Você está na página 1de 71

Contract Number: SSPI-CT-2003-503549-IMAGINE

DeltaRail
PO Box 8125
3503 RC Utrecht
The Netherlands
Telephone: +31 30 300 5100
telefax: + 31 30 3005 5150
email: margreet.beuving@deltarail.nl

IMAGINE
Improved Methods for the Assessment of the
Generic Impact of Noise in the Environment
Final Synthesis Report
Guidance on the IMAGINE methods
Project Co-ordinator: DeltaRail NL
Partners DeltaRail NL Boeing EDF LABEIN SP TUG
DeltaRail UK BUTE EMPA Leicester TML UGent
Anotec CSTB EUROCONTROL M+P TNO ULeeds
ARPAT DeBAKOM JRC MBBM TRL Volvo
Autostrade DGMR Kilde
Document identity: IMA10TR-06116-AEATNL10
Date: 2006-11-16
Level of confidentiality: public
Written by Date (YY-MM-DD) Reviewed by Date (YY-MM-DD)
Margreet Beuving, Brian
Hemsworth
2006-11-16 Steering Committee 20 03-2007

The present publi cati on only reflects the author s views. The Community is not liable for any use t hat may be made of t he information contained herein.

w
w
w
.
i
m
o
g
i
n
e
-
p
r
o
j
e
c
f
.
o
r
g
Page 2

EXECUTIVE SUMMARY
This report gives an overview of the methods, guidelines and databases produced in the IMAGINE project.
The document is aimed to guide through the deliverables from the end users point of view and it gives
information on the IMAGINE deliverables at different levels, from management information to technical
guidance.
The groups of end users defined in the project are authorities, software developers, technicians, noise
modellers and operators. The report consists of 2 parts, each focusing on different end users.
Part I of this document gives a management overview with definitions, the background of the project and
links with the HARMONOISE project, advantages of the methods and subjects for further developments.
Part II describes the technical part of the projects, it gives flow diagrams and overviews of the contents of
the deliverables, both for the source methods (road, rail, industry and aircraft) and for the propagation and
the determination of the L
den
by measurement and calculation.




Page 3

TABLE OF CONTENTS
Executive Summary ..................................................................................................................................................................................................... 2
Introduction.................................................................................................................................................................................................................. 5
PART I Management Overview.................................................................................................................................................................................. 7
I.1 List of IMAGINE deliverables .................................................................................................................................................................... 7
I.2 Links between HARMONOISE and IMAGINE.......................................................................................................................................... 10
I.3 Advantages of the IMAGINE methods ...................................................................................................................................................... 11
I.4 Subjects for further development .............................................................................................................................................................. 13
PART II Technical description of the methods in global terms.............................................................................................................................. 16
II.1 Flow diagram........................................................................................................................................................................................... 16
II.2 Methodology............................................................................................................................................................................................. 17
II.2.1 Road noise source................................................................................................................................................................................ 18
II.2.1.1 Introduction .................................................................................................................................................................................... 18
II.2.1.2 How to determine the sound power level ........................................................................................................................................ 19
II.2.1.3 Road source input data.................................................................................................................................................................... 21
II.2.1.4 Methods of data collection.............................................................................................................................................................. 21
II.2.1.5 Links with other IMAGINE Work Packages .................................................................................................................................. 22
II.2.2 Road Traffic flow modelling................................................................................................................................................................ 22
II.2.2.1 Introduction .................................................................................................................................................................................... 22
II.2.2.2 Road Traffic flow input data........................................................................................................................................................... 23
II.2.2.3 Methods of data collection.............................................................................................................................................................. 24
II.2.2.4 Links between road traffic modelling and other IMAGINE Work Packages .................................................................................. 24
II.2.3 Rail noise source.................................................................................................................................................................................. 25
II.2.3.1 Introduction .................................................................................................................................................................................... 25
II.2.3.2 How to determine the sound power level ........................................................................................................................................ 26
II.2.3.3 Rail source input data...................................................................................................................................................................... 28
II.2.3.4 Methods of data collection/measurement methods.......................................................................................................................... 29
II.2.3.5 Links with other IMAGINE Work Packages .................................................................................................................................. 30
II.2.4 Industrial noise .................................................................................................................................................................................... 30
II.2.4.1 Introduction .................................................................................................................................................................................... 30
II.2.4.2 How to determine the sound power level ........................................................................................................................................ 30
II.2.4.3 The input and output ....................................................................................................................................................................... 31
II.2.4.4 Measurement methods .................................................................................................................................................................... 33
II.2.4.5 Links with other IMAGINE Work Packages .................................................................................................................................. 33
II.2.5 Aircraft noise....................................................................................................................................................................................... 33
II.2.5.1 Introduction .................................................................................................................................................................................... 33
II.2.5.2 How to determine the sound power level ........................................................................................................................................ 34
II.2.5.3 The input and output ....................................................................................................................................................................... 35
II.2.5.4 Methods of data collection/measurement methods.......................................................................................................................... 36
II.2.5.5 Links with other IMAGINE Work Packages .................................................................................................................................. 36
II.3 Propagation.............................................................................................................................................................................................. 37
Page 4

II.3.1 The propagation method ...................................................................................................................................................................... 38
II.3.2 Propagation paths ................................................................................................................................................................................ 40
II.3.3 Source segmentation............................................................................................................................................................................ 41
II.3.4 Geometrical data model ....................................................................................................................................................................... 41
II.3.5 Input data and GIS............................................................................................................................................................................... 42
II.4 Determination of long term Lden by calculation......................................................................................................................................... 44
II.4.1 Introduction ......................................................................................................................................................................................... 44
II.4.2 Meteorological classes......................................................................................................................................................................... 44
II.4.3 Classification variables ........................................................................................................................................................................ 44
II.4.4 Calculation of D/R............................................................................................................................................................................... 45
II.4.5 Calculation of long term Lden................................................................................................................................................................ 45
II.5 Determination of Lden and Lnight using measurements ................................................................................................................................ 46
II.5.1 General ................................................................................................................................................................................................ 46
II.5.2 Description of the method.................................................................................................................................................................... 46
II.5.3 Input and output of the method............................................................................................................................................................ 47
II.5.4 Links with other IMAGINE Work Packages ....................................................................................................................................... 47
II.6 Factors affecting the quality and accuracy of end results......................................................................................................................... 48
II.6.1 Road Noise Source .............................................................................................................................................................................. 48
II.6.2 Road Traffic Flow Modelling .............................................................................................................................................................. 48
II.6.3 Rail Noise Source ................................................................................................................................................................................ 48
II.6.4 Industrial Noise.................................................................................................................................................................................... 49
II.6.5 Aircraft Noise ...................................................................................................................................................................................... 49
II.6.6 Propagation.......................................................................................................................................................................................... 49
II.6.7 Measurement of Lden and Lnight.............................................................................................................................................................. 50
II.6.8 Mapping Specifications ....................................................................................................................................................................... 50
II.7 Databases ................................................................................................................................................................................................. 51
II.7.1 Road Noise Source .............................................................................................................................................................................. 51
II.7.2 Road Traffic Data ................................................................................................................................................................................ 51
II.7.3 Rail Noise Sources............................................................................................................................................................................... 51
II.7.4 Industrial Noise Sources ...................................................................................................................................................................... 54
II.7.5 Aircraft Noise Source Database........................................................................................................................................................... 57
II.8 References ................................................................................................................................................................................................ 61
Appendix I Presentations and papers during the project..................................................................................................................................... 67
Appendix II Quick scan of IMAGINE.................................................................................................................................................................. 70

Page 5

INTRODUCTION
In June 2002, the European Directive on the Assessment and Management of Environmental Noise
2002/49/EC, (abbreviated as END) was accepted and came into force. Under this Directive, member states
are obliged to produce strategic noise maps of major roads, railways, airports and large agglomerations by
30th June 2007. These noise maps shall express the environmental noise levels caused by the above
sources, in terms of the harmonised noise indicators L
den
and L
night
. From these, other statistics such as the
total number of residents exposed to certain noise levels shall be derived. This information shall then be
submitted to the European Commission and made public. The next step will be to draft Noise Action Plans,
the first of which will have to be produced by July 2008.
It has always been the intention of the Commission to establish common assessment methods for the
production of these noise maps but until such methods are made available, the END has defined interim
methods. These interim methods or a Member States national method, if it can be shown to be equivalent
to the interim method, will be used in at least the first round of mapping in 2007.
As a first step in developing a common method, the project HARMONOISE was initiated in August 2001.
This project was partly funded by the European Commission (DG Information Society and Technology)
under the 5th framework programme. Its main objective was to develop harmonised, accurate and reliable
methods for the assessment of environmental noise from roads and railways. In order to produce a model
that was capable of predicting the long term average L
den
and L
night
the HARMONOISE philosophy was to
separate source and propagation. It thus developed source models for road and rail and propagation
models for these sources that included the effect of distance, air absorption, ground effect, barrier diffraction
and meteorological variables such as wind and temperature gradient. Following validation of the
propagation model this project was completed in August 2004.
The IMAGINE project, which commenced in November 2003, and is a Strategic Targeted Research Project
which addresses Task 3 of the Scientific Support to Policies (SSP) Call under the 6th Framework
Programme, aimed to extend the HARMONOISE source databases for road and rail and to use the
HARMONOISE methodology to develop prediction methods for aircraft and industrial noise sources. This
required the setting up of industry and aircraft source models together with such modifications to the
propagation models as was necessary to account for high sources (aircraft), large sources (industry) and
diffraction by vertical barriers (industry).
The overall objective of both projects is therefore to provide a model which meets the requirements of a
common assessment method and which eventually can be adopted for use for strategic mapping as defined
by the Environmental Noise Directive.
The main technical objectives of IMAGINE are:
1. To provide practical guidelines for data management and information technology aspects of noise
mapping (Work Package 1),
2. To provide guidelines and examples for an efficient link between traffic flow management on the
one hand and noise mapping and noise action planning on the other hand. (Work Package 2),
3. To provide guidelines and examples of how and when noise measurements can add to the
credibility and reliability of assessed noise levels (Work Package 3),
4. To provide a harmonised, accepted and reliable method for the assessment of environmental
noise levels from airports, which links well within the methods for noise propagation description
developed in HARMONOISE and at the same time has a large acceptance in the field of future
users and other stakeholders (Work Package 4),
Page 6

5. To provide default databases for the source description of road noise, i.e. vehicle category and
road surface type, for a typical fleet of European road traffic, and provide guidelines on how to deal
with situations deviating from the default value (Work Package 5),
6. To provide databases for the source description of rail noise, i.e. vehicle category and track type,
for an example sample of the European rail traffic fleet, as well as default data sets, and provide
guidelines on how to deal with situations deviating from the typical (Work Package 6),
7. To provide a harmonised, accepted and reliable method for the assessment of environmental
noise levels from industrial sites and plants, which links well within the methods for noise
propagation description developed in HARMONOISE, in combination with methods for source
description by measurements based on the existing set of standards and guidelines, together with a
default database for typical sound production for a limited but representative number of industrial
activities (Work Package 7),
8. To provide for acceptance and easy and quick implementation of the above deliverables and
those from the HARMONOISE projects, in order to allow a smooth and harmonised process of
noise mapping and noise action planning in all member states (Work Package 8).

Book-mark to this document
This report consists of two parts. Part I is meant to give a quick overview for management purposes of the
project results, the background and future developments. The advantages of the IMAGINE methods are
outlined. Part II guides the reader through the technical results by descriptions of the methods per work
package and by cross references to the project deliverables. This part is aimed at end users who need to
understand the methods in global terms and need to know the implications for practical use. Lists of
required parameters are given and implications for accuracy are briefly described.

Page 7

PART I MANAGEMENT OVERVIEW
I.1 List of IMAGINE deliverables
General
Deliverable D1 IMAGINE Website www.IMAGINE-project.org
The project website contains all relevant information about the project, the latest news and research results,
the organisations and the partners involved in this project. The website has two parts, a public part and a
restricted part. The public part is presented in popular (not too technical) language. The restricted part of the
website is only accessible for members of the project. The website is subscribed to the most popular search
engines and is linked to sites such as national acoustic societies, universities, ministries, stakeholders,
related magazines and the site of the EC.
Deliverable D2 IMAGINE State-of-the-Art, Report IMA10TR-040423-AEATNL32
This report describes the methods to determine L
den
or similar measures, and possibilities for action
planning which were available at the start of the project. The conclusions were that there was no calculation
method available anywhere to determine L
den
and that all methods dealing with similar measures shared the
deficiency that they did not make a separation between the source and propagation models. Besides
weaknesses and missing elements in existing methods were identified. These conclusions served as a
starting point for the IMAGINE project.
Deliverable D15 Final Synthesis Report, IMA10TR-06116-AEATNL10
This is the underlying report describing the main IMAGINE results and the use of the IMAGINE methods.

Road source
Deliverable D3 Assessment Programme for Parameters of the general European vehicle fleet,
Report IMA52TR-060111-MP10
This deliverable describes the road noise emission model and the data collection campaigns at the first
stage of the project. It represents the intermediate status of the model at that time.
Deliverable D11 The noise emission model for European road traffic, Report IMA55TR-060821-
MP10
This report represents a noise emission model for road vehicles that accounts for the characteristics of
different vehicle types and for the variations of the vehicle population in different European regions. It
contains the road source noise model equations and correction factors. Furthermore, validation of the model
against roadside measurements is presented, and the coupling between traffic flow modelling and road
noise source modelling is addressed.
Deliverable D7 Guidelines for the use of traffic models in noise mapping and noise action
planning, Report IMA02DR7-060531-TNO10
The purpose of this report is to assist authorities and consultants in using traffic models to produce road
traffic data for noise mapping and noise action planning. It provides information on how to improve a traffic
model for noise mapping and noise action planning. The subjects treated in the guidelines are speed,
acceleration, traffic composition, diurnal and long-time patterns, low flow roads, intersections and gradients.
Page 8

There are separate guidelines for noise mapping for main roads and for agglomerations, and for
macroscopic and microscopic models. The guidelines recommend methods for the improvement of traffic
models and give indications of the complexity, accuracy and cost of implementing the method.

Rail source
Deliverable D12/D13 Rail noise database and manual for implementation, Report IMA6TR-
061015-AEATUK10_D12/13
The report D12/D13 Rail noise database and manual for implementation accompanies the delivered rail
noise source database. It describes the general philosophy that has been applied to quantifying the
disaggregated sub-sources and storing the data. It also includes guidance on measurement, instruction on
data import and export, a method for creating the rail traffic model, advice on addressing non-standard
situations, and a default dataset. This Deliverable is a combination of the originally planned Source
database and Guidelines for typical situations.

Industry
Deliverable D6 Description of the Source Database - WP7: Industrial Noise, Report IMA07TR-
050418-DGMR01
An extensive database with a large number of industrial noise sources and their characteristic noise
emissions has been produced. For each source or industry type a typical example for the sound power
levels given for representative modes of operation, taking other aspects such as quality in terms of low
noise emissions etc. into account. The database is a powerful and convenient tool for the potential end user
who needs sound emission data for industrial noise sources for noise mapping purposes. However, the
database will have to be used with caution because there are a number of risks associated with its
application to noise mapping that are not evident - especially to the inexperienced user.
Deliverable D14 Guidelines for producing strategic noise maps on industrial sources, Report
IMAWP7D14-060811-DGMR03
This deliverable describes the work carried out for integrating industrial noise sources into the
HARMONOISE/IMAGINE model including modifications required to the propagation model for it to be
suitable for industrial source/propagation/receiver situations. Guidelines are given on how to make strategic
noise maps for industrial sources.

Aircraft
Deliverable D9 Reference and Engineering Models for Aircraft Noise Sources, Report IMA4DR-
061017-EEC-09
Deliverable D9, ref. [37], is in three volumes of which volume 3 consists of appendices. Volume 1 deals with
the propagation models:
Modifications required to the reference model in order to take high-level sources into account - the
definition of a hybrid model using both a two-ray method, for higher altitude sources, and the
original parabolic equation (PE) method for determining propagation
How the atmospheric conditions encountered in the vertical plane are classified and how wind,
temperature and humidity profiles may be created
The effects of atmospheric absorption
Page 9

The adaptations required to enable the Engineering model to be applied to aircraft noise sources.
Volume 2 treats the validation work performed using a series of flight measurement trials performed at
Ocaa, near Madrid, in Spain in September 2006. Some of the flight measurements were de-propagated
in order to produce a full noise source description of the test aircraft. The propagation models were then
applied to this source description and the results analysed.
Deliverable D10 Default aircraft source description and methods to assess source data, Report
IMA4DR-061204-Empa10
D10 provides information on how to generate sound source data and how to store it in a database. It
includes a description spectral sound powers and directivity functions. It describes the procedures to
measure and process data for the source emission model, and gives methods how to infer source emission
data from existing NPD information, how to store, exchange and use source data, proposing data tables and
a data base structure.

