Predictive Modelling Foropave - NZ RR PDF

Road
Information
Management
Systems
IMPLEMENTATION OF
PREDICTIVE MODELLING
FOR ROAD MANAGEMENT
dTIMS Research Needs
Report DT/2000/4
30 June 2000
HTC
Infrastructure Management
Specialists
HTC Infrastructure Management Ltd.

P.O. Box 177
Kumeu, Auckland 1250
NEW ZEALAND
Ph: +64-9-411-9098 Fax: +64-9-411-9096
E-mail: [email protected] Web: www.htc.co.nz
COPYRIGHT
This report is Copyright  2001 Ingenium. All rights reserved.
Brief extracts may be made from this report for technical purposes as long as they are
referenced.
Although this report is believed to be correct at the time of publication, Ingenium, HTC
Infrastructure Management Ltd., their employees or agents involved in the preparation and
publication of the report, do not accept any contractual, tortuous or other form of liability for its
contents or any consequences arising from its use. People using the information contained in the
report should apply, and rely upon, their own skill and judgement when considering a particular
issue.
© 2001 Ingenium ii
EXECUTIVE SUMMARY
dTIMS was implemented in NZ to include pavement prediction modelling to the existing pavement
management system. It is a flexible shell that provides the analytical framework while all the
models and treatments used are user defined. During Phase I of the implementation a preliminary
setup was released and tested during the Pilot Study. Participants gave feedback on the system
with regard to the application of the system on their network.
One of the tasks for Phase II of the NZ dTIMS Implementation is to establish research
requirements to enhance and refine the current system. The need for further research topics was
identified during Phase I of this project from user feedback and the decisions taken during the
Wellington Workshop held in June 1999. Additional research requirements were also identified
following on from feedback and work done during Phase II of this project.
This report has introduced and discussed the scope, the needs and the specific requirements for
predictive pavement deterioration modelling in NZ for the following research tasks required over
the ensuing years:
• Calibration Studies;
• Influence of Trench Patching and Service Covers;
• HDM-4 Road Deterioration Works Effects (RDWE) Models;
• Works Effects;
• Area Wide Treatments;
• Pavement Strength;
• Unsealed Road Deterioration Modelling;
• Texture Depth and Skid Resistance;
• Incorporation of Noise Reduction Treatments;
• Do Minimum alternative;
• Maintenance Cost Index;
• Intervention Criteria for Sealed roads; and,
• Traffic Volumes.
The research requirements can be classified into one of the following groups:
• Additional treatments to be included;

• Refinement of models; or
• Refinement to the analysis procedure.
Annex A shows a summary of the research needs. A priority is assigned to each topic based on
the urgency of the specific topic related to the refinement of the dTIMS Setup.
© 2001 Ingenium iii

CONTENTS
1. RESEARCH REQUIREMENTS FOR dTIMS IN NEW ZEALAND

.............................................................................................. 9
1.1 Background......................................................................................... 9
1.2 Identified Needs for Further Research................................................. 9
1.3 The NZ dTIMS Setup ........................................................................ 11
1.3.1 Data Items and Collection Methods ...................................................11
1.3.2 Pavement Deterioration Models .........................................................12
1.3.3 Treatment Types ................................................................................12
1.3.4 Triggers for Treatment Types .............................................................12
1.3.5 Effects of Treatments on Road Condition ..........................................15
1.3.6 Road User Costs Models....................................................................15
1.3.7 Analysis Approach..............................................................................16
2. CALIBRATION STUDIES .................................................... 18

2.1 Introduction ....................................................................................... 18
2.2 Level of Calibration ........................................................................... 19
2.3 Level 1: Basic Application ................................................................. 20
2.4 Level 2: Calibration ........................................................................... 22
2.5 Level 3: Adaptation ........................................................................... 23
3. INFLUENCE OF TRENCH PATCHING AND SERVICE
COVERS ............................................................................. 25
3.1 Introduction ....................................................................................... 25
3.2 Including Trench Patching and Service Covers to RAMM Assessments25
3.3 Including Trench Patching and Service Covers to Prediction Models 26
3.4 Further Research Needs ................................................................... 27
4. STUDY ON HDM-4 RDWE MODELS.................................. 28
4.1 Introduction ....................................................................................... 28
4.2 Pavement Characteristics ................................................................. 28
4.2.1 Adjusted Structural Number ...............................................................28
4.2.2 Construction Quality ...........................................................................29
4.3 Modification in Road Deterioration Model .......................................... 29
4.3.1 Cracking .............................................................................................30
4.3.2 Ravelling .............................................................................................31
4.3.3 Potholing.............................................................................................31
4.3.4 Edge Break .........................................................................................31
4.3.5 Rutting ................................................................................................31
4.3.6 Roughness .........................................................................................32
4.3.7 Pavement Texture ..............................................................................32
© 2001 Ingenium iv
4.4 Road Works Effects .......................................................................... 32

4.5 Testing of the HDM-4 Models............................................................ 33
5. WORKS EFFECTS ............................................................. 35
5.1 Introduction ....................................................................................... 35
5.2 Resets in NZ dTIMS Setup................................................................ 36
5.3 Resets Requiring Further Studies ..................................................... 37
5.3.1 Resets for Routine Maintenance ........................................................37
5.3.2 Roughness and Rutting Resets..........................................................39
6. AREA WIDE TREATMENTS ............................................... 42
6.1 Introduction ....................................................................................... 42
6.2 Definition........................................................................................... 42
6.3 Existing Practice................................................................................ 42
6.4 AWT in dTIMS................................................................................... 43
7. PAVEMENT STRENGTH .................................................... 44
7.1 Introduction ....................................................................................... 44
7.2 Problem Statement ........................................................................... 44
7.3 Further Research Needed................................................................. 45
8. UNSEALED ROAD DETERIORATION MODELLING.......... 46
8.1 Road Roughness .............................................................................. 46
8.2 Dust Models ...................................................................................... 50
8.3 Minimum Levels of Service ............................................................... 51
9. TEXTURE DEPTH AND SKID RESISTANCE ..................... 53
9.1 Introduction ....................................................................................... 53
9.2 Current Texture and Skid Resistance Models in dTIMS .................... 54
9.2.1 Texture Depth.....................................................................................54
9.2.2 Skid Resistance..................................................................................55
10. NOISE REDUCTION TREATMENTS .................................. 57
10.1 Introduction ....................................................................................... 57
10.2 Noise Reduction Treatments............................................................. 57
10.3 Including Noise Prediction Models..................................................... 58
© 2001 Ingenium v
10.4 Intervention Criteria for Noise Reduction Treatments ........................ 59

10.5 Maintenance Effects of Noise Reduction Treatments ........................ 60
11. DO MINIMUM ALTERNATIVE ............................................ 62
11.1 Introduction ....................................................................................... 62
11.2 Definition of Do Minimum .................................................................. 62
11.3 Status Quo ........................................................................................ 62
11.4 Problems with DOM .......................................................................... 63
12. MAINTENANCE COST INDEX ........................................... 66
12.1 Introduction ....................................................................................... 66
12.2 Maintenance Cost in the RAMM Database........................................ 66
12.3 Maintenance Cost Index Model ......................................................... 67
12.4 Issues related to MCI ........................................................................ 69
13. INTERVENTION CRITERIA FOR SEALED ROADS ........... 71
13.1 Introduction ....................................................................................... 71
13.2 Terminology ...................................................................................... 71
13.2.1 Existing Intervention Criteria in NZ dTIMS Setup...............................73
13.2.2 Treatment Selection Procedure in dTIMS ..........................................74
13.3 Issues on Intervention Criteria........................................................... 75
14. TRAFFIC VOLUMES........................................................... 76
14.1 Introduction ....................................................................................... 76
14.2 The Need for Inexpensive Traffic Data Collection.............................. 76
14.3 Principles of Moving Traffic Counts ................................................... 77
14.4 Objectives of the Study ..................................................................... 78
14.5 Benefits of the Research ................................................................... 79
Calibration Studies........................................................................................ 80
Work Effects (WE) Models ............................................................................ 82
Road User Cost Models (RUC) ..................................................................... 82
Specific Models............................................................................................. 83
Analysis and Policy ....................................................................................... 85
© 2001 Ingenium vi
LIST OF TABLES
Table 1.1: Treatment Types Used in dTIMS ................................................................................. 12
Table 1.2: VOC Model Coefficients ............................................................................................... 16
Table 2.1: Proposed Sealed Road HDM Calibration Coefficients................................................. 22
Table 4.1: Available Pavement Defects Models............................................................................ 30
Table 5.1: Condition Parameters Requiring Resets ...................................................................... 36
Table 5.2: Other Parameters Requiring Resets ............................................................................ 37
Table 5.3: Resets for Routine Maintenance .................................................................................. 38
Table 5.4: dTIMS Roughness and Rutting Resets ........................................................................ 40
Table 5.5: Roughness Reset Expressions .................................................................................... 40
Table 5.6: Rut Depth Reset Expressions ...................................................................................... 41
Table 9.1: Model Coefficients for Surface Dressings .................................................................... 54
Table 10.1: Possible Approaches for Including Noise Treatments to dTIMS................................ 57
Table 11.1: Proposed Default Triggers for Do Minimum Strategy ................................................ 63
Table 12.1 Proposed Trigger Values for MCI................................................................................ 68
Table 12.2: Resets for MCI............................................................................................................ 68
Table 13.1: Treatment Types in Maintenance Categories ............................................................ 72
Table 13.2: Strengthening Requirement (SREQ) Trigger Limits................................................... 73
Table 13.3: Smoothing Requirement (MREQ) Trigger Limits ....................................................... 73
Table 13.4: Resurfacing Requirement (RREQ) Trigger Limits...................................................... 74
© 2001 Ingenium vii

LIST OF FIGURES
Figure 1-1: Development and Refinement of the System ............................................................. 10
Figure 1-2: Flowchart of Treatment Selection Procedure.............................................................. 14
Figure 1-3: Example of dTIMS Efficiency Frontier......................................................................... 17
Figure 2-1: Streams Contributing to Modelling Capability............................................................. 19
Figure 2-2: Resources and Time Required for Calibration............................................................ 20
Figure 2-3: Data Vs Calibration ..................................................................................................... 20
Figure 2-4: Data Flow for dTIMS System ...................................................................................... 21
Figure 2-5: Model for Predicting Road Roughness ....................................................................... 23
Figure 3-1: Occurrence of Trench Patching and Services ............................................................ 25
Figure 3-2: Influence of Trench Patching and/or Service Covers on Roughness ......................... 27
Figure 5.1: Resets for Roughness................................................................................................. 35
Figure 8-1: Calculation of Total Transportation Costs................................................................... 46
Figure 8-2: Predictions of Different Unsealed Roads Models ....................................................... 47
Figure 9.1: Surface Profile Measurements.................................................................................... 53
Figure 10.1: Effects of Treatments on Pavement Condition.......................................................... 61
Figure 11.1: Trigger Limits for Minimum Level of Service ............................................................. 63
Figure 11.2: Efficiency Plot of Different Strategies........................................................................ 64
Figure 11.3: Minimum Cost Option................................................................................................ 65
Figure 12-1: Routine Maintenance Cost Increment....................................................................... 66
Figure 12-2:OPUS MCI Expression............................................................................................... 67
Figure 13-1: Strategy and Treatments for Economic Analysis ...................................................... 72
Figure 14-1:Example of Moving vs 7 Day Count from India.......................................................... 78
© 2001 Ingenium viii

1. RESEARCH REQUIREMENTS FOR dTIMS IN NEW ZEALAND
1.1 Background
The NZ dTIMS Implementation started in Mid-1998 under the full stewardship of the RIMS Group.
The RIMS Group comprises of members from Transfund, Transit NZ and members representing
Local Governments. It operates under the Association of Local Governments Engineers of New
Zealand (ALGENZ). Phase I of this project ended in June 1999. Work on Phase II started in July
1999 and ended in June 2000.
During Phase I of this project the object was to establish a preliminary system consisting of the
following aspects:
• Setup design for dTIMS;

• Pavement deterioration models based on HDM-III;
• Interface with RAMM;
• Training programme;
• Pilot testing of the setup with selected groups; and,
• Use feedback received as input to Phase II.
1.2 Identified Needs for Further Research

As a result of the feedback received during Phase I, the Preliminary NZ dTIMS Setup changed
significantly before it was released in August 1999. However, due to the short time frame of
Phase I, many unresolved issues were identified requiring further research or as issues to be
addressed as part of Phase II of the project. Many of these issues were highlighted at the
Technical Workshop held in Wellington in June 1999 and became potential research items. From
the list of unresolved issues from Phase I and the Technical Workshop, HTC has prepared this
report addressing all of the issues raised. The report discusses the required research that would
enhance the existing system and assess the benefits / added value of the research works.
The process in the development of the setup is illustrated in Figure 1.1. Note that the refinement
of the setup is an ongoing process based on the development of experience and knowledge over
time.
The report addresses and introduces the overall research issues and furthermore defines the
specific requirements of the following identified research tasks by:
• Introducing the overall topic;

• Defining the problem statement;
• Identifying the scope, issues and factors involved;
• Identifying the existing practice;
• Identifying further research required and the actions needed; and,
• Prioritising their needs in terms of timeframe by defining whether the research is required
immediately, or in the short term and an estimate of the length of time required to undertake
the research.
The issues that were raised in Phase I and the Technical Workshop have all been incorporated
by one of the following methods:
• Incorporated as system enhancements in Phase II and therefore not discussed in this report;
• Identified as tasks requiring further research and subsequently discussed in this report and
then prioritised on a list of research projects related to the NZ dTIMS Setup, the outcome
being identified in Annex A.
© 2001 Ingenium 9
The priority list of research projects was submitted to the RIMS Group for approval and this has
since been endorsed by them (see Annex A).
Develop
Does Setup
Prelimanary Train the User Test the Setup
fulfil all needs?
Setup
No
Train the User

This Report
Yes
Further Release
Identify Issues and Perform Research
Training No Updated
Further Work and Refinements
Needed? Setup
Does Setup Yes

No Test the Setup
fulfil all needs?
Yes
Use Setup on
Network
Implement Results
Figure 1.1: Development and Refinement of the System
The major research topic issues and that form subsequent chapters in this report are:
• Calibration Studies;
• Influence of Trench Patching and Service Covers;
• HDM-4 Road Deterioration Works Effects (RDWE) Models;
• Works Effects;
• Area Wide Treatments;
• Pavement Strength;
• Unsealed Road Deterioration Modelling;
• Texture Depth and Skid Resistance;
• Incorporation of Noise Reduction Treatments;
• Do Minimum alternative;
• Maintenance Cost Index;
• Intervention Criteria for Sealed roads; and,
• Traffic Volumes.
© 2001 Ingenium 10
1.3 The NZ dTIMS Setup

Pavement Management Systems (PMS) have developed rapidly during the last 30 years. New
computer applications, data measuring techniques and modelling of pavement deterioration
enabled advanced approaches in long-term maintenance planning. The implementation of the NZ
dTIMS Setup has enhanced the existing long-term maintenance capabilities of NZ Road
Controlling Authorities (RCA). One of the enhancements dTIMS has to offer is the use of
pavement deterioration models. It is now possible to predict future maintenance actions based on
predicted pavement conditions. Together with the models, many other input parameters are used
for the long-term maintenance planning performed in dTIMS. These parameters include:
• Data items and collection;

• Pavement deterioration models;
• Maintenance actions or treatment types;
• Triggers for maintenance actions;
• Effects of treatments on road condition (resets);
• Road user cost models; and,
• Analysis approach.
It is important to realise that the parameters indicated are inter-related and the inclusion or
changing of one parameter might affect all the other parameters. As an example the incorporation
of a new treatment might require the collection of additional data items, models, triggers and
resets. The input parameters are discussed in subsequent paragraphs.
1.3.1 Data Items and Collection Methods

The data items required for dTIMS analysis could be classified into the following sections:
• Road referencing;
• Traffic characteristics;
• Pavement inventory;
• Pavement history;
• Road condition;
• Road inventory;
• Environmental parameters;
• Treatment unit costs; and,
• Model and trigger coefficients.
Most of the data items required for dTIMS analysis is available in the RAMM database. Data
extracted from the RAMM database is imported to the Interface program in order to be converted
to the required format as used in dTIMS. The conversion of data is needed for two reasons.
Firstly, the condition description in dTIMS is expressed as a percentage area of the road effected
by a defect, while RAMM describes the total wheelpath length effected by a defect. Secondly,
dTIMS is using a fixed format dBASE file format, which is different to the format used in RAMM
exports. The Interface program is also used to compile additional data, which is not available in
RAMM. Additional data can be entered in look-up tables or default values are assigned for
missing data fields.
The data collection can be done visually using the RAMM assessment method and some data
items are collected using High-Speed Data collection measurements (HSD). RAMM stores both
visual assessment data and HSD. The visual condition data are stored according to the rating
sections, while HSD data are stored in short intervals such as every 10 m. Before the data are
exported for the use in dTIMS it is summarised to a treatment length.
© 2001 Ingenium 11
1.3.2 Pavement Deterioration Models

The pavement deterioration models used in the NZ dTIMS Setup consist of HDM-III models and
some locally developed models. These models include:
• Cracking initiation and progression;

• Potholing;
• Rutting;
• Roughness;
• Maintenance cost;
• Texture depth;
• Skid resistance; and,
• Gravel road models.
An incremental model form is used in dTIMS. This means that the change of a condition for the
analysis year is calculated based on the current condition. The effectiveness of the model is
therefore not only depending on the model itself, but is also a function of the data accuracy.
Models can be improved by performing calibration to local conditions.
Further developments in dTIMS may include the refinements of these models and incorporation
of new models.
1.3.3 Treatment Types

Treatments are actions taken by an authority to repair defects on a road or to reduce the decay or
deterioration of a pavement. The treatment types used in the setup can be classified into the
maintenance categories indicated in Table 1.1.
Table 1.1: Treatment Types Used in dTIMS

Strengthening Smoothing Resurfacing Unsealed Roads
AM reconstruction Thin granular overlay Seal, large chip Improvement granular to ST
ST reconstruction Thin asphalt overlay Seal, small chip Regravel
Thick granular overlay Rip and remake (ST) Seal, double chips Routine unsealed
Thick asphalt overlay Rip and remake (AM) Seal, special
Mill and replace
Recycle
Note: AM is an asphalt mix pavement
ST is a chip seal pavement
Treatments are specifically applied to address given conditions. Selecting the appropriate
treatment is done according to intervention criteria and economic principles. Triggers are
discussed in the following paragraph. The economic consideration of using a treatment will
depend on the cost and the resulting benefits from applying a treatment. The economic benefits
are a function of the improvement (reset) resulting from the treatment. These benefits include
savings in vehicle operation costs and maintenance cost.
1.3.4 Triggers for Treatment Types

A trigger specifies the condition or circumstance under which treatment is applied. Triggers
specify condition levels at which a maintenance action is performed. It can also specify practical
constraints that will prevent certain treatments to be applied for given conditions (e.g. raising of a
pavement is not allowed in urban areas).
The condition indicators used for defining triggers levels include:
• Cracking;
© 2001 Ingenium 12
• Scabbing;
• Potholes;
• Rutting;
• Roughness
• Flushing;
• Pavement strength;
• Shoving;
• Skid resistance.
• Texture depth; and,
• Maintenance cost index.
Some of these condition indicators are used in combination to define one index (composite
index). Examples of composite indexes used as triggers are Granular Overlay Need (GOVL) and
Surface Integrity Index (SII). GOVL is a function of the pavement strength and traffic. The SII is
calculated from weight factors applied to cracking, potholing, surface age and estimated surface
life.
The trigger levels based on condition indicators can be defined according to the following criteria:
• Traffic range (default or user defined, allowed for up to 7 ranges);

