area-wide noise levels, however, that are of more importance to
traffic engineers and in particular the actual levels and annoyance reaction to traffic noise. When
conducting noise measurements, in addition to the relevant traffic noise index, Austroads (2002)
recommends the recording of the following physical properties at each site:
ƒ traffic conditions, including traffic volume
ƒ speed and composition
ƒ road gradient and pavement type
ƒ topographical detail
ƒ weather conditions, including wind speed and direction, and rainfall.

In view of the random nature of noise and the intervention of many environmental factors, noise
surveys must be conducted with great care and in accordance with the relevant standards. These
include Australian Standard AS IEC 61672.1-2004 Electroacoustics – Sound Level Meters,
Australian Standard AS 1055-1997 Acoustics – Description and measurement of environmental
noise, and Australian Standard AS 2702-1984 Acoustics – Methods for the measurement of road
traffic noise.

In noise impact studies, sites should be chosen to cover the entire area expected to be affected by
changes in traffic patterns, if technically and financially feasible. Where this is not possible, it may
become necessary to use a few measurement sites to provide data for calibrating a general noise
prediction model. The calibrated model should then be used to provide sound levels at
representative sites.

In residential areas, measurements are usually taken one metre from the house or building facade
fronting the roadway. Observations should be taken when the road surface is dry and maximum
wind speeds should not exceed 7.2 km/h at a height of 1.2 m.

Austroads 2009
— 154 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

A windshield should be used on the measurement microphone and no measurements should be

undertaken when wind speeds exceed 36 km/h. Other issues that should be noted include:
ƒ Measurement positions should, in general, not be so far back from the road that the length of
visible road subtends an angle less than 150o symmetrically at the measurement position.
They should not be closer than 5 m to the edge of the nearside running lane for traffic.
ƒ Measurement positions should be located if possible at least 15 m away from any reflecting
surface, other than the ground. This is known as a free field environment.
ƒ The ground cover in the intervening area between an observation position and road affects
noise measurement results. Sites should be chosen to maintain uniformity of ground cover.
ƒ Microphones should be pointed towards the noise source, and a random incidence corrector
should be fitted if sound reflecting surfaces are nearby.

Road surface textures and grades also have some influence in noise measurements. Also,
observation sites should not be close to intersections and side-roads – unless they are being
studied specifically.

F.1.4 Surveys of Community Reactions

The degree of annoyance with noise pollution is usually ascertained from attitudinal surveys of
residents. Another method used to indicate the effects of traffic noise in residential areas is the
comparison of real estate values between neighbouring subdivisions and streets subject to
different levels of noise. If all other factors and features of residential property in different
neighbouring streets are the same, then the difference in noise pollution may be reflected in the
difference in property values. The comparison of real estate values is one way to determine the
externality cost due to noise.

F.2 Road Transport Fuels and Emissions

F.2.1 Fuel and Emission Types
Nearly all vehicles currently in use are powered by petroleum-based liquid fuels. The principal
liquid fuel for road transport use is petrol, with some use also made of diesel fuel and liquid
petroleum gas (LPG). Each of these fuels has its own properties.

Since 1986, new cars registered in Australia have been required to use unleaded petrol (ULP).
The number of vehicles using ULP has risen to almost 70% of all registered vehicles in Australia,
according to the Motor Vehicle Census of the Australian Bureau of Statistics (Catalogue No.
9309.0). ULP is designed to allow the use of catalytic converter systems in vehicle exhausts, to
minimise emissions of the gaseous pollutants carbon monoxide (CO), various hydrocarbons (HC)
and oxides of nitrogen (NOX). Lead poisons the rare metal catalysts (e.g. platinum or palladium)
used in the conversion of exhaust gases to harmless emissions, hence the need for lead-free fuel.

Super-grade (98 octane) leaded petrol results in the emission of particulate lead (Pb) in addition to
CO, HC and NOX. It is being phased out in Australia. A number of lead replacement fuels, which
are kinder to the environment, are now available on the market for cars manufactured before 1986.
Lead emissions are subsequently expected to drop further.

Diesel fuel is widely used for large vehicles and, to a limited extent, by passenger cars. Diesel
engines have the advantage that they provide greater fuel efficiency for the same volume of fuel.
Diesel fuel also offers significant reductions in CO emissions but oxides of sulphur (SOX) are
released, creating potential for acid rain. Low sulphur diesel is now available and legislation is in
place in some jurisdictions to limit sulphur in diesel emissions to 500 parts per million (ppm). Ultra
low sulphur diesel is being developed to reduce sulphur emissions further to 50 ppm.

Austroads 2009
— 155 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

LPG offers a cheap alternative to petrol, which can be readily used by most petrol-engined
vehicles. LPG is regarded as cheap because it is an unavoidable by-product of the refining
process and is subject to a different taxation regime. Gaseous emissions from LPG fuel are less
than those from petrol (for CO, HC, and NOX). There is also an absence of particulate lead. Fuel
consumption is higher in LPG engines than in petrol engines.

Guidelines are in place to further reduce fuel emissions in the Fuel Standard (Petrol) Determination
2001 and Fuel Standard (Automotive Diesel) Determination 2001. Alternative fuels to petroleum-
based ones include electricity, ethanol and hydrogen. These are slowly being developed and
introduced to the market. A private bus company in Melbourne is currently running a pilot program
with buses powered by ethanol. Car manufacturers in a number of countries have released hybrid
vehicles running on electricity and petrol, with claims of emission reductions of up to 80 per cent.

F.2.2 Air Pollution

Air pollution in urban areas consists of the primary emissions CO, HC, NOX, SOX, fine particulate
lead (Pb), dust and soot, as well as secondary or derived pollution such as photochemical smog.
Safe and annoying levels of various pollutants have been set down. Table F 1 lists the standard
levels of pollutants that are measured by the Environment Protection Authority (EPA) of Victoria,
based on either the National Environment Protection Measure (Air NEPM) or the State
Environment Protection Policy (The Air Environment) (SEPP) standard levels.

Table F 1: Pollutants, standard levels and calculation averaging times

Pollutant Standard level Source Averaging time How calculated*
Ozone 100 ppb Air NEPM 1 hour A
Nitrogen dioxide 120 ppb Air NEPM 1 hour A
Sulfur dioxide 200 ppb Air NEPM 1 hour A
Carbon monoxide 9 ppm Air NEPM 8 hour B
Fine particulates (PM10) 50 ȝg/m3 Air NEPM 24 hour C
Visibility (airborne particle index) 2.35 SEPP 1 hour A
Note: ppb = parts per billion
* A: maximum of the preceding 24 one hour averages; B: maximum of the proceeding 16 rolling eight hour averages; C: current 24 hour average result.

At high concentrations, carbon monoxide is a dangerous asphyxiant for mammals because it is

absorbed readily into the bloodstream and competes with oxygen in the circulating blood. Lead is
a cumulative poison. Hydrocarbons and oxides of nitrogen in the presence of sunlight under stable
meteorological conditions may chemically react over a period of hours and result in the formation
of photochemical smog. Ozone (O3) is one of the many components of smog.

Computer models are now commonly used to forecast area-wide environmental impact due to
traffic management measures such as lower urban speed limits (Taylor 2000) or congestion
charges (Mitchell et al. 2003). These models consider the interaction of route diversions, air
quality and energy usage.