Mapping
Deliverable D4 Specifications for GIS-Noise databases , Report IMA01-TR060526-CSTB05
D4 deals with the geometrical input for building a noise model. It starts from the best possible practice in
view of the state of the art laid down by the HARMONOISE and IMAGINE projects and works down toward
its consequences for the future end-users.
Deliverable D8 Guidelines and good practice on strategic noise mapping, Report IMA10TR-
06116-AEATNL10
Deliverable D8 mainly reflects the point of view of noise engineers confronted to the problem of carrying out
noise mapping project conform the specifications of the European Noise Directive. It deals with aspects of
noise mapping such as data collection and preparation, the estimation of exposed populations, the
assessment of quality of noise maps and innovative technologies that may help in collection cheaper and/or
more accurate input data.
In combination, deliverables D4 and D8 aim to provide some answers to questions that were put forward at
the beginning of the project, like:
What do I need to make a noise map?
How to produce a good quality noise model?
Where can I get what I need?
How to obtain descent data at a reasonable cost and effort?
What will change after the introduction of the harmonised prediction methods?

Measurements and monitoring
Deliverable D5 Determination of L
den
and L
night
using measurements, Report IMA32TR-040510-
SP10
This deliverable describes the method to determine L
den
and L
night
using measurements, and is a stand
alone document written in the format of an ISO-standard. This means that a potential user only needs this
document to carry out this type of measurements. For an ISO member body that wants to standardize the
procedure the document can be submitted as a first draft when voting for a new work item.

Page 10

I.2 Links between HARMONOISE and IMAGINE
In the predecessor of IMAGINE, the HARMONOISE Project, methods for road and railway noise and for
propagation have been developed and validated. In IMAGINE the HARMONOISE methods have been
further developed and extended for aircraft and industry. This means that the HARMONOISE deliverables
do not always contain the latest versions of the methods. In general the IMAGINE deliverables are the most
up-to-date descriptions of the methods. The next table gives a short overview of the results of both projects.


Source
description
Source data

Propagation
model
Mapping methods Action planning

Road sources



Developed in
HARMONOISE and
IMAGINE
Source data
collected in
IMAGINE. Traffic
flow models
described in
IMAGINE
Source description
and traffic flow
modelling allow
analysis at the
source
Rail sources


Developed in
HARMONOISE and
IMAGINE
Source data
collected in
IMAGINE.
Measurement
methods developed
in HARMONOISE
and IMAGINE
Developed in
HARMONOISE and
IMAGINE


Source description
and traffic flow
modelling allow
analysis at the
source
Aircraft
sources

Developed in
IMAGINE, based
on separation
technique between
source and
propagation
Methods developed
in IMAGINE to
deduce data from
existing models
Adapted in
IMAGINE for high
sources
Meteo classes and
mapping guidelines
in IMAGINE

Source description
and traffic flow
modelling allow
analysis at the
source
Industrial
sources
Overview of
measurement
methods in
IMAGINE
Source database
(SourcedB)
developed in
IMAGINE
Adapted in
IMAGINE for lateral
diffraction
Meteo classes in
IMAGINE
Categorisation of
different industrial
activities allow
action planning

The HARMONOISE propagation method has been extended in IMAGINE for practical use and for the
calculation of L
den
at the receiver. The points which have been added and improved are:
Use of external GIS data in the noise modeling, ref. [4]
Adaptation of the propagation model to industrial noise sources, ref [75]
Adaptation of the propagation model to aircraft noise, ref [37]
Estimation of populations exposed to noise, ref [7]
Determination of L
den
and L
night
as long term averaged quantities, ref. [78]
Pre-processing of meteorological data to calculate the L
den
and L
night
, ref [78]
In the HARMONOISE project it was shown how noise levels are influenced by meteorological conditions
such as wind speed, wind direction and thermal inversion effects. For the determination of long term
averaged noise levels by means of measurement or calculations, short term levels, valid under specific
meteorological conditions, must be weighted according to their frequency of occurrence. The step-by-step
method to define the occurrence of meteorological conditions is an output of IMAGINE.

Page 11

I.3 Advantages of the IMAGINE methods
Road source method
The road noise source method represents the noise emission of the average European road vehicle. It is
more suitable for noise mapping purposes than currently used methods because:
it is based on extensive and recent measurement sets from the most important European regions,
and therefore is truly representative of the European average
heavy duty vehicles (trucks) and powered two-wheelers (mopeds, motorcycles) have received
special attention and are now supported with extensive measurement data;
it provides correction factors to adapt for local variations in road surfaces and vehicle fleet.

Traffic Modelling
Traffic models are needed because in most cases it will be impossible to construct a noise map without any
form of traffic modelling when there is not enough measured traffic data. Also, in noise action planning, a
traffic model is needed to determine the expected effects of measures.
With the guidelines given in Deliverable D7 (Guidelines for the use of traffic models in noise mapping and
noise action planning), more accurate traffic flow data can be produced when using a traffic model. The
purpose of the report is to assist authorities and consultants in using traffic models to produce road traffic
data for noise mapping and noise action planning. Separate guidelines are given for noise mapping and
noise action planning. For noise mapping, a further distinction is made between noise mapping for main
roads and agglomerations, and between macroscopic and microscopic traffic models.
In addition to the guidelines, the technical report of task 2.4 Collection Methods for Additional Data, ref.
[16] provides guidance on how to collect traffic data to improve traffic models or their output, or to measure
the traffic flow data needed for making a noise map.

Railway source method
In the HARMONOISE/IMAGINE rail model the combined roughness of the wheels and the rails is a key
parameter for rolling noise. The inclusion of the combined roughness leads to a major improvement in
modelling accuracy, especially because local track roughness can cause rolling noise to vary over a range
of up to 20 dB. At present, most national rail noise models include overall rolling noise data (ie vehicle +
track contributions combined) that have been acquired from pass-by measurements on track that is not
excessively rough or corrugated.
In the HARMNOISE/IMAGINE model, rolling noise is split into the vehicle and track contribution. In addition,
the rail model takes into account all other potential noise sources, such as traction elements (exhaust, fans,
compressors), braking noise (including brake squeal), curve squeal and aerodynamic noise. The level of
detail of these extra sources is significantly greater than is the case with other available mapping models.
This complete separation allows a detailed apportionment of the rolling noise sub-sources and it allows an
evaluation of the effects of noise mitigation applied to sub-sources. This is a valuable tool in Action
Planning, and is also useful for identifying whether vehicles or track are the prime emitters of noise, helping
with such considerations as track access charging or determining responsibility when levels are exceeded.
The source term database and associated procedures for acquiring and storing data provide a standardised
and efficient method both for accessing appropriate data and for adding new data. In its delivered form it
Page 12

holds example data from France, the Netherlands, Sweden, Hungary and the UK, representing a broad
range of generations of rolling stock, configurations and operating speeds. The database also provides
default data. Noise modellers who wish to model vehicle types that are not already within the database are
provided with guidance on how to acquire the data. Such a comprehensive pan-European rail source
database has not been available previously and represents a significant improvement in knowledge as well
as in the harmonisation of rail noise modelling.

Industry
If industry is located next to housing development, noise can become a problem. Unlike roads, where a
large number of different vehicles pass by, the noise may be the product of just a few sources. Although
these sources may be common to a large variety of industries, their usage (operating speed-maintenance
level-operating hours) may differ greatly from company to company. So it is often not enough to have a
general knowledge of the source, it is better to measure the source.
Deliverable D14 [75] gives a very large number of measurement possibilities, in order to obtain the sound
power levels of machine and areas. If no measurement data are available, the source database SourcedB
[72] can be used containing a large variety of noise sources, based on measurements or formulae. This
database which contains more than 1500 entries is available from the DGMR website and has already been
downloaded more than 200 times.
The noise propagation part is superior to the interim methods because of the possibilities of including
meteorological conditions. The method proved to be valid for many meteorological situations. A method is
proposed in ref. [28] and [78] for calculating the L
den
and L
night
, based on the statistics over a year.

Aircraft
Currently used aircraft noise models, including the Interim model, are based on the "integrated" modelling
guideline defined in European Civil Aviation Conference (ECAC) Doc. 29 [40]. These models have all noise
sources (engine and airframe) and propagation combined into a set of "Noise-Power-Distance" (NPD)
curves for each of the two phases of flight - approach and departure. This method does not allow for taking
into account:
- realistic directivity of the aircraft source under various operating conditions
- temperature gradients and other meteorological effects;
- ground characteristics and shielding by relief.
The IMAGINE model described in ref. [37]and ref. [38] is a source-propagation model that takes all these
factors into account, enabling much more accurate analysis to be undertaken.
Work performed in IMAGINE has demonstrated the creation of example source data either by measurement
or by reverse-engineering of NPD data. For the noise calculations, the aircraft is modelled as a point source
with three-dimensional spectral directivity to account for the different contributions from individual sound
sources like fan, engine, jet and aerodynamic effects.

Propagation Method
The interim propagation method and most national propagation methods are derived from the ISO 9613-2
standard. This standard provides an empirical model to estimate the propagation effects in a very simplified
situation (i.e. a flat terrain with a single screen). When applied to more complex situations, the real geometry
Page 13

has to be matched to this simplified description. How to do this is not part of the ISO standard. Even
though some national methods provide rules for the determination of the geometrical parameters, a large
degree of freedom is left to the software developers. As a consequence, different implementations based on
these common standards often lead to significantly different results, even in moderately complex situations,
and differences up to 5 or even 10 dB(A) are not unusual.
With HARMONOISE/IMAGINE methods a harmonised approach is possible through the following steps:
1. The HARMONOISE model relies on explicit geometrical modelling of propagation paths; i.e. it is the
complex geometry of the paths that is used as the input to the model. Because the software
developers do not need to interpret the geometrical data, this will clearly result in higher
reproducibility of the results.
2. The IMAGINE project provides a complete and explicit set of specifications for the geometrical
model (i.e. the data) on which the HARMONOISE methods (i.e. the algorithms) operate. Such
specifications define the end-user requirements for the collection of data and for the construction of
the geometrical model. As much as possible, these requirements are expressed in quantitative
terms.
3. Sensitivity analysis of the HARMONOISE/IMAGINE method provides explicit links between the
accuracy and the level of detail of input data and the expected accuracy of the results. This also
allows for objective rank ordering of the data items to be collected.
As part of the common models and methods as prescribed by the END, the IMAGINE project also looked
into common methods for the estimation of populations exposed to noise ref. [4].

Measurement of L
den
and L
night

There is no interim method for the measurement of L
den
nor is there any international or national method
addressing all the relevant parameter related problems resulting from the use of the HARMONOISE source
and meteorological models. Thus the measurement method developed within IMAGINE introduces many
new features compared to available methods. The most important new features are
measurements are classified into meteorological classes. The measurement method is harmonised
with the IMAGINE calculation methods for road, rail, aircraft and industry noise.
stratification of measurement according to meteorological conditions
measurement uncertainty is dealt with in compliance with the guidelines given in ISO GUM
the method is general and can be adapted both to industrial noise and road, rail and air traffic
both short term and long term measurements are dealt with
correction of measured values to be representative for yearly averages is dealt with
the method is flexible, different measurement efforts will yield different measurement uncertainties.

I.4 Subjects for further development
Road Traffic Noise Mapping
Although the predicted noise levels represent the European average, and some variations between
countries and regions are accounted for, reliable data from different countries showed differences in the
Page 14

lower frequency range that could not be explained. Some ideas about the origin of these differences
described in ref. [52] should receive more attention.
Furthermore, the corrections for vehicle acceleration could be improved. Gathering data for this correction is
difficult, especially for heavy duty vehicles. A more dedicated measurement campaign would definitely
improve the results.
Many techniques and data have been collected and investigated that are similar to the work done in the field
of traffic air pollution. A combined approach is more efficient and beneficial for both fields, and for the end
users, and should definitely receive attention in the future.

Traffic Modelling
No specific road traffic model is recommended to produce traffic flow data. In principle, many different types
of (commercially available) traffic models can produce the data required for the road noise source model.
Deliverable 7 Guidelines for the use of traffic models in noise mapping and noise action planning can
assist in choosing the most appropriate model in the context of noise mapping and noise action planning. It
also gives guidance on how to improve traffic models in the context of noise mapping and noise action
planning. It is up to the users of the traffic model to use a suitable traffic model, provide good quality input to
the model and be aware of the limitations of the model and its output.

Railway Noise Mapping
The rail noise database framework has been written in Access and is linked with a standardised input
spreadsheet so that data is automatically transferred to the database. The algorithm for converting sound
pressure level, as stored within the database, to sound power level as required for assembling the traffic
model, is presented, but this process is not automated. Similarly, the algorithms for building the traffic model
are available, but this process is also not automated. Therefore these two elements require attention during
the implementation of IMAGINE. A tool to link the database with GIS software, to interface the database
with noise mapping software automatically, would be desirable, possibly by the creation of an appropriate
.dll.
The example data included within the database has been designed to cover the full range of typical stock
that operates across Europe. This allows noise modellers to identify examples that are identical or similar to
the specific stock at locations of interest to them. There is also the option of using the default data provided
within the database.
However, it will always be necessary to add new data to the database, both where new, relevant, examples
are needed for specific locations and networks, and where new trains come into service. The database has
been designed specifically to allow this to happen efficiently in the future. The user interface is clear and
straightforward, but it is necessary for the database to be administered and maintained by an expert central
administrator who will be able to judge the appropriateness of the available data and ensure that the data
has been imported correctly. The administrator will also be able to manage the release of data to modellers,
including the provision of advice on which datasets will be of use. This system of administration and
maintenance requires setting up before the IMAGINE model can be fully implemented.

Industrial noise Mapping
The measurement methods based on measurements on a certain distance from the source, should be
evaluated for the propagation part. This should be done in agreement with the HARMONOISE/IMAGINE
propagation method.
Page 15

The industrial source database (SourcedB) should be maintained as new data (development of equipment,
changes in measurement methods, more data) becomes available. Preferable this should be maintained by
a group of international recognized specialists.
The HARMONOISE/IMAGINE model was validated over almost flat terrain. Further investigation in
propagation over/through build up areas would be very useful in order to further enhance the method.

Aircraft Noise Mapping
The major problem facing the IMAGINE model is the availability of data. In IMAGINE an example set of data
has been reverse-engineered from NPD curves. These data are fairly approximate and incomplete. A
complete set of data needs to be supplied for each aircraft that will be needed in the database.
This is most easily done by the manufacturers themselves. These manufacturers are, however, very reticent
to supply what they consider confidential intellectual property. If the political will can be found to encourage
them to supply these data, there will be a significant financial price to pay.
Another method of producing these data is to perform a series of measurement campaigns, measuring and
converting data from a number of fly-pasts of each aircraft. This will be extremely costly.

Propagation models
At present the propagation model is only available as an executable dll-file and as a number of reports. It
still remains to describe the model systematically including all algorithms used in order to make it possible
for software writers to make their own optimizations of the algorithms used.
It turned out that most available software packages already include most features of the extended three-
dimensional modelling required for the new methods. However their different approaches can easily lead to
5 or 10 dB(A) differences in complex situations, so harmonisation and/or standardisation is required after
the IMAGINE project in order to promote common modelling techniques for 3D geometry, to guarantee the
reproducibility of propagation path detection methods or to increase the interoperability of software.
Geographical Information Systems are already largely used to collect, store and manipulate the input data
for noise mapping. One of the main challenges when using currently available data with the new prediction
methods is to handle inaccurate, incomplete or missing data. Modelling solutions for these situations are
provided in ref. [3] and ref. [4], in such a way that the new methods can in principle be used with todays
data. In order to take advantage of the increased accuracy and the extended modelling capacities of the
new methods however, new data must be collected or existing data refined to higher levels of detail.

Measurement of L
den
and L
night

As the IMAGINE method for the measurement of L
den
and L
night
introduces many new features it would be
desirable to evaluate it after having been in use for a few years. It would then be possible to learn from the
experiences and improve it further. It would be desirable to establish more default values for some
uncertainty components and to get further experience from the combination of measurements and
calculations.
Page 16

PART II TECHNICAL DESCRIPTION OF THE METHODS IN GLOBAL
TERMS
II.1 Flow diagram


The global structure of the HARMONOISE and IMAGINE methods is given in the schematic overview
above. A clear separation is made in the model between the source descriptions for road, rail, industry and
aircraft sources, and propagation to the receiver. The result of the source models is a sound power level per
source type for each source height relevant to that source, together with a certain directivity. The source
methodologies are described in chapter II.2.
The propagation method describes the transmission of sound along a set of propagation paths, linking the
source positions to the receiver point. The number and type of the propagation paths depend on the
complexity of the site. The Point -to-Point module estimates the effects of ground and obstacles on the
propagation of the sound along these paths, under various meteorological conditions.
The result of the propagation model is an L
eq
at a specific receiver point for a given propagation class (the
meteorological influence on the propagation paths is divided into 4 different propagation classes). The total
L
eq
from a source is the summation of the contributions of the different sub sources (located at different
heights) via their respective propagation paths.
The L
den
and L
night
values are calculated from the L
eq
values by determining the occurrence of the different
propagation classes within the time period of assessment, and summing the relative contributions of each
class. The determination of long term L
den
(L
night
) by calculation and the propagation classes are described in
chapter II.4, the measurement of long term L
den
(L
night
) is described in chapter II.5.
Chapters II.6 and II.7 describe the quality of the end results and the databases.
PropagaIion 5ource
8asic source
properlies
Dalabases
Measuremenl
melhods
Trallic llow P2P model
Meleorological
condilions
Ceomelry
Surroundings
8uildings
Lden
Propagalion classes
PropagaIion 5ource
8asic source
properlies
Dalabases
Measuremenl
melhods
Trallic llow P2P model
Meleorological
condilions
Ceomelry
Surroundings
8uildings
Lden
Propagalion classes
8asic source
properlies
Dalabases
Measuremenl
melhods
Trallic llow P2P model
Meleorological
condilions
Ceomelry
Surroundings
8uildings
Lden
Propagalion classes
Page 17


II.2 Methodology

In the following diagram the steps of the mapping process for each source type are outlined.