• Road type (urban, rural and motorway); and,
• Functional class (user defined, allowed for up to 10 ranges)
The triggers for practical constraints prevent the use of a trigger in an illogical manner. These
triggers include:
• Road hierarchy;
• Surface thickness;
• Possible raising of the pavement;
• Existing chip size;
• Second coat seal required;
• Bridge sections;
• Surface type: and,
• Traffic volume.
The triggers are used in a selection procedure shown in Figure 1.2 (HTC, 1999a).
© 2001 Ingenium 13
Start
Analysis
SPEC_TRT Use Specified

Yes
defined Treatment
No
Gravel? Yes AADT > limit No HS < limit
No
Second Seal
Required
No
Bridge?
No Yes
MCI > limit

Yes IRI > IRIstrength
Yes
No
Yes
IRI > limit

RDM > limit No
ASH > limit Yes GOVL > limit Yes Bridge?
ACA > limit
No
SI I> limit No
SFC < limit
TD < limit
Yes
No
IRI > IRIresurf

Yes
Yes
No
Strengthenin
Resurfacing Smoothing
g
AC surfacing AC surfacing Raising AC surfacing Raising

No No
required? required? Allowed? required? Allowed?
Yes Yes Yes Yes

No
No Raising HS < 100 Raising

No Bridge ?
Allowed? Bridge ? Allowed? Yes
HS > limit
Sealcycle >
Yes limit No
Yes No No No Yes
No
Thin Overlay Mill & Replace Yes Thin Overlay Yes Thick Overlay No Thick Overlay
AC AC Granular + ST AC Granular + ST
(MOA) (MMA) (MOL) (SOA) (SOL)
Chip >= 5
Rip & Remake Rip & Remake Reconstruct Reconstruct

Yes AC ST AC ST
(MRA) (MRL) (SRA) (SRL)
Reseal No
Large chip Yes
(RSL) Improvement
Granular + ST
(IOL)
Reseal
Small chip
(RSS) IRIstrength - Maximum roughness which when exceeded will result in strengthening Regravel
(ROG)
IRIresurf - Maximum roughness where a pavement can be resurfaced
Routine
Sealed
Routine
(RTS)
Unsealed
(RTU)
Figure 1.2: Flowchart of Treatment Selection Procedure
© 2001 Ingenium 14
1.3.5 Effects of Treatments on Road Condition

Applying a treatment will have an immediate effect on the pavement condition and the long-term
effects on the pavement performance. The attributes that are effected by treatments include:
• Pavement condition;
• Pavement history; and,
• Maintenance cost.
Resets for different treatment types will be different for each road characteristic. For example with
the overlay of a pavement it is expected that the roughness will improve significantly, however,
applying a reseal on the same pavement will result in a negligible improvement on roughness.
The parameter resets in dTIMS can be done in three ways:
• Absolute reset - a given parameter is reset to an absolute value (e.g. Chip size);
• Relative reset - a given parameter is reset to a percentage of the existing value (e.g.
pavement strength improves by 10% following a rip and remake); and,
• Expression reset - an expression is used to define a reset (e.g. the reset for an overlay is a
function of the overlay thickness).
The resets will ultimately determine the benefits associated with using a specific treatment. The
benefits manifest in the reduction of maintenance costs and road user costs.
1.3.6 Road User Costs Models

The road user cost (RUC) models used in the dTIMS setup for economic analyses are based on
the Transfund NZ Project Evaluation Manual (PEM). The road user costs are calculated
according to six vehicle classes that include:
• PC - passenger cars;
• LCV - light commercial vehicles;
• MCV - medium commercial vehicles;
• HCV-I - rigid heavy commercial vehicles;
• HCV-II - heavy commercial vehicles towing; and,
• BUS - heavy buses.
The road user cost models were derived for three different roughness levels shown in the
following expressions (HTC, 1999a):
a1
RUE = UF [a0 (NAASRA – 60) ] NAASRA < = R1 (1.1)
a3
RUE = UF [a2 (NAASRA – 60) + a4] NAASRA <= R2 (1.2)
RUE = UF [a5 NAASRA - a6] NAASRA > R2 (1.3)
where RUE is the PEM vehicle operating cost in cents/km

NAASRA is the roughness in NAASRA counts/km
UF is the PEM cost Update Factor which has a value of 0.80 in the
October 1999 PEM.
a0 to a6 are regression coefficients
R1 and R2 are transitional roughness’
© 2001 Ingenium 15
The value of the roughness levels R1, R2, and R3 varies by vehicle class and the equations used
in the expressions are shown in Table 1.2 (HTC, 1999a).
The update factor of 0.8 has been used, as the maintenance / rehabilitation cost for the treatment
length sections rarely exceed 1 million. The PEM user cost tables are now under review and
therefore the VOC expressions used in dTIMS will be updated once these tables have been
reviewed and made available.
Table 1.2: VOC Model Coefficients

Vehicle RUE Model Coefficients
Class a0 a1 a2 a3 a4 a5 a6 R1 R2
PC 0.0017 1.986 0.0061 1.564 0.13 0.2014 28.84 85 350
LCV 0.0019 1.966 0.0064 1.564 0.00 0.2085 28.00 85 350
MCV 0.0120 1.210 0.1986 1.000 -3.215 - - 75 600
HCV-I 0.0045 1.879 0.4375 1.000 -7.530 - - 80 600
HCV-II 0.0052 1.897 0.4098 1.000 -6.773 - - 80 600
BUS 0.0003 2.657 0.0614 1.164 -1.200 0.2093 26.43 80 350
1.3.7 Analysis Approach

The analyses in dTIMS are undertaken in two phases namely strategy generation and
optimisation. During the strategy generation all possible treatments are generated for each road
section and the prediction being how each strategy would effect the road condition over the
analysis period. The road deterioration models are used to calculate the future condition of the
roads following the treatment. To find the appropriate treatments to generate on each section the
ranges of triggers applicable to that section are used. The cost stream for the total analysis period
is calculated for each strategy and depreciated to present value cost. The user has a selection of
three methods to define the benefits of the strategies including:
• Area under condition curve;

• Saving in vehicle operating cost; and,
• Net present value of total transportation cost.
The area under the condition curve calculates the benefits by summing the present value of the
difference between the condition index resulting from a strategy and the condition index for a do-
nothing strategy for each year in the analysis period.
Savings in vehicle operating costs are the difference in vehicle operating cost between using a
strategy and the do-nothing strategy. The saving in vehicle operating costs suggest that the cost
would be lower driving on a road that received a strategy compared with the vehicle operating
cost experienced on the do-nothing strategy.
The Total Transportation Costs (TTC) comprises the sum of construction, maintenance and
vehicle operating cost. During the analysis the benefits are calculated for the difference between
the TTC of the strategy compared with the TTC of the do-nothing strategy. There is a theoretical
optimum maintenance level calculated according to the TTC.
The optimisation in dTIMS is based on the incremental benefit-cost ratio technique. The benefit-
cost ratio is defined as the ratio between the increase in the benefits to the increase in cost for
successive strategies. Figure 1.3 illustrates the efficiency frontier, which is also, one of the
outputs provided in dTIMS. The vertical axis shows the present value of benefits and the
horizontal axle the present value of the costs. Each dot on the graph represents a strategy. The
segmented line connecting the uppermost dots is the efficiency frontier. Note that the line starts at
the do-nothing alternative and connects the dots in such a way that no line segment has a slope
© 2001 Ingenium 16
larger than the slope of the previous segment. No dots exist above the efficiency frontier. The
bottom line indicated in the graph represents the efficiency envelope. Only those strategies within
the efficiency envelope will be selected during the optimisation process.
Figure 1.3: Example of dTIMS Efficiency Frontier
The heuristic optimisation in dTIMS is undertaken performing the following steps (Deighton,
1998):
1. All strategies are sorted in descending order of incremental benefit cost regardless of the
section they are on;
2. dTIMS starts at the top of the list and checks whether there is enough money in the budget in
each year to cover the yearly cost of that strategy. If there is, that strategy is selected for that
element;
3. The available budget in the respective category is reduced by the annual yearly costs of the
treatments for the selected strategy;
4. dTIMS continues down the list doing the same process for each strategy on this sorted list;
5. As the analysis continues, a strategy is replaced by another for the same element, only if the
next strategy provides a greater benefit and the budget is available; and,
6. The process is finished when there are no more strategies, or there is no more available
budget.
© 2001 Ingenium 17
2. CALIBRATION STUDIES
2.1 Introduction
During long-term maintenance planning, models are used to simulate the actual behaviour of
pavements over time. The accuracy of this simulation depends on two conditions (Bennett and
Paterson, 1999):
• How well the data provided represents the real conditions being analysed as understood by
the model; and,
• How well the predictions of the model fit the real behaviour and the interactions between the
various factors for the variety of conditions to which it is applied.
The need for calibration originates from the fact that models are developed during studies with a
range of conditions influencing the outcome of the models. These conditions might differ for the
area in which the models are applied. A further difference can be found in the condition
description used. Derived models are based on condition data defined in a specific format.
Typically the HDM condition description uses the percentage of the total pavement area effected
by a defect. In the dTIMS context calibration is needed for the data, pavement deterioration
models and the road user effects. This chapter will highlight the need for calibration studies and
recommendations will be made for studies to be performed in NZ. A calibration guide will also be
done as part of Phase II of this project, which will cover the technical details of calibration studies.
dTIMS is one of the most powerful software applications of its kind as all the parameters within
the programme are user defined. This strength might also be seen as the downside of using the
software, because it is not “plug and play” software that can be used without a fair amount of
effort spent on the system in order to customise it for a network. It must be appreciated that long-
term maintenance planning is a dynamic field and although NZ authorities can pride themselves
in the compatibility of their systems, conditions and practises used do differ from one authority to
the next. The NZ dTIMS Setup was developed with the vision of it being able to be used by all
authorities, but it was also realised that this setup needs to be customised for each authority. This
customisation is needed for two reasons:
• The short time frame allowed for the development of the preliminary system and therefore in
providing a working system, but it is accepted that many unresolved issues can only be
addressed during a later phase of the project. These issues include items such as data and
calibration.
• One system has been developed in the NZ dTIMS Setup, which makes provision for the
diversity of all the networks of this country. It also makes provision for all network types such
as urban, rural and state highways. Therefore, it is only detailed input parameters that have
to be changed to make the system applicable for a specific network.
Implementation of any long-term maintenance planning systems usually has a development

stream towards true modelling capability as shown in Figure 2.1 (HTC, 1999a). At this stage of
the implementation NZ is well on the way of modelling capability but it must be realised that there
is still much to be done. It is also important to note that calibration studies should not only be
seen as research, but involves data appropriateness and the research stream shown in Figure
2.1. Refinement of the system and especially input parameters is an ongoing process that will
address more detail as confidence and experience with the system grows.
© 2001 Ingenium 18
DATA OPERATIONAL
INTELLECTUAL RESEARCH
APPROPRIATENESS ADOPTION
Development of the Condition and inventory Refinement of models Control mechanisms,

ability to use the skills data suitable for based on appropriate consumption and
(training, upskilling) modelling data utilisation of outputs
Research based on
Most data are inappropriate data
Principles understood Principles understood
inappropriate (concerned with
principles)
Control mechanisms
Sensitivities understood
established
Developing
understanding
Management
Changed data collection
methodologies can
and QA principles
absorb outputs
Model refinement
Ability to interpret Appropriate data Outputs used in setting
capability based on
outputs becoming available policy
appropriate data
True Modelling Capability
Ongoing Refinement
Figure 2.1: Streams Contributing to Modelling Capability
2.2 Level of Calibration

The degree of local calibration will depend on the type of application and the recourses available
for an authority. Bennett and Paterson (1999) defined three levels of calibration:
Level 1 Basic Application. Calibration of most sensitive parameters with best estimates,
desk studies or minimal field surveys;
Level 2 Calibration. Key models are calibrated using moderate field surveys. The
collection of additional input parameters are required; and
Level 3 Adaptation. Major field surveys are conducted and controlled experiments are
performed. This level can enhance existing models or new models can be derived.
The effort spent on the three levels are shown in Figure 2.2 (Bennett and Paterson, 1999).
© 2001 Ingenium 19
Time Required
Experimental
Years
Surveys and
Research
Months
Field Surveys
Weeks
Desk Studies
Resources
Limited Moderate Significant Required
General Planning Project Appraisal Research and

Quick Prioritisation Detailed Feasibility Development
Preliminary Screening
Coarse Estimates Reliable Estimates
Figure 2.2: Resources and Time Required for Calibration
The levels of calibration required to be performed by authorities include at least the desk studies
and some field surveys. A well planned and co-ordinated national or regional level calibration
programme with the involvement of various local authorities is required for Level 2 calibration.
Fragmented research based on individual initiatives in this regard will lead to a disjointed
approach and application of dTIMS. In subsequent paragraphs the work required for each
calibration level is discussed. Technical detail on the calibration studies will be covered in the
Calibration Guide to be prepared as part of Phase II of this project.
2.3 Level 1: Basic Application

All users of the NZ dTIMS Setup should perform basic application ca libration on data. The most
important component of basic application calibration is the validation of existing data. Although
data is often not seen as important as calibration, the opposite is true. The data provided must
represent the real condition and interpret (understand) as such the models and the analyses
used. Bennett and Paterson (1999) explain the difference between data and calibration using the
analogy of ‘The depth of the sea and the height of the waves’ as illustrated in Figure 2.3. Once
the user feels comfortable with the data, time and effort can be spent on calibration. It is not
helpful to attempt calibration on corrupt data.
Calibration
'The Depth of the Sea and

Data
the Height of the Waves'
Figure 2.3: Data Vs Calibration
The aim of this level study is to satisfy the question: “Does my data make sense?” and “Does the
predictions make sense?” The NZ dTIMS System comprises of many components as illustrated in
Figure 2.4. Data validation has to be performed at all the stages of the dTIMS System, namely:
© 2001 Ingenium 20
• Existing data in RAMM;

1
• Conversion factors and default values used in the Interface program ; and,
2
• Strength program calculations.
A detailed report on data issues (HTC, 1999c) was prepared for Phase II of this project.
Data Storage Conversion Analyses
Strength program
Interface program
RAMM
Figure 2.4: Data Flow for dTIMS System
Desk studies for calibrating model predictions can be performed on various methods of which
simulation of the past can be the easiest to perform. This calibration method can be used for
Level 1 and Level 2 calibration. Analysing historical data with prediction models can give a good
indication of the accuracy of predictions achieved. It will not be possible to calibrate all models on
this level as the data collection method might be inappropriate for calibration. Bennett and
Paterson (1999) suggest that roughness progression and the age term of roughness can be
successfully calibrated on Level 1 calibration.
Level 1 model calibration has been performed for Phase I of this project on a limited scale. This
calibration was based on work done by Riley (1998) and Hallet (1998). The recommended
calibration coefficients from this study is shown in Table 2.1 (HTC, 1999a). This study was
seriously hampered by the data quality. It is recommended that supplementary Level 1 calibration
has to be performed by authorities using data off their own networks.
1
The Interface program was developed to convert RAMM data to the required format used in dTIMS.
2
The Strength program is used to calculate pavement strength based on measurements such as FWD,
CBR, Benkelman Beam, ARRB and Typical pavement method.
© 2001 Ingenium 21
Table 2.1: Proposed Sealed Road HDM Calibration Coefficients
Distress Mode Pavement Type Coefficient Value Recommended Value

Adopted
Riley Hallett for dTIMS
(1998a) (1998)
Crack initiation Surface treatments Kci 1.0 1.0
on granular base
Crack initiation Reseal on surface Kci 1.0 (1) 1.0
treatment
Crack progression All Kci 1.0 1.0 1.0
Ravelling initiation and All Kvi 1.0 (2) 1.0
progression
Pothole progression All Kpp 1.0 1.0 1.0
Rut depth progression All Krp 0 (3) 1.0 1.0
Roughness age term All Kge 2.2 1.0 1.0
Roughness progression All Kgp 1.0 1.0 1.0
NOTES: 1/ Coefficient 0.22 HSNEW replaced by 0.15 HSNEW

2/ Model not adopted for dTIMS project
3/ Based on State Highways where rutting was not observed. Not appropriate for local authority
roads.
The calibration of unsealed road models is discussed in Chapter 8.
2.4 Level 2: Calibration

HTC (1999a) recommended that Level 2 and 3 calibration should supplement the initial
calibration done during Phase I of this project. Level 2 calibration extends the scope of calibration
done for Level 1. It will require additional measurements performed on the network to validate
and adjust the prediction models used in dTIMS. It is anticipated that the Level 2 calibration study
will be performed by interested parties involved in the dTIMS project. The research will be
managed and co-ordinated from one body and the actual work will be distributed amongst
participants. This approach will ensure that the workload is evenly distributed, recourses are
effectively utilised and the whole country is involved with the calibration. A further benefit is that
the total range of influencing factors such as climatic conditions are included in the research.
One of the obstacles identified during Level 1 calibration was the lack of data, especially for
progression models. This is as a result of maintenance policies in NZ. Most defects are repaired
before significant progression takes place. Many people argue whether progression models are
that relevant to the NZ conditions. The answer to this can be illustrated by Figure 2.5. This figure
illustrates the deterioration of roughness for the do-nothing scenario and a treatment done in
2004. Economic analyses in dTIMS is based on the benefits and costs experienced over a 20
year analysis period. Therefore it is necessary to know what the condition would be if no, or very
little maintenance is done on the pavement. The progression rates of defects are thus essential
for the economic analyses performed in dTIMS. This further suggests that calibration should be
performed on pavements with no or very little maintenance done.
© 2001 Ingenium 22
20
18
16
14
12
10
8
6
FMRL, 2004
4
2
0
1999 2004 2009 2014 2019
Analysis Year
1, Test, Start, End
Figure 2.5: Model for Predicting Road Roughness
The experimental design for this level of calibration should include all the following items (Bennett
and Paterson, 1999):
• Effect of traffic on pavement deterioration;

• Environmental effects on the pavement deterioration;
• Deterioration rate by pavement type;
• Pavement strength effects; and,
• Surface distresses.
Calibration sites should therefore be located throughout NZ in order to include all these items to
the experimental design. Only limited maintenance (e.g. pothole patching) can be performed on
these sections. Calibration sections can be as short as 200 - 500 m. It is essential to provide
proper road signs for the calibration sections. This will not only inform the public on the
experiment but also alert maintenance teams to the position of the sections.
Factors to consider during the selection of calibration sections include:
• The section must fall in the range set by experimental design (e.g correct traffic);
• The total length of the section needs to be homogeneous;
• The location of the section should be safe for the assessors and the public traffic; and,
• Flat gradients are preferred in order to exclude factors such as drainage effects.
The measurements and assessment procedure will be specified in the Calibration Guide.
2.5 Level 3: Adaptation

Level 3 calibration, adaptation, involves improved data collection and fundamental research. With
this level enough data are collected to derive new relationships for models based on local
conditions. This suggests that the data are collected on more sites and for longer periods
compared with Level 2 calibration. Pavement deterioration research is typically done over a 5-
year period (Bennett and Paterson, 1999).
© 2001 Ingenium 23
The HDM recommendation does not include pavement deterioration models for this calibration
level but some road user effects models have to be done on this level (Bennett and Paterson,
1999). However, not all the pavement deterioration models in dTIMS have an HDM origin. Some
models such as the texture depth models were developed in NZ. The development of new
models can only be performed on Level 3 calibration.
All Level 3 calibration topics currently needed for dTIMS are discussed separately in this report
and include:
• Incorporation of trench patching and service covers in analyses;