F.2.3 Surveys of Fuel Consumption and Emissions

Numerous factors exist that affect fuel consumption. These include vehicle size, shape and mass,
engine power and capacity, and types of tyres, as well as factors independent of the car such as
acceleration and deceleration, stopping frequency, traffic conditions, and road surface and
curvature. Surveys of fuel consumption and emissions can be divided into three basic categories:
ƒ individual vehicle-based surveys

Austroads 2009
— 156 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

ƒ area-wide (system-wide) surveys

ƒ attitudinal surveys of households or individuals.

It is possible to observe individual vehicle performance, fuel consumption, and emissions in a

laboratory using a dynamometer. A dynamometer is a test bed on which a vehicle is fixed, with the
driving wheels in contact with a system of flywheels that simulate the inertia of the vehicle. The
wheels are also in contact with a power absorption unit, which can be adjusted to reproduce
different road conditions as speed is varied. The vehicle’s acceleration, deceleration and speed
performance can then be recorded directly. Measuring instruments are attached to the vehicle’s
engine/carburettor and exhaust to measure instantaneous fuel consumption and exhaust
emissions respectively. A computer is attached to directly record the data.

Figure F 2 is a schematic representation of a dynamometer set up to record fuel consumption and

exhaust emissions. A driver’s aid indicates the desired speed-time curve for the test run, and
provides feedback to the driver on his success in following the curve. Samples of exhaust gas are
collected by the constant volume sampling (CVS) method. The CVS unit has a constant flow pump
system, with a capacity much greater then the maximum flow rate of the vehicle exhaust. The
difference is made up by ambient air fed via the filters, mixing chamber and heat exchanger
(Figure F 2), so that the concentration of exhaust gases in the dilute mixture is proportional to the
exhaust flow rate. Continuous samples of the dilute mixture are extracted by a smaller constant
flow rate device (the sampler pump). Samples are stored for analysis of CO, HC, NOX and carbon
dioxide (CO2). Samples of filtered ambient air are also taken, so that corrections may be made for
contamination from other sources.

Figure F 2: A vehicle dynamometer using constant volume sampling

Fuel consumption data are taken by one of two methods: fuel flow meters or the carbon balance
method. Fuel flow meters are best suited in instrumented vehicles, where they can be installed for
long periods of time. Dynamometer testing usually involves short duration tests on large numbers
of vehicles, so that fuel meters are inconvenient. The carbon balance method uses equipment and
procedures similar to those used for emissions testing and can be conducted simultaneously.
Dynamometer testing and the carbon balance methods are described in further detail in Australian
Standard DR 01043, Methods of test for fuel consumption of motor vehicles designed to comply
with Australian Design Rule 37.

Austroads 2009
— 157 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

A major limitation of the dynamometer test is whether or not the results of the laboratory tests
represent observed field data. The alternative experimental method is then to use an instrumented
vehicle, driven in a traffic stream, to record these on-road data. Usually only fuel consumption can
be recorded, as data logging of exhaust emissions requires some device to be placed over the
exhaust system.

The extrapolation of individual fuel consumption and emission data to more aggregated levels in a
traffic system is difficult. The characteristics of the individual test vehicle need to be related to all
of the driver/vehicle combinations in a traffic stream to permit this aggregation. Experimental
design methodology developed by the CSIRO (Johnston et al. 1982) may make the extrapolation
possible. Chase cars can also provide some data. The chase car pursues a vehicle selected at
random along a survey route, logging speed-time performance. The driving cycle can then be
used on the dynamometer to yield laboratory estimates of fuel consumption and emissions.

Fuel consumption data over long periods can be collected using conventional traffic survey
techniques such as interviews, questionnaires and diaries. These methods can provide data on
fuel purchase and travel behaviour, travel distances and trip destinations. The quarterly ABS
(2007) Survey of Motor Vehicles Usage (Catalogue no. 9208.0) employs the questionnaire method
to estimate fuel consumption and other data.

Air pollution can also be measured on an area-wide level. For example, the Environmental
Protection Authority of Victoria (EPA) has various permanent recording stations set up across
metropolitan Melbourne, which provide hourly air quality data to a central computer for monitoring
and analysis. The area-wide systems record absolute levels of air pollution from all sources, not
just road traffic, providing continuous base data. To gather more detailed local data, mobile testing
stations can be used. The CSIRO Division of Atmospheric Research has developed a portable
laser-based device called lidar (for light detection and ranging) to measure vehicle emissions and
other measurements. The lidar emits brief and intense beams of coherent light. The air pollutants,
dust and water droplets in the atmosphere absorb or backscatter the light, or rotate the plane of the
polarised beam. The intensity and polarisation of the backscattered light yields details of the
composition of pollutants in the beam’s narrow path.

F.3 Data Analysis and Presentation of Results

In the case of traffic noise surveys, NAASRA (1980) and Australian Standard AS 2702-1984
provide guidelines for the presentation of results. These include a description of the following
details in the reporting of results:
ƒ location sketch
ƒ detailed site geometry
ƒ microphone location
ƒ weather conditions
ƒ time of day
ƒ traffic flows and composition
ƒ speed limits
ƒ instrumentation used
ƒ traffic noise descriptors (e.g. LAeq) and sampling methods used to determine them.

Austroads 2009
— 158 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

There are no specific rules for data analysis and the presentation of results from fuel and
emissions surveys, because of the specialised nature of the different survey types. Each survey
needs to be treated on its merits, using the principles for analysis and reporting cited in Section 1.

The EPA in Victoria measures pollutant levels on an hourly basis, and issues air quality bulletins.
The air quality bulletins are expressed in terms of an air quality index:
Pollutant Concentrat ion
Index x 100
Pollutant Standard Level

The standard level of pollutant used is either the Air NEPM or SEPP standard, as already shown in
Table F 1. The level of pollutants is assigned a category, based on the calculated index value for
the relevant period. The categories are shown in Table F 2.

Table F 2: Environment Protection Authority of Victoria (EPA) air quality categories

Category Index range
Very good air quality 0-33
Good air quality 34-66
Fair air quality 67-99
Poor air quality 100-149
Very poor air quality 150 or greater

In summary, the systems planning process for traffic survey methods places a requirement on
traffic engineers to investigate the impacts of traffic on other systems. Traffic surveys need to
account for the environment and energy consumption, as these are areas in which traffic has
significant impacts. The present methodology is still not completely satisfactory. It is apparent that
noise, fuel and emissions surveys need to be undertaken through a multi-disciplinary approach, to
which the traffic engineer has much to contribute.

Austroads 2009
— 159 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX G VEHICLE MASS AND DIMENSIONS

SURVEYS
This section describes the methodologies employed by member authorities for the measurement of
vehicle mass, and surveys undertaken by the Australian Bureau of Statistics that are relevant to
heavy vehicles.

G.1 Vehicle Dimensions

Notwithstanding the high cost and possible bias that may be involved, there will continue to be a
need for data on vehicle dimensions for which it will be necessary to stop and measure a sample of
the vehicles at a suitable site. The data collected usually involves vehicle dimensions such as
those on which legal limits are set, e.g. overall length, width, height and other internal dimensions
such as axle spacing and front and rear overhang, which may be important for determining vehicle
mass regulations, vehicle off-tracking and turning circle. The latter is important for geometric
design of intersections and parking and loading areas.

Automatic vehicle classifiers are now readily available to classify vehicles by length or axle
configurations. Loop sensors on freeway and arterial roads, or virtual loops using video imaging in
urban environment, have also enabled the real-time collection of classified counts. Intelligent
Transport Systems (ITS) can potentially provide useful continuing and permanent classified counts,
although the quality of these data needs to be checked.