The result is a sound power level (SWL) in 1/3 octave bands per source height or per source, per hour for
the day, evening or night period. The methods for gathering data and calculating the SWL are described in
the following sections. The sound power level feeds into the propagation calculations which are described in
section II.3 of this report.

Calculate SWL (1/3
oct bands, source
height, per m of
source line, per hour)
or use railway
database
Sum of SWLs per
source height
Site measurements,
then calculate SWL
(1/3 oct bands, per
source, per hour
Define locations of
source lines with end
points in more detail
Identify roads and
define road source
lines with end points
Define locations of
source lines with end
points in more detail
Identify road sources,
vehicle classes
Define operating
conditions per unit of
time (day, evening,
night), #per hour,
speed, acceleration,
braking, etc.
Calculate SWL (1/3
oct bands, source
height, per m of
source line, per hour
Correction factors,
directivity, regional
variations, road
surface type, winter
tyres, etc.
Sum of SWLs per
source height
Road
Identify railways and
define railway source
lines with end points
Identify rail vehicle
type, track types,
traction noise, rolling
noise, aerodynamic
noise
Define operating
conditions per unit of
time, (day, evening,
night), # per hour,
track roughness,
speed, acceleration,
braking, etc.
Correction factors,
directivity, curves,
joints, bridges, etc.
Rail Industry
Identify airport runway usage
and ground track
Identify aircraft types and
engines
Determine number of flights of
each aircraft type on each
flight path, day, evening, night
Convert each flight path into
short segments. Calculate the
time t the aircraft is within
this segment. Calculate the
position within this segment to
place the point source
representing the aircraft.
For a specific receiver point,
determine the emission angles
(including bank angle) and
determine from the sound
source model the sound power
spectrum for the appropriate
angles, thrust and operation
(landing/departure)
Describe flight profiles (height,
speed, engine thrust as a
function of flown distance) for
each aircraft type and for
classes of take-off weight.
Aircraft
For curved flight, calculate the
bank angle of the aircraft (=
lateral inclination)
Define source
location or area,
including source
height (variation)
Identify source types
(point, line, area, or
volume source),
source classes, plant
types
Define operating
conditions, #per unit
of time (and day,
evening, night),
working hours,
Correction factors,
directivity
Define locations and
sizes/heights of
sources in more detail
Measure SWL (1/3
oct bands) or use
SourcedB
Page 18

Definitions used in the scheme
Source lines tracks, roads, flight paths
Source locations location of industrial sources (point sources, line sources, area sources or volume
sources)
Source segments parts of source lines on which L
w
is assumed constant or almost constant
Source types train vehicle types, track types, industrial sources, aircraft types in combination with engine
types
Operating conditions numbers per hour (yearly average), speeds, acceleration, braking, working hours
Unit of time period of time during which the operating conditions are considered constant, normally this will
be an hour
Correction factors correction for road type, directivity of the source, curves, bridges, road surface types,
winter tyres, etc.

II.2.1 Road noise source
II.2.1.1 Introduction
The road noise emission model describes the noise emission of an "average" European road vehicle in
terms of a sound power level. The emission model interfaces with the propagation calculation method
described in II.3. The road source methods are described in detail in [ref [52].
The emission model consists of a set of mathematical equations representing the two main noise sources:
a. rolling noise due to the tyre/road interaction; (combined with aerodynamic noise)
b. propulsion noise produced by the driveline (engine, exhaust, etc.) of the vehicle;
The mathematical formulae exhibit the following general form:
) ( ) (
, , ,
v f B A v L
m i m i m i
+ = , (1)
with f(v) being either a logarithmic function of the vehicle speed v in the case for rolling and aerodynamic
noise, or a linear function with v in the case of propulsion noise. The sound power level L
i,m
is calculated in
1/3-octaves from 25 Hz to 10 kHz, where the subscript i indicates the spectral frequency band. The index m
represents the vehicle type.
The rolling and propulsion noise production of the road vehicle at the reference speed of 70 km/h is
represented by the values A
i,m.
B
i,m.
f(v) represents the change in noise production due to a difference in
vehicle speed relative to a reference speed.
The structure of this model, i.e. the form of the equations, as well as the source height definition were
developed within the HARMONOISE project. Within IMAGINE, the coefficients needed to calculate the
emission have been established accurately, reliably, and representative of the European average, by
means of measurements. Also, the category of Powered Two-Wheelers has been added, which had
received no attention in HARMONOISE.
Regional correction factors have been added to account for deviations of the local or national vehicle fleet
from the European average, and other correction factors have been added, extended, or reviewed.
Page 19

II.2.1.2 How to determine the sound power level
Vehicle classes
The road noise model predicts the source emission levels for four main vehicle classes:
1. Light motor vehicles (passenger cars and light vans)
2. Medium heavy vehicles (heavy vans and trucks with 2 axles and 6 wheels)
3. Heavy duty vehicles (trucks with more than 2 axles)
4. Powered Two-Wheelers (scooters/mopeds and motorcycles)
The vehicle classification scheme was developed within the HARMONOISE project ref. [54], With respect to
this project, the category of Other heavy vehicles has been deleted.


Point sources

source at 30 cm



figure 1 Noise source positions
For the calculation of the noise emission L
W
, each vehicle is represented by one or two point sources, which
are depicted in
figure 1 above. The lowest source is located at 0,01 m above the road; the highest source is located at
0,3 m for light motor vehicles and at 0,75 m for heavy motor vehicles. For two-wheelers, only the 30 cm
source height is used; since the contribution of rolling noise for these vehicles is assumed to be negligible,
the main noise sources are located between 20 and 50 cm.
It is assumed that rolling noise is distributed 80% to the lower position and 20% to the higher position and
that propulsion noise is distributed 20% to the lower position and 80% to the higher position.
The horizontal resolution of the point sources is not relevant since a traffic stream will be represented by a line
source. This line source is located at the centre of the road lane.
Page 20

Source equations
For rolling noise, the general accepted and widely validated logarithmic relation between sound power and
rolling speed is used. The emission L
WR
is formulated as follows:

+ =
ref
R R WR
v
v
B A L log
,
where, as stated above, the coefficients A
R
and B
R
are given in 1/3-octave bands for each vehicle class, and
v
ref
= 70 km/h.
The propulsion noise emission L
WP
includes all contributions from engine, exhaust, gears, air intake, etc. For
propulsion noise, the emission L
WP
is formulated as follows:

ref
ref
P P WP
v
v v
B A L

+ = + C
p
.a,
where the coefficients A
P
B
P
and C
p
are given in 1/3-octave bands for each vehicle class, v
ref
= 70 km/h and
a
1
is the vehicle acceleration in m/s
2
.

Correction factors
The source model has correction factors for the following parameters:
for rolling noise:
o road surface type (mostly speed dependent and per frequency band)
o road surface age and wetness
for propulsion noise:
o vehicle acceleration and deceleration, frequency dependent
o ramps, see ref. [52]
on the overall level:
o horizontal and vertical directivity of the sources, for certain frequency ranges
for variations in vehicle fleet:
o % of Diesel vs. petrol engines
o use of winter and studded tyres, and traction tyres for trucks
o vehicle weight and tyre width, which are related
o illegal replacement exhaust systems
Detailed descriptions of these correction factors can be found in ref. [52].
Page 21

II.2.1.3 Road source input data
To predict the total sound power level of a traffic stream on a road, the necessary input is:
the number of vehicles per class (light motor vehicles, medium heavy vehicles, heavy duty
vehicles, powered two-wheelers), per unit of time (usually one hour), per day, evening and night
period. The input comes from databases, statistics or traffic models (described in II.2.3);
the speed of the vehicles in each class, in km/h (also input from traffic flow models).
To use the corrections factors for certain additional parameters, the following input is needed:
the acceleration / deceleration of each vehicle class in m/s
2
, again with as much detail as possible;
the type of road surface, and/or its material parameters (stone size, porosity, etc.) see ref.[58]
and ref [59];
for regional corrections: statistical data of the regional vehicle fleet, i.e on a national or local level:
o the % of Diesel and petrol engines;
o the % of vehicles with winter tyres;
o the % of vehicles with studded winter tyres;
o the average vehicle weight or the average tyre width.
All correction factors are optional, their default value is 0 dB(A). Therefore, the model can be used without
any correction factors. However, their effect may be significant, and it is therefore strongly advised to use
these corrections.
The effect of road surfaces on the total source emission can be up to 10 dB(A), especially at road sections
with a constant vehicle speed of 40 km/h or more. Therefore, it is strongly advised to correct the results for
road surface effects.
Secondly, vehicle acceleration/deceleration should not be neglected at intersections. The effect on a single
vehicle may be 8 10 dB(A); the effect on the total vehicle stream, however, is estimated to be 3 4 dB(A).
Guidelines for default corrections for certain types of intersection are described in ref. [52], the use of which
is advised if no other acceleration data are available.
Finally, it is possible to make regional corrections if statistical data are available. Regional effects are
estimated to be from 0,5 to 2 dB(A) for each term.
The sound power level is calculated in 1/3 octave frequency bands from 25 Hz to 10 kHz, of a single road
vehicle of a certain class, for each point source. The basic equations and guidelines are given to combine
the sound power levels of all the vehicles in a traffic stream, see ref. [52].
II.2.1.4 Methods of data collection
The road noise emission model for the vehicle classes developed in IMAGINE requires no additional data
collection, apart from the traffic data already mentioned, to calculate the European average noise levels.
Only in the development of more accurate correction factors is additional data needed, as identified in
section I.7.1.5,
For the road surface correction, some default types have been included in ref. [52]. However, since the
influence of the road surface on traffic noise is large, especially at speeds higher than 40 km/h, the quality
and relevance of the noise map is greatly improved by including road surface data. The 5
th
Framework
Page 22

Programme SILVIA has developed data collection methods for the influence of road surfaces, the results of
which can be directly integrated with the IMAGINE road noise model.
An ISO recommendation for the assessment of the road surface effects is to use a CPX measurement, that
can determine the effect of the road surface while driving over the road (ref. [56] ).
Statistical data for the vehicle fleet can usually be gathered from national traffic bodies. For instance, the
national database of license plate registrations usually includes data on engine fuel type and vehicle weight
or engine displacement volume. Relations between the latter two and the tyre width, which is the parameter
of influence on rolling noise, can be found in our deliverable. Gathering data on winter or studded tyres can
be more challenging, unless the use of these tyres is obligatory in certain periods of the year, as is the case
in some Nordic countries. Otherwise, counting the number of winter tyres on vehicles in parking lots has
proved to be a relatively reliable and easy method of gathering these data.
II.2.1.5 Links with other IMAGINE Work Packages
The road noise model calculates the instantaneous sound power level of a single vehicle, as a function of its
vehicle class, speed, and many other parameters. To calculate the sound power level of an entire traffic
stream on a road section, traffic data are needed.
If this information is not available from other sources, guidelines for gathering these traffic data are
described in II.2.3 and [16].These guidelines describe the best way of obtaining the data for several levels of
traffic models, and where to improve currently existing models, if so desired.
The processes involved for the final noise mapping exercise using GIS are described in II.3 and ref. [3] and
ref. [4].
Methods on how to carry out additional short-term and long-term road noise measurements, have been
developed in ref. [27].

II.2.2 Road Traffic flow modelling
II.2.2.1 Introduction

For the input of the road noise source model it is possible to use traffic flow models. Traffic flow modelling is
treated as a separate subject in IMAGINE to establish an efficient link between traffic flow management and
road noise mapping. The resulting requirements and guidelines for noise mapping and action planning are
described in ref. [16].
The requirements of the HARMONOISE/IMAGINE method for traffic flow (numbers, classes, vehicle speeds
etc) were met by reviewing alternative traffic modelling strategies. This allows the end user to:
judge whether the traffic model available can deliver the desired data, to an acceptable accuracy;
choose an appropriate traffic model (for noise mapping of main roads or of agglomerations, or for
noise action planning); and
review the possibilities for refinement of the traffic model and ways of implementation of these
improvements.
Also, guidance is given for the situation in which no traffic model is available.
Page 23

Traffic models can, depending on the type of model, provide aggregated flow data (e.g. flows and speeds
per hour or per period of the day) or individual vehicle data (e.g. vehicle type, speed and acceleration) for
the road sections included in the noise calculations. Both can be handled by the source model, but when the
traffic data is provided to the source model in the form of individual vehicle data, the resulting noise levels
from individual vehicles need to be aggregated to obtain the noise levels for the desired periods of the day.
For noise mapping, traffic flow data can be provided by traffic models, or measured on the road. For main
roads, measuring traffic flow data might be feasible. For an agglomeration using a traffic model is the more
logical option. For action planning, a traffic model is needed to estimate the effect of measures to reduce
noise.
The Good Practice Guide for Noise Mapping V2 (see General References) provides default values if no, or
very limited traffic data is available.
II.2.2.2 Road Traffic flow input data
Input required for road traffic models
Traffic models require quite a large amount of information. In most cases, an existing model will be used, in
which most of the data needed is already incorporated. Generally, the traffic demand is first estimated,
based on:
socio-economic data: land use (number of household/inhabitants, jobs, schools, etc.), income, car
ownership, travel and parking costs, etc.;
road network representation;
travel data for the periods modelled (for noise, that would be day, evening and night, ideally).
The traffic then needs to be assigned to the road network. For this, data are needed on network layout.
Depending on the model type (macroscopic vs. microscopic), this may be more or less detailed, and include
ata on:
zones with properties (e.g. socio-economic data, parking costs);
nodes with properties (e.g. crossing penalty);
links with properties (from node, to node, length, capacity/number of lanes, free flow speed).
For some traffic models, input is needed on traffic control systems (e.g. traffic lights), and traveller/driver
properties and vehicle properties (vehicle stock).
Checks on input of road traffic models
Traffic models were not developed for environmental analyses, so it should be checked to what extent an
available model is suitable for noise mapping. Common problems are:
the traffic model does not distinguish between the vehicle types as used in the noise model;
the traffic model does not cover the periods (day-evening-night) as used in the noise model;
the traffic model for an agglomeration does not include all roads in the network;
the traffic model does not have information on road gradients;
the traffic model does not provide data on speed distributions and/or accelerations (which may be
important in urban areas, near intersections).
Page 24

Output of road traffic models
The output of traffic models differs per type of model, and even between different models of the same type.
All of them can provide data on:
traffic volumes;
average speeds.
Some models (mainly microscopic models) provide additional data on speed distributions and/or
accelerations.
II.2.2.3 Methods of data collection
Road Traffic flow data
There are many different ways to obtain traffic flow data. Inductive Detector Loops (IDLs) are the current,
de-facto standard for in-situ detectors, and may be expected to provide a considerable volume of
information for mapping. Other technologies, whilst not yet routinely deployed, provide comparable
performance to IDLs. Almost all reviewed technologies may prove adequate for noise modelling purposes
with some minor caveats regarding deployment location, performance in inclement weather conditions and
cost of operation. The use of video techniques and fusion of multiple sensors are rapidly developing areas.
Regarding in-vehicle systems, the operations of commercial service providers and continuing
advancements of satellite tracking technologies offer the possibilities of collecting journey information
across extensive areas of the continent.
There is a growing trend for traffic (and other) information from both main roads and in agglomerations to be
checked processed and stored in unified databases. These databases are housed at Traffic Management
Centres (TMCs) or Urban Traffic Control Centres (UTCs) and offer analysis of long-term traffic patterns, and
the potential for continual validation of mapping exercise results.
It is recognised that, by the time the HARMONOISE methods are first used in practice, considerable
experience will have been developed in handling large volumes of traffic information from the interim round
of mapping (2007). This experience will add to knowledge gained through the continued operation of air-
quality management systems and supplement additional traffic information published by relevant authorities
or agencies.
Technical report 2.4 - Collection Methods for Additional Data provides more information on methods to
collect traffic flow data.
II.2.2.4 Links between road traffic modelling and other IMAGINE Work Packages
When traffic and noise modellers discuss potential improvements to the traffic modelling process, to obtain
the best possible data for the noise model, they will need to weigh costs and benefits. In that respect, it is
important to know to what traffic parameters the closest attention should be paid. The following list gives the
traffic parameters in order of importance for the noise modelling process:
vehicle speeds & traffic composition;
vehicle flows;
acceleration/deceleration;
speed distributions;
(data regarding the above parameters on) low flow roads.

Page 25

II.2.3 Rail noise source
II.2.3.1 Introduction
Railway noise comprises the combination of a number of complex sources that can all be dominant under
different modes of operation. At low speed, traction noise, eg the noise from engine exhaust, the engine
carcass, compressors, fans, will dominate. At very high speed, aerodynamic noise, generated at
discontinuities in the vehicle body and at the current-collecting pantograph, is dominant. However, over a
very wide speed range, rolling noise is dominant. The figure below illustrates this.
Not only do these sources have different characteristics, but they also occur at different heights above the
rail head, as can be seen in the following figure.