• Work effects models;
• Area wide treatments;
• Unsealed road deterioration models;
• Texture depth and Skid resistance models; and,
• Noise models.
© 2001 Ingenium 24
3. INFLUENCE OF TRENCH PATCHING AND SERVICE COVERS
3.1 Introduction
Trench patching and service covers need to be considered in the Pavement Management System
(PMS), especially for city councils. It has an effect on maintenance programming as authorities
co-ordinate the scheduling of trenching and pavement rehabilitation in order to cause as little
traffic disruption as possible. Significant cost savings can also be achieved in proper planning of
trench and rehabilitation contracts. Also, trenches and services will effect the performance of the
pavement in various ways.
It is unavoidable that trenches and services will increase the roughness of a pavement. It could
also be the reason for poor repeatability of successive roughness surveys in the urban areas The
effect on the roughness will depend on the construction quality during reinstatement and the
physical properties of the trench patch such as size, location and orientation relative to the
direction of travel. Poorly reinstated trench patching can also cause secondary and premature
failure to the pavement. If the patch does not provide sufficient water proofing, the area around
the patch will develop some defects such as cracking and deformation. Often these secondary
defects will ultimately result in higher roughness.
Hence, the following aspects are needed to be considered when undertaking a study on the effect
of trench patching and service covers:
• Determine the significance of the influence of trench patching and service covers to the
pavement deterioration and road user costs;
• Level of data details required to be collected and stored; and,
• Quantifying the condition and occurrence of the trench patches and services.
Currently trench patching and services are not considered during the RAMM assessments. There
is therefore a need to incorporate these assessment items to the assessment procedures,
database structure and analyses process. This chapter will discuss the research need with regard
to this incorporation. It will highlight the aspects to be investigated and recommendations will be
made in terms of the scope and the extent of the study required.
3.2 Including Trench Patching and Service Covers to RAMM Assessments

To quantify trench patching can become complicated as illustrated in Figure 3.1.
Trench Patching in
Longitudinal Direction
Service Inspection Hole

Cover
Trench Patching in
Transverse Direction
Figure 3.1: Occurrence of Trench Patching and Services
Figure 3.1 illustrates items to note during inspections, which include:
© 2001 Ingenium 25
• Orientation of the trench patch;

• Location of patch or service hole cover, especially whether it is in the wheel path or not; and,
• Dimensions of the trench patch.
The orientation of the trench patch will have a significant effect on the resulting roughness
increase caused by the bad patches. A transverse patch and service cover cause a sudden jump
in the roughness profile, while a longitudinal patch cause a distortion of the roughness profile for
the length of the patch. Roughness measurements are normally only done in the wheelpaths and
hence there might be a need to indicate the location of the trench patch or service inspection
cover.
2
Normal pothole patching is only rated with RAMM if the area of the patch is less than 0.5 m . The
merit of using a similar approach for the trench patches is uncertain and this needs to be
investigated.
The condition of the trench patch will determine the effect it has on the deterioration of the
pavement in that area. Defects of the trench patch will result in water ingression into the
pavement layers that will ultimately lead to the weakening of that layer. Typical defects occurring
on trench patches can include cracking, deformation, shoving and delamination (loss of surface
material). The RAMM rating method does not consider the condition of pothole patching. The
inclusion of trench patching to the rating method raises some questions with regard to the
compatibility of the existing method. If pothole patch conditions are not rated, a similar approach
should be followed with trench patching. However, this decision should only be based on
research.
The requirements of the data on trench patches and service covers will result in additional cost
for data collection. Hence, before deciding whether or not to collect the data on the trench
patches and service covers and, also, to what details the data should be collected, it is essential
to clearly define how and whether we can use these data.
3.3 Including Trench Patching and Service Covers to Prediction Models

The benefit of inclusion of the predictive modelling of the Trench Patching and service covers
have to be very carefully weighted before deciding to incorporate for into the dTIMS setup. The
following aspects will have to be studied:
• The effect of the trench patches and service cover on the roughness. Factors mentioned
earlier such as patch size, location and orientation might have a significant effect on this
relationship; and,
• The influence on pavement deterioration, if research finds this influence to be significant.
A study into the effect of trench patching and service cover on the roughness could be done by
establishing the relationship between the numbers of trench patches and/or service covers versus
the increase of roughness as indicated in Figure 3.2. This study will offer unique challenges such
as the method of establishing the contribution of the patching to the roughness in addition to the
roughness caused by pavement deterioration. It would also establish the effect of patches to the
repeatability of the roughness survey. Besides which, during the study other real issues in urban
streets related to the dTIMS analysis will be identified.
© 2001 Ingenium 26
Increase in Roughness
Number of Trench Patches

and/or Service Covers
Figure 3.2: Influence of Trench Patching and/or Service Covers on Roughness
To know in advance where trenching and service installations will occur could assist in the
modelling process. However, this will require full integration with other management systems
such as sewer management. It should therefore not be expected to achieve better predictions
than what can be achieved by probabilistic models. With this type of model the likelihood of an
occurrence are predicted using statistical methods such as Marcov chains.
If trench patching and service covers are included into the NZ dTIMS Setup modelling strategies
need to be developed for treatments applicable to trench patches and service covers. Criteria
have to be developed for determining intervention or trigger levels for treatments to be
undertaken as a response on certain conditions. For example, a trench patch might significantly
influence roughness for a section, while the remaining part of the section is still in a satisfactory
condition in terms of the roughness. For such a section, reinstatement of the trench can be
triggered based on the roughness.
However, the needs for including trench patching and service covers in predictive pavement
deterioration modelling needs to be confirmed by further research for as Riley, (2000b) has
stated:
“I am not convinced that including utility patches will significantly improve the modelling,
while increasing the data needs and modelling complexity.”
3.4 Further Research Needs

A comprehensive study needs to be undertaken to determine whether including trench patching
and service covers to the NZ dTIMS Setup is appropriate. It must cover all the aspects including:
• Development of an assessment procedure for trench patching and service covers. This will
also include the definition of these items in the RAMM database;
• Investigation of modelling aspects such as the influence of trench patching and service
covers on other defects such as roughness; and,
• Establishing the incorporation of these items into the dTIMS analyses process that includes
items such as defining triggers and resets for treatments related to the improvement of trench
patching and service covers.
© 2001 Ingenium 27
4. STUDY ON HDM-4 RDWE MODELS
4.1 Introduction
NZ dTIMS setup uses the HDM-III road deterioration and works effects (RDWE) models for
predicting the road condition. Since it’s release, HDM-III models have been used in numerous
projects covering a range of climates and conditions. The basic structure and predictions of the
relationships have been widely confirmed. However, various limitations of HDM-III in terms of
scope were identified during the implementation in NZ. For example, road safety issues are not
included; the RDWE models do not encompass all of the pavement types (e.g. rigid pavements);
the range of climates is primarily tropical and temperate; vehicle emissions and similar
environmental effects are not included.
The ISOHDM study was conducted to collate the results of various research works throughout the
world done in HDM-III and to come up with a specification for improved models (NDLI, 1995).
Based on the recommendation of this study the World Bank HDM-4 models have been recently
released. Most of the HDM-III pavement deterioration and works effects models have been
improved and some new models have been added. HDM-4 now includes (Greg M & et al, 1999):
• A greater range of physical environments (climatic zones including cold freezing climates);
• Rigid/concrete and semi-rigid pavements, and a wider range of flexible pavements;
• Deterioration and maintenance of side-drains and their effects on pavement strength;
• Texture depth and skid resistance models;
• Edge break, particularly on narrow roads;
• A broader range of routine maintenance operations and effects; and,
• A broader range of improvement/new construction works options.
With the improved models of HDM-4 the accuracy of the road deterioration can be predicted more
accurately than in the case of the HDM-III models used in the NZ dTIMS setup. However, it
should be appreciated that HDM-4 models have not been tested in NZ conditions. This chapter
briefly discusses the improvements in various HDM-4 models and the required research works
needed before it can be implemented in the NZ dTIMS setup.
4.2 Pavement Characteristics

Some of the parameters defining the pavement characteristics have been modified in HDM-4 to
improve the reliability of the HDM-III deterioration models. These changes are discussed in the
following paragraphs with the emphasis on the implications it might have on the NZ dTIMS setup.
4.2.1 Adjusted Structural Number

The modified structural number (SNC) used in HDM-III models has the following shortcoming:
wrong estimation of the pavement performance when the pavement depth exceeds 700 mm. The
deficiency is corrected in HDM-4, which uses the adjusted structural number (SNP) as a strength
parameter.
Expression used to calculate SNP is already included in the Strength program. However, the
seasonal factor representing the effect of climate and drainage on the strength parameter has not
been modelled in the program. The expressions and coefficient used to define the seasonal factor
has to be verified and calibrated in the NZ context.
© 2001 Ingenium 28
4.2.2 Construction Quality

In HDM-III, two construction quality indicators were used :
• Surfacing construction quality indicator (CQ); and,

• Construction compaction indicator (COMP).
CQ flags if the quality of construction is good (1) or bad (0). The relative compaction in the base,
subbase and selected subgrade layers – COMP is expressed in the percentage. HDM-4 retains
the same expression for the COMP. However, the concept of construction defects has been
extended by using parameters that are continuous variables for the surfacing and the base:
• CDS construction defects indicator for bituminous surfacing; and,

• CDB construction defects indicator for the base.
CDS is a factor indicating the general level of binder content and stiffness relative to the optimal
material design for the specified bituminous mixture. It is used as an indicator to illustrate whether
a bituminous surfacing is prone to cracking and ravelling (low value of CDS), or prone to rutting
through plastic deformation (high value of CDS). CDS generally range in value between 0.5 and
1.5.
CDB is a continuous variable ranging between 0 (no construction defects) and 1.5 (several
defects). It is used in the potholing models. Poor gradation of material, poor aggregate shape and
poor compaction are the types of defects that should be considered in setting a value of CDB.
Continuous defect indicators CDS and CDB in HDM-4 instead of CQ have made the models more
flexible. Using these parameters will make it possible to include the different levels of
construction quality. However, back analysis might be needed to verify that the plastic
deformation and cracking predictions are valid.
Review of the pavement characteristics shows that the parameters used in HDM-4 can make the
models much more flexible in comparison to the parameters used in the NZ dTIMS models. It will
make the model much easier to customise for local conditions. However, the parameters together
with the models should be tested and calibrated thoroughly for NZ conditions.
4.3 Modification in Road Deterioration Model

Individual pavement defects modelled in different systems are given in the Table 4.1.
© 2001 Ingenium 29
Table 4.1: Available Pavement Defects Models
Defects HDM-III HDM-4 NZ dTIMS2 HDM-4 Models differ from HDM-III models
Bituminous Pavement
Drainage •
Cracking • • • • Continuous construction defect
indicator is used
• Transverse thermal cracking has been
added
Ravelling • • •1 • Continuous construction defect
indicator is used
Potholing • • • • Introduction of the time lapse factor
Edge repair •
Rutting • • •3 • Rut depth measurement based on 2 m
straight-edge rather than 1.2 m
• New relationship for modelling the
plastic deformation of the pavements
Roughness • • • • Potholing component modified
Texture depth • •
Skid resistance • •
Concrete
Cracking •
Joint spalling •
Faulting •
Failures •
Serviceability loss •
Roughness •
Unsealed
Gravel loss • • •
Roughness • • •
Note: 1/ Ravelling model is currently turned off in the NZ dTIMS setup
2/ HDM-III models are used in NZ dTIMS setup
3/ HDM-III approach used in NZ dTIMS setup
Most of the HDM-4 models are the improved version of the HDM-III models in one way or
another. The following sections explain some comparison between the pavement deterioration
models applied in the NZ dTIMS setup with the HDM-4 models.
4.3.1 Cracking
Cracking models used in the dTIMS setup are primarily based on the HDM–III models. Some
improvements on these models are carried out for the HDM-4 by introducing continuous
construction defects indicator (CDS) for bituminous surfacings, The use of the HDM-4 models
with CDS in dTIMS in the case of cracking will enable to distinguish between pavements that are
more likely to crack and those that are prone to plastic deformation and therefore less likely to
crack (Greg M. et al, 1999).
Additional cracking type - transverse thermal cracking has been introduced as a new type of
distress in HDM-4. It is modelled as cracking intensity expressed as number of cracks per km
(Riley, 1997). In addition to that in the future version of HDM-4 it has been proposed to add the
Reflection cracking model (Riley et al 2000). This will allow to predict the reflection cracking in the
bituminous pavements in NZ.
© 2001 Ingenium 30
4.3.2 Ravelling
The models for predicting the initiation and progression of ravelling in HDM-4 are based on those
used in HDM-III, but with a traffic variable as proposed by Riley (1999). The other alteration in the
initiation model is that the discrete construction quality variable used in HDM-III has been
replaced by continuous variable CDS.
The ravelling model is turned off in the NZ dTIMS setup as it was not giving a reliable prediction
(Riley 1998). It was suggested in (Riley 1998) that with CDS being introduced in the HDM-4
models, this could be used for the NZ situation. More detailed study on the appropriateness of the
HDM-4 ravelling models will need to be checked for NZ conditions.
4.3.3 Potholing
HDM-4 uses the construction defects indicator for the base, CDB, as a variable. Besides, in the
2
models the potholing is expressed in terms of the number of ‘pothole units’ of area 0.1 m . The
potholing progression model included in HDM-4 has introduced a time lapse factor (TLF) which
considers the early response between the occurrence and patching of potholes which could result
in fewer potholes occurring during the course of a year. Inclusion of this model in the NZ dTIMS
setup will allow the user to take account of the different response time used in NZ for the roads
with different hierarchies.
4.3.4 Edge Break

Edge break is the loss of surface, and possibly base materials, from the edge of the pavement,
and commonly arises on narrow roads with unsealed shoulders. Loss of material at the pavement
edge can be caused by two mechanisms:
• Shear failure; and,

• Attrition.
Shear failure occurs in the upper layers due to vertical wheel loads at, or close to, an edge that is
lacking lateral support from the shoulder. Parameters governing this mechanism are drop height
from pavement to shoulder, the strength of the pavement material and the number of wheel loads
that pass close to or over the pavement edge. Attrition occurs when wheels travel on and off the
pavement edge, as happens when vehicles pass on narrow roads (carriageway width less than 4
– 5 m) or when parking on the shoulder. As with shear failure, the extent of material loss is a
function of seal width, treatment types, wheel passes, edge step and, possibly, the speed of the
vehicles. Usually edge break will occur on pavements with the carriageway width less than 6.2
metres (NDLI, 1995). Including the Edge break model in the NZ dTIMS system will allow to model
the narrow bituminous roads in NZ.
4.3.5 Rutting
Rutting is the permanent or unrecoverable traffic-associated deformation within pavement layers
which, if channellised into wheelpaths, accumulates over time and becomes manifested as a rut
(Paterson, 1987). Traffic associated permanent deformation results from a rather complex
combination of densification and plastic flow mechanisms.
Some of the limitations of the HDM-III rutting model have been addressed in the model available
in HDM-4. In particular these are:
• Separate relationships to model the various phases of the progression of structural

deformation;
• A new relationship for modelling the plastic deformation of pavements;
• A new relationship for modelling the seasonal surface wear which occurs in countries where
vehicles use snow chains on the roads covered with snow and ice;
• Standardising the rut depth predictions to those measured under a 2.0 metre straight-edge.
© 2001 Ingenium 31
The HDM-4 rut depth model is based on four components of rutting, the rut depth at any time
being the sum of the four components. These four components are as follows:
• Initial densification;
• Structural deformation;
• Plastic deformation; and,
• Wear from snow chains.
Modelling of plastic deformation will allow taking account of the deterioration of the multiple seal
layers found in NZ roads. As the problem is quite prominent in almost all parts of NZ, a detailed
study to calibrate and use the model in the NZ dTIMS setup should be undertaken. Similarly the
deterioration of the pavement surface in those regions of NZ with snowfall in winter could be
considered.
4.3.6 Roughness
The roughness model in HDM-4 is based on the HDM-III model and has the same five
components of roughness. The structural, cracking, rutting and environmental components are
similar to the HDM-III versions. However, as some of the parameters are modified and added
these models need to be tested and calibrated to NZ conditions.
It should be noted that the potholing component has been extensively modified. It now also
includes the effect of road width on the freedom to manoeuvre to avoid the potholes and
response time between occurrence of potholes and patching.
With the improved parameters and expressions the new HDM-4 model should be much more
reliable. Roughness being the most important factor determining the vehicle prediction model in
NZ and therefore should be well calibrated.
4.3.7 Pavement Texture

Texture models were not modelled in HDM-III. Models for texture depth and skid resistance is
available in HDM-4. The model used by HDM-4 for skid resistance has been included in the
setup. The HDM-4 texture depth model has been customised to the NZ condition and used in the
setup. A further study is needed to verify the reliability of the models (Please refer to HTC,
2000a).
4.4 Road Works Effects

The effects of the roadworks are more comprehensively modelled in HDM-4. Bituminous
pavement works are classified into 4 different groups:
• Routine maintenance;
• Periodic maintenance;
• Improvement works; and,
• Construction.
Routine maintenance operations in HDM-III models include only pothole patching. Other routine
maintenance activities were supposed to be included in the deterioration models which assume
adequate levels of the routine maintenance. Thus HDM-III did not allow the evaluation of the
beneficial effects of other routine maintenance works.
HDM-4 with more explicit modelling of pavement strength, allows the effects of several routine
works to be evaluated. Reactive routine maintenance works which are user-specified are:
© 2001 Ingenium 32
• Drainage maintenance and rehabilitation;

• Sealing of wide structural cracking;
• Sealing of transverse thermal cracking;
• Surface patching of ravelled areas;
• Patching of potholes;
• Deep patching of wide structural cracking; and,
• Repair of pavement edge break.
Periodic treatments in HDM-4 includes:
• Preventative treatments;
• Resurfacing;
• Overlays; and,
• Reconstruction.
All four types of periodic maintenance were modelled in HDM-III. Some modifications on the
expressions used in HDM-III to reset various parameters are undertaken based on the
experience in different countries.
Road Improvements (widening and geometric realignment) and construction works are also
modelled in HDM-4.
Road works effect models in HDM-4 are a substantial improvement from the HDM-III works
effects models used in the NZ dTIMS setup. However it is essential that these models are well
calibrated as the reset values will affect the road deterioration model as well.
4.5 Testing of the HDM-4 Models

Most of the models included were developed by various agencies based on the extensive
research works in different countries. But only limited information on testing of HDM-4 models is
available.
PIARC-ISOHDM review workshop on HDM-4 road deterioration and works effect models was
organised by Austroads (Roberts J, 1999). A review of the HDM-4 road deterioration models
done by ARRB TR was presented in the forum. It was agreed that some of the models have to be
adjusted for the implementation of HDM-4 in Australia. At the workshop it was generally agreed
that the local calibration of the models is essential and sufficient for the implementing of the
models effectively. Only a few minor modifications to the models were identified and considered
for further testing and future development of the models. A similar study is needed to be
conducted in NZ for the customisation and calibration of the models to NZ conditions.