G.2 Vehicle Mass

Apart from the important aspect of monitoring and enforcing gross vehicle mass and axle load
regulations to control excessive damage to roads and bridges, vehicle mass data are particularly
important in bridge design and in pavement design and management. This is illustrated in the
shaded area of Figure A 11 in Section A.5.1 indicating the requirement of axle spacing, axle load
and load frequency (repetitions) for the structural design of bridges and pavements. In addition
vehicle mass data plays an important part in economic analysis, transport planning and freight
route planning.

Vehicle mass is often expressed in terms of the number of equivalent standard axle (ESA) loads.
A standard axle load is treated in Australia as the load of 8.2 tonnes on a single axle using dual
tyres. The number of equivalent standard axles (NESA) for a particular load and axle configuration
is defined as the number of passes of the standard axle load which could cause the same damage
on a pavement as a single pass of the load and axle configuration in question (Lay 1998).

To calculate the NESA, first assume that deflection is used as the criterion of damage. It is common
to employ a fourth power law to relate the deflection and hence damage to the fourth power of the
applied load. Let Pa be the load on the axle configuration in question, and PESA be the standard
equivalent axle load (= 8.2 tonne). Then one pass of a load of Pa on the axle configuration in
question produces the same deflection as NESA passes of the standard axle load. The equation is
therefore:
N ESA x (PESA ) 4 1 x (Pa ) 4
or
N ESA (Pa / PESA ) 4

Austroads 2009
— 160 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

For example, the pavement deflection due to a 8.2 tonne load on a single axle, single tyre is 1.5
times the same load on a standard axle (dual tyre single axle). The number of equivalent standard
axles is therefore (1.5)4 or 5.

G.3 Vehicle Mass Data Collection Systems

Many different types of low and high-speed weigh-in-motion (WIM) systems are in operation in
Australia. Austroads (2000) provided a review of technologies in use. Table G 1 summarises the
attributes of representative sensor technologies. These attributes include:
ƒ installation mode
ƒ maximum number of lanes
ƒ wheels weighed
ƒ tolerances
ƒ mass sensor life
ƒ accuracy.

Table G 1: Weigh-in-motion systems used and available in Australia

(Adapted from Austroads (2000) with updates)
Mass sensor type Bending plate Load cell Piezoelectric cable Strain gauge
WIM system name PAT DAW 100 ARRB TR ARRB TR ARRB TR
HSEMU Express-weigh CULWAY
Installation mode Semi-permanent flush Semi-permanent flush Permanent flush Semi-permanent
mounted mounted mounted within pavement
culvert installed
Max no. of lanes 4 User-defined 2 1 to 4
Wheels weighed All All All All
Tolerance of axle load (tonnes) 20 80 - 50
Temperature (°C) -40 to +70 -20 to +70 -50 to +80 -10 to +70
Sensor life span (years) - 15 - 10
Accuracy
Vendor *GVM GVM GVM GVM
(± 10% at 95%) (± 5% at 95%) (± 10% at 95%) (± 10% at 95%)
3rd party - GVM - GVM
(± 3% at 66%) (± 7% at 95%)
* GVM (± 10% at 95%) means that the accuracy of measuring gross vehicle mass is ± 10% for 95% of vehicles weighed.

Stationary or low-speed WIM systems using load cells have long been used to monitor vehicle or
axle loads at weighbridges for compliance to legislation on mass limits. The research in the last
two decades has been on high-speed WIM systems.

Of the high-speed systems, temporary surface-mounted systems offer the advantage of portability
but at a lower level of accuracy. A Golden River surface-mounted capacitive pad system was
tested by Koniditsiotis (1990) and Austroads (2000), and the accuracy was found to be ± 18% for
gross vehicle mass (GVM) and even lower for individual axle mass. Most WIM sites in Australia
now employ semi-permanent or permanent flush-mounted systems, and strain gauge/transducer
systems. Brief descriptions of these systems are given below.

Austroads 2009
— 161 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

G.3.1 Flush-mounted Systems

The flush-mounted sensors used in Australia include the following:
ƒ bending plate, as used in the LAW-100 system from the PAT Group
ƒ load cells, as used in the High Speed Electronic Mass Unit (HSEMU) from ARRB Group Ltd
(Samuels 1988)
ƒ piezoelectric cable, e.g. the Kistler quartz sensor and used in the Express-weigh system from
ARRB Group Ltd.

The bending plate and load cells can be installed on-site and thus offer some portability. The
piezoelectric cable has to be permanently mounted in-ground.

The piezoelectric cable has been used as an axle detection sensor (Section A.1; Luk and Brown
1987), especially in the CULWAY system described below. It produces a measurable electrical
response when pressure is applied to it. The magnitude of the electrical response is related to the
pressure applied so that this device can be used also as a weighing device.

Figure G 1 illustrates the cross-section of a quartz piezoelectric WIM sensor. The quartz-sensing
material has shown good reliability in yielding electric charges proportional to the applied force. It
is practically not affected by temperature changes and is suitable for very slow-moving traffic.

Figure G 1: Quartz piezoelectric sensor

G.3.2 Strain Gauge/Transducer Systems and CULWAY

The primary example of the strain gauge/transducer system currently used in Australia is
CULWAY. Earlier types included systems such as FASTWEIGH (Moses 1979) and
AXWAY(Peters 1984) for bridges. CULWAY has been installed in many culvert sites in most
states and territories (Figure G 2). It is a high-speed weigh-in-motion system and provides data on
axle loads as well as vehicle classifications and speeds.

Austroads 2009
— 162 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure G 2: A CULWAY data logger

Figure G 3 shows a schematic layout of a CULWAY installation with the following main
components:
ƒ Strain sensors – these are attached to the roof section of an appropriately sized box culvert.
ƒ Two axle detectors – piezoelectric cables are embedded in the road surface. These are
placed at a set distance apart, one over the centre of the culvert in which the strain sensors
are placed and the other at a known distance (e.g. 10 m is used in Western Australia)
upstream. The axle detectors provide data for vehicle classification and speed and are also
used to trigger the recording of strain data as each axle passes over the culvert.
ƒ Data logger – this is either battery or mains powered. The logger stores vehicle count, strain
measurements and other data on vehicles heavier than a minimum gross mass threshold.
This threshold is generally 2.5 tonnes gross vehicle mass, although a 4.5 tonne threshold is
sometimes used to reduce the amount of stored data and maintain accuracy in the critical
higher mass range.
ƒ PC – it is used to set up the data acquisition system and retrieve the stored data. It may also
be used to carry out some processing and reporting functions.

Figure G 3: Elements of a CULWAY axle weight measuring system

Austroads 2009
— 163 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The CULWAY system generally operates unattended with data being periodically retrieved when
the recording capacity of the unit is reaching its limit. This is generally about 10,000 vehicles or
about two weeks’ duration, depending on battery life and other factors. The data may also be
retrieved by appropriate telemetry using data modems, avoiding the need for regular site visits.
Some CULWAY systems now operate with solar power and are particularly useful at remote
locations.

G.4 Tonnage Estimation

It is also possible to estimate the tonnage of a heavy vehicle from its axle configurations using
historical WIM data, say, from CULWAY. The relationship would be site-dependent but is useful
when WIM systems are not available. Some results reported in Pearson and Foley (2001)
suggested the following relationship from data collected at 16 CULWAY sites in Victoria from 1995
to 1998 (Table G 2).