In the IMAGINE Railway method the rolling noise mechanism is divided into an excitation element and a
noise emission element. The track and wheel roughness, which form the excitation element, are separately
described, and can be summed to represent the combined roughness (taking into account contact filter
effects).
The noise emission is generated by a forced vibration mechanism which transforms the total excitation
(combined roughness) into noise. This mechanism is represented by transfer functions. Distinction is made
between the transfer functions from total roughness to noise for the vehicle and for the track, thus resulting
10 20 50 100 200 300 400
70
80
90
100
110
120
130
Soundpressure level as functionoftrain speed
S
o
u
n
d

p
r
e
s
s
u
r
e

l
e
v
e
l

d
B
(
A
)
Trainspeed[km/h]
Tractionnoise
Rollingnoise
Aerodynamic noise
Total
10 20 50 100 200 300 400
70
80
90
100
110
120
130
10 20 50 100 200 300 400
70
80
90
100
110
120
130
Soundpressure level as functionoftrain speed
S
o
u
n
d

p
r
e
s
s
u
r
e

l
e
v
e
l

d
B
(
A
)
Trainspeed[km/h]
Tractionnoise
Rollingnoise
Aerodynamic noise
Total
0n, rolling: track
0.Sn, rolling: vheel, aero,
traction, brake/curve squeal,
braking
2n, traction
3n, traction
4n, traction, aero
0n, rolling: track
0.Sn, rolling: vheel, aero,
traction, brake/curve squeal,
braking
2n, traction
3n, traction
4n, traction, aero
Page 26

in separation of the contributions of vehicle and track. The separate contributions are summed to give the
total rolling noise.
Rail head roughness, wheel roughness, and the contact filter at the wheel/rail interface, as well as vehicle
speed, are therefore required in order to predict these rolling noise values from vehicle and track. The
roughness values can be measured, but if this is not possible, default values both for roughness and
transfer functions are available within the supplied guidance document.
Traction noise tends to be independent of train speed, but there are other causal parameters that can be
used to predict this element of railway noise. For example, fan noise is dependent on fan speed and diesel
exhaust noise is dependent on engine speed. Therefore, a traction noise model, comprising a set of
algorithms, has been created within the project to enable source term predictions to be made. Default data
is provided for traction noise elements, including default speed-dependency factors.
Aerodynamic noise is also divided between the element centred at the bogie area, and that at the
pantograph, via measurement techniques outlined in the guidance document.
The main deliverable for the rail source method is D12/13 ref. [60]. It contains guidance on the database
framework including example data and default data, together with guideline documentation on data
acquisition, storage and retrieval, and on the aggregation of individual sources into a traffic model for
interfacing with the Engineering propagation method. This document also includes guidelines on how to
model situations and sources that are not straightforward main line railway scenarios, together with advice
on the modelling of light rail and trams.
II.2.3.2 How to determine the sound power level
The fundamental requirement of the rail source database is to provide spectral sound power level inputs to
the propagation method from each potential source and at each potential source height.
The data output from the database for all of these elements is in terms of sound pressure level. The input to
the Engineering Propagation Model must be, however, in terms of a source sound power line. It is therefore
necessary to define this source line in terms of start and end points for each source height, over a section of
railway that is acoustically homogeneous.
Each acoustically homogeneous source line is defined by its height, the type and number of vehicles
passing along it in a defined time period, the speed and operating mode of the vehicles, the type of track
and track-supporting structure (only for h=0.0m) and the wheel and rail roughness. These roughnesses are
directly measurable with specialist equipment, but it is often not practical to do so for wide-scale noise
mapping. The combined effective roughness (logarithmic sum of rail and wheel roughness including the
contact filter) at the wheel/rail interface is also measurable indirectly by measuring track vibration at
specific locations, but again this is not very practical for mapping purposes. Therefore it will often be
necessary to draw on the default roughness data provided in the database to provide this information.
Expressions are given in the accompanying guidance for converting sound pressure levels at 7.5m, as
extractable from the source term database, into sound power per unit length, taking into account directivity
as appropriate.
Page 27

Data is acquired in a standardised manner, and logged on a supplied spreadsheet format, so that it can be
automatically input to the database, as shown in the following Figure:
The spreadsheet not only allows automatic import of the data to the database, but also acts as a prompt to
data-gatherers to ensure that all essential data and metadata is acquired and logged during measurement
exercises.
The data is output from the database via an export function.
Page 28

II.2.3.3 Rail source input data
The input data required for rail noise source modelling, as acquired in a standardised way and stored within
the rail noise source database, is as follows:
Noise data on vehicles and track
For rolling noise transfer function between combined effective roughness (including contact filter)
at wheel/rail interface and the rolling noise contribution from the track and from the vehicle. Details
of track type (rail, sleepers, support structure, ballast, slab etc) and vehicle type (usually defined by
its class from which can normally be inferred all required detail on braking type, wheel type and
diameter, axle number and spacing etc)
These transfer functions should include special cases of elevated structures and non-standard
track constructions
Combined effective roughness obtained either by measuring track vibration during pass-by (IDR
method) or by measuring roughness directly or by using default values
This roughness should also include the roughness effect of discontinuities such as track joints
Traction noise data (diesel exhaust, fans, compressors, braking)
Rolling noise enhancement due to braking
Squeal from braking
Squeal on curves
Aerodynamic noise data (pantograph or bogie area)
Curve radius
Traffic flows and behaviour
Vehicles per day, evening, night, over a defined long term time period if necessary (eg annual
averages) broken down by individual vehicle types
Whether accelerating or braking
Traffic speeds
Speed of each type of train over a segment (from line speeds, train performance, discussion with
operators/drivers/train performance experts, GPS, radar, light beams, wheel detectors etc)
Priority Ranking and crucial input data
Rolling noise factors have the biggest impact across typical networks
Diesel traction noise can be a major contributor locally where trains accelerate or are on positive inclines
Aerodynamic noise is the key parameter at very high speeds because of its strong speed-dependence
If relevant data is not available
The key data for rolling noise is rail head roughness. A smooth wheel running on corrugated track can be
20dB noisier than on smooth track. If roughness data is not available it is probably most appropriate to
assume roughness is around the mode of a typical distribution, found to be around 4 dB noisier, for smooth
wheels, than models would normally assume. The default data helps with this by providing a range of
potential roughness spectra. Speed and flow data are less critical, provided estimates can be made that are
Page 29

within approximately 10%. Squeal noise can have a major impact, and measurement is advised wherever
possible, so the default values will only be used as a last resort.
For most other situations (rolling noise, traction noise, aerodynamic noise) choice from the database of
either example data from stock or track similar to that of interest, or of appropriate default data, will usually
provide an adequate compromise.
Standards of quality for input data
The measurement techniques advised for data-gathering are largely based on ISO 3095, which is broadly
accepted as the key standard for railway environmental noise measurement.
In order to produce rail maps via the application of the IMAGINE approach, information (together with
potential sources) is required as follows:
Noise source term data on vehicles and track: Vehicle manufacturers or owners, government,
consultants on behalf of such organisations, the future custodians and maintainers of the IMAGINE rail
source term database
Track physical characteristics: Infrastructure owners, designers, installers or maintainers
Traffic types and flows: Infrastructure owners, train operators
Traffic speeds: Infrastructure owners, train operators, trackside measurements, on-board measurements,
discussion with train performance experts and/or train crew
Train operating modes: Train operators, observation, discussion with train performance experts and/or
train crew
Rail head roughness: Infrastructure owners or maintainers, consultants

II.2.3.4 Methods of data collection/measurement methods
The different sources of railway noise are measured in a variety of ways, all of which are described in ref.
[60].
Rolling noise is acquired via microphones and track-mounted accelerometers, so that the vehicle
and track contributions can be separated (via MISO or PBA or VTN), and also so that the combined
effective roughness (ie wheel + rail + effect of contact filter) can be measured dynamically. This
information allows transfer functions between this roughness and track, or vehicle, contribution to
be calculated and stored as core data in the database.
Traction noise elements are measured using defined ISO3095-based approaches for a variety of
modes of operation.
Squeal noise and braking noise are measured in a standardised manner.
Aerodynamic noise requires a dedicated technique to be applied. Here, pass-by noise level is
measured at a range of speeds, allowing the relationship between rolling noise and speed to be
determined and hence predicted at higher speeds where aerodynamic noise is also a contributor.
The energy difference between total noise and the predicted rolling noise is the aerodynamic
element. Further separation of aerodynamic noise into that which arises around the pantograph and
that which is emitted in the bogie region can be achieved by measuring under identical conditions
with and without a noise barrier in place.
Page 30

II.2.3.5 Links with other IMAGINE Work Packages
During the project it has been important to liaise with IMAGINE Work Package 1, to ensure that the rail
traffic model is fully compatible with the input requirements of the Engineering Propagation model.
Discussions have also taken place with Work Package 3 to ensure compatibility, where appropriate,
between the measurement requirements defined within that Work Package and those of Work Package 6.
It should be noted, however, that the detail required within Work Package 6 is considerably greater than that
considered within Work Package 3.

II.2.4 Industrial noise
II.2.4.1 Introduction
Industrial noise sources represent a large variety of source types, mostly located in industrial sites and
plants. Many industrial sound sources are complex in terms of geometry, sound generating mechanisms
and radiation properties. It is therefore necessary to simplify the real, physical properties into a source
model that can realistically be handled in a calculation process.
The acoustic properties of industrial noise sources like machinery or a piece of equipment producing sound
can usually be defined in terms of four key parameters:
- the "type" of sound source (point, line, area),
- the source height(s)
- the total sound power produced by the source
- the spatial distribution of the sound radiation (the "directivity" of the source)
These parameters, and particularly the last two, are frequency dependent.
The sound power and the directivity usually need to be determined through measurements, but can
sometimes be calculated from theory. Data from previous measurements may be available in literature or in
a data base. But there are many products, variants and versions of the "same" machine, and maintenance
standard can also have a clear influence. The relevance, i.e. the accuracy, of the data obtained from
literature will therefore strongly depend on the quality of the documentation of the original measurements.
II.2.4.2 How to determine the sound power level
To determine the sound power level of an industrial source, a large number of methods exist, ranging from
measurements in an anechoic room to sound intensity measurements. Most methods are based on
measuring the sound pressure level at a certain distance from the source in relatively free field conditions.
These noise levels are used in a formula to calculate the sound power level, together with other input data
like distance to source, ground reflections, etc. The report "Measurement Methods" IMA07TR-050418-
MBBM03 reviews many of these methods. In the methods a distinction is made between
Whole plants (methods to determine the overall noise emissions of whole plants or industries or
parts thereof without having to take measurements of individual noise sources),
Individual noise sources (methods for measuring the noise emissions of specific individual noise
sources), and
General methods (methods that can be applied - in principle - to any type of noise source, at least
within a certain class of sources).
Page 31

The user is then led through a three-level approach to find a suitable measurement method for the particular
problem as shown schematically below:


no
yes
method found
(category A)
"individual source method"
available ?
"whole plant method"
available ?
determine individual sources
no
yes
apply best general method
method found
(category B)
method found
(category C)
no no
yes yes
method found
(category A)
"individual source method"
available ?
"individual source method"
available ?
"whole plant method"
available ?
"whole plant method"
available ?
determine individual sources determine individual sources
no no
yes yes
apply best general method apply best general method
method found
(category B)
method found
(category C)

At the point "method found", the end user will find all the information necessary to decide if the method is
suitable to solve the measurement task as well as the bibliographical data on the respective standard or
guideline in which he/she can find the specific details about the measurement method.
A database has been developed containing more than 1500 entries to assist the user with a comprehensive
source of industrial noise source sound power levels.
II.2.4.3 The input and output
The following information, in the form of sound power data, constitutes a complete set of input data for
sound propagation calculations with the methods developed in HARMONOISE and is to be used for noise
mapping:
Location (coordinates) and elevation of the noise source
Dimensions and orientation
Type of source (point, line, area, volume)
Operating conditions of the source
Emitted sound power level spectrum for each relevant operating condition
Directivity
Working hours (day, evening, night)

Note that not all of the information above is equally important as the impact on the noise mapping results is
different. This becomes important if only limited or no data for individual quantities is available. For example,
while - obviously - knowing the sound power level of the source is crucial, lacking knowledge of the
directivity may still lead to an acceptable noise map as a result of many sources.
Page 32

To assess the sound power levels for industrial noise sources, such as electric motors, pumps and
compressors, fans, furnaces and boilers, coolers, piping and valves, stacks and flares, construction and
building machinery, and many others, the following approaches can be used:
Determination of the noise sound power level of the sources from measurements near the source or
further away in the field.
Noise data obtained from the manufacturer of the noise source in question.
A database with default data for typical sound emission values for individual industrial noise sources.
Since the large variety in industrial noise sources and particularly in their individual state of noise
control, does not allow a database of generic validity to be set up, the data provided in such a database
will inevitably need to be limited to a representative number of typical examples.
Prediction/computation using parameter formulae usually based on the principle of acoustic efficiency.
Different classes of noise reduction, e.g. good, average and poor can be distinguished.
Default values for the typical sound power level radiated by a specific industrial activity per unit area,
taking into account classes of sound reduction applied. This is possible for many industrial activities, for
example petrochemical plants, power plants, ship yards, timber plants, etc.
The best way to gather the relevant sound power data is to take noise measurements at each source. Apart
from determining the sound power, information can be gathered about location, dimensions, operating time
and conditions. If there is not enough time to measure the relevant sound power data, the sound power
levels might be obtained from manufacturers of the equipment, for instance the CE-label. These sound
power levels are of course for very strict operating conditions, probably described by the measurement
method used. In addition to this data information such as the location and working hours must be obtained.
If the manufacturers cannot supply these sound power levels, one can use data from the industrial source
database, called SourcedB.
Like other calculation schemes, the same data is expected: topography such as building height lines and
meteorological data, etc.
The priority in making noise maps is as follows
1 sound power levels
2 working hours
3 topography
4 meteorological situation
Global information on working hours and the way to calculate the yearly averaged working hours can be
found in ref. [72]. The output of the propagation modal is validated in ref. [73].
Calculations and measurements were made to give input for the propagation. Next this three weeks of noise
measurements were carried out in France near Clamart, The main conclusion is that the difference between
the two is in 75% of the studied cases smaller then the measured standard deviation.
The main reasons for differences in measured and calculated values are:
Incomplete knowledge of micro-meteorological phenomena; e.g. similarity theory and determination of
Monin-Obukhov length do not work well for low wind speeds. Therefore, we get the input data wrong;
Classification of meteorological effects by a single parameter. For some values of the independent
parameter, the measurements show a large spread. This suggests that the single parameter is not able
to take into account the full complexity of reality. The spread is strongly reduced in the
HARMONOISE/IMAGINE calculations that aim to produce the averaged effect for each situation. Under
Page 33

such situations, calculations should only be compared to averages over sufficiently long measurements
periods;
P2P is validated using Reference Model as a golden standard. But reference models do not take into
account the full complexity of reality; e.g. the assumption of a completely stratified atmosphere with
range-independent wind and temperature profiles is an idealisation of reality.
Some of these limitations are:
Horizontal line of measured values around 55 dB is mainly due to inaccurate estimation of the sound
speed gradient. Theory underlying the estimation of sound speed gradient goes wrong in cases where
there is little or no wind.
Larger spread for lower levels: this is due to turbulent scattering, which has not been taken into account
in the model.
II.2.4.4 Measurement methods
WP7 has developed a database describing many measurement methods. These measurement methods
together with report "Measurement Methods" ref. [71] give guidance on how to determine sound power
levels. The methods are divided into single source measurements, whole plant methods or generalized
methods. If no time is available for making noise measurements, the next option is to use the IMAGINE
source database, called SourcedB. Many sources are covered in this database. The program SourcedB can
be downloaded from the IMAGINE internet site, including the report "Description of the source database"
ref. [72].
II.2.4.5 Links with other IMAGINE Work Packages
Driving with a truck on the road is considered as road noise, while driving on an industrial site becomes
industrial noise. The same holds for railway yards and taxiing of a plane on an airfield. So many parts of the
outcome of other WP's can be used in WP7. WP1 is useful for gathering information for making a model
while WP3 can be used to make validation measurements.