The review of HDM-4 models shows that the models have been greatly improved on since the
HDM-III models used in the NZ dTIMS setup. After the release of HDM-4 in February 2000 it is
going to be used extensively, the world over, and will soon become the industry standard. All the
models require good calibrations and continuous improvements based on the experience. It
should be noted that HDM-4 cracking, ravelling, potholes, edge break, rutting and roughness
models can not be separately implemented.
It is essential that the research works required for the implementation of HDM-4 models in the
dTIMS should be started as soon as possible. But it is important that the research is carried out
phase wise in a well planned manner. The research required for HDM-4 models should be
undertaken in the following phases:
© 2001 Ingenium 33
• Study on the appropriateness of the HDM-4 RDWE models for the NZ condition and possible
modification by the desktop study by comparing the results of the condition prediction
undertaken by NZ dTIMS and HDM-4 using the data of various regions in NZ. Calibration to
Level 1 is required for this study;
• Study to Level 2 calibration of the HDM-4 models for different regions of NZ;
• Replace the HDM-III models with HDM-4 models in the NZ dTIMS setup;
• Use the Long term performance prediction sites for calibrating and modification of the model.
© 2001 Ingenium 34
5. WORKS EFFECTS
5.1 Introduction
When maintenance work is performed it will have an immediate effect on the pavement condition
and long-term effects on the subsequent pavement performance and road use.
Different types of treatment have different effects on the road inventory and condition. Figure 5.1
illustrates that the different category of treatment resets the roughness to different extents. It is
very important that all the parameters which are affected by the treatment should be reset to a
correct value.
Roughness Progression Maintenance Category
Routine
Reset
Resurfacing
Reset
Smoothing
Reset
Strengthening
Figure 5.1: Resets for Roughness
The variable that describes the road characteristics needs to be reset to reflect the effects of the
works in terms of the following:
• Pavement condition;
• Pavement history;
• Road deterioration factors; and,
• Road use.
© 2001 Ingenium 35
This chapter outlines the details of the method used for resetting the variables that describe the
road characteristics and necessary improvements and the study needed to be carried out in this
issue.
5.2 Resets in NZ dTIMS Setup

The parameters for resetting have been divided into two groups for simplicity:
• Condition parameters; and,

• Other parameters.
Maintenance treatment when applied affects the pavement condition. All the condition parameters
have to be reset when a treatment of strengthening or smoothing maintenance categories is
applied (see Table 5.1). On the other hand, in the case of resurfacing of the pavement all the
condition parameters are reset except ‘RDM’ and ‘RDS’.
Table 5.1: Condition Parameters Requiring Resets
Field Name Description Maintenance Category

Strengthening Smoothing Resurfacing Routine
ACA Area of all cracking (%) • • • •
ACW Area of wide cracking (%) • • • •
ARV Area of ravelling (%) • • • •
APT Area of potholing (%) • • • •
APH Area of pothole patches (%) • • • •
RDM Mean rut depth (mm) • •
RDS STD of rut depth (mm) • •
IRI Roughness (m/km) • • •
TD Texture depth (mm) • • •
SFC Sideways force coefficient • • •
ASH Area of shoving (% of wheelpath length) • • • •
AFL Area of flushing (%) • • •
PCA Previous all cracking (%) • • •
PCW Previous wide cracking (%) • • •
Treatments of strengthening, smoothing and resurfacing categories correct all the surface
defects. Therefore, resets of the surface distress parameters such as ‘ACA’, ’ACW’, ‘ARV’, ‘APT’,
‘APH’, ‘ASH’ and ‘AFL’ are just equal to a zero value. Surface deformation parameters like ‘RDM’,
‘RDS’ and ‘IRI’ are mainly affected by the shape correction treatments, which come under the
strengthening and smoothing categories. Resets for a treatment can also have different values in
various places depending upon the local construction practice. For example, asphalt concrete
reconstruction can have ‘IRI’ of 2.0 m/km if the modern asphalt layer is used, but at the same
time could have 2.5 m/km if old asphalt layer is used. Hence, a study should be carried out to
determine these resets for the dTIMS setup.
Only the parameters indicating the pavement surface distress such as ‘ACA’, ‘ARV’, ‘APT’, ‘APH’
and ‘ASH’ have to be reset after the routine maintenance treatment.
‘TD’ and ‘SFC’ models are still in the development phase. Study on the resets required for ‘TD’
and ‘SFC’ parameters should be carried out together with the research on improvement of such
models.
© 2001 Ingenium 36
The other parameters requiring resets in dTIMS are shown in Table 5.2. All of the listed
parameters have to be reset if a treatment of strengthening or smoothing category is applied.
‘SNP’ and ‘AGE3’ do not require any reset if a resurfacing category of treatment is applied. No
resets to any of these parameters are needed if only routine maintenance is applied.
Table 5.2: Other Parameters Requiring Resets
Field Name Description Maintenance Category

Strengthening Smoothing Resurfacing Routine
SNP Structural number • •
AGE2 Years since last (re)surfacing • • •
AGE3 Years since last (re)construction of • •
base
SLIF Predicted seal life (years) • • •
PAVETYPE HDM-4 pavement type code • • •
SURFTYPE Surfacing type code • • •
BASETYPE Base type code • •
HOLD Thickness of old surfacing (mm) • • •
HNEW Thickness of last surfacing (mm) • • •
CQ Construction quality indicator (0/1) • • •
MCOMP Relative compaction of the pavement • •
CHIP Chip size (code) • • •
SS_FLAG Second seal needed (0/1) • • •
MCI Maintenance cost index • • •
‘MCI’ model is in the development stage and the values for resets should be established in the
process of improvement of the model.
5.3 Resets Requiring Further Studies

Default values for various resets available in dTIMS are based on observations and experience.
The expressions used are mostly derived from HDM-III models. More recent and improved
expressions are available in HDM-4, which should be checked and incorporated in to dTIMS.
The following issues have to be in the priority list:
• Resets for routine maintenance; and,

• Roughness and rutting resets.
5.3.1 Resets for Routine Maintenance

Routine maintenance in the NZ dTIMS Setup is based on the actual maintenance practice
followed in NZ. It includes:
• Patching of potholes;
• Digout for shoving;
• Sealing of narrow cracks; and,
• Crack filling of wide cracks.
Patching of potholes is surely small on most networks, ditto treatment of wide cracks except
where they are caused by longitudinal shear failure, something we are not modelling. The digout
of shoving is common but we do not have a shoving progression model worthy of the name. The
initial values in RAMM give no real indication of the true shoving problem if the RAMM survey has
been made immediately after the crew has been around and dealt with the shoved areas. A
© 2001 Ingenium 37
better indication is the annual area of digouts, but this data is said to be unreliable. All these facts
should be considered while carrying out the study on the routine maintenance resets.
Besides, there are other minor maintenance activities which are carried out under routine
maintenance, but are not modelled under dTIMS. Hence the resets are not required for these
treatments.
The default values for resets of the condition parameters due to routine maintenance is given in
Table 5.3. The default criteria for resets were based on observations and on the feedback from
the pilot study. It is based on the assumption that all the above mentioned defects will be
corrected and there is no budget constraint for the routine maintenance works.
Table 5.3: Resets for Routine Maintenance
Field Name Description Resets1

ACA Area of all cracking (%) P = 50
ACW Area of wide cracking (%) P = 20
ARV Area of raveling (%) N
APT Area of potholing (%) P = 10
APH Area of pothole patches (%) N
RDM Mean rut depth (mm) N
RDS STD of rut depth (mm) N
IRI Roughness (m/km) X = R_IRI_PTCH
ASH Wheelpath length of shoving (%) P = 30
AFL Area of flushing (%) N
Note: 1/ Resets A = Absolute resets
X = Expression
P = Percentage reset
N = No resets needed
However, the budget is not always sufficient to carry out all the activities included in routine
maintenance work. The decision on how much of the defects are to be treated by routine
maintenance could depend on:
• Condition of the roads;

• Functional class of road;
• Maintenance regime applied by the given RCA; and,
• Funding available.
For example it may not be economical to fill all the potholes in a badly deteriorated road with a
large amount of shoving. Hence, while defining a standard, a maximum limit for a routine
maintenance effort (in terms of cost or area) is usually defined. Such provision is not available in
the current setup.
The functional class of a given road also defines the level of routine maintenance. Remedial
works for the surface defects are done more promptly and with higher quality in the case of the
State Highways and Motorways. It will obviously affect the reset of condition parameters.
Similarly, some Road Controlling Authorities (RCA) give preference to routine maintenance and
try to fix most of the problems through routine maintenance, while others, prefer to wait for
periodic maintenance once the condition deterioration reaches a given level.
Obviously the funding available also affects the level of the routine maintenance carried out.
© 2001 Ingenium 38
It must be noted, that explicitly modelling different routine operations and their effects will
significantly increase the complexity of the dTIMS setup and is generally not recommended.
However, it is more effective to customise the resets applied to the general routine maintenance
regime for a given RCA based on the separate study on modelling appropriate routine operations
and effects for different classes of road and traffic volume.
No published research work carried out in NZ on this issue could be found. Therefore a
comprehensive study will be needed to define the average trend and calibrate the resets for the
routine maintenance works. Besides, models in dTIMS have to be enhanced to handle the annual
maximum limits of area of different defects (i.e. potholes, cracking) per km or maximum cost limits
of routine maintenance per km. The expressions for the resets should be modified to calculate the
reset values even when the partial corrections of the defects are carried out.
5.3.2 Roughness and Rutting Resets

As stated earlier, the roughness of the road is affected by all the treatments whereas only
treatments of smoothing and strengthening categories affect the rutting condition. The resets for
roughness and rutting used in the NZ dTIMS Setup are given in Table 5.4. The reset expressions
used in some cases is shown in the Table 5.5 and
© 2001 Ingenium 39
Table 5.6.
Table 5.4: dTIMS Roughness and Rutting Resets
Maintenance category Treatment Type Resets for1

IRI RDM RDS
Strengthening AM reconstruction (SRA) A = 2.0 A=0 A=0
ST reconstruction (SRL) A = 2.5 A=0 A=0
Thick granular overlay (SOL) A = 2.5 A=0 A=0
Thick asphalt overlay (SOA) X = R_IRI_OL A=0 A=0
Smoothing Thin granular overlay (MOL) A = 2.5 A=0 A=0
Thin asphalt overlay (MOA) X = R_IRI_OL X = R_RDM_OL X = R_RDS_OL
Rip and remake ST (MRL) A = 2.5 A=0 A=0
Rip and remake AM (MRA) A = 2.5 A=0 A=0
Mill and Replace (MMA) A = 2.0 A=0 A=0
Recycle (MYL) A = 2.5 A=0 A=0
Resurfacing Seal – Large chip (RSL) X = R_IRI_ST N N
Seal – Small chip (RSS) X = R_IRI_ST N N
Note: 1/ Resets A = Absolute resets
X = Expression
N No resets needed
Table 5.5: Roughness Reset Expressions
Treatment Expression dTIMS expression

name
Routine/Patching R_IRI_PTCH IRI - 0.38 * (DPC + DPR + DPE)
Reseal R_IRI_ST IRI + MIN(0, MAX(0.3 * (5.4 - IRI), -0.5)) - 0.0066 * (0.62 * PCA + 0.39 * PCW)
- 0.38 * (DPC + DPR + DPE)
AC overlay R_IRI_OL MIN(IRI, 3.85 - (MIN(HNEW, 80) + MIN(HNEW, 40)) / 52 + 28 * MAX(IRI -
3.85, 0) / MAX(HNEW, 28))
Note: IRI is Roughness in m/km
DPC, DPR, DPE is annual increment in potholes due to cracking, ravelling and pothole extension
PCA, PCW is area of previous all and wide cracking
© 2001 Ingenium 40
Table 5.6: Rut Depth Reset Expressions
Treatment Expression dTIMS expression

name
AC overlay R_RDM_OL 0.15 * RDM
R_RDS_OL 0.15 * RDS
As these resets are only based on the observations the default values provided and the
expressions used should be calibrated based on a comprehensive study. An assessment of the
condition of the pavement before and after a given treatment could be done for the calibration. It
is essential that all the possible conditions in NZ should be covered during the calibration
process.

The research should be focused in the following directions:
• Study on possible inclusion of HDM-4 work effects models;

• Calibrate the surface distress resets for the routine maintenance including the effect on reset
while following different routine maintenance regimes by different authorities; and,
• Calibrate the resets for roughness and rutting.
© 2001 Ingenium 41
6. AREA WIDE TREATMENTS
6.1 Introduction
During the Wellington Technical Workshop the issue regarding the inclusion of Area Wide
Pavement Treatment (AWT) in the dTIMS setup was raised. It was identified that a lot of
maintenance works are funded under this Transfund structural maintenance cost category.
6.2 Definition
The Transfund’s definition for AWT is as follows:
The area-wide pavement treatment category provides for all the pavement
maintenance techniques where the least-cost maintenance option is an area-
wide treatment and there is no geometric improvements
Examples of the qualifying work include:
• Overlays;
• Rip and relay; and,
• Chemical stabilisation.
Rule: To qualify for inclusion in this work category the work must:
• Be the long-term, least-cost option for the road controlling authority

calculated in terms of present value; and,
1
• Not include any geometric improvements. .
The above definition for AWT highlights the fact that it should be a long-term, least-cost option for
the agency maintaining the road. That is, the economic appraisal is based on the difference of the
present values for the proposed maintenance strategy and existing maintenance strategy. Net
present value (NPV) is calculated based on the following expression (Transfund, 1999a):
NPV = A – B (6.1)
where NPV is the net present value

A is the present value of the existing maintenance strategy
B is the present value of the alternative (AWT in our case) strategy
For the project to be justified NPV must be positive.
6.3 Existing Practice

AWT category is newly introduced by Transfund and there are still some misunderstanding as to
when the fund will be allocated under this category. Some of the RCA’s are using this category to
fund the treatments which could not be requested under any of the remaining structural
maintenance categories. Others are using this to request funds for the project when the required
benefit cost ratio (B/C) based on total transportation (TTC) cost is less than the threshold required
by the Project Evaluation manual (Transfund, 1999a).
1
Transfund (1999a). Transfund’s Programme and Funding Manual. Amendment 5 of September 1999.
Wellington.
© 2001 Ingenium 42
The only criteria followed is that the treatment should be totally related to the pavement only and
no geometric improvement is carried out. For most of the low traffic volume roads, for example,
in the Southland District Council are usually justified economically by lowest NPV (Marshall,
1999).
6.4 AWT in dTIMS

Two major issues related to incorporating the AWT in dTIMS are:
• Defining the criteria when a section should be considered for AWT category of funding; and,
• Modifying the dTIMS software/setup so that it can be optimised based on NPV of the agency
cost.
Most of the other Transfund structural maintenance categories use the total transportation cost
(TTC) as the basis of their economic appraisal. TTC includes construction, maintenance and road
user costs. That means the benefit of reduction of road user cost with improvement of roughness
is also taken into account. The current NZ dTIMS Setup does the optimisation based on the
Incremental Benefit Cost ratio (IBC) as a function of TTC. This is not in the case of AWT where
only the NPV of the agency cost for maintaining the road has to be considered.
The criteria for when a section should be considered as AWT and evaluated based on NPV, must
be clearly defined. A survey with all the relevant authorities should be carried out to develop this
criteria.
For the high traffic road the maintenance strategy can easily be justifiable using optimisation
based on TTC. For very low traffic volumed roads there could be some problems with justification
based on TTC as the benefit of roughness improvement will be marginal. Economic appraisal
based on NPV of the agency cost could be appropriate in such cases. Therefore, one of the
options could be to use AADT threshold values below which the economical evaluation will be
based on NPV. Similarly a possibility of using a threshold of BCR or other parameters should be
considered. Besides, we also have to consider the fact that if the maintenance includes geometry
as well as pavement maintenance it will not fit into the AWPT criteria in the first cut of decision
making.
It should, however, be appreciated that some kind of minimum level of service should be provided
even for the economically non-viable roads. It is essential to define this minimum level of service.
In this regard it is preferable to do this study with the Do-Minimum alternative study. Furthermore,
a study should be carried out establishing the locally used, low cost maintenance treatments for
such roads. These treatments should be included in the NZ dTIMS Setup if needed.

• Defining the criteria for when a section should be considered for AWT category for funding;
• Defining the minimum service level performance standard (if needed) for those sections
considered as AWT;
• Defining additional treatments which (if needed) are required for low traffic roads; and,
• Development of a procedure to modify the dTIMS software/setup so that it can also be
optimised based on NPV of the agency cost.
© 2001 Ingenium 43
7. PAVEMENT STRENGTH
7.1 Introduction
The strength of a pavement is a complex function of the interactions between material properties,
layer thickness and depth, subgrade stiffness, and pavement condition. Paterson (1987) found
that the modified Structural Number (SNC) was the most statistically significant measure of
pavement strength influencing the deterioration of pavements. SNC was therefore adopted as the
primary strength parameter in the HDM paved road pavement deterioration relationships used in
dTIMS.
Recognising that pavement strength can be difficult to establish on a network wide basis, the
Pavement Strength program has been developed by the project to use up to seven different
methods for establishing SNC based on different levels of data availability (HTC, 1999a).
The methods are:
• Falling Weight Deflection (FWD) processed data method with each layer thickness known;
• FWD method with total layer thickness known;
• FWD without layer thickness;
• Californian Baring Ratio (CBR) with layer properties method;
• Benkelman Beam Deflection method;
• ARRB method; and,
• Typical Pavement Design method.
Based on what data is available SNC is calculated using one or more of these methods. Most of
the methods make use of layer properties and the in-situ CBR to establish the pavement strength.
It is important to realise that the pavement strength is determined by two factors, namely:
• The stiffness of the layers itself; and,

• The resistance to deformation of the sub-grade, which is determined by the CBR and the
moisture condition.
It is a complex issue to quantify these two factors and the interrelationship. The methods currently
in use do not always take account of all the factors that can influence the pavement strength.
Therefore, it is excepted that not all the methods will result in similar outcomes, but a certain trust
can be associated with each method. Provision is made in the Strength program to assign a
hierarchy of methods and according to the available data the best method’s results will be
assigned for a section (see HTC, 1999b).
7.2 Problem Statement

Non Destructive Test (NDT) methods like the Benkelman Beam and FWD methods are used
frequently these days, because of the high speed of these measurements. Concerns were raised
as to the appropriateness of the expressions developed overseas for determining SNC. Tonkin
Taylor Ltd. have developed the appropriate equation for Benkelman and FWD data for NZ (HTC,
1999a). These methods are not yet compared with the traditional method of determining the SNC
based on the subgrade CBR, pavement layer properties and thickness’ using the destructive
methods.
Furthermore, research done in the New Plymouth District showed the interesting behaviour of
pavements built on volcanic soils. The NZ NDT expressions for determining SNC do not include
the effect of volcanic soils at this stage.
© 2001 Ingenium 44
Similarly, doubts on the appropriateness of using SNC as the pavement strength indicator for the
predictive modelling has been expressed on the observation of some of the pavements in the
South Island (Salt, 1999). Pavements with strong rock subgrade and weak basecourses usually
gives relatively high SNC. But it was found that the failure of the pavement was mainly due to
shallow share failure (manifested as shoving and rutting). The FWD processed data (layer modulli
defined with back calculation) method with each layer thickness can identify sub standard base
materials and then subsequently exclude the layers below from the consideration in the structural
number calculation.
Furthermore, the default expressions in the existing NZ dTIMS setup used to reset the structural
number after application of a treatment is based on the HDM models. These reset expressions
need to be checked for NZ conditions to make sure that the estimates are reliable.
SNC is one of the most sensitive parameters used in the pavement deterioration predictive
modelling. Hence, any anomalies similar to those explained earlier should be thoroughly
investigated and remedies should be found.
7.3 Further Research Needed