Table G 2: Mean axle group and gross vehicle mass

Mean mass (tonne) Trend (% growth
1995 1996 1997 1998 per annum)

Class 9 articulated vehicle [0 00 000]

Axle group mass 2 8.9 9.4 9.5 9.2 1.3%
Axle group mass 3 10.1 10.7 10.7 10.5 1.3%
Gross vehicle mass 23.5 25.3 25.3 24.9 1.8%
Class 10 B-double [0 00 000 000]
Axle group mass 2 8.6 9.5 9.5 9.3 2.4%
Axle group mass 3 9.7 10.8 10.9 10.7 3.3%
Axle group mass 4 9.2 10.4 10.4 10.2 3.3%
Gross vehicle mass 32.0 35.9 36.0 35.6 3.4%

G.5 Compliance of Weigh-in-motion Technologies

According to Austroads (2000), there were 171 highway-speed WIM systems installed in Australia.
Of these, 140 installations adopt the CULWAY system that employs the strain-gauge as the mass
sensor. Due to the dominance of CULWAY for WIM applications, both the quality and reporting of
data is quite uniform amongst different users. Table G 3 shows the accuracy requirements of WIM
applications.

Table G 3: WIM Accuracy Requirements for Various Applications

Applications WIM accuracy (gross vehicle mass
for 95% of vehicles)
Economic analysis
Transport studies <20%
Classification of vehicles
Road and bridge management
Overload warning
<15%
Preselection for static weighing
Road Safety
Enforcement
<5%
Compliance

Austroads 2009
— 164 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

WIM systems installed in Australia achieve an accuracy of better than 10%, although some
systems specify the accuracy for only 66% of vehicles rather than 95% of vehicles. The single-
lane or multi-lane CULWAY system provides an accuracy of 10% for 95% of vehicles. Note that
there is currently no standard Australian specification or method to evaluate WIM systems.

WIM technologies have not been as widely adopted as anticipated since the early 1980s. The
maximum potential of WIM technologies can only be realised with a mass sensor that promises a
good accuracy and a low cost both for the sensor itself and its installation. CULWAY systems are
cost-effective but can only be installed at sites where culverts are available.

High-speed WIM systems have been integrated with other systems for a range of applications.
Two examples are as follows (Pearson and Foley 2001):
ƒ High-speed Electronic Mass Unit (HSEMU) – this system automatically classifies, measures
dimensions (length, height and width) and weighs each vehicle as it passes a HSEMU site at
highway speed. In NSW, the unit is used as a screening system. Every heavy vehicle is
diverted with a roadside advisory sign to a left-hand HSEMU lane and checked for
compliance to pre-defined weight and dimension parameters. Vehicles that conform to the
HSEMU check are diverted back to the traffic stream without stopping. Only those that fail
the compliance tests are diverted to a static weighbridge for further checking.
ƒ Safe-T-Cam – this system is a Roads and Traffic Authority initiative and consists of a number
of video cameras to recognise the image of a heavy vehicle and identify the numberplate
using image processing techniques. It is used to detect heavy vehicles that violate one or
more regulations such as speeding, unregistered or stolen vehicles, pollution if integrated
with pollutant sensor, or overload vehicles if integrated with WIM systems.

G.6 Australian Bureau of Statistics Surveys

The monitoring of vehicle usage, especially that of heavy vehicles, is of national significance. The
Australian Bureau of Statistics (ABS) has provided various types of data related to heavy vehicles.
Note that vehicle classes used to describe vehicle types differ from the Austroads classifications
described in Appendix A. Three relevant data sets are as follows:
ƒ Survey of Motor Vehicle Usage (ABS Catalogue No. 9202.0) – The publication provides the
annual statistics for private and commercial vehicles registered for road use with a motor
vehicle registration authority, by state or territory of registration or area of operation. It
provides the number of vehicles, total and average kilometres travelled, driver
characteristics, vehicle usage, fuel consumption and load carried. The publication also
contains statistics on buses, including kilometres travelled by bus type and main type of
service. Data on capital city travel have also been available in recent years. The publication
was first released in 1963.
ƒ Freight Movements Survey (ABS Catalogue No. 9220.0) – The publication provides
statistics on tonnes and tonne-km of freight by Statistical Divisions, commodity type and
transport mode. The commodity type is also classified by whether it was bulk, dangerous,
refrigerated or containerised. The data were initially published on a quarterly basis from the
June quarter in 1994 but the survey was suspended for a while. A different survey
methodology for the road freight component was introduced in 2000. The issue in
September 2002 covered all freight movements in the period from 1 April 2000 to 31 March
2001.

Austroads 2009
— 165 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

ƒ Motor Vehicle Census (ABS Catalogue No. 9309.0) – Results are available yearly and was
first issued in 1955. It provides data on the number of vehicles on register for each state and
territory. Vehicle classification is by type of vehicle (passenger, light commercial, rigid trucks,
articulated trucks, non-freight carrying trucks, buses and motorcycles), year of manufacture,
make, gross vehicle mass, fuel type and average age of vehicle.

G.7 Data Analysis and Results

In addition to the software associated with the data acquisition and retrieval units, a range of PC-
based software has been developed for processing the large amount of raw data derived from
CULWAY sites. Analysis programs have been developed at ARRB Group, and reporting programs
are available from member authorities.

The analysis software creates files containing summaries over various time periods from which
graphical outputs including histograms of mass and vehicle speed can be generated. Figure G 4
and Figure G 5 show examples of WIM reports from the Queensland Department of Main Roads.

Source: Queensland Department of Main Roads.

Figure G 4: Example of a WIM report at Yandina, Sunshine Coast’

Austroads 2009
— 166 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Source: Queensland Department of Main Roads.

Figure G 5: An overloading report from a WIM site at Loganholme, Brisbane Parking Surveys

Austroads 2009
— 167 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX H PARKING SURVEYS

H.1 Data Types
Many different procedures are available for collecting the variety of data types required to
understand the parking process (Young 1996). Possible information required for parking analysis
is:
ƒ the number and location of existing parking spaces, both kerbside and off-street
ƒ existing parking practices, including usage and available spaces, parking duration and illegal
parking
ƒ the need to impose or vary parking time limits or to install parking meters
ƒ the adequacy of existing enforcement.

Information is required in respect of the supply of, and the demand for, parking facilities.
H.1.1 Parking Supply Data
A fundamental part of any parking survey is an inventory of the parking facilities and the
possibilities for new development in the area of concern. Such an inventory should detail the type
of parking and its location. Supply studies seek inventory information on:
ƒ parking spaces
ƒ type of parking
ƒ method of operation of off-street facilities
ƒ parking restrictions
ƒ parking fees.

In compiling a record of street facilities, the data can first be entered on prepared sketch plans in
the field. These sketch plans may be based on existing local maps or each street can be sketched
onto a graph paper to the correct scale, with respect to length. The use of a suitable key enables
the exact location of parking and parking restrictions to be marked on the map. The location of off-
street parking could be marked on the map but the layout of these facilities is usually detailed
separately. Normally the road network is coded in relation to road lengths with each block given a
number key. This coding provides a basis for recording and analysing the data. An example of a
typical inventory map is shown in Figure G 6.

Austroads 2009
— 168 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure G 6: Typical inventory map

The inventory should be carried out by a two-person crew, equipped with a base map, kerbside
and off-street inventory forms, and a 20 m tape. Portable computers installed with geographic
information systems (GIS) are also readily available. Map data can be edited in the field to update
parking inventory data.

The inventory may also involve the inspection of signs and markings, particularly those relating to
the location of and access to parking facilities, or to parking restrictions. This inspection should
include a record of deviations from standard signs as well as poorly located, confusing, conflicting
or misleading signs and kerb markings. Information on signs placed without authorisation should
be included in the inventory.