II.2.5 Aircraft noise
II.2.5.1 Introduction
Different types of aircraft noise models currently exist, which differ in their level of complexity in modelling
aspects of sound emission and propagation.
The modelling principle of integrated (segmentation-based) models like INM ref. [39] , which is also fully
described in ECAC Doc.29 ref. [40], combines source characteristics and propagation in the well known
Noise-Power-Distance (NPD) data. These provide immission levels at specific distances perpendicularly
beneath the flight path. Source longitudinal directivity effects are implicitly included in the SEL-NPD data by
the sound integration over the whole flight event.
Another modelling approach is simulation. Simulation models developed during the 90s have proven to
have advantages in complex situations. In these models, the sound source and the propagation model may
be treated separately. In addition, the sound source directivity is also present.
The advantage of the IMAGINE method is that it treats the sound source and the propagation models
separately. As described in ref. [46] to increase accuracy and versatility of aircraft sound calculations, a
more complex description of aircraft sound source is needed than the ones used mostly today. The sound
Page 34

radiation characteristics of an aircraft vary both in level, frequency contents and directivity during a flight.
Therefore the IMAGINE method is be based on 3D directivities for the sound source.
II.2.5.2 How to determine the sound power level
If possible source sound power levels are deduced from measurements of fly-bys of aircraft, either in the
context of certification or from other well controlled measurements. If measurement data are not available,
NPD curves may be analysed using the procedures of SAE 1845 to estimate a source spectrum.
For measurements of fly-overs the next points describe the process:
The aircraft is treated as a point source positioned consecutively at discrete points along the flight
path
The sound emission characteristics represent free field conditions
The aircraft is in motion (in-flight conditions, not data from test stands)
Levels are only specified outside the immediate vicinity of the source (i.e. in the far field), that is
there is no distinction between different sound source components of the aircraft (e.g. location of
individual engines or different sources like fan inlet, fan exhaust, jet exhaust etc)
Doppler effects are included in the source description as part of the apparent directivity


The relevant geometric values are: distance, emission angles theta and phi and at the receiver the
immission angle beta (see Figure 2-2). For finding the specific location on the flight path the geometry has
to be linked with the corresponding 1/3 octave spectrum, based on the time stamps.
For further explanation the reader is advised to read the IMAGINE deliverable D9 ref. [37] for a description
of the flight test measurements carried out in Spain, or Deliverable D10 ref. [38] for a description of the
aircraft source method.

For the estimation of the sound power spectrum L
w,dir
from Noise Power Distance data the steps are as
follows:
calculate the level-time history of a flyover
optimise the source spectrum to produce the best fit for the SEL values for all 10 distances of the
NPD
Page 35

convert the optimised spectrum to a linear spectrum by removing the A-weighting
subtract the ground interference from the source spectrum
convert the spectrum of L
p
(1m) to sound power L
w,dir
, by adding 11 dB
account for the number N of engines by adding 10* lg (N) to the result
The resulting source spectrum reflects average sound emission close to L
ASmax
, the maximum value during
an aircraft flyover of the A-weighted sound pressure level. Neither the variations from fan-noise (approach)
to jet-noise (leaving) nor the Doppler shift are accounted for. Due to lack of information it is assumed that
the spectrum at L
ASmax
applies for all longitudinal directions. These assumptions result in a lack of accuracy,
but apart from the real IMAGINE method described earlier, this is the best result obtainable from available
data.
II.2.5.3 The input and output

The source module for IMAGINE has the following input parameters to generate the requested output of the
sound power spectrum:












specific location on the flight path. For this point, the geometry has to be linked with the corresponding 1/3
octave spectrum.

The sound emission of the aircraft is described by the sound power spectrum L
w,dir
which results from the
combination of the following input parameters:
aircraft type, including specification of the engine types
operation (departure or arrival)
especially for approach, the position of flap, slats and gear, which have an influence on airframe
noise
noise-related engine power parameter (thrust or equivalent)
speed (optional parameter, which influences the airframe noise)
Page 36

longitudinal angle : angle between the longitudinal axis of the aircraft and the direction towards the
receiver
lateral angle : angle between the vertical plane (perpendicular to the wing plane) and the direction
towards the receiver.
The type of propulsion system is one of the main factors. Jet noise (or turbofan) is one of the main causes
of noise pollution in the communities around airports; however the non-propulsive noise or airframe noise
also plays a significant role, especially during approach and landing.
With a clean configuration, the airframe noise is mainly produced by the main wing structure and is mostly
broadband in nature. The airframe noise increases during the approach phase of operation due to the
turbulence induced by fuselage, tail plane, wings, flaps and landing gear. They combine together and
exceed the level of the engines running at low power. The overall airframe noise can increase on the
landing approach by about 10 dB ref.[47]. In airframe noise, the landing gear noise is the most important; it
is broadly spherical in directivity and has spectral characteristics slightly higher in frequency than those for
clean airframe. In general, the airframe noise produced depends on the dimensions of the aircraft and the
speed of the airflow over the structure.
Notes:
There are typical combinations of input parameters for the most common flight operations (example:
full landing configuration with gear down is normally associated with a low - final - approach speed)
There may not be available data to characterise all operational states of the aircraft.
The list of input parameters may not be complete.
Ground operations like taxiing and operations at the terminal are not covered by the IMAGINE
aircraft method, but in principle they can also be calculated, provided the appropriate source data
are available
For immission calculations, only a few typical combinations of speed, thrust, flaps and gear are
needed, e.g. for departure sound emission is dominated by the engine operation.
Intermediate thrust levels may be interpolated as for NPD data.

II.2.5.4 Methods of data collection/measurement methods
See the ref. [37] for a description of the flight test measurements carried out in Spain, or ref. [38] for a
description of the aircraft source method and conversion of NPD data to IMAGINE source data..

II.2.5.5 Links with other IMAGINE Work Packages
For long term aircraft noise measurements, the measurement method of L
den
and L
night
described by Work
Package 3 are relevant.
Work Package 7 includes airports as an industrial source. This encompasses noise from ground vehicles,
including passengers cars, auxiliary power units (APUs) used while an aircraft is on stand, and aircraft
taxiing operations.
Work package 1 does not include aircraft in its mapping guidelines. Aircraft noise mapping methodology is
defined in ECAC Doc. 29 ref. [40].
Page 37

II.3 Propagation
Below is the general flow diagram for the propagation phase of the HARMONOISE/IMAGINE method.

Notes
(1)
Thematic data includes geometrical shapes (e.g. centrelines of roads and railways, footprints of buildings)
and other attributes (width of the road, nature of the carriageway, height of the building, number of
inhabitants,). Operating conditions and traffic data is part of the source models.

Build geometrical 2.5D model
Assign material properties (impedances) to ground and walls
Create point, line and surface sources
Assign source properties (sound power and directivity)
Collect geometrical / thematic data
(1)
for terrain, buildings,
screens, road and railway infrastructures, industrial sites,
Input from
source models
for road, rail,
aircraft and
industry
Select receiver points / grid
Calculate L
w,k
for k = day, evening and night

Meteorological
data records for a
relevant year
Calculate L
geo
(3)

Calculate averaged excess attenuation
L
excess,k
for k = day, evening and night
Calculate partial contribution for this propagation path
(5)

L
eq,k
= L
w,k
+ L
geo
+

L
excess,k
for k = day, evening and night
For each source/receiver combination
(2)
- decompose source in segments and/or equivalent point
sources
- search direct, reflected and laterally diffracted
propagation paths

Sum partial contributions for k = day, evening and night at the
receiver. Keep track of levels per sources/per source type
(4)

Choice of day,
evening and
night hours &
choice of
relevant year
For each receiver and each source type,
calculate L
DEN
and L
NIGHT
.
Page 38

(2)
This is in fact a triple loop over receivers, sources and possible propagation paths. The exact order of
execution of the loops depends on implementation in software (ref. [3] details). Propagation paths include
direct, reflected and possibly laterally diffracted paths. By definition, a propagation sector is a sector in
which the propagation conditions are homogenous (i.e. the excess attenuation doesnt vary too much), but
L
w
and L
geo
do not need to be constant in each sector.
(3)
By definition L
geo
= L
p,free
- L
w
. For a point source L
geo
= D(,) -10 log (4d), where D is the directivity
of the source in the direction of the receiver and d the propagation distance. For extended sources, adapted
formulae rely on integration over the source segment or surface. Numerical and analytical formulae (large
angle integration) or approximations (small angle integration and point source decomposition) are possible
implementations. See ref. [3]. Accuracy of each proposed solution should be demonstrated by means of
benchmarking, by comparisons with the exact solutions or by convergence tests.
(4)
The Directive requires the production of separate maps for each source type (road, rail, aircraft and
industry). This also applies to agglomerations. Summation of noise levels for different types of sources is
not covered by the current version of the Directive. For the preparation of action planning it may be better to
keep track of partial contributions per source (i.e. for each road/railway section, separation of highways/local
roads, high-speed/passenger/freight trains, per aircraft type, etc.).
(5)
A-weighting and frequency band integration can be done at this level, i.e. to save memory storage. It may
also be delayed till the last step of this procedure, i.e. keep track of spectral information about the total noise
exposure.
II.3.1 The propagation method
The HARMONOISE project provided two kinds of propagation method: a reference model based on the
numerical solution of the wave equations and an engineering model based on analytical formulae and
heuristics. Both models were validated against experimental data. The numerical models were then
intensively used to provide numerical reference results against which the engineering model could be
further validated and fine-tuned. The IMAGINE project followed a similar approach. One of the main
objectives was to extend the domain of application of the propagation models to the particularities of aircraft
and industrial noise. Special attention was given to higher sound sources, larger propagation distances and
to lateral diffraction by obstacles close to the source.
The numerical propagation models developed by the HARMONOISE project implemented the Boundary
Element Method and the Parabolic Equation formulation. BEM was used to describe complex geometry in
the vicinity of the source, PE for the propagation at longer distances taking into account refraction effects
due to meteorological stratification. Both methods are highly time consuming and could only be applied to
relatively simple geometries with a high degree of symmetry and for sources close to the ground.
In the IMAGINE project, two new reference methods were introduced and validated. A numerical ray tracing
method has been developed in order to extend the PE methods for higher sources. The method was
validated both numerically and experimentally during the measurement campaign in Ocaa. Special
attention was also given to the characterisation of the atmospheric conditions at greater heights. For the
Industry propagation method an analytical ray tracing method was used based on the Uniform Theory of
Diffraction. This method allows the detection of direct, reflected and diffracted ray paths in a complex three
dimensional scene. The model was validated numerically by means of a 3D BEM code before being
compared to an experimental setup.
Page 39

The HARMONOISE project developed and validated an engineering propagation model applicable to typical
road and railway source configurations; i.e. for sources close to the ground and propagation distances up to
1 km. This model also needed adaptations in order to cover the extended domain of application and the
specific situations related to aircraft and industrial noise. The main adaptations and improvements of the
engineering propagation model include:
The determination of the transition frequency between the low and high frequency calculation of
ground reflections was fine-tuned to allow both octave and third-octave band calculations.
Following the findings of the Nord 2000 project, a frequency dependent Fresnel factor was
introduced. This provides a more realistic behaviour of the model in the case of low sources, large
propagation distances and stronger sound speed gradients.
Conformal mapping was introduced as a more rigorous derivation of curved ground analogy.
Conformal mapping applied to the wave equation removes the sound speed gradient and allows
the use of the homogeneous propagation model over the transformed ground profile.
Loss of coherence was introduced in order to smooth out strong interference effects that are
considered an artefact of the point source and frozen atmosphere approximations. The model
accounts for uncertainties in the source and receiver positions, for the extended size of the source,
for variations of the sound speed and for frequency band integration.
Loss of coherence and scattering due to turbulence were introduced in a simplified and
comprehensive manner; both depend on a single parameter that characterises the turbulence
strength.
The determination of an equivalent linear sound speed gradient from the more general lin-log
approximation has been adapted to the case of higher sound sources. Good agreement was found
for moderately high sources; i.e. industrial sources up to 10m above ground. For aircraft sources
some systematic differences between engineering and reference model still persist. It was found
that there is no clear correlation at favourable propagation conditions and higher noise levels for
sources more than 25m above ground. As a consequence, the effect of meteorological conditions
on long-time averaged noise levels from aircraft sources is not fully answered by the IMAGINE
project.
The HARMONOISE propagation models use specific meteorological conditions: i.e. wind speed, wind
direction and an indicator for thermal stability of the atmosphere. For validation purposes, the
HARMONOISE project relied on 120 meteorological classes (8 wind directions, 3 wind speed classes and 5
stability classes). The IMAGINE project developed a methodology that reduced the complexity of this
problem to the calculation of 4 propagation classes, defined in terms of equivalent linear gradients. This
classification of propagation conditions applies to prediction models as well as to measurements. From
either or both, long term averaged L
den
and L
night
values are obtained by considering the frequency of
occurrence of each propagation class over a (climatologically) relevant year.

Page 40





Figure: validation of the engineering propagation model at EDF nuclear site in Clamart (left).
measured and calculated excess attenuations versus equivalent linear sound speed gradient.
II.3.2 Propagation paths
It was a initial choice of the HARMONOISE and IMAGINE teams to limit the determination of propagation
paths to a 2.5-dimensional description of the real world. In application of this modelling solution, propagation
paths are first determined in a 2-dimensional projected view of the site and the third dimension is added
only to calculate the excess attenuation along these paths (see ref. [10]). Propagation paths can be
classified as direct (when the source is in view of the receiver), reflected by (vertical) obstacles or laterally
diffracted (around corners of vertical obstacles).
The construction of direct, reflected and diffracted propagation paths is usually at the very heart of a noise
mapping software package. Historically, two families of strategies have been developed to tackle this
problem: image methods and discrete ray tracing. Image methods provide exact answers for the optical
paths connecting sources and receivers but become highly time consuming when higher orders of
reflections are to be taken into account. Discrete ray tracing techniques are generally faster and better
suited for the detection of propagation paths between point receivers and extended sources, but suffer from
false hits and missing paths, introducing some random variations in the results. More recently, beam
tracing methods have been developed that combine faster calculations and exactness of the results. These
techniques are common to other branches in computer graphics and their implementation and optimisation
is the object of intensive research in that domain. For more details on path detection algorithms, see ref. [3].
The IMAGINE team recognises that the choices made by independent software developers with respect to
the path detection problem are currently one of the main causes of the lack of reproducibility of noise maps,
at least as important as the lack of common source and propagation models. However, path detector
algorithms are considered of strategic importance (or equivalently, as a commercial argument) by the
software editors. In view of this, it does not seem advisable to impose a unique and harmonised solution for
path detection. Moreover, standardisation of common practice may hinder future innovation benefiting from
ongoing research in computer graphics and virtual reality applications. Guidelines for source segmentation
and for implementation of loops over sources and receiver only make sense in the framework of a specific
path detection method and must not be generalised to other (existing or future) technologies.
To promote improved reproducibility without imposing mandatory implementation details, the IMAGINE
project has explored two approaches. On one hand, a generalised formulation is given for the definition of
homogeneous propagation sectors and for the integration over extended sources. On the other hand,
harmonised requirements have been established with respect to the data model used on input of the path
detectors. In this sense, the geometrical modelling and the path detection problem are described in terms of
desired functionality and accuracy rather than in terms of (mandatory) implementation details. Finally,
Page 41

possible simplifications of the data model have been considered taking into account the levels of
applications already identified in ref. [9]. In fact, it seems reasonable to impose less stringent modelling
requirements for strategic noise mapping tasks than for expert usage of the model by scientists; noise
engineering task such as the preparation and the evaluation of action plans fall somewhere in between
these two extremes.
II.3.3 Source segmentation
The propagation from an extended source to a receiver point is based on the construction of homogeneous
propagation sectors; i.e. sectors that can be described by similar topography (see HARMONOISE D16 and
D18). Within each propagation sector, the transfer function between the source power and the level at the
receiver is decomposed into two parts: L
geo
and L
excess
. The term L
excess
includes the effects of ground
reflections, diffraction by horizontal edges and meteorological refraction; it is typically calculated using the
point-to-point propagation modules. The term L
geo
includes the spherical spread, the directivity of the
source and the integration over the extended sources (which may be lines, surfaces or volumes).
From the theoretical point of view, L
geo
is defined as an integral, see ref. [10]. Implementers of the methods
need to estimate this integral either analytically or numerically. The implementation and the level of
approximation will depend on the choice of a particular path detector technology and on parameter settings
by the end-users. The accuracy of different implementations can however be examined by benchmarking,
convergence tests or by comparison against well known analytical solutions. As a result of these
comparisons, software developers may recommend specific parameter settings to be applied by end-users,
taking into consideration the expected accuracy and the foreseen exploitation of the results.
II.3.4 Geometrical data model
The path detection algorithms operate on a geometrical model that is a simplified description of reality. This
model must fit the assumptions of the 2.5D propagation model. Ref. [3] shows how this geometry can be
decomposed into a set of connected surface elements so that their projections form a partition of the
domain under study. The surface elements can be classified as vertical and almost horizontal. The
projections of vertical elements on the horizontal plane are represented by line elements. Such elements
are used to model barriers, parapets, retaining walls and external walls of buildings. Within the path
detection algorithms these elements are used to construct candidate reflected propagation paths. The
connection between two vertical surface elements is a vertical edge which is projected into a single point.
Such points may generate laterally diffracted propagation paths. The almost horizontal surfaces are used
to describe the terrain, carriageways, rail beds, as well as for the roofs of buildings. These elements do not
interfere with the determination of propagation paths but enter into the calculation of the propagation effects
as they are intersected by the vertical plane corresponding to the propagation path.
The requirements for the geometrical model can be expressed in three ways:
Criteria for inclusion or omission of elements in the model;
Choice of control points (i.e. points defining the simplified shape of the element);
Positional accuracy of the control points
Within the IMAGINE project, the HARMONOISE propagation method has been used to check the sensitivity
of the results with respect to variations and simplifications of the geometrical input parameters. Results were
expressed both as noise levels and in terms of numbers of exposed populations. As an outcome,
geometrical modelling requirements were related of the desired accuracy of the results. The results of this
study are summarised in the table below. The criteria apply to the accuracy of horizontal and vertical
Page 42

coordinates of control points, to the minimal size of geometrical details to be included in the geometrical
model, and to possible simplification of geometrical shapes.