The three focus areas for further research could be to:
• Verify the results of the expressions developed for estimating the SNC using non destructive
method with the traditional destructive methods;
• Refine the New Zealand FWD methods to include the influence of volcanic soils;
• Check the reliability of the SNC reset expressions; and,
• Check the effectiveness of SNC as a strength parameter for pavement deterioration
modelling in the different conditions in NZ.
All three of the above-mentioned issues has been highlighted during the Wellington meeting as
important issues to be addressed during Phase II.
Existing data should be used in this investigation as far as possible. However for comparing the
SNC estimating methods using non-destructive and destructive techniques some field and
laboratory testing might be needed. The most important thing is that one must ensure that the
data used will be representative of the conditions around New Zealand.
Research on pavement strength should also be a part of a long term pavement performance
(LTPP) study. The effects of surface water ingress and drainage on the strength of different
pavement layers should be accessed by periodic FWD measurements. As for the LTPP sections
there should be more accurate data on layer thickness than the back calculation method which
should be more accurate than at present for network-wide data.
This study can be time consuming and it is therefore recommended that the scope of the study is
kept simple. Resulting from this study, recommendations can be made with regard to the
accuracy of the methods and future research needs. It is also important to take into account other
ongoing research done in the country on this topic.
© 2001 Ingenium 45
8. UNSEALED ROAD DETERIORATION MODELLING
Unsealed roads represent a large proportion of the road network of NZ. It is estimated that 39 %
of NZ roads consist of unsealed roads. Although these road types are normally low hierarchy and
low volume roads, they still fulfil an important function in the total road network. Tourists, farmers
and timber logging companies use these roads and these three industries are large contributors
to the NZ economy.
High costs are involved with unsealed roads. Not only do they require regular grading and re-
gravelling but high vehicle operating costs are also experienced on these roads. Because of the
nature of unsealed roads higher roughness levels than on sealed roads are unavoidable.
Managing maintenance on unsealed roads is critical, as the condition of unsealed roads changes
relatively quickly. This can be illustrated by comparing the total transportation costs (see Figure
8.1). Total transportation cost is a method to determine the optimum maintenance level of a road
network. It is calculated by the sum of initial construction cost, maintenance costs and road user
cost. Too little maintenance undertaken on unsealed roads will cause a drastic rise in road user
costs. However it would also not be economical to spend too much effort and funding on
maintenance.
Optimum Maintenance
Total Transport Costs

COSTS
Road User Costs
Maintenance Costs
Construction Costs
MAINTENANCE LEVEL
Figure 8.1: Calculation of Total Transportation Costs
Total transportation cost is one of the methods used in dTIMS to perform long-term maintenance
planning. Input parameters to this analysis include amongst others condition data, pavement
deterioration models and road user costs models. This chapter discusses the research
requirements for prediction models of unsealed roads.
8.1 Road Roughness

Road roughness is used to calculate road user costs and is therefore a critical input to the
analysis performed in dTIMS. Many researchers have derived models for roughness progression
on unsealed roads (Visser, 1981; Page-Green, 1989 and Paterson, 1985) but these models are
© 2001 Ingenium 46
based on studies performed outside NZ and limited studies have been performed in NZ. Cenek
(1998) presented Figure 8.2 to compare available models for NZ roads. Note that that the
different models gave significantly different predictions for road roughness.
600
NAASRA Roughness [counts/km]
500 Ferry
HDM-III Avg.
HDM-III Calc.
Paige-Green
400
300
200
100
0
0 50 100 150 200 250 300 350
Measured Average Roughness [counts/km]
Figure 8.2: Predictions of Different Unsealed Roads Models
Various unsealed road models are used in different countries and their experience of using these
models for predictive modelling should be reviewed to determine the best possible models for NZ
conditions. However to carry out such a study will take some time. Hence, it is advisable to as a
first step carry out the calibration of the existing model already available in the NZ dTIMS setup
for use in the interim.
The HDM unsealed road models have been adopted in dTIMS for Phase I. The decision on this
model was based on the fact that the HDM model is less data intensive than the Paige-Green
model.
The equations used in the current dTIMS Setup are (HTC, 1999a):
The slope mean material gradation is calculated as:
P _ MGR = MIN(0.36, MGM,1 − MGM) (8.1)
(MG075 j + MG425 j + MG02 j )

P _ MGM = (8.2)
3
 P075 
P _ MGD = IF P425 = 0,1,  (8.3)
 P 425 
where: MGR is the slope of the mean material gradation

MGD is the material gradation dust ratio
P425 is the amount of material passing the 0.425 mm sieve
P075 is the amount of material passing the 0.075 mm sieve
© 2001 Ingenium 47
The value of parameter MG075j is obtained as follows:
if D95j > 0.4 then:

 P075 j 
log e  
 95 
 
MGM075 j = (8.4)
 0.075 j 
log e  
 D95 j 
 
otherwise MG075j= 0.3
if D95j > 1 then:
 P425 j 
log e  
 95 
 
MGM425 j = (8.5)
 0.425 
log e  
 D95 j 
 
otherwise MG425j= 0.3
if D95j > 4.0 then:
 P02 j 
log e  
 95 
 
MGM02 j = (8.6)
 2.0 
log e  
 D95 j 
 
otherwise MG02j= MG425j
where: MG j is the slope of the mean material gradation

D95 is the maximum particle size of the material, defined as the equivalent
sieve opening through which 95% of the material passes in mm
The minimum and maximum roughness are calculated as follows:
P _ IMIN = MAX(1, MIN(7.7,0.36 × D95 × (1 − 2.78 × MGR ))) (8.7)
P _ IMAX = MAX
(150,279 − 421× (0.5 − MGD) 2
+ 0.22 × CV − 9.93 × RF × MMP )
13
(8.8)
where: P_IMIN is the minimum roughness in IRI

P_IMAX is the maximum roughness in IRI
CV is the average horizontal curvature of the road in deg/km
RF is the average rise plus fall of the road
MMP is the mean monthly precipitation in mm/month
© 2001 Ingenium 48
The grading frequency is calculated by:
P _ BLFQ = 165.52 − 19.1× LOG(ADTT ) (8.9)
where: ADTT is the annual average daily traffic (AADT) in veh/day
The average roughness during the analysis period is calculated as:
(
P _ IRIG = IMAX + (0.447 − 0.23 × MGD) × 1 − e (COEF×BLFQ ) )
 
 COEF 
×
(IMAX − IMIN)
×
e × BLFQ 
(8.10)
(1 − (0.53 + 0.23 × MGD))   COEF  
  BLFQ  
   
The coefficient used is calculated as:
COEF = −0.001 × (0.461 + 0.0174 × (ADT1 + ADT2 + ADT3 ) + ) (8.11)
where: P_IRIG is the average roughness of the unsealed road during the analysis year
IMAX is the maximum roughness for a material
IMIN is the minimum roughness of a material
BLFQ is the grading frequency
ADT is the daily traffic per vehicle class
In order to perform roughness calibration on unsealed roads extensive data collection is required.
The experimental design should consider all the important factors influencing the deterioration
rate of unsealed roads. During the original HDM study in Kenya and Brazil 46 and 48 sections
were monitored respectively. The factorial design consisted of the following factors (Paterson,
1987):
• Vertical geometry;
• Horizontal curvature;
• Daily traffic;
• Rainfall; and,
• Surface type;
A similar approach can be followed for the calibration of unsealed roads for NZ conditions. The
definition of rainfall area and material types classification needs careful consideration as these
factors will largely determine the numbers of sections needed for the design matrix. Normally the
sections are between 500m and 1 km long. The roughness survey frequency depends on the
maintenance practise and the deterioration rate of the particular roads. Enough surveys have to
be performed between grading cycles to provide the required data points for regression analyses.
The survey frequency can vary from 14 days up to as long as 30 days. It is recommended that an
accelerated measuring interval be used for NZ unsealed roads as this approach could shorten the
study period. The accessibility to roughness survey vehicles will dictate where the study can be
undertaken as a lot of unsealed roads requiring surveying will be very costly to monitor.
© 2001 Ingenium 49
8.2 Dust Models

Unlike roughness dustiness does not only contribute to the discomfort of the travelling public but it
also has a negative effect on the surrounding countryside in various ways. Some of the negative
effects of dust from unsealed roads include:
• Excessive dust leads to poor visibility and can become a safety hazard for drivers;
• Dust has harmful effects on agricultural activities. The effect on animal production seems to
be non-existent, but on crop yield the effects can be significant. Transfund (1997) uses a
value of $ 300 per kilometre per annum for the effects of dust from unsealed roads on
horticultural land;
• Higher maintenance cost on vehicles travelling on unsealed roads can be attributed to
increased roughness and dustiness; and,
• Dustiness is one of the main attributes for driver discomfort on unsealed roads. Together with
other factors such as roughness and safety aspects, this contributes to drivers being willing to
pay an additional 9 - 22 cents per vehicle kilometre to avoid travelling on these roads
(Transfund, 1997).
Limited research work has been undertaken to date on the prediction of dust levels on unsealed
roads. Paige-Green (1989) presented a dust model based on a correlation between a dust
severity rating and material properties. Typically the severity rating is on a five point scale ranging
from Dust Level 1 (No visible dust behind the vehicle) to Dust Level 5 (Extremely dusty-
surroundings obscured to a dangerous level). The downside of a severity rating is that it is based
on the subjective opinion of the assessor. However, later research work started using dust
measurements by the means of infrared beam and a transducer system (Jones, 1998). This
technique measures the air opacity in a duct behind the rear wheel of the test vehicle.
Dust effects should be considered in the economic analyses of unsealed roads as it has a
significant effect on the acceptance and performance of unsealed roads. The benefits of
upgrading an unsealed road to a sealed road are therefore, not only determined by savings made
on road user costs due to roughness improvements.
In order to include dust models to the analyses of unsealed road sections the following have to be
established:
• An effective measurement technique to quantify dust levels;

• Models to predict dust levels based on material and environmental factors; and,
• A relationship between dust levels and associated costs attributed by dustiness. These costs
can also include social costs such as driver and passenger discomfort.
A further aspect to consider is the effect of using dust palliatives. Dust palliatives are chemical
agents mixed into or sprayed on top of the wearing course. It acts as a bonding agent keeping the
fine material particles intact under the action of traffic. In preventing or depressing dust loss
roughness progression is slowed down. The benefits of using a dust palliative therefore include:
• Cost savings due to lower maintenance levels needed;

• Cost savings in road user costs; and,
• Cost saving experienced due to lower dust levels.
It is therefore recommended that the influence of using dust palliatives is included in the unsealed
road calibration study.
© 2001 Ingenium 50
8.3 Minimum Levels of Service

The land use of the area through which the unsealed road passes is also an important factor in
determining the required level of service for a given unsealed road. At the same time the
minimum level of service for an unsealed road will largely be reflected in the level of the
maintenance work carried out on the road. The cost for maintaining the road should be justified
economically and based on minimising the total transportation cost, which is directly related to the
traffic volumes.
At present, no hierarchical standard exists in NZ for the minimum level of service required for
different unsealed roads. The research study should develop such standards, which should be
based on economic as well as social considerations.

The HDM unsealed road deterioration models have currently been adopted into the NZ dTIMS
Setup. Other models were initially considered but preference was given to using the HDM
models based on the fact that it is less data intensive. However, an evaluation and review of the
models that have been used in different countries and the countries experiences in using them for
predictive modelling and for economic analysis should be undertaken as an integral part of the
detailed study of unsealed road models.
As the final output and results of the study will only be available after a minimum of two years the
research should be planned carefully and in such a way that the results of the research could be
used in earlier stages to at least improve existing model outputs. For example preliminary
calibration of the HDM unsealed roads models in dTIMS could be undertaken in the early stage of
the research programme. This evolutionary approach would help to utilise the benefits of the
research in the shortest possible timeframe.
The existing data for unsealed roads is limited for calibration purposes and therefore calibration of
these models will be data intensive and demanding, as it will require collecting material properties
for each unsealed road studied. Therefore, as an interim measure, an attempt should be made to
identify generic types of surfacing materials throughout NZ which can be used in the Interface
Program, to define the more specific material properties required within dTIMS.
It will not be possible in the near future to perform research for the calibration of all the unsealed
road models. The most critical models to calibrate include roughness and dust. Although material
loss is also an important model, however the calibration is much more difficult and takes longer
than roughness and dust. It is nonetheless desirable, as resurfacing can be a major part of
unsealed road maintenance costs.
The design matrix for the calibration of the roughness models should include factors such as:
• Vertical geometry;
• Horizontal curvature;
• Daily traffic;
• Rainfall; and,
• Surface type.
It is also recommended that dust models be developed as these models also contribute to user
costs and costs to agricultural activities. The development of dust models requires investigation
into the followings aspects:
• An effective measurement technique to quantify dust levels;

• Models to predict dust levels based on material and environmental factors; and,
© 2001 Ingenium 51
• A relationship between dust levels and associated costs attributed by dustiness. These costs
can also include social costs such as driver and passenger discomfort.
© 2001 Ingenium 52
9. TEXTURE DEPTH AND SKID RESISTANCE
9.1 Introduction
Both texture depth and SFC (Side Force Coefficient) are used as triggers in the NZ dTIMS Setup.
With these triggers, minimum standards or intervention levels are set to determine the timing of
resealing actions on a network. In this chapter future research needs will be identified for texture
depth and skid resistance models. These two models will be discussed simultaneously as there is
a strong interrelationship between skid resistance and texture depth and that research in the one
field might also need consideration of the other model.
Skid resistance is influenced by the following road associated factors (HTC, 1999c):
• Road geometry;
• Mix design for Asphalt;
• Seal design for chip seals;
• Seasonal affects;
• Road temperature;
• Chip size;
• Aggregate properties;
• Macro and micro texture; and,
• Drainage condition.
Of these factors the micro and macro texture have the strongest influence on the skidding
properties of a road. The definition of the micro and macro texture can be done according to the
wavelengths of road profiles where (Cenek, 1996):
• Macro texture - profile wavelengths between 0.5 mm and 50 mm; and,

• Micro texture - profile wavelengths less than 0.5 mm.
The profile wavelength can be defined as the distance between two following peak (amplitude)
measurements of a profile. This is shown in Figure 9.1 (HTC, 1999c).
Mean Profile Depth (MPD) Wavelength
Profile
Profile Depth (PD)
Figure 9.1: Surface Profile Measurements
The relationship between texture depth and skid resistance is complex, as it is a function of the
vehicle travelling speed. With the simultaneous measurement of the macro texture and the
friction, a friction-speed relationship can be derived that will give an accurate indication of the slip
properties of a road. This approach has been followed in the definition of IFI (International Friction
© 2001 Ingenium 53
Index). The IFI is reported in the form IFI(F60,Sp), where F60 is the friction at 60 km/h and Sp is
the speed constant for the instrument used (Lund, 1997).
9.2 Current Texture and Skid Resistance Models in dTIMS
9.2.1 Texture Depth

The incremental model for texture depth given by Cenek (1999a) is:
  1 − TD  
 k 2 k1 k 2 
∆TD = k1 − TD − k1 k 2 log10 10   + ∆NELV  (9.1)
 
 
where: TD is the sand patch derived texture depth in mm at the beginning of the
analysis year
∆TD is the incremental change in texture depth for the analysis year in mm
∆NELV is the number of equivalent light vehicle passes for the analysis year,
where one heavy vehicle is equivalent to 10 light vehicles
The model coefficients k1 and k2 for surface dressings are given in Table 9.1.
Table 9.1: Model Coefficients for Surface Dressings
Chip Grade
Coefficient AADT 2 3 4 5 6
k1 25 5.953 4.835 3.717 2.512 1.537
k1 125 6.208 5.044 3.880 2.639 1.601
k1 625 6.463 5.252 4.042 2.765 1.664
k1 3000 6.711 5.456 4.200 2.888 1.726
k1 15000 6.966 5.664 4.363 3.015 1.790
k2(1) All 0.127 0.126 0.123 0.118 0.110
k2(2) All 0.130 0.125 0.120 0.110 0.100
Source: Cenek (1999a)
Note that the texture depth model does not include a construction quality factor as used by other
HDM models. The construction quality factor was used by HDM-III to make provision for certain
defects that are attributed to problems during construction rather that to normal pavement
deterioration parameters. A similar behaviour of pavements is experienced in terms of the texture
depth models. The inclusion of a construction quality factor can reflect the poor performance of
some seals and the variability of seal life experienced in the field. It is thus recommended that a
constructed quality factor should be determined (Cenek ,1999a).
The texture model shown assumes that 1 heavy commercial vehicle is equivalent to 10 light
vehicles. Research done in South Africa showed that this ratio is closer to 1:40 (Cenek, 1999a).
Similar studies in NZ showed that this ratio for logging trucks could be as high as 1:100.
Therefore it is required that the current model should be expressed in terms of standard design
axles or that the equivalent light vehicle passes are defined for the different truck classes
operating in NZ (Cenek, 1999a).
The current model predicts the seals will retain sufficient texture depth well beyond 14 years,
depending on the Average Least Dimension (ALD) and traffic volumes. However, the texture
depth decreases much sooner than the models predict due to other factors such as flushing and
chip loss. Binder softening attributes flushing that manifests as chip loss and embedding into the
existing seal layer. The softening of binder is more likely to occur in areas with extreme
© 2001 Ingenium 54
temperatures. Therefore, calibration of the texture models should be performed for on a range of
temperatures and types of existing seals. By selecting a range of locations it would be possible to
calibrate the model constants K1 and K2 to different environmental conditions and surface types.
9.2.2 Skid Resistance

The SFC model used in the NZ dTIMS Setup is based on the original model developed by Salt
(1977). The modified model used in the NZ dTIMS Setup is:
∆SFC = SFC − 0.663 × 10 −4 GROWTH _ H / 100 × QCV (9.2)
where: ∆SFC is the incremental change of SFC in the analysis year

SFC is the SFC at the beginning of the analysis year
GROWTH_H is the traffic growth rate in %
QCV is the number of commercial vehicles/lane/day
The model developed by Salt (1977) was based on research work on asphalt with large 20 mm
chips. It is thus recommended that this work is validated for NZ asphalt and chip seals. There are
further influencing factors on SFC that need further research such as the effect of (Cenek,
1999a):
• Geometry;
• Chip size; and,
• Commercial vehicle equivalency.
Cenek (1999b) also proposed research in the following areas, which also involve texture depth:
• Deriving of crash risk and surface characteristic relationships;

• Skid resistance/roughness relationships; and,
• Predicting water pond depth / aquaplaning potential based on predicted rut depth.
There has been conflicting views on whether SFC should be used as an intervention in dTIMS
per se. The reason is that SFC is dependent upon the Polished Stone Value (PSV) of the stone
and volume of heavy traffic. As PSV of the chips are not kept in the RAMM database it is not
possible to use the model using PSV. Using a common expression for the whole nation without
the PSV data of chips may not result in the right prediction of SFC. The solution is that, where a
treatment is triggered, the need for higher quality aggregate should be flagged if SFC is deficient
and may then be allowed for in the treatment costing.