H.1.2 Parking Demand Data

It is important to distinguish between revealed demand at the time of observation and latent
demand for parking. Revealed demand is the observed use of the facility. Latent demand is a
measure of the total desired use of the facility. Latent demand will only be exposed when supply
exceeds demand. Most parking demand data collected in the field are revealed demand data,
whereas future planning data require the latent demand.

The type of parking demand data needed includes:

ƒ spatial distribution of parking demand and parking demand generators
ƒ total number of parking events in study area over a study period
ƒ parking duration and turnover
ƒ trip purpose and destination
ƒ trip origin.

The data collection procedures may involve interview surveys and observational surveys.

Austroads 2009
— 169 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

H.2 Interview Surveys

If the parking demand is to cover a large geographic area, and it is expected that changes in
parking supply could cause substantial change in the total number of parkers or their spatial
distribution, data collection by interviews may be necessary. Four techniques commonly used are
parking person interviews, reply paid questionnaires, home interview surveys and site-specific
interview surveys.

H.2.1 Parking Person Interview

This approach involves assigning an interviewer to a predetermined number of parking spaces.
The interviewer records each parking incident and attempts to interview people parking in this
area.

Questions asked in the interview relate to the following items:

ƒ trip purpose
ƒ final destination of trip
ƒ origin of trip
ƒ places visited
ƒ duration of parking
ƒ alternative parking locations considered
ƒ frequency of parking in the study area.

Other details that can be collected by observation are:

ƒ the vehicle registration number
ƒ vehicle classification (car, taxi, truck, etc.)
ƒ nature of parking (legal, kerbside, off-street, garage, etc.)
ƒ time of arrival or departure.

The information obtained can be recorded onto an appropriate survey form (e.g. Figure G 7) and
then transferred to a computer for further analysis. Alternatively, the answers can be coded
directly into a hand-held computer. The latter approach reduces the number of times the data
must be manipulated, reducing the chance for errors in coding.

The personal interview can be used to obtain data on people’s attitude to various parking policies
(e.g. changes in parking fees, parking restrictions). Care should be taken to keep the length of the
interview to tolerable limits.

Interviews of on-street parkers can be carried out on parker arrival or departure. The departure
interview has a number of advantages. Firstly, places visited can be reported more accurately
since the parker has already visited them. Secondly, accurate information on the duration of stay
can be available. Thirdly, the parker is less likely to be in a hurry and therefore more likely to
complete the interview. The major disadvantage is that the interviewer has less time to catch
parkers before they leave.

Austroads 2009
— 170 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Non-response to the interview could introduce bias into the estimation of population statistics. The
bias may result from the people who will not answer the questions since they are in a hurry, being
unique in character and not typical of the people who complete the questions. Some attempt
should always be made to determine the character (age, sex, occupation, etc.) of the people not
completing the questionnaire. This may enable corrections to be made to the results of the
responding sample. Sampling of parking vehicles can be carried out but care must be used not to
introduce bias into the analysis.

In the case of off-street facilities, interviews can be carried out when the vehicle is entering or
leaving the facility. During peak periods, sufficient interviewers are needed to avoid the
unnecessary build up of vehicles. Interviewing people as they leave has the advantage of avoiding
vehicles having to queue onto the adjoining roads. If interviews are carried out upon entry,
provision for queuing vehicles should be made to prevent disruption to traffic flows.

Facility Reference: __________________ Interviewer’s Name: _________________

Vehicle Time Trip Trip Purpose Parking

Time In Trip Origin
Type Out Destination (Circle one) Type No.
1. Shop 2. Work
3. Friend 4._____
1. Shop 2. Work
3. Friend 4._____
1. Shop 2. Work
3. Friend 4._____
1. Shop 2. Work
3. Friend 4._____
1. Shop 2. Work
3. Friend 4._____

Date: __________________ Supervisor: ____________________

Figure G 7: Typical personal interview survey form

H.2.2 Reply-paid Questionnaire

When detailed information is not required, reply-paid questionnaires may be used. A typical
example of the questions asked in a card-type questionnaire is shown in Figure G 8. These
questionnaires can be inserted under the windscreen wipers of parked vehicles. Personnel costs
for this method are smaller than for a personal interview since one person can cover a larger
number of parked vehicles. Information on the parking location and arrival time of the vehicle can
be obtained by marking or pre-coding the questionnaire. If pre-coding is used, the person
distributing must note the time of distribution and location. This can be recorded in a logbook or
hand-held computer.

The response rate of self-completion questionnaire surveys can be quite low, at about 30%. The
response rate can be increased with the offer of a prize-draw or similar enticements, and
subsequent reminder mails.

Austroads 2009
— 171 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Parking Survey
The rapid growth of population in Knox has reached a point
where parking facilities are inadequate. Your answers to the
following questions will aid in determining the need for
parking facilities.

1. Where did your trip start before parking here?

______________________________
No. Street
_______________________
Suburb

2. Where did you go after parking?

______________________________
No. Street
_______________________
Suburb

3. What time did you park today?

_______am _______pm

What time did you move your car from this parking space?

_______am _______pm

4. What did you do while your car was parked here?

(e.g. work, shop, visit friend etc.)

________________________________

Figure G 8: Questions in a Typical Reply-Paid Questionnaire

H.2.3 Home Interview Survey

The above surveys do not measure the latent demand for parking, but rather the revealed demand.
Many people wishing to visit the area may be turned away by the lack of parking facilities.
Indications of latent demand could however be obtained by a home interview survey. The large
cost associated with such an approach usually results in questions on parking being grouped with
other questions on a large transport questionnaire. Sampling of the population may be a problem if
the study is concentrating on improvements in a particular part of the urban area.

H.2.4 Site-specific Interview Survey

The home interview survey addresses the entire population of an urban area and is the only
approach that can determine the latent demand in multi-use parking lots. However, some parking
lots are only used by people working at specific locations. In such a case the total population of
possible users can be defined. This population can then be used as a basis for determining the
latent demand for the particular site. This approach will not provide an indication of the latent
demand of visitors for parking.

H.3 Observational Surveys

Many parking studies are not concerned with information on the parker’s overall trip. In this case a
simplified type of survey can be warranted. These can be either cordon counts or patrol type
surveys. Technologies are available for carrying out observational surveys by video with automatic
reading facilities or by electronic tagging.

Austroads 2009
— 172 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

H.3.1 Use of Parking Revenue Collection Data

Computerised revenue collection systems are now common in the operation of parking facilities,
and large amounts of data on entry, exit and parking duration times can easily be collected as a
by-product. Bonsall (1991) describes a typical system for an off-street carpark as one that
monitors the entry and exit of vehicles detected via inductive loops and/or barrier movements, and
all financial transactions. Arrivals of special types of users such as season ticket holders can be
logged separately. Data produced by a central computer could include (per hour intervals):
ƒ entry flows
ƒ exit flows
ƒ accumulation
ƒ fees paid – disaggregated by class of user
ƒ numbers of tickets sold
ƒ distributions of lengths of stay.

Data on the usage of unrestricted-entry off-street carparks such as pay-and-display facilities and
on-street parking spaces have traditionally been harder to collect apart from levels of revenues.
The introduction of electronic and computerised meters with various payment methods (e.g. stored
value cards, credit cards, mobile phones) has meant that information such as durations of stay and
distributions of usage can be recorded. The reading of meter data has also been revolutionised
through use of portable enquiry terminals via infrared or optical scanners, and the networking of
meters. Data from parking meters relates to transactions rather than actual usage, so correction
factors may need to be used.