Level of detail Position accuracy for geometrical items
High Medium Low
Horizontal position 1 m 2 m 5 m
Vertical position in general 0.5 m 1 m 2.5 m
Height of obstacles near low sources 0.25 m 0.5 m 1 m
Height of other obstacles 10 % 20 % 50 %
Typical accuracy of L
propagation
< 0.5 dB < 1 dB < 3 dB
positional accuracy requirements for the geometrical model

The geometrical model needs to be completed with acoustical properties. For source elements this
corresponds to the sound powers and directivities as calculated with the HARMONOISE and IMAGINE
source modules for road, rail, aircraft and industrial noise. Moreover, all surface elements must be assigned
acoustical impedance values. For the terrain, the ground classes as defined in the Nord2000 were found
sufficiently precise for most engineering applications. For strategic noise mapping a simplified method has
been proposed based on the overall land cover in the area of interest. For man-made obstacles like
buildings and screens the IMAGINE project has worked out a classification scheme based on EN 1793-1
standard, ref. [4].
II.3.5 Input data and GIS
End-users will rarely be confronted with the basic tasks of building the physical data model. Building the
data model is mainly implemented by application software. From the end-users point of view, the data is
organised in themes; i.e. collections of items from the real world that have similar properties and
behaviours. From a point of view of software analysts and developers, themes have uniform symbolic
representations and data structures. For noise mapping applications, themes include the natural terrain,
road and rail infrastructures, buildings, barriers etc. Application software is responsible for transforming the
thematic data into the physical model by constructing the appropriate geometrical items and by assigning
their acoustical properties.
Links to the domain of Geographical Information Systems are made at this thematic level. The IMAGINE
project has investigated this link from two points of view: on one hand, well formalised end-user
requirements for the domain of noise mapping can be set up by applying the techniques and principles of
geo-spatial data modelling (see ref. [3]) on the other hand available geo-spatial data is one of the main
sources of input into noise mapping projects and the best way of using these existing data have been
investigated, ref. [4].
Formal requirements and clear product specifications may form the basis for checking the usability of
available data sets with respect to a noise mapping project.
The world of GIS is rapidly changing and becoming more and more subject to international standards. The
main goal of these standards is to provide a common framework for the formal expression of end-user
requirements and product specifications to be used by data producers and consumers. In IMAGINE ways
were defined for expressing the abstract specifications of the IMAGINE model in terms of these emerging
standards, ref [3] or [4]. In the future, noise engineers and software developers may use the findings of
Page 43

IMAGINE as a starting point to establish common standards for the contents and the structure of input data
sets to be used in conjunction with the HARMONOISE/IMAGINE models, comparable to the ongoing
development of common specifications in the field of Virtual Reality.
The growing market for geographical information stimulates innovation and new technologies for the
production of high quality data at lower costs. Some of these new technologies and their application to noise
mapping were reviewed in the IMAGINE project.




Page 44

II.4 Determination of long term L
den
by calculation
II.4.1 Introduction
In the HARMONOISE project it was shown how noise levels are influenced by meteorological conditions
such as wind speed, wind direction and thermal inversion effects. For the determination of long term
averaged noise levels, short term levels, valid under specific meteorological conditions, must be weighted
according to their frequency of occurrence. Even in the case of long-term monitoring some corrections may
be required in order to take into account seasonal variations of traffic flow or atmospheric absorption. The
IMAGINE project worked out an operational scheme linking the HARMONOISE propagation model to a
statistical (climatological) description of local meteorology.
II.4.2 Meteorological classes
For validation purposes, the HARMONOISE project relied on 200 meteorological classes (8 wind directions,
5 wind speed classes and 5 stability classes). In order to limit the calculation efforts needed for the
estimation of long term averaged noise indicators, it was proposed to reduce this problem by sampling all
possible lin-log sound speed profiles down to 25 propagation classes.
The final aim of the engineering methods, however, is to provide a pragmatic and unified prediction scheme
to be used for noise mapping according to END. In view of this, the IMAGINE project developed a
methodology that further reduced the complexity of the problem to the calculation of 4 propagation classes,
defined in terms of equivalent linear gradients. This classification of propagation conditions applies to
prediction models as well as to measurements. From either or both, long term averaged L
den
and L
night

values are obtained by considering the frequency of occurrence of each propagation class over a
(climatologically) relevant year.
In order to calculate a value for the long-term L
den
it is first necessary to be able to evaluate the propagation
of noise levels for a variety of meteorological conditions and then determine the frequency of occurrence of
those meteorological conditions over a yearly period. In certain circumstances particular meteorological
conditions will also affect the source level e.g. winter conditions requiring the use of chains on road vehicle
tyres or the operation of aircraft in different wind conditions. In these cases further stratification of the
measurements is recommended.
II.4.3 Classification variables
For the development of this methodology, it was first noted that, in any given situation, the importance of
meteorological effects on sound propagation can be assessed by the single dimensionless
parameter R D/ , the ratio of the (horizontal) propagation distance relative to the equivalent curvature
parameter as defined above. Using this parameter, the meteorological effects (excess attenuation relative to
free field) may then be quantified using a limited number of propagation classes, as indicated in the table
below.
Page 45


Propagation
class
1)
D/R
Range
D/R
Representative
value
Verbal
description
Typical values of wind
speed component
2),3)

m/s
M1 < - 0.04 -0.08 Unfavourable < 1 (day)
< -1 (night)
M2 -0.04 0.04 0.00 Neutral 1-3
M3 0.040.12 0.08 Favourable 3-6
M4 > 0.12 0.16 Very favourable > 6 (day)
4)

> -1 (night)
Notes:
1)
Propagation Class M0 has been removed from the propagation class calculation because it is very
unfavourable and unlikely to influence the year long average noise level.
2)
The wind speed component is the projection of the wind speed vector on the direction of propagation.
Negative values represent upwind conditions, blowing from the receiver to the source. The wind speed
should be determined at 10 m high in a sufficiently large open space between the source and the receiver.
3)
The values given in the table are only indicative and were obtained assuming moderate thermal effects.
Clear sky conditions may change the estimation one class up (night time) or down (day time).
4)
It may be difficult to carry out reliable measurements under such conditions.
II.4.4 Calculation of D/R
For the calculation of the equivalent noise levels per meteorological class see ref. [78].
II.4.5 Calculation of long term L
den

Once the equivalent noise level in each meteorological class has been determined, these values shall be
combined together based on the frequency of occurrence of each class during a relevant average year:
10 /
5
1
,
10 log 10
i eq
L
i
i den
p L

=
=
The frequency of occurrence of each propagation class can be derived from meteorological data records,
including wind speed, wind direction and cloud coverage (or other ways for determining stability) over a
sufficiently long period. Except in mountainous area where local effects may be dominant, the frequencies
of occurrence are stable over large areas and data from the nearest weather station or airport can be used.
It may even be possible to provide generic values that can be used within the main climatological regions
covering Europe. Frequencies of occurrence must be determined separately for day, evening and night
periods. More information can be found in report ref. [78] and ref. [28].
Page 46

II.5 Determination of L
den
and L
night
using measurements
II.5.1 General
The IMAGINE measurement method describes how to determine L
den
and L
night
, as defined by the European
directive 2002/49/EC, by direct measurement or by extrapolation of measurement results by calculation.
The measurement method is intended to be used outdoors as a basis for assessing environmental noise in
individual cases. Typical applications would be measurements to support the credibility of calculations, to
validate calculations in well-defined situations and to assess situations not properly covered, or believed not
to be properly covered, by the available prediction model. The method can also be used for monitoring
purposes.
II.5.2 Description of the method
The method is flexible and to a large extent the user determines the measurement effort and, accordingly,
the measurement uncertainty, which has to be determined and reported in each case. Often the
measurement results have to be combined with calculations to correct for operating or propagation
conditions different from those during the actual measurement. In each case the long term equivalent sound
pressure level is determined by taking into account the frequency of occurrence of the different operating
and propagation conditions. For each of these conditions the sound pressure level is measured or
calculated. The method can be applied to all kinds of noise source, such as road and rail traffic noise,
aircraft noise and industrial noise.
The method distinguishes between long and short term measurements although there is no clear distinction
between the two. Ideally a long term measurement encompasses such a long measurement period that the
measured data taken during this period can be stratified into different samples in such a way that all
operating and meteorological conditions are represented. These samples can then be combined and added
in such a way that the equivalent sound pressure level during an average year is obtained. A short term
measurement on the other hand will only include a very limited number of measurements and some
operating and meteorological conditions will not be represented at all. Instead missing data are calculated
by applying corrections to the few data measured. These corrections are calculated by using a prediction
method. In practice some corrections always have to be applied even for a long term measurement.
The preferred method is long term measurements, which means that measurements are carried out for
several weeks or months. The meteorological conditions are monitored during the measurements and the
measurement results are stratified into 4 different meteorological classes. These 4 classes are determined
using the same procedures as those described for sound propagation calculations.
If the number of statistically independent measurements is small it may become necessary to make a
further stratification with respect to source operating conditions. Guidance is given on how to achieve
independent measurements. The measurement values are then corrected to reflect the yearly average of
the source and meteorological conditions. Measured values are also used to estimate the measurement
uncertainty.
Alternatively a limited number of short term measurements is made. In such cases also the source
conditions have to be monitored and each measurement has to be corrected to represent the yearly
average of the source conditions. In the case of short term measurements, the measured values are not
always sufficient to determine the measurement uncertainty which is instead is determined from default
values and simplified formulas.
Page 47

The method provides detailed guidance on how to determine the measurement uncertainty following the
general guidelines given by the ISO GUM. Numerous equations and default values are given. Calculated
examples are given in an annex.
II.5.3 Input and output of the method
Necessary input is the frequency of occurrence of different meteorological conditions, and, in the case of
short term measurements, information about source operating conditions. Output is L
den
or L
night
including
information about measurement uncertainty.
II.5.4 Links with other IMAGINE Work Packages
WP 3 is linked with other work packages as follows:
WP 1 handles input data for noise mapping and different ways to get these values. Most of these
input data are the same as those required to process measurement results to get L
den
or L
night
. Thus
the identification of data needed for measurement is included in the work of WP1.
WP 2 deals with flow management of road traffic noise. Some of these results can be used to
estimate the measurement uncertainty of road traffic noise measurements.
WP 4 deals with air traffic noise. Some of these results can be used to estimate the measurement
uncertainty of air traffic noise measurements
WP 5 deals with noise emission of road vehicles. Some of these results can be used to estimate the
measurement uncertainty due to uncertainty in the operating condition of road vehicles.
WP 6 deals with noise emission of rail vehicles. Some of these results can be used to estimate the
measurement uncertainty due to uncertainty in the operating condition of trains.
WP 7 deals with industrial noise. Some of these results can be used to estimate the measurement
uncertainty due to the operating conditions.
Page 48

II.6 Factors affecting the quality and accuracy of end results
II.6.1 Road Noise Source
The quality of the end result will be mainly determined by the available input data. The amount of
information about traffic volumes, speeds and accelerations, are of direct influence on the end result, see
ref. [52] for details. Secondly, the data on road surfaces is important; though the model can be used without
it, it is one of the main parameters determining the difference between model results and real noise levels.
II.6.2 Road Traffic Flow Modelling
Road traffic modelling is not based on fundamental physical laws. The modelling of traffic involves the
description of numerous decisions of travellers. Thus, the nature of traffic brings along unreliability in the
output.
Also the current practice can cause unreliability in the traffic model results needed for noise calculations.
Current applications of traffic models generally focus on transport related objectives and problems, such as
accessibility of areas and levels of congestion. This means that the set-up and application of these models
is focused on specific items within the policy framework of transport (e.g. capacity, travel times). Using the
output of these models for other purposes will introduce additional uncertainties and difficulties, because the
models are not calibrated for this use, or do not contain the right parameters, e.g. certain times of day.
Using traffic data from a traffic model for noise models should therefore involve additional quality checks on
the traffic parameters which are important for noise modelling and which the traffic model should ideally be
optimised for. These parameters are speeds, traffic composition and flows.
There are acceptability guidelines for flows (see section 2.2.1 of ref. [18] and section 3.3 of ref. [17]) which
describe the maximum difference between flows resulting from the traffic model and flows from traffic
counts. Using these acceptability guidelines should ensure that flows will meet the accuracy standards of
the noise model. This will, however, not guarantee the accuracy of other traffic parameters.
For speeds, the same section in the report mentioned above contains an acceptability guideline. However,
this guideline focuses on the accuracy of calculated journey times, which gives little indication of the
accuracy of linked speeds.
As for traffic composition, while micro-simulation models (nearly) always offer the possibility of modelling
several vehicle categories, this is not necessarily the case for macroscopic models. If the transport demand
and traffic assignment models do not (or only partly) distinguish between vehicle categories, this will
inevitably cause unreliability in the traffic model results that are used in the noise model. The available
acceptability guidelines for flows and speeds (travel times) can also be applied per vehicle category (if data
is available), but this is not common yet.
The influence of these uncertainties on the L
den
/SWL of road noise is described in the report Review of data
needs for road noise source modelling, ref. [20].
II.6.3 Rail Noise Source
The quality of the end results is dependent on the accuracy of the data-gathering exercise and is
determined by the closeness of the IMAGINE example data or default data to the stock and track in
question, and also by the choices that are made in measurement procedures. For example the IDR
(determining combined effective roughness from track vibration) and VTN (separation of vehicle and track
contribution) methods give different uncertainties under different circumstances.
Page 49

The source database can either comprise measurement of individual events, or an averaging of a set of
nominally identical events. It is the responsibility of the user to determine the accuracy they require for these
sources. For the database the data gathering has been carried out strictly according to the quality required
both by ISO 3095 and the procedures laid down within IMAGINE. It is recommended that control over the
accuracy and the associated uncertainty be obtained by appropriate statistical tests.
It has to be kept in mind that the effect of uncertainties in sound power levels also depends on the size of
the mapping exercise. In a large scale noise mapping process over large geographical areas on an annual
average basis, the overall impact of uncertainty is likely to be reduced.
II.6.4 Industrial Noise
In principle, taking measurements of the noise emissions of the individual sources will yield the highest
accuracy of the sound power levels calculated from the obtained results if the measurements are performed
correctly. A compromise is sometimes needed between the high accuracy of the obtained noise data on the
one hand and the demand for as few source input/descriptive parameters as possible on the other, because
of the large variety in industrial noise sources and particularly in their operating conditions and individual
state of noise control.
There are various possibilities for taking noise measurements which make use of different methods and
equipment and produce results of different quality and degree of detail and accuracy. However, for noise
mapping it is essential that the input data sets be consistent with each other and with the noise propagation
calculations to be used. This must be ensured for all input data, in spite of being acquired by a large number
of different end users for a huge variety of noise sources in often completely different locations, situations
and operating conditions.
II.6.5 Aircraft Noise
The accuracy of the end results obtained with the IMAGINE method is higher than obtained with the
methods used mostly today.
The method takes into account many parameters that are ignored by traditional prediction schemes.
Particularly, the importance of source directivity and meteorological conditions have been demonstrated
through the experimental setup used for validation of the proposed prediction method.
The data obtained by reverse engineering NPD data reflect an average sound emission close to the
maximum A-weighted level. Neither the variations from fan-noise (approach) to jet-noise (leaving) nor the
Doppler shift are accounted for. Due to lack of information it is assumed that the spectrum at L
ASmax
applies
for all longitudinal directions. This is definitively wrong, but as the most important contribution to the
resulting L
AE
-level is in general produced by the levels around L
ASmax
this simplification is tolerable.
II.6.6 Propagation
The accuracy of the propagation methods depends on the quality of the geometrical input data and on the
representativeness of the meteorological description of the lower atmosphere. Both the reference and
engineering models correctly predict the averaged effects within large classes of similar propagation
conditions. Long time averaged noise indicators are obtained as weighted averaged over sound levels in
each propagation class, based on a statistical estimate of the frequency of occurrence of each class. Under
specific conditions, e.g. as for a single short time measurements, the deviations between measurements
and predictions can be as large as 7 dB. Because these differences behave as random variable, they
cancel out for longer periods of integration. The remaining uncertainty on the calculated values is then
within the objectives as initially settled by the HARMONOISE project.
Page 50

II.6.7 Measurement of L
den
and L
night

The measurement uncertainty will depend on the measurement effort. A long term measurement carried out
during several months may end up with a standard uncertainty of about 0,5 dB yielding an error (Level of
confidence 95%) of + 1 dB. On the other hand a single measurement carried out during half an hour under
favourable propagations conditions may end up with a standard uncertainty of about 2,0 dB yielding an error
of + 4 dB.
II.6.8 Mapping Specifications
The HARMONOISE/IMAGINE methods correspond to the best possible practice today in the sense that
they represent a practical compromise between the current understandings of the physical phenomena
involved in the generation and the propagation of noise in the outdoor environment, on one hand, and the
possibilities and limitations of information technology, on the other hand. These limitations refer both to the
available computational power and to the availability and suitability of digital data needed to run the models.
It was found that todays common practice with respect to input data collection is compatible with the
requirements and specifications of the harmonised methods. However, in order to achieve this compatibility
some of the new parameters, absent in existing data, must be defaulted to average conditions. As a
consequence there may be a concern that, if input data is specified at the same level of detail as today, the
introduction of the new methods will not significantly change the accuracy of the predicted noise levels.
In reality, the existing prediction schemes are only weakly sensitive to the level of detail and the accuracy of
the input data. Improving the quality of the input model will only partially affect the accuracy of the
predictions and is therefore considered by most stakeholders as an operation with a high cost/benefit ratio.
The HARMONOISE/IMAGINE methods on the other hand take full advantage of the higher quality of the
input data in the sense that they exhibit correct sensitivity to all changes in the relevant input parameters.
This advantage clearly makes the HARMONOISE/IMAGINE methods the better choice for the evaluation of
action plans.
Page 51

II.7 Databases
II.7.1 Road Noise Source
The road source database consists of a single Excel file containing:
the coefficients A
R
and B
R
for rolling noise (see II.2.3), which are the source power level at 70 km/h
and the speed dependence respectively, in 1/3 octave bands from 25 Hz to 10 kHz, for each
vehicle class; similarly,
the coefficients A
P
and B
P
for propulsion noise;
the coefficients C
P
for the acceleration correction, also in 1/3-octave bands;
These coefficients can then be used with the main source equations, see II.2.3, to calculate the propulsion
and rolling noise separately, and then to combine the two into the total sound emission at the two point
sources. Details can be found in the deliverable. All the other vehicle correction factors are described in ref
[52]. The road surface correction factors are described in the European project SILVIA, ref. [58] and [59].
II.7.2 Road Traffic Data
Useful datasets and statistical information on road traffic may already exist (or will exist by 2012 onwards)
for all of the basic traffic parameters required by the HARMONOISE/IMAGINE model. Such datasets may
be in the possession of local, regional or national authorities, or commercial enterprises. The exact sources
will undoubtedly vary between individual member states, given the diverse political and organisational
structures across the continent. Potential sources are:
1. Data collected from local or national traffic monitoring, available from authorities;
2. Data provided by commercial or Non-Governmental Organisations (NGOs);
3. Data collected, or derived, for the interim round of noise mapping;
4. Data collected, or derived, for ongoing air-quality management exercises.
It is recognised that sources 3 and 4 will most likely include substantial elements derived from sources 1
and 2.
The Good Practice Guide ref. [2] provides default values in the case where no or very little information is
available.
II.7.3 Rail Noise Sources
The rail noise database is described fully in ref. [61] and the structure is shown schematically in the figure
below. It is a relational database with strict referential integrity and has been implemented in Access 2000
(SR-1).