Skid resistance and texture depth are important triggers for resealing actions in NZ and likewise
in the NZ dTIMS Setup as more and more resealing in the last few years have been triggered
because of the surface texture. The refining of the current prediction models for SFC and texture
depth are therefore high on the priority list of research needs.
In the context of the dTIMS implementation the most critical issues to address are the calibration
of the existing models. This includes calibration of texture depth models as a function of:
• Environmental conditions;
• Construction quality; and,
• Surface type.
Besides, high speed texture data collected for the State highway network are in terms of mean
profile depth. Hence, it is suggested that there is a need to develop additional models and build
them in to the setup, so that it can use MPD data without conversion.
© 2001 Ingenium 55
Likewise the skid resistance models should be calibrated to local conditions as a function of
geometry and chip size. For both the texture and the skid resistance models the appropriate light
vehicle equivalency factors should be established for the heavy vehicle classes operating in NZ.
As a second priority further research is also needed in the following fields:
• Deriving crash risk and surface characteristic relationships;

• Establishment of skid resistance/ roughness relationships; and,
• Predicting water pond depth/ aquaplaning potential based on predicted rut depth.
Cost effective research can be done on both the skid resistance and texture depth models as
high-speed data does exist for state highways. Different regions can be used in the design matrix
to ensure that the required environmental conditions are included in the study.
© 2001 Ingenium 56
10. NOISE REDUCTION TREATMENTS
10.1 Introduction
The reduction of noise pollution became an important consideration in the design of maintenance
actions in the urban environment. Noise pollution is also one of the important aspects to consider
during environmental impact studies for new road construction projects. However, with long-term
maintenance planning the engineer has to provide solutions for existing roads. This resulted in
the review of seal designs and maintenance practices used in these areas. The challenge was to
provide surface layers that reduce noise levels and at the same time still providing the required
skid resistance needed for safety aspects of the road. Products such as open graded asphalt,
slurry and two coat chip seals instead of single coat chip seals have become more popular to
achieve lower noise levels.
During Phase I of this project the incorporation of noise reduction treatments to the NZ dTIMS
Setup was identified as an issue to be addressed during Phase II. This incorporation necessitates
the resolving of the following aspects:
• Definition of treatment types used for noise reduction;

• Modelling of noise prediction;
• Formulation of noise levels intervention criteria; and,
• The effect of noise reduction treatments.
At this stage the extent of incorporating noise reduction treatments to the NZ dTIMS Setup is still
unclear. There are two possible approaches to follow, which are shown in Table 10.1.
Table 10.1: Possible Approaches for Including Noise Treatments to dTIMS
Parameter Sophisticated Approach Simplistic Approach

Treatments All treatment types used for noise All treatment types used for noise
reduction are separately defined reduction are categorise in one group
Models Noise prediction models included No model for noise
Intervention criteria Based on noise levels and practical Only based on existing triggers such
considerations of the locality where as texture depth. Once a reseal is
these treatments are used triggered practical consideration will
determine the use of these treatments
Effects Consider the effect on noise levels Consider only the effects on other
and other condition effects condition indexes such as cracking
Benefits Quantification of the benefit of the Use the treatment as a constraint to a
noise reduction and in comparison given section with low noise level
with the extra cost required for the requirements.
special treatment and future
maintenance
Cost implications and practical considerations will influence the decision on the approach to
follow. It is also possible to incorporate an approach somewhere in between the approaches
shown in Table 10.1.
10.2 Noise Reduction Treatments

The definition of treatment types used in long-term maintenance planning needs careful
consideration. The level of detail for the treatment definition has to be appropriate for the level at
which the outputs are used. This level of detail depends on the following factors:
© 2001 Ingenium 57
• The level of detail for the input parameters will determine the output level of detail. It should
be appreciated that long-term maintenance planning does give project level information.
However, not enough information is available to address detailed design aspects of
treatments such as type of asphalt used in overlays. The NZ dTIMS Setup typically
distinguishes between thick and thin overlays where a thick overlay is an overlay with
thickness more than 100 mm and 50 mm for granular and asphalt pavements respectively;
• The number of treatments defined for the dTIMS Setup will determine the complexity of the
analyses. This will not only have a significant effect on the analysis time but also on the
maintainability of the system; and,
• The effect of each treatment type on pavement behaviour must be predictable. Therefore the
effect of a specific treatment on individual condition parameters must be quantified. These
effects will ultimately determine what the resulting benefits are, of using a treatment.
The current version of the NZ dTIMS Setup only caters for three types of surfacing namely
asphalt surfacing, small chip seal and large chip seals. The distinction between large and small
chips are defined according to the grading of the seal where seal codes 5 and 6 represent small
chip seals and seal codes 1 to 4 large chip seals. Similarly no distinction is made for different
asphalt mixes. It must be noted that if porous asphalt is applied, then future maintenance costs
(cleaning) must be allowed for if the noise reduction and other properties are to be sustained.
The inclusion of noise reducing surface types is needed as the cost of these seal types differ
significantly from the seal types already in use. The application criteria of noise reducing seal
types also differ from the seal types mentioned. Seal types with lower noise reduction properties
are widely used around the world and can include:
• Open graded asphalt (also known as porous asphalt);

• Nova chip;
• Two coat chip seals; and,
• Slurry.
However asphalt, two coat seals and slurry seals are the most commonly used in urban areas
such as Auckland (Rambisheswar, 1999). Of these surface types slurry seals are not used in the
NZ dTIMS Setup.
10.3 Including Noise Prediction Models

Prediction models in the NZ dTIMS Setup mainly consist of HDM-III and some local developed
models. This section will discuss available noise models developed for NZ and HDM-4.
Barnes and Ensor (1994) proposed noise prediction models for NZ based on UK CRTN
(Calculation of Road Traffic Noise) formulae (DOT, 1988). These formulae are:
L10 (18Hr )Chipseal = 1.66 − 5.57 log(S / V ) + 1.65(P / r ) (10.1a)
L10 (18Hr )Asphalt = −2.63 − 1.65 log(P / r ) (10.2b)
L10 (18Hr )Friction − course = −2.63 − 1.65 log(P / r ) (10.3c)
where: L10(18Hr) is the sound level (in dB) that is equal or exceeding for 10% of the 18
hours
S is the sand circle reading in mm
V is the mean vehicle speed in km/h
P is the percentage of trucks
r is the ratio of medium:heavy trucks
© 2001 Ingenium 58
Correction factors are applied to these formulae in order to establish the predicted noise at a
specific point measured alongside the road. These correction factors resulted in the final equation
for predicting noise from uninterrupted traffic flow (Barnes and Ensor, 1994). This equation is:
L 10 (18Hr ) =
  500  
26.5 + 10 log(Q) + 33Log V + 40 +   
  V 
 5P   A 
+ 10Log 1 +  − 68.8 + 0.3G + 10Log 
 V   180 


− 10Log
((D + 3.5)
2
+ (h − 0.5 )
2
)  + F × 5.2
 13.5 
 
 6H − 1.5  P  S
× Log  + 1.65Log + 5.57C 0.77 − Log 
 D + 3.5  r  V
−3.4 X (10.4)
where: D is the distance from the reception point of the noise to the nearest
trafficked part of the carriageway in m
G Average longitudinal gradient of the section of road in %
A Angle of view in degrees. A clear view of the road has an angle view
of 180 degree
H Average height of propagation in m
h Relative height between the reception point of noise and the carriageway
in m
V Average vehicle speed in km/h
F Ground cover variable between 0 and 1 (see Table 20.4 in DOT (1988))
S Sand circle in mm
It can be observed that this equation is not in an incremental form as used in HDM-III. However
the texture (sand circle) variable in this model is incremental.
The incorporation of this model into the dTIMS Setup will also need to consider the following
aspects:
• Noise levels associated with some surface types could decrease with time. The worst noise
levels for some types of seal might be experienced when the seal is newly constructed. It is
therefore expected that some noise reduction treatments will only be triggered based on an
initial noise level of the road sections; and,
• Using these types of models requires the collection of significantly more data items than are
currently collected for roads such as average vehicle speeds.
10.4 Intervention Criteria for Noise Reduction Treatments

A trigger or intervention level is the criteria used to apply a treatment based on certain pavement
conditions or practical constraints. As an example one can trigger resealing actions based on a
© 2001 Ingenium 59
texture depth limit such as TD > 0.35 mm. Or, an alternative would be to trigger actions based on
practical constraints such as traffic volume and urban environment.
Currently authorities only consider the practical criteria in the selection of noise reduction
treatments. These criteria include (Rambisheswar, 1999):
• Business or shopping areas;

• Traffic volume > 10 000 veh/day;
• Road hierarchy (strategic, regional and district arterial); and,
• Hospitals or other special care areas.
These criteria could be utilised as the practical constraints for setting noise triggers in dTIMS.
Depending on the incorporation approach used, additional trigger limits for noise levels could be
added. Note that the incorporation of these triggers will require the addition of extra fields in
dTIMS such as noise level, business areas and special care areas. As no specific study has been
carried out and no definite guidelines are available on this method it was decided to take a
pragmatic approach in the refined dTIMS setup. The user will explicitly assign a flag to the section
where the noise restriction is required and dTIMS will, based on the treatment selection tree,
apply a treatment characterised by low noise level.
In setting the noise level triggers, some guidelines must also be set. Hunt (1988) found that 50%
of residents expressed a significant negative reaction on road noise levels higher than 69 dB
(L1018Hr). He proposed a limit of 60 dB (L1018Hr) to be used as only 35% of residents expressed
a significant negative reaction to traffic noise at this level.
As a general rule it is undesirable to use trigger parameters for which no benefit can be
estimated. In the case of noise reduction, quantification of noise reduction is problematic. The
initial cost will probably cost more and future maintenance needs may be higher. In an economic
analysis such treatments may be disadvantaged unless on high volume roads.
10.5 Maintenance Effects of Noise Reduction Treatments

In order to perform long-term maintenance, planning the effect each treatment has on the
condition and the deterioration of pavements must be quantified. The benefits used in the
economic analyses are based on the difference of pavement performance between applying
different treatments. Maintenance effects can be quantified by the immediate improvement on the
pavement condition as a result of applying a treatment. This is illustrated in Figure 10.1. The
effect or reset following on Treatment A is different from the reset for Treatment B. This figure
also shows that the condition deterioration rate can also differ following a treatment. This may be
as a result of different models associated for each treatment type.
© 2001 Ingenium 60
Do-nothing
Different Slope
Different
Treatment A Model
Index
Jump
Treatment B
Reset
Independent Variable
Figure 10.1: Effects of Treatments on Pavement Condition
This approach is used in the NZ dTIMS Setup where some condition indexes are immediately
reset following a treatment (e.g. ravelling is zero for a reseal action). The condition deterioration
rate following a treatment differs depending on the treatment type (e.g. crack initiation on a
resealed road will be sooner than crack initiation for overlays).
The resets for noise reduction treatments has to be defined and the following need to be
considered:
• Resets, as a function of noise level for all seal types, needs to be specified; and,
• Noise reduction treatment types should have resets for all other condition indexes such as
cracking.

This chapter discussed the aspects involved with the incorporation of noise reduction treatments
to the NZ dTIMS Setup. It showed that one of two approaches could be followed for this
incorporation. A simple approach can be followed, including practical triggers for the use of noise
reduction treatments. This approach will not consider any noise modelling. Secondly a more
sophisticated approach can be followed that would include noise models and the triggering of
noise treatments based on noise levels. A study should be conducted to establish the appropriate
approach for long-term maintenance planning in NZ. The aim of this study would be to investigate
current practises and available data at the relevant authorities. This study should also be
supplemented with discussions with role players in this field.
Further research areas for the incorporation of these treatments into dTIMS will include:
• Definition of treatments to use;

• Selection of the appropriate model taking account of cost implications such as collection of
additional data for these models;
• Establish the triggers to be used, that might also result in the inclusion of additional data field
in dTIMS;
• Defining the effects of noise reduction maintenance actions including the effects on other
condition indexes such as cracking and roughness; and,
• Quantification of the benefit of noise reduction, so that the treatment could be economically
justifiable.
© 2001 Ingenium 61
11. DO MINIMUM ALTERNATIVE
11.1 Introduction
Transfund’s Project Evaluation Manual (PEM) requires Benefit Cost ratio (B/C) to be calculated in
comparison with the “Do Minimum” option. Besides, for certain roads with very low traffic it may
not be economically viable to carry out maintenance as it won’t have a required benefit. However,
it is required that a certain minimum level of service should be provided by these roads to the
road users.
11.2 Definition of Do Minimum

Transfund’s definitions for Do Minimum option are included in the following statements
(Transfund, 1999a):
“….Minimum level of expenditure required to maintain a minimum level of

service…”
“….Only include in the do minimum work which is absolutely essential to

preserve a minimum level of service.”
“… For some projects on low volume roads the existing level of maintenance
expenditure may not be the minimum. In such cases, particularly where the
existing level of maintenance expenditure is high, the maintenance expenditure
shall be justified as an option along with other improvement options and the do
minimum shall only be the work necessary to keep the road open.”
“… The do minimum generally should not include rehabilitation/shape correction

treatment of a road to an improved standard. The only exception is when the
present value of the cost of the project and its future maintenance is less than the
present value of continued maintenance of the existing situation.”
The above definition clearly indicates that the Do Minimum option should be based on the
minimum level of service. But there is no clear definition on what should be considered as the
minimum level of service.
The routine maintenance and resurfacing is generally applied. There is some provision for
rehabilitation/shape correction treatment. However, only smoothing treatment could be
considered (strengthening treatment not applicable).
11.3 Status Quo

In the NZ dTIMS Setup for each road section a special Do Minimum strategy (DOM) is generated
based on the absolute minimum level of service defined for each traffic class. Triggers for DOM
are based on only two major condition parameters:
• Roughness(IRI); and,
• Surface Integrity Index( SII).
The IRI triggers are used to assign the shape correction treatments. Only the smoothing type of
treatments are considered in DOM.
The SII triggers are used for triggering the resurfacing treatments.
The proposed IRI and SII trigger limits are given in Table 11.1.
© 2001 Ingenium 62
Table 11.1: Proposed Default Triggers for Do Minimum Strategy
Traffic AADT IRI SII

Standard1 Do Minimum Standard1 Do Minimum
>20,000 3.5 5.8 8 30
10,000 – 20,000 3.8 6.3 10 32
4,000 – 10,000 4.2 6.9 11 36
2,000 – 4,000 4.6 7.6 13 40
500 – 2,000 5.0 8.3 15 45
100 – 500 5.5 9.4 18 55
<100 6.1 10.6 20 62
Note: 1/ The trigger limits used for Smoothing in the dTIMS performance based analysis.
The IRI trigger limits are kept very high for DOM strategy, so that expensive shape correction
treatments would be triggered as little as possible (see Figure 11.1).
18
16
14
IRI Limts in m/km
12
10
0
50 300 1250 3000 7000 15000 30000
AADT in veh/day
Performance Standard Minimum Level of Service
Figure 11.1: Trigger Limits for Minimum Level of Service
11.4 Problems with DOM

The proposed triggers for the DOM Strategy are based on observation as no definite criteria are
available at present to define the minimum level of service. Therefore, a nationwide study on the
current practice used to define the minimum level of service should be carried out.
The following parameters should be considered while defining minimum level of service:
• Economic justification;
• The perception of the road user on the minimum level of service (based on user surveys,
could differ from RCA to RCA); and
• The availability of funding.
© 2001 Ingenium 63
For some of the roads, the minimum level of service standard may not be economically viable.
The extra cost for such maintenance should be within the limit established by the policy decision
of the funding authority.
Benefit Cost Ratio (BCR) in dTIMS is calculated in comparison with the “Do Nothing” strategy
which causes some reporting problems. This is because, as mentioned earlier, Transfund
requires the BCR to be compared based on the DOM. It should be appreciated that this
shortcoming does not affect the optimisation process for the economically viable road as
optimisation in the dTIMS is based on the Incremental Benefit Cost (IBC) ratio. However, the
problem arises when a given road is not economically viable. During the optimisation process
dTIMS will choose the least cost option except the Do Nothing strategy (we explicitly exclude the
Do Nothing strategy to be selected). However, sometimes the strategy consisting of only routine
maintenance could be selected (See Figure 11.2). Such a strategy may not be able to maintain
the minimum level of service specified.
20000
PV Savings in TTC
15000
ROUTINE SMOOTHING
RESEAL
DOM
10000
STRENGTHENING
5000
0
0 50000 100000 150000
PV Agency Cost
Efficiency Frontier Lower Limit of Efficiency Frontier Strategies
Figure 11.2: Efficiency Plot of Different Strategies
In meetings with Transfund it was decided to apply a temporary solution in the refined dTIMS
system, until the findings of further study is made available. The least cost option in terms of the
present value of the agency cost for the analysis period of 20 years (see Figure 11.3) is
considered as the Do Minimum in the Reporter program. However, as discussed earlier, this
approach does not allow the option to maintain a predefined minimum level of maintenance
standard.
© 2001 Ingenium 64
Do Nothing
Routine
Maintenance
Do Minimum
PV Benefit
PV Benefit
PV Benefit
Strategies
PV Cost PV Cost PV Cost
Figure 11.3: Minimum Cost Option

• Define more precisely the criteria for defining the minimum level of service;
• Sensitivity analysis of trigger limits of various condition parameters to define the DOM
strategy; and,
• Developing the procedure for improving the dTIMS software/setup so that the BCR can also
be calculated based on the DOM strategy and the Do-Nothing, not any other strategy, is
selected when the road is not economically viable to maintain.
It must be noted that such research should be carried out in conjunction with the Area Wide
Pavement Treatment (AWPT) and Maintenance Cost Index (MCI) research work which is closely
related to the Do Minimum research.
© 2001 Ingenium 65
12. MAINTENANCE COST INDEX
12.1 Introduction
During the life cycle of road maintenance, actions such as reseals and rehabilitation are
performed in order to keep the road condition at the required level of service. During the periods
between major maintenance works, routine maintenance is applied to repair localised failures.
Routine maintenance also includes preventative maintenance actions such as crack sealing.
As the road deteriorates with time the routine maintenance cost of the road also increases. Once
a rehabilitation is carried out the cost of routine maintenance decreases (Figure 12.1). When the
lifecycle cost of the maintenance is calculated there is always a point on the road deterioration
model when it is economically beneficial to do rehabilitation rather than continuing to do the
routine maintenance works. Maintenance Cost Index (MCI) represents the annual maintenance
cost threshold value to warrant rehabilitation work.
Routine Maintenanace Cost
Rehabilitation
Years
Figure 12.1: Routine Maintenance Cost Increment
Ideally one would be able to model the distresses that give rise to routine maintenance needs.
With such models, the appropriate levels of routine maintenance could be evaluated and costed.
MCI is just a surrogate for such models. Unfortunately, the only current way to handle annual
quantities of a distress is via the annual amount of repair.
12.2 Maintenance Cost in the RAMM Database

Expenditure related to ‘General Maintenance’ is held in the RAMM Maintenance Cost table. The
cost is categorised into several cost groups which include pavement, surfacing, environment etc.
The maintenance cost for the pavement and surfacing group for each year from the RAMM
maintenance cost table is summarised in a separate table based on the treatment length section
in RAMM v 3.7. The total annual maintenance cost is defined as the “Maintenance Cost Index”
(MCI).
© 2001 Ingenium 66
12.3 Maintenance Cost Index Model

Cost of the maintenance gives an indication of the over-all condition of the road. The worse the
condition of the road the higher the routine maintenance costs become. The MCI which is related
to the actual pavement and surfacing repairs has been identified as an indicator for triggering a
strengthening treatment once it exceeds the required limit.
A study carried out by Opus International Consultants Ltd showed that the maintenance cost is
basically the function of the pavement age. Resurfacing has very little effect on reducing the
maintenance cost. Models developed by Opus International Consultants Ltd and supplied to the
project to estimate the Cumulative Maintenance Costs are:
2
For pavement maintenance costs in $/m :
1.3183
P(A) = 0.0797*A (12.1)
2
For surfacing maintenance costs in $/m :
S(A) = 0.077*A0.8952 (12.2)
where A is the pavement age in years
Figure 12.2 illustrates the cumulative maintenance cost for the surfacing and the pavement
group.
4.5
4.0
Cumulative Maintenance Cost, $/m2
3.5
3.0
1.3183
0.0797 x A
2.5
2.0
1.3183
0.0797 x A
1.5
1.0
0.5
0.0
0 2 4 6 8 10 12 14 16 18 20
Pavement Age
Pavement Surfacing
Figure 12.2:OPUS MCI Expression
© 2001 Ingenium 67
These models are converted into an incremental model for setting up in the dTIMS system:
For AGE3=1:
DMC = 0 (12.3)
For AGE3>1
DMC =
(PAV _ WIID) × (AGE3M12 − 2 × (AGE3 − 1)M12 + (AGE3 − 2)M12 )+
8
(M21× (AGE3 M22