H.3.2 Cordon Counts

In this type of survey, the study area is surrounded by a closed cordon. Counting stations are
established on all crossroads entering and leaving the cordon. At each station, a separate count is
made of vehicles entering and leaving the area hourly or in shorter periods. The algebraic
summation of entering and leaving traffic gives the accumulation of vehicles in the area. This
accumulation represents the sum of vehicles parked and on the move in the study area. After
identification and removal of the moving vehicles, a measure of the required parking facilities is
obtained. The accumulation revealed can also act as a control on the usage of a carpark.

Counting can be carried out either manually or by automatic counters. Manual methods are more
expensive but may be required in special surveys, or to check and/or make corrections to the
automatic counts. More detailed information can be obtained by recording the numberplates of the
vehicles entering and exiting the cordon. The numberplate survey technique has already been
discussed in previous sections.

The following information can be obtained from cordon counts;

ƒ total number of parkers
ƒ arrival and departure rates
ƒ composition of population
ƒ parking accumulation
ƒ duration of parking.

Austroads 2009
— 173 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

H.3.3 Patrol Surveys

The patrol survey approach involves an observer patrolling along a predetermined route at fixed
intervals. The location of parked vehicles and/or their registration numbers are recorded. Each trip
around the section enables the accumulation for the parking facility to be estimated. Further, the
number of times a vehicle is observed in the same parking place multiplied by the observation
interval gives an indication of the parking duration.

The information that can be obtained using this approach is:

ƒ total number of parkers
ƒ arrival rate
ƒ departure rate
ƒ parking accumulation
ƒ parking duration
ƒ spatial distribution of parkers within the parking lot.

The study area must be divided into sections sufficiently small for each to be toured once every
hour, half-hour or other small interval of time.

Patrolling by car enables longer sections to be considered in a given interval but both a driver and
an observer are required. The observer would record the location and numberplates of parked
vehicles, preferably using a portable computer to reduce analysis time and transcribing errors. It is
important that the recording of the numberplates is done as inconspicuously as possible. Drivers
of the cars being observed may change their normal habits if they are aware they are being
observed.

Patrol surveys can be used to locate the spatial distribution of particular users. This requires the
users to be identified by either a parking sticker or predetermined numberplates.

An obvious disadvantage of the patrol method is that many short-term parkers may be missed.
The number missed depends on the interval of observation and the distribution of parking duration.
Corrections can be made by carrying out a small survey of a particular location to find the general
shape of the distribution. Adjustments can be made using this sample survey. Richardson (1974)
also described a statistical method to determine these corrections.

H.4 Presentation and Analysis of Parking Data

The presentation and analysis of parking data has been facilitated by PC software such as
spreadsheets, statistical/graphical analysis packages, high quality graphics functions and
databases such as GIS. Typical results would include summaries of:
ƒ variations in parking demand throughout the day and comparisons with available parking
space on the kerb-side and in off-street areas
ƒ variations in traffic flow and vehicle accumulation in the study area throughout the day
ƒ parking duration and comparison with existing and proposed time limits
ƒ time series, spatial and comparative analyses.

Austroads 2009
— 174 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

H.5 Summary on Parking Surveys

The introduction of electronic and computerised revenue collection equipment for parking systems
has greatly increased the ease with which parking data can now be collected. The ease of
undertaking observational surveys has also increased with the introduction of video and voice
recognition technologies.

Advanced technology has been developed for remotely monitoring space occupancies in a
carpark, but these technologies are still relatively expensive. Photoelectric and infra-red beams, or
inductive loops, could be installed to monitor the occupancy of each parking space. Electronic
chips energised by radio signals identify individual vehicles. When economical, these technologies
will aid in the automatic collection of parking data, but privacy issues will again need to be
addressed.

Austroads 2009
— 175 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX I TRAFFIC GENERATION SURVEYS

I.1 Data Collection
The techniques for collecting traffic generation data are of two broad types:
ƒ techniques for observing traffic movements at a site
ƒ interview or questionnaire surveys of people involved in a given activity in a given study area.

These techniques involve the application of other traffic survey methods, such as traffic counting
(Appendix A), origin-destination surveys (Appendix D) and parking surveys (Appendix H), or
interviewing (Richardson et al. 1995). Observed data on traffic flows, numbers and travel patterns
of people, etc. may then be analysed to produce rates or absolute levels of traffic generation for a
given land use activity.

Most surveys of traffic generation are conducted as observational surveys, particularly when the
land use activity can be ascribed to a well-defined area or a site, such as a shopping centre or fast
food restaurant. When the land use activity is more dispersed, such as the traffic generation from
a residential area, then the interview or questionnaire survey should provide more useful data.

I.2 Observational Surveys

Observational surveys of traffic generation involve the recording of traffic movements across a
cordon line drawn around the site. The surveys involve the collection of traffic count data, either
manually or automatically. These data indicate the accumulation of traffic at a site over time.
Further data, such as the duration of stay in the survey area, may be collected using the methods
for parking surveys. Observational methods have been used to collect data for the Roads and
Traffic Authority (1995) and Institute of Transportation Engineers (2008) data sets. An example of
their use for analysing traffic generated by shopping centres in an Australian city is described in
Foley (1981).

A summary of the procedures for collecting traffic generation data is as follows (Institute of
Transportation Engineers 2006):
ƒ Select traffic generators with similar development characteristics.
ƒ Conduct automatic counts for at least a three-day period. The counts should be directional in
15 min time periods, and the count locations should avoid through traffic and double
counting.
ƒ If peak traffic hours are not known, choose a typical week of the year to provide the weekday
and weekend peak hours. If the development exhibits large seasonal variations, ‘design
days’ should be selected, reflecting the 30th highest hour.
ƒ If only peak period counts are needed, conduct manual counts on one or more typical
weekdays. Vehicle classification and occupancy data should be collected where appropriate.
In some cases, weekend counts may be necessary to cover developments with peak activity
on Saturdays or Sundays, e.g. shopping centres, markets and museums.

Observed behaviour, particularly that specific to one transport mode (such as private vehicles) and
a particular site (such as the off-street car park at a shopping centre), provide a measure of
revealed demand. Thus, a count of vehicles entering or leaving the shopping centre indicates the
actual use made of the facility. It does not indicate the overall demand to use that facility the
observational survey will provide no information on the latent demand (Appendix H).

Austroads 2009
— 176 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

I.3 Interview and Questionnaire Surveys

If there is reason to believe that a development has unique traffic generation characteristics,
interview or questionnaire surveys should be used to determine the average weekday person trip
ends by transport mode, and the number of trips produced by and attracted to the site. Information
concerning the survey generator should be obtained through interviews with the site owner or
manager, telephone conversations and mail-out/mail-back questionnaire.

The survey should obtain as many variables as possible to determine those closely correlated to
traffic generation. Some essential variables are (Institute of Transportation Engineers 2006 and
2008):
ƒ current number of employees
ƒ gross, net or leasable floor areas
ƒ number of occupied rooms and dwelling units
ƒ population and acreage of the development
ƒ hours of site operation
ƒ work shifts
ƒ available public transport.

Interview and questionnaire surveys should theoretically provide better information than from
observations of vehicle movements because they can provide a better understanding of travel
demand. These surveys are, however, more expensive and difficult to design and administer than
the observational surveys.