Page 52

Schematic structure of IMAGINE rail noise database


The heart of the database is formed by tables containing data in the following categories (represented by
the boxes labelled spectrum1, spectrum2 etc in the centre of the Figure):
roughness (wheel, rail and total effective ie including the filter effect of the contact patch at their
interface)
contact patch filter
rolling noise transfer functions between combined effective roughness and sound pressure level
emission (vehicle, track and total)
traction noise (idling, acceleration, deceleration and constant speed)
aerodynamic noise (bogie, pantograph and total, plus speed exponents for bogie and pantograph)
braking and braking squeal noise
curve squeal noise
impact noise
Each of these tables has a common structure containing fields that (a) describe the data and (b) contain the
data in one-third octave band spectra. The roughness and contact patch filter spectra are in terms of one-
third octave band wavelengths (0.8 mm to 1000 mm) whilst the remainder are defined in terms of frequency
(20 Hz to 25 kHz).
In the above figure, the processes shown beneath the broken line indicate the nature of data that is input to
the database, obtained either via measurement according to the IMAGINE rail source measurement
procedures [60] or via an alternative approach.
IM vehicle/track
combinations
measurement set
info
s
p
e
c
t
r
u
m

1
s
p
e
c
t
r
u
m

2
s
p
e
c
t
r
u
m

N
IM vehicles IM tracks
s
p
e
c
t
r
u
m

s
p
e
c
t
r
u
m

s
p
e
c
t
r
u
m

measurements
vehicle info track info
IM vehicle/track
combinations
measurement set
info
s
p
e
c
t
r
u
m

1
s
p
e
c
t
r
u
m

2
s
p
e
c
t
r
u
m

N
IM vehicles IM tracks
s
p
e
c
t
r
u
m

s
p
e
c
t
r
u
m

s
p
e
c
t
r
u
m

measurements
vehicle info track info
IM vehicle/track
combinations
measurement set
info
s
p
e
c
t
r
u
m

1
s
p
e
c
t
r
u
m

2
s
p
e
c
t
r
u
m

N
IM vehicles IM tracks
s
p
e
c
t
r
u
m

s
p
e
c
t
r
u
m

s
p
e
c
t
r
u
m

measurements
vehicle info track info
Page 53

Metadata, such as vehicle and track descriptors, is included as additional tables, and related to the relevant
spectra by connections within the database. Thus it is possible to trace back measurement data to its
source. All this data is then stored as appropriate spectra for use by modellers.
When a railway scenario is to be modelled with the IMAGINE method, an appropriate combination of
spectra for each rail vehicle and track type being considered is selected from the database. These spectra
are chosen to represent the various subsources (eg traction noise, rolling noise from the vehicle, rolling
noise from the track, aerodynamic effects) that apply for the particular scenario. The process of
reassembling a scenario for IMAGINE application is represented in the above figure by the processes above
the broken line, labelled as IM (ie IMAGINE). The constituent parts of the reassembled vehicle and track
combination will often derive from several separate measurements on a variety of trains at different
locations.
Use of database
The user front end for the rail noise database is described in ref. [61] and is illustrated graphically in the
figure below. The front end can be operated at three levels:
Level 0 predefined track and trains
Level 1 define new track or train with default track and train characteristics
Level 2 import new track and train characteristics

Rail noise database and user front end

Database
Level 0
Level 1
Level 2
calculation
module
Source
model
Engine
L
W1,z=0m
L
W5,z=5m
Import new data
Front end
User interface
Back end
Database
Level 0
Level 1
Level 2
calculation
module
Source
model
Engine
L
W1,z=0m
L
W5,z=5m
L
W1,z=0m
L
W5,z=5m
Import new data
Front end
User interface
Back end

Within IMAGINE, the detailed data-gathering and importing processes described above comprise the Level
2 approach, which will normally require expert operators to carry out. An administration tool has been
developed to import acquired data from the standard data input forms into the database framework.
The source model algorithms (available at all levels of user) take the traffic information for the required
segment and the appropriate data from the database to calculate the sound powers at the required five
heights. The details of this calculation, including the traffic model and the conversion from sound pressure
to sound power, is described in ref. [65].

Page 54

Data currently in the database
Currently the database contains default and measured data. The default data is described and presented in
ref. [65]. It includes information on wheel and rail roughness, traction noise, transfer functions for rolling
noise from track and vehicles, impact noise, aerodynamic noise, brake squeal and curve squeal.
The database currently includes an initial dataset as follows:
Data from the Netherlands IPG Project
Data from the STAIRRS project, reworked within IMAGINE to be compatible with the IMAGINE
database, from the Netherlands, France, and the UK
UK Traction Noise data
UK Aerodynamic Noise data from the Channel Tunnel Rail Link
Data from an extensive measurement campaign in Hungary
Data from Sweden
The database framework is designed to be added to readily, provided new data is acquired and logged
using the processes and spreadsheets developed within IMAGINE. It is therefore envisaged that, post-
IMAGINE, the database can be used both as a source of existing example and default data for railway noise
modellers, and as a permanent repository for new data as it is obtained over time.
A full description of the rail source model, the rail traffic model, the required processes for the acquisition,
storage and retrieval of data, and the application of the database both to standard situations and to those
that are non-standard, is provided in ref. [60].

II.7.4 Industrial Noise Sources
The number of parameters taken into account and also the number of parameter combinations for a specific
type of noise source, needs to be limited to keep the database usable. For this purpose, the following was
decided upon:
Source types are grouped into the most important categories only (e.g. process furnaces will be
categorized into furnaces with floor and with sidewall burners only, although many other special designs
exist).
The quality of the source in terms of noise emissions and the extent of applied noise control are
categorized into "poor", "average" and "good only.
Only the most important parameters will be varied from dataset to dataset.
Parameter variations are made in steps that lead to differences in the overall noise level of at least 2 dB.
The terms poor, average and good have the following interpretation:
Poor: no noise mitigation measures have been taken and the machine/source is either old or badly
maintained. The sound power levels are not always based on detailed measurements, but some
examples have been found.
Average: some noise mitigation measures have been taken and the machine/source is adequately
maintained. If no other information of a specific machine/source is available, the source is called
average.
Page 55

Good: the sound power levels of these machines/sources are among the lowest that can be bought or
manufactured. No research has been done into whether these sources may be called Best Available
Techniques.
Using an extensive set of parameters and many small variation steps would increase the number of entries
in the database dramatically but would at the same time bring about only a minor improvement in accuracy:
even for a technically "comprehensive" set of input parameters, the sound power level of many source types
will still vary within a considerable range. This is due to various reasons, such as the influence of minor
construction details or the noise generating process itself.
In contrast to this, a limited parameter set and a coarser variation step will limit the number of entries in the
database, making it easier to use while still maintaining a level of accuracy that is only insignificantly lower
than for the approach described above. This is especially true for situations where more than a few noise
sources are present, i.e. especially for typical industrial plants with a high number of different sound
sources.
Source dB consists of seven screens, which can be accessed easily. The opening screen looks like this:


Clicking on the tab called Spectra gives the third octave spectral information.



Page 56

The first tab is always called Source, which shows all the input data, including the table showing in which
industry types the source may be found.

For some sources formulas are derived and the source power can be calculated.

The structure of the database is as follows:
Page 57


II.7.5 Aircraft Noise Source Database
The aircraft noise source description is based on sound power, including directional information: L
w,dir
. There
are various factors influencing sound emission; these are listed in ref. [38]. Each hemisphere represents, for
an aircraft/engine combination, a unique operating condition that depends on the factors mentioned. There
is one table for each 1/3 octave frequency. This is a multi-dimensional database where a large matrix of
sound hemispheres is therefore required to model an aircrafts sound emission.
The universal way to describe emission is to use tables for spectral sound
power in the polar co-ordinates theta and phi. (For a definition see section
II.2.6.2 ). Lateral symmetry can be assumed for fixed wing jet aircraft, i.e. phi
has only to be listed in the range from 0to 90.
The polar co-ordinates theta and phi indicate locations on a sphere as shown in
the figure to the right. When L
w,dir
(,) is listed in a rectangular table, there exist
Page 58

combinations of and , which must have the same values. For instance, at
= 0 and at = 180all values of coincide in the same point (the poles of the sphere). This is indicated
in the table below by the grey fields. It is recommended, that these fields have the interpolated (or constant)
value of neighbouring cells.
The resolution (or the amount of data) of the database will depend on the available data. A 5 resolution in
both and is recommended. Based on the experience from the IMAGINE flight tests the benefits of a
higher resolution (if any) do not justify the additional cost involved with the increased number of flights
required.

Aircraft type: Engine type:
1/3 octave band frequency = f Operation
Corrected net thrust (CNT)= T True air speed (TAS)= v
Flaps / slats angle = Landing Gear = up/down
Longitudinal angle Sound power
L
w,dir

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
0
10 XX XX XX XX XX XX XX XX
20 XX XX XX XX XX XX
30 XX XX XX XX XX XX
40 XX XX XX XX XX XX
50 XX XX XX XX XX XX
60 XX XX XX XX
70 XX XX XX XX XX XX XX XX
80 XX XX XX XX XX XX
Lateral
angle

90 XX XX XX XX XX XX XX XX

Proposed structure of data table
Proposed database structure
The proposed star schema model is one of the most common systems for data analysis. It permits the
implementation of multi-dimensional views of data using a relational database as well as easing
understanding. The star schema model (see below) is composed of a single fact table which contains the
final values of sound power. The values of L
w,dir
will depend on the values of the dimension tables which
surround the fact table. The dimension tables are used to group data when performing data queries that will
allow representative tables like those of the previous paragraph: i.e. directional levels of a certain
combination of thrust, speed, flaps and landing gear position and frequency.

Page 59


Star-scheme of a data base structure
The data base would consist of a single table. The following example was produced with Access 2000


Example of Access Program

Sound
power
including
directivity
Aircraft ID /
Engine /
Operation
Thrust

Aircraft
Speed
Angle
Phi
Flaps
angle
Angle
Theta
Landing
gear pos.
Frequency

Page 60


Example of a data sheet in the Access program
Page 61

II.8 References
General
[1] Directive 2002/49/EC of the European Parliament and of the Council of 25 June 2002 relating
to the assessment and management of environmental noise, Official Journal of the European
Communities L 189 dated 18/7/2002.
[2] WG-AEN, Second position paper on Good practice guide for strategic noise mapping and
the production of associated data on noise exposure.

WP1
IMAGINE documents
[3] IMAGINE, Deliverable 4, IMA01-TR060526-CSTB05, GIS noise specification.
[4] IMAGINE, Deliverable D8, IMA01-TR22112006-ARPAT12 Guidelines and good practice
on strategic noise mapping.
[5] IMAGINE, IMA10-TR-24012006-ARPAT10.doc, Survey on noise mapping input data
specification and availability.
[6] IMAGINE, IMA1TR060426-Autostrade01.doc, Case studies concerning Autostrades urban
road noise mapping methodology.
[7] IMAGINE, IMA1PEN-311006-Labein03, Estimating the number of people exposed to noise.
[8] IMAGINE, Technical report IMA1TR060426Autostrade01, LIDAR innovative technology
extracted from Case studies concerning Autostrades urban road noise mapping
methodology,.

HARMONOISE documents
[9] HARMONOISE, D18_WP3_HAR32TR-040922-DGMR20, Engineering method for road
traffic and railway noise after validation and fine-tuning.
[10] HARMONOISE, Deliverable 16, report D16_WP2_HAR29TR-041118-TNO10, Description of
the Reference model.

General references
[11] ECAC Doc 29 3rd edition, Volume 1: Applications guide
[12] Adrienne, SMT Project MAT1-CT94049, Final report. Test methods for the acoustical
performance of road traffic noise reduction devices, Final report of the Adrienne project,
1988.
[13] NT ACOU 104, Ground surfaces: Determination of the acoustic impedance,
http://www.nordtest.org
[14] EN 1793-1.Road traffic noise reducing devicesTest method for determining he acoustical
performance, Part 1:Intrinsic characteristics of sound absorption, 1997.
[15] Timothy Van Renterghem and Dick Botteldooren, Influence Of Correlation Between Diurnal
Traffic Pattern And Meteorological Conditions On Long-Term Average L
den
, Proceedings of
Euronoise 2006, Tampere, Finland 2006.

Page 62

WP2
IMAGINE documents
[16] IMAGINE, Deliverable 7, IMA02DR7-060531-TNO.10, Guidelines for the use of traffic
models for noise mapping and noise action planning, 31 May 2006.
[17] IMAGINE, report IMA02TR-060525-UL10, Collection methods for additional data, 25 May
2006.
[18] IMAGINE, report IMA02TR-060131-UGent10, Development of strategies for the use of traffic
models for noise mapping and action planning, 31 January 2006.
[19] IMAGINE, IMA02TR-050112-TML10, Suitability of traffic models for noise modelling, 12
January 2005.
[20] IMAGINE, IMA02TR-040615-M+P10, Review of data needs for road noise source
modelling, 15 June 2004.

General references
[21] Wilmink, I.R. (2004) How to link road traffic and noise modelling (IMAGINE WP2: demand
and traffic flow management) in: Proceedings IMAGINE South Europe Workshop, Pisa, 19
November 2004.
[22] Wilmink, I., E. Versteegt, P. Goodman and M. Bell (2005), The role of ITS in noise mapping
and noise action planning, In: Proceedings of the 5th European Congress and Exhibition on
Intelligent Transport Systems and Services, Hannover, Germany, 1-3 June 2005.
[23] Bert De Coensel, Filip Vanhove, Steven Logghe, Isabel Wilmink and Dick Botteldooren
(2006), Noise emission corrections at intersections based on microscopic traffic simulation,
paper presented at Euronoise 2006, Tampere Finland, 30 May 1 June 2006.
[24] Timothy Van Renterghem and Dick Botteldooren (2006), Influence of correlation between
diurnal traffic pattern and meteorological conditions on long-term average L
den
, paper
presented at Euronoise 2006, Finland, 30 May 1 June 2006.
[25] Paul Goodman, Phillip Skelton and Margaret C. Bell (2006), Understanding the effects of
uncertainties in the HARMONOISE road source model, paper presented at Euronoise 2006,
Tampere Finland, 30 May 1 June 2006.