− 2 × (AGE3 − 1)
M22
+ (AGE3 − 2)
M22
)) (12.4)
where DMC is the annual increment in maintenance cost in $/km

M11 and M12 are coefficients for surfacing maintenance cost (Default M11=616
and M12 = 0.8952)
M21 and M22 are coefficients for pavement maintenance cost (Default
M21=637.6 and M22 = 1.3183)
PAV_WID is the pavement width in m
Maintenance Cost index for a given year is calculated as:
MCIa = MCIb + DMC (12.5)
where MCIa is the Maintenance Cost index in $/km

MCIb is the maintenance cost of the pavement and surfacing
maintenance in last year in $/km
These MCI expressions and triggers are based on the Opus experience on State Highway
maintenance works in the Napier region. Trigger limits for triggering a Strengthening treatment is
given in Table 12.1. It is recommended that the coefficients of MCI and triggers should be
customised based on local experience as they are likely to vary between road controlling
authorities.
Table 12.1 Proposed Trigger Values for MCI
Traffic Range Proposed Trigger

(AADT) ($/km)
>20,000 + Motorway 6,000
10,000 – 20,000 7,000
4,001 - 10,000 8,000
2,001 - 4,000 10,000
501 – 2,000 12,000
101 - 500 15,000
< = 100 18,000
Source: HTC (1999b)
Table 12.2 shows the resets used in the NZ dTIMS setup based on the experience of Opus for
the Napier region.
Table 12.2: Resets for MCI
© 2001 Ingenium 68
Resets after Resurfacing

Strengthening Smoothing Resurfacing
MCI Resets 1400 X = 0.2 x MCI 0.4 x MCI
12.4 Issues related to MCI

There are 2 main issues related to MCI:
• Defining the initial data of MCI for a section; and

• Rate of increment of the MCI.
The initial value of the MCI is based on the data available in the RAMM database. To reduce the
factor for the given year, maintenance cost of the last 3 years is averaged to get MCI. However, it
should be appreciated that no rehabilitation/reconstruction works should have been undertaken
within this period.
As described earlier the maintenance cost data of ‘pavement’ and ‘surfacing’ groups are only
used in MCI. However, it was found that most of the RCA’s are not recording the maintenance
cost properly. The following are some of the exceptions found while skimming the RAMM
Maintenance Cost Data tables of some of the RCA’s:
• maintenance cost to be charged against the ‘pavement’ and ‘surfacing’ group is claimed
against the environment or other groups;
• some of the actual expenditures are not entered into the RAMM database. This will cause an
underestimation of the maintenance cost expenditure actually occurring under the surfacing
and the pavement group;
• There was a large jump in the annual amount of ‘digout’ where there was a reseal. This
suggests that some ‘routine’ maintenance is accumulated until the time of reseal. Whether
there was a real need for this in the preceding years is difficult to know. Besides, it could also
be related to:
The strategy followed by the RCA;
Being linked with waiting for a rehabilitation/resealing programme
The execution of an edge repair programme for a given year etc.
The rate of increment of MCI using the current expression is very slow. And, hence, the proposed
trigger limits will never be achieved with this rate. The expression was derived based on the State
Highway data of the Napier region. To make the expression more generic additional research
was carried out by OPUS using the much larger database. The latest set of graphs from OPUS
does not inspire much confidence in the numbers – some go up with age, some go down. It is
hard to believe that the maintenance needs can vary so much within NZ. Hence, the conclusion
that the existing RAMM maintenance data seems to be unreliable for developing a reasonable
model for MCI.

The further research studies should include:
• Defining the works operations that are related to pavement condition digouts, rut filing,
surface patching etc.;
• Defining the modes of distress that trigger these operations – shoving, cracking, scabbing
etc.;
© 2001 Ingenium 69
• Maintaining records on selected roads for both works operations and distress triggers. The
road should have FWD data so that the properties of the pavement layers can be estimated.
• Creating a model form for distress modes using the records of annual amounts of distress on
the study roads;
• Calibrating the model coefficients using the records of annual amounts of distress on the
study roads; and,
• Combining the new models with other existing models and carrying out a series of tests to
see how much routine works operations mitigate general pavement deterioration defined by
IRI at different traffic volumes.
If reliable models can be developed for other distress modes it must be decided whether to use
these to trigger various routine works or whether to exogenously use them to make a refined MCI.
There is also a need for the development of guidelines on what type of data are to be collected,
how it should be processed and in what format it should be stored in.
© 2001 Ingenium 70
13. INTERVENTION CRITERIA FOR SEALED ROADS
13.1 Introduction
It is essential that various types of road maintenance activities (treatments) be well defined in
terms of the pavement type to which they can be applied, under what circumstances they should
be applied. For each treatment, user-specified intervention criteria are used to determine the
timing and the limits of the work to be carried out.
The intervention criteria may differ from one RCA to another based on the local conditions and
availability of funding. The level of service, the long-term performance standard and consequently
the funding requirement will depend on the intervention criteria used. Hence, it is essential that
intervention criteria are correctly chosen and well calibrated.
This chapter outlines the intervention criteria used in the NZ dTIMS Setup together with the
suggestion for the further improvements.
13.2 Terminology
The following terminology is used in this project:
• Maintenance Categories classify the treatments to four categories based on the level of
intervention (see Table 13.1):
Routine/General maintenance;
Resurfacing;
Smoothing; and,
Strengthening.
• Treatments are actions an agency takes on a given section to either reduce the pavement
deterioration rate (prevention) or to repair the effects of deterioration (reaction). A treatment
can be considered as a response to the certain condition. When a treatment is applied, it
usually affects the condition and performance of the pavement section. Different treatments
affect the condition and performance in different ways;
• A Strategy is a course of actions taken over the analysis period in order to keep the road
section in good condition or within pre-set performance limits. A strategy consists of one or
more treatments applied on the section at a specific point in time during the analysis period.
One strategy may differ from another simply because the same treatment is applied in
different years. For example, one strategy may apply overlay in 2002, reseals in 2007 and
2011, and another overlay in 2017 (Figure 13.1). Another strategy for the same section may
overlay in 2001, reseals in 2008 and 2012, and overlay again in 2018;
• Triggers or Intervention Criteria are the conditions which initiate a maintenance treatment.
The criteria may be scheduled at regular time intervals (e.g. reseal every 7 years) or
condition responsive (e.g. reseal at 5% cracking);
• Constraints are used to exclude certain treatments from being applied under certain
conditions. For example, reseals would not be applied to motorways;
• Filters are the combination of the triggers and constraints so that dTIMS selects a given type
of treatment for a specific section; and
© 2001 Ingenium 71
• Resets are the condition of the pavement after a treatment. For example, if the pavement is
resealed the cracking would be reset to 0 but the roughness may be unchanged.
Table 13.1: Treatment Types in Maintenance Categories
Maintenance Category Treatment types

Routine/General Maintenance Routine
Resurfacing Seal – Large chip (RSL)
Seal – Small chip (RSS)
Smoothing Thin granular overlay (MOL)
Thin asphalt overlay (MOA)
Rip and remake ST (MRL)
Rip and remake AM (MRA)
Mill and replace (MMA)
Recycle (MYL)
Strengthening AM reconstruction (SRA)
ST reconstruction (SRL)
Thick granular overlay (SOL)
Thick asphalt overlay (SOA)
First Treatment Subsequent Treatments
FMOL SRSS SRSL FMOL
Strategy FMOL2000
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
FMOL SRSS SRSL FMOL
Strategy FMOL2001
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Note:
FMOL - Thin Granular Overlay
SRSS - Reseal with Small Chips
SRSL - Reseal with Large Chips
Figure 13.1: Strategy and Treatments for Economic Analysis
© 2001 Ingenium 72
13.2.1 Existing Intervention Criteria in NZ dTIMS Setup

During the analysis, dTIMS uses the triggers to determine if these limits have been exceeded to
trigger a treatment in a given section. The following condition parameters are used by NZ dTIMS
Setup as triggers:
• Cracking;
• Rutting;
• Potholes;
• Roughness;
• Flushing;
• Shoving;
• Texture depth; and,
• Skid resistance.
Besides, maintenance cost index (MCI) and the granular overlay requirements (GOVL) which
also represent the performance of the pavement are used as a trigger.
Different maintenance categories uses different sets of triggers. Table 13.2 gives the trigger sets
and trigger limits for the strengthening. Similarly Table 13.3 and Table 13.4 shows the trigger sets
and limits for smoothing and strengthening requirements.
Table 13.2: Strengthening Requirement (SREQ) Trigger Limits
Traffic Range IRI GOVL MCI

(AADT) (m/km) (mm) ($)
>20,000 + Motorway 4.2 100 6,000
10,001 – 20,000 4.6 100 7,000
4,001 – 10,000 5.0 100 8,000
2,001 – 4,000 5.5 100 10,000
501 – 2,000 6.1 100 12,000
101 – 500 6.9 100 15,000
<= 100 7.6 100 18,000
Note: IRI Strengthening Roughness limits
GOVL Granular overlay thickness
MCI Maintenance cost index
Table 13.3: Smoothing Requirement (MREQ) Trigger Limits
Traffic Range IRI RDM ASH ACA

(AADT) (m/km) (mm) (%) (%)
>20,000 + Motorway 3.5 0.8 5.0 12.0
10,001 – 20,000 3.8 10.0 5.0 15.0
4,001 – 10,000 4.2 10.0 7.0 15.0
2,001 – 4,000 4.6 15.0 10.0 20.0
501 – 2,000 5.0 15.0 15.0 30.0
101 – 500 5.5 20.0 20.0 35.0
<= 100 6.1 30.0 30.0 40.0
Note: IRI Smoothing roughness limits
RDM Mean rut depth limits
ASH Shoving limits
ACA All cracking limits
© 2001 Ingenium 73
Table 13.4: Resurfacing Requirement (RREQ) Trigger Limits
Traffic Range IRI SII TD SFC

(AADT) (m/km) (mm)
>20,000 + Motorway 3.1 8.0 0.75 0.52
10,001 – 20,000 3.4 10.0 0.75 0.52
4,001 – 10,000 3.7 11.0 0.75 0.52
2,001 – 4,000 4.1 13.0 0.75 0.52
501 – 2,000 4.5 15.0 0.75 0.52
101 – 500 5.1 18.0 0.75 0.52
<= 100 5.7 20.0 0.75 0.52
Note: IRI Resurfacing roughness limits
SII Surface integrity index
TD Texture depth limits
SFC Surface friction limits
In some cases a single trigger can warrant a treatment where as in other cases a combination of
two or more treatments should be applied. The flowchart (see Figure 1.2) illustrates how different
triggers are used in the treatment selection procedure.
dTIMS also requires “Constraints” to trigger a treatment. A constraint restricts the treatment to
given criteria, so that treatment is not applied in an illogical manner. The following sealed
pavement attributes are used as constraints:
• Traffic Volume;
• Road Hierarchy;
• Surface thickness;
• Possible to raise pavement level;
• Existing Chip Size;
• Second coat seal required;
• Section is a bridge; and,
• Surface type.
13.2.2 Treatment Selection Procedure in dTIMS

dTIMS requires that treatments should be divided into:
• First treatment (called rehabilitation in Deighton’s terminology); and,

• Subsequent treatment (called periodic in Deighton’s terminology).
Only the first treatment can activate a strategy. In the performance standard analyses a dummy
treatment called ‘PANA’ is applied in the year 1 so that it can activate strategies. In the economic
analyses the first 3 treatments – one treatment from each of the strengthening, smoothing and
resurfacing maintenance categories are considered for a road section every year for the first 5
years for the economic comparison. The treatment for each maintenance category is selected
using only constraints explained earlier. Neither the existing road condition nor the long-term
standard is considered while selecting the first treatment.
Subsequent treatments are applied after the first treatment. Selection procedure for subsequent
treatment is based on the decision tree consisting of both constraints and trigger limits (Figure
1.2). Once the condition of the roads exceeds certain trigger limits, one of the subsequent
treatments is considered. The selection of a particular treatment will be based on the constraints
specified for the given treatment. Hence, the triggering of the subsequent treatments are totally
based on performance standards. No economical consideration is made while triggering these
treatments.
© 2001 Ingenium 74
13.3 Issues on Intervention Criteria

A pragmatic approach was followed while defining the constraints and triggers during Phase I of
the NZ dTIMS project. The constraints used in the treatment selection procedure have been
defined based on the most used construction and maintenance practice. Similarly the default
trigger limits for the triggers were proposed based on the available information and results of the
testing using a test database. Limited testing of these intervention criteria were carried out during
the pilot studies. The feedback of the study was incorporated in the preliminary NZ dTIMS Setup.
However, full testing of all the triggers and constraints with real data and the field verifications has
yet to be carried out. Full implementation of the NZ dTIMS system by selected RCA, in a planned
way, under proper guidance HTC could help to get much of the data required for such study.
The trigger limits could depend on:
• Traffic volume;
• Road functional class;
• Urban or rural area;
• Local perception of level of service required;
• Availability of funding;
• Existing average network condition; and,
• Policy of the government and RCA.
Therefore trigger limits could vary from one RCA to another. Hence, comprehensive guidelines
are required on the calibration of these triggers for local condition. Although some flexibility is
preferable, but for the consistency level of service and for the funding decision there must be
some kind of national standard for maintenance intervention criteria. Such standards should
cover all possible conditions in NZ. The users should not be allowed to change the standard
unless there is good reason for it.
Development of such a standard is a long-term process. One of the first steps would be sensitivity
analysis of various triggers. In most of the cases, different triggers can be used for triggering the
same treatment. In such a situation the trigger limits for corresponding condition parameters
should be defined in such a way that they have similar economic benefits. The observation further
shows that some of the condition parameters were more frequently used than others during the
treatment selection procedure. So a priority list has to be developed to suggest which of the
trigger limits should be calibrated first.

Further research should be focused in the following directions:
• Sensitivity analysis of the triggers used in the NZ dTIMS Setup;

• Calibration of trigger limits for roads of different functional classes and traffic ranges; and,
• Develop national performance standard for road networks.
© 2001 Ingenium 75
14. TRAFFIC VOLUMES
14.1 Introduction
This research task is required to empirically confirm a theoretical method for providing low-cost
traffic count data on an entire network of roads and to establish simple guidelines for the
application of the method in the field.
Traffic count data have always been recognised as important in the design and planning process.
Bennett and Paterson (1999) show that the AADT is much more significant than any other data
item in a life-cycle cost analysis of a road. AFRICON (1999) in analysing the key data items for
the Gautang road network found that the AADT was the only input variable with a high sensitivity
on the optimisation output. The variables which are often considered as being important, such as
pavement strength and roughness, were found to have a low to moderate impact.
In spite of its importance in asset management, traffic data is often lacking. This arises because it
is based on short-term or permanent traffic count stations which, by their nature, are expensive to
establish/operate and only have a limited coverage. For many local authorities roads are
monitored infrequently at best, and some not at all.
To circumvent this problem one needs to have a simple method which could be used to collect
data rapidly and inexpensively. Such a method does exist, and it is this method that needs to be
validated for NZ. The method is termed a ‘Moving Traffic Count’ and is based on recording traffic
from a moving vehicle driving along the road.
14.2 The Need for Inexpensive Traffic Data Collection

Road Controlling Authority managers are very familiar with the need for good traffic data required
for planning, and the areas where existing practices fall short. Predictive pavement deterioration
modelling, as in the NZ dTIMS system is highly sensitive to having good traffic data. It is
therefore very important that inexpensive methods of collecting traffic data are researched and
adopted to allow all territorial authorities in NZ (both small and large) the opportunity of gaining
inexpensive but relatively good quality traffic data. The RIMS Group have also indicated the key
importance of improving traffic count data to the successful implementation of predictive
modelling of pavement deterioration using dTIMS.
The reasons why the ‘Moving Traffic Count’ data collection method would be ideally suited for
these requirements can be summarised under the following terms:
• Coverage; and
• Ease of collection.
It is simply impossible to collect traffic count data on all links in a road network using traffic
counters. A typical RAMM database contains approximately 2000 links. To collect data on all links
would be prohibitively expensive. So for this reason what is usual practice is that traffic counts are
estimated with very little empirical basis.
However, since most road controlling authorities conduct regular roughness surveys, a moving
traffic count could be conducted for very little additional marginal cost. Thus, instead of having no
data they would at least have an estimate which is based on something besides what is too often
guesswork. Were they to also do moving counts as part of their regular monitoring of the road
network this would serve to further improve the estimate since the sample sizes would be larger.
© 2001 Ingenium 76
However, it should also be noted that Moving Traffic Counts will not work under all situations.
There are always situations where the traffic varies so much over the course of the day or
between days and over weeks and in these situations it is necessary to resort to more traditional
traffic counting techniques. However, overseas experience suggests that these are the
exceptions rather than the rule and so the method should be applicable over large portions of the
road network.
14.3 Principles of Moving Traffic Counts
To calculate the ADT from a moving vehicle, consider the following proof where:
• X is the oncoming flow rate in veh/h
• So is the average oncoming vehicle speed in km/h
• Sr is the speed of the survey vehicle in km/h
• L is the distance travelled by the survey vehicle in km
• C is the number of vehicles counted travelling in the opposite direction in veh
The average headway (km/veh) between vehicles is So/X. The length of the observed traffic
stream (in km) is C So/X.
The observed duration of the survey (in h) is given by L/Sr. Hence the length (km) of the traffic
stream that has been observed is (Sr+So) L/Sr.
Equating these two expressions:
X = C So/[(Sr + So) L/Sr]
C 1
X=
L 1 1
+
So Sr
If the survey vehicle travels at the same speed as the oncoming traffic, and it is assumed that
there are no speed differences between classes, the above expression reduces to:
C Sr
X=
L 2
Since the duration of the survey is given by t = L/Sr, this can be expressed as:
C
X=
2t
We assume that the ADT is twice the flow in the opposing direction so the ADT is given as:
C
ADT =
t
Or, for different survey vehicle speeds to oncoming traffic:
C 2
ADT =
L 1 1
+
So Sr
The accuracy of the ADT count can be improved by classifying the vehicles and using the above
expression to calculate the ADT by class and added to get the total ADT. It is not necessary to
have the precise speeds of these other vehicle classes in the calculations. The speeds for the
buses and trucks can be expressed relative to car speeds, for example as travelling 10 and 20
km/h slower respectively:
Cpc 2
ADTpc =
L 1 1
+
So Sr
© 2001 Ingenium 77
Chb 2
ADThb =
L 1 1
+
So − 10 Sr
Ctk 2
ADTtk =
L 1 1
+
So − 20 Sr
ADT = ADTpc + ADThb + ADTtk
Figure 14.1below demonstrates how reliable this method can be in calculating the derived AADT
from moving survey data from India, corrected for time of day and other factors, compared to the
AADT from 7-day counts for the same road sections (Bennett and Paterson, 1999). It can be
observed that the moving survey method gave an excellent correlation with the 7-day counts.
8,000
7,000 Line of equality

y = 1.02x
R2 = 0.97
AADT Derived from Moving Observer Count
6,000
5,000
4,000
3,000
2,000
1,000
0
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
AADT: 7 - day Count
Figure 14.1:Example of Moving vs 7 Day Count from India
In order to apply the moving traffic count method it is necessary that it first be validated and then
it requires the establishment of the principles by which the moving survey data can be converted
to ADT and AADT. This can be undertaken by doing surveys over road sections where normal
traffic count data are also being collected. Correction factors are established by time of day and
day of week to convert the moving count data to ADT and AADT.
14.4 Objectives of the Study
The technical objectives of the study are:

1. To conduct a series of moving traffic counts on a range of roads which are also being
monitored with traffic counters—either temporary or permanent counters.
2. To develop functions which convert the moving count data to ADT and AADT.
3. To establish simple yet practical guidelines for applying this method.
The output from the project will be the the ability for Road Controlling Authorities to inexpensively
and quickly obtain improved AADT data. Without adequate traffic data it is impossible to ensure
the correct allocation of funds. The project will also address the following specific objectives:
© 2001 Ingenium 78
Life Cycle Costs

It is impossible to accurately estimate the life-cycle costs without having an adequate estimate of
the AADT. The implementation of pavement deterioration modelling in NZ through dTIMS
increases the importance of having adequate AADT data.
Improving Data-Capture Technology and Analysis/Utilisation

This research task will help determine a simple, non-technology based method for acquiring
AADT data along with the method of how the data can be applied.
14.5 Benefits of the Research

The specific benefits which will come from the outcome of the research consist of a simple, cost
effective method to obtain AADT data for roads where data would not be readily available. If
combined with annual roughness surveys, it is possible to have an AADT ‘snap shot’ of all roads
in a RCA’s network so any investment decisions will have a good starting point for traffic data.
With regard to the magnitude of the benefits, AFRICON (1999) found that a 10% error in the
AADT estimate resulted in >5% change in the NPV of the life-cycle cost of a pavement.
It is presently not possible to estimate the accuracy of the existing AADT data, or the implications
of not having adequate data for the entire network available when RCA’s are looking at their
investment programme. However, this low cost method, once it has been proven for NZ
conditions, when compared to the potential large mis-estimates from using inappropriate AADT in
life-cycle costing analyses, ensures that the project would have a very high BCR.