I.4 Summary of Survey Requirements

The general principles for traffic survey design and management are of crucial importance in traffic
generation studies. The following particular requirements should be noted:
ƒ Exact definitions must be provided for the traffic variables to be measured. For example, in a
shopping centre survey, is it all people arriving at the centre, all vehicles arriving, or those
vehicles that park there?
ƒ Decisions must be made as to whether the measured generation includes all vehicles, or just
private vehicles. For example, should service and goods vehicles be included in, say, a
shopping centre or residential area survey?
ƒ Area-wide surveys need to account for through traffic movements separately. Through traffic
would be defined as that part of the traffic not stopping in the survey area.
ƒ The possibility of internal trip movements, i.e. trips not crossing the external cordon, must be
considered in area-wide studies. The larger the survey area, the more likely are such trip
movements. Cordon-line observations will miss these movements.
ƒ Precise definition of traffic generation rates is needed, e.g. the measurement of trip ends (the
sum of trip productions and attractions) is preferred to a measurement of trips per se. This is
particularly important when examining time profiles of traffic activity.
ƒ Collection of long-term data is recommended to provide information on the variations in traffic
generation over days of the week and weeks of the year.
ƒ As traffic generation is often related to measures of activity, such as floor area, care is
needed to ensure consistent measurements and definitions are used. There are several
alternatives for area, e.g. gross area, net floor space, leasable floor space, etc.

Austroads 2009
— 177 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

I.5 Forecasting Traffic Impacts

Computer models are often used to forecast the traffic generation impacts on a road system due to
new developments such as a shopping centre or other land use changes in general.

The computer program SIDRA (Akçelik and Associates 2002), is a tool often employed in traffic
impact studies. It is suitable for the evaluation of alternative intersection designs in terms of
capacity, level of service and other performance indicators such as delay, queue length, fuel
consumption and pollutant emissions. The software is particularly useful as an evaluation tool to
investigate alternative traffic growth scenarios in traffic impact studies. Users supply traffic counts,
the geometric details of the intersection and the types of intersection control, e.g. signal, give-way
signs, roundabout, etc. and the program analyses capacity and performance of the intersection
lane-by-lane.

A model with traffic assignment capability is necessary to predict changes in the flow pattern of a
study area. The use and application of microsimulation traffic models in assessing changes in
network traffic patterns is reviewed in Austroads (2006). See also Part 8 of Guide to Traffic
Management for a summary of their use in local area traffic management studies (Austroads
2008c).

Austroads 2009
— 178 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

APPENDIX J ERROR IN SEASONAL ADJUSTMENT

FACTOR
Suppose counts are available from three permanent stations, which have similar AADTs and
exhibit similar patterns.

Calculate ADT for each month for the three stations and then 12 average seasonal adjustment
factors to convert each ADT to AADT. Select a random, seven day count in each month for each
station and calculate the ADT. Apply the corresponding Seasonal Adjustment Factor for the
particular month to obtain an estimated AADT, and compare each with the actual AADT for the
particular station.

Calculate the number of estimates that are within nominated accuracy. This proportion of
estimates will represent the confidence limit corresponding to the accuracy. A typical tabulation is
shown in Table J 1.

Table J 1: Typical values for the estimation of errors in seasonal adjustment factor
Relative % error Number of Accumulated %
in Estimated AADT estimates of estimates
d 1.0 1 2.8
1.1 – 5.0 14 41.7
5.0 – 6.8 11 72.2
6.8 – 8.5 3 80.6
8.5 – 16.0 6 97.2
> 16.0 1 100.0
Sum 36 -

The data from Table J 1 is plotted in Figure J 1, from which the errors at different confidence limits
can be estimated. For example, at the 75% confidence level, the % error due to the Seasonal
Adjustment Factor is 7.4%.

In the example given in Appendix A (Section A.4.3), the estimated AADT is 7051 vehicles from the
Short Term Station. Applying this error to the estimated AADT gives:
7.4
Absolute error = u 7051 = 522 vehicles
100

Austroads 2009
— 179 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Accumulative % of estimates
100

80

60

40

20

0
0 5 10 15 20
% error in AADT

Figure J 1: Determination of errors due to seasonal adjustment factor

The absolute error from the count duration is 818 vehicles (Section A.4.3). This error can be
combined with the above error from Seasonal Adjustment Factor as follows:

Combined error = 818 2  522 2 = 970 vehicles

970
Combined % error = u 100 = 13.8%
7051

Austroads 2009
— 180 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

COMMENTARY 1 CAPACITY ANALYSIS – EXAMPLES

C1.1 Uninterrupted Single-lane Flow
C1.1.1 Example
In order to carry out major maintenance works on one carriageway of a four-lane divided rural
arterial road in rolling terrain, it has been decided to temporarily restrict traffic to a single lane with a
pavement width of 3.2 m, with minimum lateral clearances of 1 m on each side of it. Calculate the
capacity of a single lane under these conditions, assuming uninterrupted flow, and 10% trucks.

Solution

The base equation is C = 1,800 fW fHV

In this case:

fW = 0.8 (from Table 4.1)

EHV = 4.0 (from Table 4.2)

fHV = 1 / [1 + PHV (EHV – 1)] = 0.77

C = 800 x 0.8 x 0.77 = 1,110 veh/h

C1.2 Metered Flow

C1.2.1 Example
A total of 600 vehicles per hour arrive at a toll booth, the average service time being 4 seconds per
vehicle. Calculate the expected mean queue length and mean waiting times. What is the
probability of the queue length exceeding 4 vehicles?

Solution

Use the relevant equations from Section 5.1.

In this case:

r = 600 / 3,600 = 1/6 vehicles per second

s = 1/4 vehicles per second

ȡ = 1/6 / 1/4 = 2/3

The mean queue length, including the vehicle being serviced:

nq = ȡ / (1-ȡ ) = 2/3 / 1/3 = 2

The mean waiting time in the system, including the time being serviced

wm = nq/qa = 2 / 1/6 = 12 seconds

Austroads 2009
— 181 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

The probability of n (= 4) vehicles in the system, including the one being serviced:

= (1 – ȡ) ȡ n = (1 – 2/3) ( 2/3)4 = 0.066 or 6.6%

C1.3 Signalised Intersections

A simple example is given here. Refer to Akçelik (1989) for more comprehensive examples.

C1.3.1 Example

Figure C1 1: Example: traffic flows at signalised intersection

The signalised intersection approach in Figure C1 1 is located in a small shopping area where
numbers of pedestrians, and the effects of loading and unloading of goods are small. It is
designed to desirable geometric standards and the approach grade is flat. Calculate the saturation
flow for the approach, and its degree of saturation, assuming the green time ratio is 0.5.

Solution
Using the descriptions in Section 6.4.2, it can be assumed that:
ƒ environment class = B
ƒ lane type = 2 (for each lane).
From Table 6.4, the basic saturation flow for each lane is 1,670 through car units per hour.

The basic saturation flows should be adjusted to take account of lane widths, grades and traffic
composition.

In this case:
As the lane widths are 4 m, then fw = 0.83 + 0.05 w = 1.03

As the approach grade is flat, then fg = 1.00

The traffic composition factor is calculated as follows:

Austroads 2009
— 182 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Adopt an average value of eo = 3, then the data can be tabulated as follows:

Left Through Right Total

Car Truck Car Truck Car Truck
Flow (v/h) 80 20 720 70 120 10 1,020

Through car 1.25 2.5 1 2 3 4

units/vehicle
Weighted flow 100 50 720 140 360 40 1,410
(tcu)
Thus:
fc= 1410/1020=1.38

Hence, the estimated saturation flow for the approach:

= fW fg (1 / fc )Sb

= 1.03 x 1.0 x (1/1.38) (1,670 + 1,670)

= 2,490 vehicles per hour

The flow ratio for the approach equals the ratio of the arrival flow to the saturation flow. i.e.
y = 1,020 / 2,490 = 0.41

Given that the green time ratio u = 0.50, the movement degree of saturation:
= y / u = 0.41/0.50 = 0.82

Austroads 2009
— 183 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

COMMENTARY 2 PERFORMANCE OF ROUNDABOUTS

The performance of roundabouts can be explained by evaluating the critical acceptance
parameters. Their measurement does need to be explained.