Further research should be focused in the following directions:
• Proving the Moving Traffic Count survey as an appropriate method for gaining good traffic
data for varying conditions for RCA networks;
Furthermore, but not as high priority, there is a need for better traffic data in the following areas:
• Regional traffic growth rates for varying classification of roads;
• Areas of growth in population, land use and industry (ie logging) that will influence the change
of mix and /or change of growth rates in traffic;
• Arithmetic growth rates vs geometric and / or other growth rates;
• Variation about the mean from existing Weigh In Motion (WIM) sites and establishing portable
WIM surveys for areas that do not currently have WIM stations.
© 2001 Ingenium 79
ANNEX A : Priority List of Research Projects Related to the NZ dTIMS Setup
NR Topic Item Description Actions Needed Start Priority Research

Immediate After 1-2 period
years (years) 1
Calibration Studies
1 Calibration Studies Establish the Establish the sensitivity of calibration • Perform sensitivity analysis based • <½
all network types sensitivity of coefficients on the outcome of model on the NZ dTIMS Setup
calibration predictions and/or financial outcome • The study should be structured to • <½
coefficients provide the information to
planning and structuring of future
calibration work performed in NZ
2 Calibration Studies Level 1: Basic This level of calibration should be • Perform checks on the • v
(Sealed Pavements) Application performed by all the RCA’s. The aim of reasonableness of prediction • v
these studies is to verify the data models (This was done on
Note: Transfund appropriateness and models to local national level for Phase I of the
should not conditions RIMS’ dTIMS project but need to • v
necessarily fund all be repeated for local calibration
three levels. The on RCA level)
levels are however
inter related and
could overlap with
each other
Level 2: Calibration This level of calibration should be • This level could include both desk • <½
performed by most RCA’s. The aim of studies and collection of additional
these studies is to validate the data.
calibration coefficients appropriateness • Establishment of monitoring • <½
and models to local conditions “benchmarking” (BM) sites to
monitor future deterioration and /
or works effects of region. The
bench marking sites could turn
into LTPP sites at a later stage if
the data collection was
appropriate for the next level
calibration. • <1
• Monitor BM sites • <1
• Perform calibration based on desk
study data and or additional data
from BM sites
1
V - Indicates that the research period can vary depending on the level to which the research is conducted. The time frame indicated can also vary depending on the actions done
simultaneously.
© 2001 Ingenium 80

years (years) 1
Calibration Studies Level 3: Adaptation Calibration of the RDM and RUE. This • Develop calibration policy • <1
(Sealed Pavements) level of calibration will require the framework for NZ
…..Continued collection of additional data and the • Establish calibration sections • <1
establishment of calibration sections. • Monitor sections • 3-5
(LTPP) Although some results from this • Calibrate models (specific models • 3-5
study could be obtained from the first discussed elsewhere)
year, it should be structured to provide
data for long term studies (i.e. 3 - 5
plus years) The quantity of data on this
level should be sufficient to derive
uniquely defined models.
3 HDM-4 Road RDWE models Study on HDM-4 RDWE models for • Study on the appropriateness of • <1
Deterioration & appropriateness and possible inclusion in NZ dTIMS setup HDM-4 RDWE models
Works Effects Model calibration to NZ • Calibration of HDM-4 models • >3
(RDWE) conditions (level 2) for various regions in NZ
• Replace HDM-III PDME models • 1
with HDM-4 RDWE models in the
NZ dTIMS setup
• Use LTPP sites for calibrating and • v
modification of model
4 Calibration Studies Level 3: Adaptation A comprehensive study is needed to • Develop calibration policy • ½
(Unsealed of unsealed roads calibrate unsealed road deterioration framework for NZ
Pavements) deterioration models models. The road roughness models • Establish monitoring sections • ½
Note: Although a (Limited data and were identified to be the most critical according to the design matrix for
time frame is models exist for NZ for dTIMS analysis an unsealed roads study
assigned to the that could be • Perform an accelerated • 1
individual task this calibrated. monitoring programme
project should be set Therefore level 1 • Derive an unsealed road • ½
up to be performed and 2 calibration roughness and re-gravel model
within 1,5 years. studies will only be • Establish dust prediction models • ½
possible at later • Establish skid resistance models • v
stage)
© 2001 Ingenium 81
Work Effects (WE) Models

years (years) 1
5 Work Effects models Calibration of WE Calibrate the WE models in conjunction (SEE NR 1&2)
models with the PD models mentioned in NR 1
-3
Calibration of WE Calibrate the WE models of the existing • Calibrate resets for roughness • <1
models (in addition system (Limited study done in Phase II)
to normal calibration • Calibrate surface resets for • <1
studies) routine maintenance actions
• Investigate possible inclusions of
HDM-4 WE models • v
Work Effects models Effect of drainage Calibrate the effect of drainage and • Assess the benefits of including • <1
…..Continued moisture to pavement deterioration drainage effects to pavement
deterioration modelling
• Investigate the applicability of • <1
work done by HDM-4
• Model and/or model effects of • v
drainage
• Investigate issues for • <1
incorporation of models in dTIMS
Evaluate the effect of CQ has a significant effect on • Assess the sensitivity of this • ½
Construction Quality pavement deterioration. It is however parameter on PD modelling
(CQ) on pavement difficult to quantify and to incorporate in • Investigate the principles used in • ½
deterioration a PMS with deterioration capabilities. HDM-4
• Evaluate methods to incorporate • <1
into PDM
Road User Cost Models (RUC)

6 Road User Costs HDM-4 RUC Models Study the appropriateness of HDM-4 • Study the appropriateness of • v
Models RUC models to NZ conditions HDM-4 RUC models to NZ
conditions
1
V - Indicates that the research period can vary depending on the level to which the research is conducted. The time frame indicated can also vary depending on the actions done
simultaneously.
© 2001 Ingenium 82
Investigate other Many treatments have additional • Determine the influence of • v

road user costs benefits such as safety aspects that pavement defects such as
are not currently considered in the roughness, potholes and rutting
analysis. on the risk of crashes
• Determine the inclusion of time • v
savings of certain treatments
• Quantify users preference on
treatments such as surfacing of • v
unsealed roads
Specific Models
7 Pavement Strength Validate Calculation Various methods are available for the • Investigate the effect of subgrade • <1
Methods calculation of pavement strength. All strength to the structural number
these methods should be validated for of thin flexible pavements for the
NZ conditions NZ conditions
• Verify the results of the • <1
expressions developed for
estimating the SNC using non
destructive methods with the
traditional destructive methods
• Refine the New Zealand methods • <1
to include the influence of volcanic
soils
• Check the effectiveness of SNC • <1
as a strength parameter for
pavement deterioration modelling
in the different conditions in NZ
Effect of seasonal Investigate the effect of seasonal • Investigate the effect of seasonal • <1
variation and variation on SN variation on SN
moisture
8 Trench Patching & Assessment Development of an assessment The following issues should be
Service Covers (TP requirements procedure for TP & SC. This will also resolved with regard to the
& SC) include the definition of these items in assessment:
the RAMM database • Quantifying the occurrence (i.e. • <1
number of TP)
• Quantify the condition of TP • <1
• Orientation, location and • <1
dimensions of TP & SC
Model influence of Investigating of modelling aspects such • Perform measurements of TP & • <1
TP & SC as the TP & SC on other defects such SC together with HSD surveys
as roughness • Derive appropriate models to • v
predict the effect of TP & SC on
condition parameter such as
roughness
© 2001 Ingenium 83
Define treatments Establish the incorporation of these • Define treatments and triggers • <1
and intervention items into the dTIMS analyses process based on practises used
criteria that includes items such as defining • Establish the effects of treatments • v
triggers and resets for treatments on pavement condition
related to the improvement of TP & SC
Define benefits of Define the benefits from improving the • Establish the effects of trench • v
new treatment types bad trench patches and services improvements on savings of RUC
covers and maintenance cost
9 Texture depth and Calibrate existing The current texture model has to be Calibration has to be performed as a
skid resistance texture models calibrated on a wide range of NZ function of the following:
conditions • Environmental conditions • v
• Construction quality • v
Calibrate existing The skid resistance models were Calibration has to be performed as a
skid resistance developed on limited data with regard function of the following:
models to surface types. It should therefore be • Chip size • v
calibrated including the common • Geometry • v
surface types used in NZ
Vehicle equivalency The light vehicle equivalency factor • Establish light vehicle equivalency • v
factors should be established for the heavy factors
vehicle classes operating in NZ
Further skid/texture Further skid resistance and texture • Derive crash risk and surface • v
research depth research is needed for better characteristic relationship
understanding and application of • Establish skid resistance/ • v
models and construction practices roughness relationships
• Predicting water pond depth/ • v
aquaplaning potential based on
predicted rut depth
10 Noise Noise Treatments Include noise reduction treatments to • Define treatments to include • ½
and Triggers the dTIMS Setup current setup
• Establish the triggers •
• Define the effects of noise • ½
reduction maintenacnce actions v
Noise Models Include noise model (if required) • Select the appropriate model • ½
taking account of the cost
implications such as collection of
additional data for these models
11 Maintenance cost Maintenance Costs Maintenance Costs Index Models were • Calibrate / improve models to • v
index (MCI) Index Model developed on a limited database and include other regions within NZ
should be calibrated on a wider scale • Develop various routine • <1
maintenance standards for MCI
© 2001 Ingenium 84
Trigger values used Calibration of triggers used to trigger • Sensitivity analysis of the triggers • ½
with MCI model maintenance work based on MCI used in the NZ dTIMS Setup
• Calibration of trigger limits for • ½
roads of different functional
classes and traffic ranges
• Develop national performance • <1
standard for road networks.
Analysis and Policy

12 Applications of dTIMS as funding How will Transfund evaluate / validate • Define the criteria for validation • <1
dTIMS tool funding applications submitted by RCA • Establish requirements to be set • <1
by Transfund
• Sensitivity analysis of various • <1
parameters used in analysis
Evaluating Demonstrate the use of dTIMS for • Show viability of present • <1
Maintenance evaluating pavement maintenance maintenance strategies
strategies strategies • Investigate the sensitivity of • ½
network condition to maintenance
expenditure
Evaluate national The current funding regime needs to be • Set criteria for evaluation • ½
funding regime evaluated for all the RCAs in NZ • Compile database with sampling • ½
method
• Develop setup • ½
• Conduct analysis • ½
© 2001 Ingenium 85
REFERENCES
AFRICON Consultants Ltd. (1999). Sensitivity Analysis on the Input Variables of the HDM Pavement
Prediction Models. Internal Report, Pretoria.
Barnes, J. Ensor, M. (1994). Traffic Noise from Uninterrupted Traffic Flows. Transit New Zealand
Research Report No. 28. Beca Carter Hollings & Ferner Ltd, Hegley Acoustic
Consultants Ltd. Transit New Zealand, Wellington.
Bennett, C.R. and Paterson, W.D.O. (1999). Guidelines on Data Collection and Calibration of the HDM
Model. Report to the International Study of Highway Development and Management Tools,
University of Birmingham, UK.
Cenek, P.D. (1996). Review of Instruments for Measuring the Texture Profile of Road Surfaces.
Transit New Zealand Research Report No 63, Transit New Zealand, Wellington.
Cenek, P.D. (1998). Presentation to Seminar on Multi-Year Programming, Wellington. University

of Auckland, Department of Continuing Education.
Cenek, P.D. (1999a) Email to M Riley: Texture Depth Model
Cenek, P.D. (1999b) Email to T Henning: SCRIM.
Department of Transport (DOT). (1988). Calculation of Traffic Noise. Her Majesty’s Stationary
Office, London.
Deighton Associates Limited, (1998), dTIMS 6.1 Reference Manual, Bowmanville, Ontario,
Canada.
Greg M and et al (1999). HDM-4 Series, Volume 6, Modelling Road Deterioration and Works
Effects.
Hallett, J.D. (1998). Presentation to University of Auckland Seminar on Multi-Year Programming,

Wellington. University of Auckland, Department of Continuing Education.
HTC, (1999a). Implementation of Predictive modelling for Road Management. Phase I:

Preliminary dTIMS System. Final Report DT/99/F2. HTC Infrastructure Management Ltd.,
Auckland.
HTC (1999b). Pavement Strength Software Manual. Report DT/99/12. HTC Infrastructure
Management Ltd., Auckland.
HTC, (1999c). dTIMS Data Issues. Report DT/99/16. HTC Infrastructure Management Ltd.,
Auckland.
Hunt, M. (1988). A Synthesis of Survey on Reaction to Road Traffic Noise in New Zealand. Road
Research Unit. National Roads Board. Wellington.
th
Jones, D. (1998). Dust Prediction. 19 ARRB Conference. Sydney, Australia.
Lund, B. (1997). Friction Test: Comparative Testing with 3 Different Equipment Carried Out
During Summer 1996. Danish Road Institute. Report 82. Danish Road Directorate,
Roskilde.
© 2001 Ingenium 86
Marshall, (1999). Email to Mark Gordon on 4 November 1999. Southland District Council,
Invercagill.
NDLI (1995). Modelling Road Deterioration and Maintenance Effectsin HDM-4. Final Report
Asian Development Bank Project RETA 5549. N.D. Lea International, Vancouver.
Rambisheswar, V. (1999). Email to T Henning on Noise Treatments.
Riley, M.J. (1998). Calibration of HDM-III for New Zealand. Report to Opus Central Laboratories,
Riley Partnership, U.K.
Riley, M J (1999). Notes on inclusion of the effect of traffic on the progression of ravelling.
Communication to the ISOHDM, University of Birmingham.
Riley, M J (2000a). Modelling road deterioration and work effects for bituminous pavements –
Discussion document, University of Birmingham.
Riley, M J (2000b). Review of the Draft dTIMS Research Needs Report - Discussion document,
Riley Partnership, U.K.
Paige-Green, P. (1989). The influence of Geotechnical Properties on the Performance of Gravel

Wearing Course Materials. Unpublished PhD Thesis, University of Pretoria.
Paterson, W.D.O. (1985). Prediction of Road Deterioration and Maintenance Effects: Theory and
Quantification. The Highway Design and Maintenance Standards Study. Vol 3.
Transportation Department, World Bank, Washington, DC.
Paterson, W.D.O. (1987). Road Deterioration and Maintenance Effects: Models for Planning and
Management. World Bank Publications, Washington, D.C.
Roberts, J D (1999). Report on PIARC-ISOHDM Review Workshop on HDM-4 RDWE Models,

Austroroads.
Salt, G.F. (1977). Research on Skid Resistance at the Transport and Road Research Laboratory
(1927 to 1977). Transport and Road Research Record 622. Transportation Research
Board, Washington, D.C.
Salt (1999). Email to Theuns Henning on 15 September 1999. Tonkin & Taylor Ltd., Dunedin.
Transfund New Zealand. (1997). Project Evaluation Manual. First Revision. Transfund New
Zealand, Wellington.
Transfund (1999a). Programme and Funding Manual. Transfund NZ Publication, Wellington
Visser, A.T. (1981). An Evaluation of Unpaved Road Performance and Maintenance. Unpublished
PhD Thesis, University of Texas at Austin.
© 2001 Ingenium 87

Predictive Modelling Foropave - NZ RR PDF

Uploaded by

Copyright:

Available Formats

Predictive Modelling Foropave - NZ RR PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Predictive Modelling Foropave - NZ RR PDF

Uploaded by

Copyright:

Available Formats

Road

dTIMS Research Needs

HTC Infrastructure Management Ltd.

This report is Copyright  2001 Ingenium. All rights reserved.

• Additional treatments to be included;

© 2001 Ingenium iii

1. RESEARCH REQUIREMENTS FOR dTIMS IN NEW ZEALAND

2. CALIBRATION STUDIES .................................................... 18

4.4 Road Works Effects .......................................................................... 32

10.4 Intervention Criteria for Noise Reduction Treatments ........................ 59

© 2001 Ingenium vii

© 2001 Ingenium viii

1. RESEARCH REQUIREMENTS FOR dTIMS IN NEW ZEALAND

• Setup design for dTIMS;

1.2 Identified Needs for Further Research

• Introducing the overall topic;

Train the User

Does Setup Yes

Figure 1.1: Development and Refinement of the System

1.3 The NZ dTIMS Setup

• Data items and collection;

1.3.1 Data Items and Collection Methods

1.3.2 Pavement Deterioration Models

• Cracking initiation and progression;

1.3.3 Treatment Types

Table 1.1: Treatment Types Used in dTIMS

1.3.4 Triggers for Treatment Types

The condition indicators used for defining triggers levels include:

• Traffic range (default or user defined, allowed for up to 7 ranges);

SPEC_TRT Use Specified

Gravel? Yes AADT > limit No HS < limit

MCI > limit

IRI > limit

IRI > IRIresurf

AC surfacing AC surfacing Raising AC surfacing Raising

Yes Yes Yes Yes

No Raising HS < 100 Raising

Rip & Remake Rip & Remake Reconstruct Reconstruct

Figure 1.2: Flowchart of Treatment Selection Procedure

1.3.5 Effects of Treatments on Road Condition

1.3.6 Road User Costs Models

RUE = UF [a5 NAASRA - a6] NAASRA > R2 (1.3)

where RUE is the PEM vehicle operating cost in cents/km

Table 1.2: VOC Model Coefficients

1.3.7 Analysis Approach

• Area under condition curve;

Figure 1.3: Example of dTIMS Efficiency Frontier

Implementation of any long-term maintenance planning systems usually has a development

Development of the Condition and inventory Refinement of models Control mechanisms,

True Modelling Capability

Figure 2.1: Streams Contributing to Modelling Capability

2.2 Level of Calibration

General Planning Project Appraisal Research and

Coarse Estimates Reliable Estimates

Figure 2.2: Resources and Time Required for Calibration

2.3 Level 1: Basic Application

'The Depth of the Sea and

Figure 2.3: Data Vs Calibration

• Existing data in RAMM;