C2.1 Measuring the Critical Acceptance Parameters

Troutbeck (1989) found that the critical gap parameters were dependent on the circulating flow. At
higher circulating flows the critical acceptance parameters were generally shorter. The critical
acceptance gaps were determined using the maximum likelihood method, which has been found to
have little bias. Most other techniques were significantly affected by the traffic flow in the
circulating stream (refer to Brilon et al 1999).

In the analysis of roundabouts Troutbeck found that all entering drivers gave way to all circulating
vehicles. It was on this basis that the critical acceptance headways were evaluated. If it had been
found that drivers, entering from the left lane, did not give way to all circulating drivers then the
critical acceptance gap would be evaluated on a lesser number of headways (per time period) and
the critical acceptance gap values would have been larger. The net result would be about the
same entering capacity. Similarly, if it was found that exiting vehicles affected the drivers critical
gap parameters, then the total number of vehicles (per time period) affecting drivers’ ability to enter
the roundabout would be increased; it would be the circulating and the exiting vehicles. With this
increased number of vehicles (per time period), the average gap size would be reduced and the
recorded critical gap would be similarly reduced.

C2.2 Development of the Australian Analysis Procedure

Early research by Horman and Turnbull (1974) established the critical acceptance parameters that
represented the performance of all roundabouts. Towards the end of the 1980s, it was decided
that these parameters should be related to the geometry of the intersection (Troutbeck, 1989) and
later described in previous Austroads guidelines on roundabouts.

Troutbeck’s data were collected at sites with little or no congestion. It was not possible to record
the capacity of the roundabout or an approach but only record the traffic operation through the
headway acceptance parameters.

Akçelik (1998) was able to observe congested or near congested roundabouts and concluded that
the use of the headway acceptance values recorded in previous Austroads guidelines on
roundabouts, with the absolute priority capacity theory equations, resulted in an overestimate of
capacity for some roundabouts particularly those with higher circulating flows. Akçelik included a
process to account for increased usage of one circulatory lane and for increased bunching of traffic
from a dominant approach. He included the product, pqdpcd, which was ‘the proportion of total
circulating stream flow that originated from and is queued on the dominant approach’. This term is
important in explaining how traffic from one approach affects the performance of other approaches.

Another significant innovation from Akçelik was the concept that the critical acceptance parameters
would be reduced if the entry flow were much larger than the circulating flow. This effect increased
capacity for circulating flows less than 900 pcu/h. This is discussed below.

The changes to the Austroads process by Akçelik can best be explained with graphs of the entry
capacity against circulating flow. In this example, it was assumed that the two-lane roundabout
had an inscribed diameter of 60 m, an entry lane width of 4 m and the ratio of the dominant entry-
lane flow to the subdominant entry-lane flow is 1.20.

Austroads 2009
— 184 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Figure C2 1 shows three curves: the predicted values from the previous Austroads guidelines and
the two curves based on Akçelik (1999) where the ratio of the entry flow to the circulating flow is
low and the circulating traffic originates from all entries and not from one dominant approach.

The ‘recalibration’ curve uses revised equations from Akçelik (1999) but excludes the change to
the minimum headway between circulating vehicles; this remained at 1.0 as recommended and
measured by Troutbeck (1989). The curve labelled ‘Minimum headway = 1.2 s’ includes all the
recalibration curves recommended by Akçelik. The most significant effect resulted from changing
the minimum headway between major stream vehicles from 1.0 to 1.2, which in turn then reduces
the maximum circulating flow from 3600 veh/h to 3000 veh/h.

3600
Austroads
3200
Recalibration
2800
Minimum headway = 1.2s
2400

2000

1600

1200

800

400

0
0 600 1200 1800 2400 3000 3600
Circulating flow (veh/h)

Figure C2 1: Effect on capacity of recalibration for two lane roundabout

(inscribed diameter 60 m, minimum circulating headway changed from 1.0 to 1.2 s)

As explained above, the Akçelik analysis process has two major additional features. The first is
that capacity is increased when the ratio of the entry flow to the circulating flow is high and the
second is that capacity is reduced when there is a high flow from a dominant approach and
vehicles from this approach were queued. Applying these two attributes provides the range of
possible capacities for a particular circulating flow. These are shown in Figure C2 2 below. This
figure shows a minimum capacity at high circulating flows.

Akçelik appropriately introduced time dependent delay equations, and geometric delays. These
aspects improved the evaluation process. Akçelik re-analysed the Austroads data and introduced
limits on variables; however, most of the equation forms are based on the Austroads data. In
keeping with the philosophy that SIDRA INTERSECTION is both a design and research tool,
Akçelik has enabled the user to change values from defaults.

Austroads 2009
— 185 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

Troutbeck and Kako (1999) discussed a phenomenon of ‘limited priority’. This process allows for
entering drivers to accept a short headway with the following circulating vehicle slowing a little to
accommodate the entering vehicle. This effect is more likely to occur at roundabouts with lower
operating speeds and it is why the critical acceptance gap declines with increasing flow. Drivers
are more comfortable in accepting shorter headways when relative speeds are low. Troutbeck and
Kako presented field data to demonstrate that this effect exits.

Equations to account for this limited priority behaviour are given in Troutbeck (2002). Figure C2 2
shows the predicted capacities for a two-lane roundabout with a 60 m inscribed diameter.
Figure C2 3 shows a similar curve for a single lane roundabout with a 40 m inscribed diameter and
4 m entry lanes. The limited priority process explains a likely cause for the headway acceptance
data not being able to predict capacities at roundabouts.

3600
Limited priority
3200
SIDRA Lower
2800
SIDRA Upper
2400

2000

1600

1200

800

400

0
0 600 1200 1800 2400 3000 3600
Circulating flow (veh/h)

Figure C2 2: Upper and lower bound estimates from SIDRA and the limited priority estimates of capacity

Austroads 2009
— 186 —
 GUIDE TO TRAFFIC MANAGEMENT PART 3: TRAFFIC STUDIES AND ANALYSIS

2000
Limited Priority

1600 SIDRA Lower

SIDRA Upper
1200

800

400

0
0 300 600 900 1200 1500 1800
Circulating flow (veh/h)

Figure C2 3: Upper and lower bound estimates from SIDRA and the limited priority estimates of capacity

The Australian research is held in high regard overseas. Criticism has been directed at the
research on the basis that it generally lacks a credible data set that correlates the capacity or
delays predicted from a model, or theory, with data collected on the road. The data used to define
the process in previous Austroads guidelines were collected on the road and the theory was the
best available at that time. The limited priority theory does enable the data to be used to establish
more rational predictions, but again there is only limited data to indicate where it can be applied.
Akçelik (1998) indicated anecdotally that the Austroads process overestimates capacity at times
and used a simulation model to identify suitable e values for the calibration parameters.
Unfortunately a simulation program cannot be guaranteed to give real world predictions. The
processes would be more robust if the correlation between field data and predicted values was
established.

Austroads 2009
— 187 —