WP3
IMAGINE documents
[26] IMAGINE, report IMA31TR-040322-SP05, State-of-the-art-report, Measurement Methods for
Environmental Noise
[27] IMAGINE, Deliverable D5, IMA32TR-040510-SP010, Determination of L
den
and L
night
Using
Measurements
[28] IMAGINE, Report IMA03TR-060610-CSTB01, Processing of meteorological data and
determination of long time averaged noise indicators L
den
and L
night

General references
[29] Iacoponi, M. Paviotti, G. Licitra & T. Carnevale, Track-Train Railway Noise Measurement
Accuracy, Symposium on Managing Uncertainty in Noise Measurement and Prediction, Le
Mans (France) 27-29 June 2005
Page 63

[30] H.G. Jonasson, Uncertainties in Measurements of Environmental Noise, Symposium on
Managing Uncertainty in Noise Measurement and Prediction, Le Mans (France) 27-29 June
2005
[31] H.G. Jonasson, Determination of L
den
Using Measurements, Euronoise 2006, Tampere 30
May 1 June, 2006
[32] S. Kephalophoulus, D. Knauss and M. Brengier, Uncertainties in Long-Term Road Noise
Monitoring Including Metorological Variations, Symposium on Managing Uncertainty in Noise
Measurement and Prediction, Le Mans (France) 27-29 June 2005
[33] D. Kuehner, Minimising Uncertainties of Long term Measurements Using Stratification,
Uncertainties in Long-Term Road Noise Monitoring Including Metorological Variations,
Symposium on Managing Uncertainty in Noise Measurement and Prediction, Le Mans
(France) 27-29 June 2005
[34] M. Paviotti, Evaluation of the Meteorological Uncertainties of the HARMONOISE and
IMAGINE Project Noise Measurements, Euronoise 2006, Tampere 30 May 1 June, 2006
[35] N. Tellado & Iziar Aspuru, Uncertainty Analysis in Trains Noise Emission Characterisation
due to Simplifications in Official procedures, Symposium on Managing Uncertainty in Noise
Measurement and Prediction, Le Mans (France) 27-29 June 2005
[36] See published papers for presentations made at Euronoise 2006 and the Le Mans
Symposium on Managing Uncertainty in Noise Measurement and Prediction apart from at the
final IMAGINE workshop in Budapest on October 25, 2006 WP 3 has been presented at one
IMAGINE South European workshop in Terrassa, Spain, on October 20, 2005 and at another
one in Pisa on November 19, 2004. It has also been presented at one workshop in Prague in
connection with Internoise 2004 and at a road traffic noise workshop in Bors, Sweden on
Oct 27.

WP4
IMAGINE documents
[37] IMAGINE, Deliverable 9, Reference and Engineering Models for Aircraft Noise Sources,
Report IMA4DR-061017-EEC-09
[38] IMAGINE, Deliverable D10 Default aircraft source description and methods to assess
source data, Report IMA4DR-061204-Empa-10

General references
[39] INM Integrated Noise Model,
www.faa.gov/about/office_org/headquarters_offices/aep/models/inm_model/
[40] DOC 29, 3rd edition, Report on Standard Method of Computing Noise Contours Around Civil
Airports Volume 1: Applications Guide; Volume 2: Technical Guide: www.ecac-ceac.org,
December 2005
[41] ISO 9613 (1996): Attenuation of sound during propagation outdoors:
- Part 1: Calculation of the absorption of sound by the atmosphere
- Part 2: General method of calculation
[42] FLULA: Pietrzko S.J., R.F. Hofmann: Prediction of A-Weighted Aircraft Noise Based on
Measured Directivity Patterns, Applied Acoustics 23 (1988), pp. 29-44.
[43] Krebs, W.; Btikofer, R.; Plss, S.; Thomann, G.: Sound source data for aircraft noise
simulation, Acta Acustica - Acustica, 90 (2004), 91-100
[44] NASA: Engine Installation Effects of Four Civil Transportation Airplanes: The Wallops Flight
Facility Study; October 2003 http://techreports.larc.nasa.gov/ltrs/PDF/2003/tm/NASA-2003-
tm212433.pdf
Page 64

[45] SAE ARP 866A "Standard values of atmospheric absorption as a function of temperature and
humidity, Revised 1975
[46] R. Btikofer, Concepts of Aircraft Noise Calculations, Acta Acustica united with Acustica,
Vol. 93, No 2, 253-262, March 2007
[47] Michael J.T. Smith, Aircraft Noise, Cambridge Aerospace Series, Cambridge University
Press, 1989.
[48] Wyle Research Report WR 01-04: Acoustic Repropagation Technique, Version 2 (ART2),
prepared for NASA-Langley Research Center, 2001, (Part of documentation to NASA's RNM
- Model)
[49] Aircraft Noise and Performance (ANP) database: http://www.aircraftnoisemodel.org
[50] ICAO, Annex 16 to the convention on international civil aviation: Environmental Protection,
Volume I: Aircraft Noise, Fourth Edition 2005
[51] SAE AIR 1845 "Procedure for the calculation of airplane noise in the vicinity of airports",
March 1986

WP5
IMAGINE documents
[52] IMAGINE, Deliverable D11, IMA55TR-06021-MP10, he Noise Emission Model For European
Road Traffic, January 11
th
2007
[53] IMAGINE, Deliverable D3, IMA52TR-060111-MP10, Assessment Programme for
Parameters of the "general" European vehicle fleet, January 11th 2006

HARMONOISE documents
[54] HARMONOISE, Deliverable D9, HAR11TR-020614-SP09v4, Source modelling of road
vehicles, March 2004

General references
[55] Guidance manual for the implementation of low-noise road surfaces, SILVIA Final
Deliverable, FEHRL report 2006/02, 2006 (see http://www.trl.co.uk/silvia)
[56] ISO/CD-11819-2, Method for measuring the influence of road surfaces on traffic noise part
2: The Close Proximity Method

Papers / Publications Work package 5
[57] H.M. Peeters, G.J. van Blokland, D.F. de Graaff, Development of an European road traffic
emission model within the framework of the EU 6th Framework project IMAGINE, Forum
Acusticum 2005, paper no. 279-0, August 2005
[58] F. Anfosso, W. Bartolomeus, G.J. van Blokland, P. Morgan, C. Padmos, SILVIA-DWW-025-
04-WP2-080205_classification system
[59] BASt, DWW, LCPC, M+P, TRL, Classification Scheme and COP Method, SILVIA-DWW-
025-14-WP2-141005 (2005)

Page 65

WP6
IMAGINE documents
[60] IMAGINE, Deliverable D12 /D13, IMA6TR-061015-AEATUK01 , Rail noise database and
manual for implementation, 2007
[61] IMAGINE, report IMA6-150606-AEATNL06, Description of the IMAGINE Rail source
Database and administration software, 23 November 2006
[62] IMAGINE, report IMA6TR-041114-AEATNL03, IMAGINE practical measurement
guidelines & analysis, 10 October 2005
[63] IMAGINE, report IMA6MO-041014-AEATNL02, User interface database, 24 November
2004
[64] IMAGINE, report IMA6TR-050905-AEATNL06, Data input sheet guidance, 5 September
2005
[65] IMAGINE, report IMA6TR-050912-TNO01 Version 5, IMAGINE railway noise source model,
default source data and measurement protocol, 20 July 2006

General references
[66] Jones, RRK IMAGINE Rail Noise Sources, Institute of Acoustics Autumn Conference,
Oxford, Oct 2004
[67] Jones, RRK Quality Considerations For Railway Rolling Noise Data Within IMAGINE,
Conference on Managing Uncertainty in Noise Measurement and Prediction, Le Mans, Jun
2005
[68] Hardy, AEJ and Jones, RRK Rail Head Surface Quality in Environmental Noise Prediction
and Noise Mapping Forum Acusticum, Budapest, September 2005
[69] Dittrich, MG The IMAGINE model for Railway Noise Prediction Forum Acusticum, Budapest,
September 2005
[70] Block, JR and Jones, RRK IMAGINE Progress towards a comprehensive description of
railway noise sources, Euronoise, Tampere, May 2006

WP7
IMAGINE documents
[71] IMAGINE, IMA7TR-050418-MBBM03, Measurement Methods.
[72] IMAGINE, Deliverable 6, IMA7TR-050418-DGMR02, Description of the source database.
[73] IMAGINE, IMA7TR-060614EDF02, Reference results for validating the engineering model.
[74] IMAGINE, IMA-0660610-CSTB01, Processing meteorological data and determination of long
time averaged noise indicators L
den
AND L
night
.
[75] IMAGINE, Deliverable D14 Guidelines for producing strategic noise maps on industrial
sources, Report IMAWP7D14-060811-DGMR03

WP8
IMAGINE documents
[76] IMAGINE Project web site, http://www.IMAGINE-project.org

Page 66

WP10
IMAGINE documents
[77] IMAGINE, Deliverable 2, Report IMA10TR-040423-AEATNL32 The State of the Art in
Noise Mapping.
[78] IMAGINE extra document, IMA010TR-070222-DeltaRail10, Determination of L
den
by
calculation - definition of meteorological classes.
Page 67

APPENDIX I PRESENTATIONS AND PAPERS DURING THE PROJECT
Presentations Description Location Date
All WP leaders Presentations kick-off meeting IMAGINE Callantsoog,
Netherlands
20 January 2004
WP3, Hans
Jonasson
Seminar on environmental noise with 25 participants from
Swedish authorities and consultants
Sweden 9 March 2004
DGMR Presentation about future models (over 200 people) At a Geonoise
software clients day
12 May 2004
AEA NL Presentation at Road Authority Noise Contact meeting Utrecht, The
Netherlands
10 June 2004
WP3, Hans
Jonasson
Workshop in connection with Internoise in Prague Internoise 2004,
Prague, Czech
Republic
23-25 August 2004
WP4, Peter Hullah Research Into Aircraft Noise Nuisance; presentation on new
developments in aircraft noise modelling, including IMAGINE
Internoise 2004,
Prague, Czech
Republic
23-25 August 2004
WP7, Rob Witte Presentation on the IMAGINE objectives on industrial noise Internoise 2004,
Prague, Czech
Republic
23-25 August 2004
IMAGINE Presentations at the first I IM MA AG GI IN NE E workshop Prague, Czech
Republic
26 August 2004
WP6, Rick Jones Institute of Acoustics meeting Oxford, UK 6-7 October 2004
IMAGINE ,
Edwin Verheijen
General project presentation at the HARMONOISE final
conference
Rhodes, Greece 21-22 October 2004
all WPs Presentations at the second IMAGINE workshop Pisa, Italy 19 November 2004
WP7, Rob Witte Dutch Acoustical Association NAG Utrecht, The
Netherlands
24 November 2004
WP1, Dirk van
Maercke
4mes assises nationales de lenvironnement sonore, session
on HARMONOISE & IMAGINE
Avignon, France 18 January 2005
WP10, Margreet
Beuving
Presentation at Workshop on Implementation of Directive
2002/49/EC on Environmental noise
Vilnius, Lithuania 9, 10 March 2005
WP2, Isabel
Wilmink
Presentation at ITS in Europe Hannover 1-3 June 2005
WP 2, 5, 6 Forum Acusticum Budapest, Hungary 29 August 2
September 2005
WP2 WP2 workshop for end users, during Forum Acusticum Budapest, Hungary 29 September 2005
WP1, 3, 6 INCE symposium on Managing Uncertainties in Noise
Measurements and Predictions
Le Mans, France 27-29 June 2005
Page 68

WP1, 3, 6, 7, 10 Southern European Workshop on Environmental Acoustics and
the IMAGINE Project
Barcelona, Spain 21, 22 October
2005
WP4, 8 X2Noise conference on Noise Mapping and Aviation Vilnius, Lithuania October 2005
WP3 Workshop on road traffic noise Boras, Sweden October 2005
WP2, 5, 7, 8 Seminar at the Dutch Road Organisation DWW Delft, The
Netherlands
1 November 2005
WP6, Rick Jones,
Paul van der Stap,
Michael Dittrich,
WP8, Margreet
Beuving
Workshop on Railway Noise in cooperation with the Working
Group AEN, organized by Work Package 6
Brussels, Belgium 1 December 2005
WP leaders
WP3, 7
Euronoise 2006 Euronoise, Tampere,
Finland
30 May - 1 June
2006
WP2 Paper on Noise emission corrections at intersections based on
microscopic traffic simulation, Bert De Coensel, Filip Vanhove,
Steven Logghe, Isabel Wilmink and Dick Botteldooren
Euronoise, Tampere
Finland
30 May - 1 June
2006
Timothy van
Renterghem, Dick
Botteldooren
(WP2)
Paper on Influence of correlation between diurnal traffic
pattern and meteorological conditions on long-term average
Lden, Timothy Van Renterghem and Dick Botteldooren
Euronoise, Tampere
Finaland
30 May - 1 June
2006
Paul Goodman,
Phillip Skelton,
Margaret Bell
(WP2)
Understanding the effects of uncertainties in the
HARMONOISE road source model

James Block and
Rick Jones (WP6)
IMAGINE progress towards a comprehensive description of
railway noise sources
Euronoise, Tampere,
Finland
30 May 2006
WP 3 Hans
Jonasson
Determination of Lden using measurements
Euronoise 2006
Tampere, Finland 1 June 2006
WP 3 Marco
Paviotti
Evaluation of the meteorological uncertainties of the
HARMONOISE IMAGINE project noise measurements,
Euronoise 2006
Tampere, Finland 30 May-1 June
2006
Rob Witte (WP7) Meeting of noise specialists of the provinces of the
Netherlands
Utrecht Netherlands May 2006
Renez Nota (WP7) Meeting of Geonoise software users Utrecht, Netherlands May 2006
Dirk van Maercke
(WP1), Peter
Hullah (WP4), Gijs-
Jan van Blokland
(WP5), Rick Jones
(WP6), Rob Witte
(WP7), Margreet
Beuving (WP8,
Workshop on Software Development with strategic software
developers, organized in cooperation with the WG-AEN
Brussels, Belgium 19 June 2006
Page 69

10), Paul van der
Stap (WP9)
WP3 Hans
Jonasson
Seminar of the Australian Acoustical Society on measurement
methods in the HARMONOISE, IMAGINE and Nord2000
projects.
Australia 29 June 2006
AEA UK (WP6) Silence Workshop, IMAGINE rail noise sources Brussels 21 June 2006
WP4, Peter Hullah Aeronautics Days 2006, IMAGINE aircraft sources Vienna 19 21 June 2006
WP4, Peter Hullah Presentation to Friendcopter workshop Maastricht 13 September 2006
WP6, WP10, Rick
Jones and Brian
Hemsworth
General overview of IMAGINE on the Workshop on Presenting
Noise Mapping Data to the Public
London 5 October 2006
WP2, Uleeds and
TNO
ITS World Congress London 2006 London, UK October 2006
WP 3,5 and 6
Hans Jonasson
Xuetao Zhang
Thematic day Road traffic noise
Thematic day Road traffic noise
Thematic day Rail traffic noise
Bors;Sweden
Malm, Sweden
Bors, Sweden
31 October 2006
19 October 2006
17 November
WP 6, Xuetao
Zhang
Implementation of the IMAGINE source model on Swedish
trains, Baltic Nordic Acoustical meeting
Gteborg, Sweden 10 November 2006
All WP leaders General overviews of Work Packages, structure, input data and
benefits, IMAGINE Final Conference
Budapest, Hungary 25 October 2006
WP 1 and 3
partners
Presentations in the Mapping session of the Final Conference
on Meteorological model, data & preprocessing, Measurement
of Lden, The P2P show, Modelling requirements, data collection
& GIS, Estimation of exposed populations
Budapest, Hungary 25 October 2006
WP 2, 5, 6 and 7
partners
Presentations in the Data session of the Final Conference on
the databases for Road, Rail and Industry, needed input data
for Road Rail and Industry and measurement methods
Budapest, Hungary 25 October 2006
WP4 partners Presentations in the Aircraft session of the Final Conference on
the structure of the model, conversion of data, the aircraft
propagation model and flight tests.
Budapest, Hungary 25 October 2006
B Hemsworth UK Institute of Acoustics, Midlands Branch Walsall, UK 15 Nov 2006
WP3, J. Ejsmont &
G. Ronowski
Short term versus long term traffic noise
measurements, Internoise 2006
Honolulu, Hawaii,
USA
3-6 Dec 2006
ulf.sandberg The concept of virtual reference pavement for noise prediction
and comparison purposes
Honolulu, Hawaii,
USA
3-6 Dec 2006
WP 6, Xuetao
Zhang
Applying the TSI formulae together with the multiple-
interpolation method to determine track decay rates using train
pass-by measurements, Internoise 2006
Honolulu, USA 3-6 December 2006
Dirk van Maercke Ministry of Environment: Overview of the results
HARMONOISE and IMAGINE projects.
Paris, France 6 December 2006
Table 1: IMAGINE presentations
Page 70

APPENDIX II QUICK SCAN OF IMAGINE
This section describes the main outcomes of IMAGINE in a few words.

IMAGINE end users
The end users of the IMAGINE methods can be split into authorities, noise modellers, technicians, operators
and software developers. The IMAGINE deliverables give information for all groups of end users.

Mapping and Action planning
The methods developed in HARMONOISE and IMAGINE can be used for mapping and action planning. For
strategic mapping projects the accuracy requirements may be lower and the amount of data may be larger,
whereas for action planning purposes high detail is needed. By its modular structure and the flexible set up
of the methods and the databases the IMAGINE methods can be used both for very detailed and for more
global assessments.

Determination of L
den
The determination of L
den
is done by calculation of the source noise emission and the propagation of the
emitted noise to the receiver. The long term average value is calculated by including propagation classes
which are typical for the considered location. The data needed for the calculations are source data (location
of source, source types, numbers, running/driving/flying conditions, working hours, etc.) topographical data
(ground surface, buildings, surroundings, etc.) and meteorological data.

Measurements
For the determination of the L
den
calculations are highly recommended because of the reproducibility of the
results and the better possibilities for analysis. However noise measurements can add to the credibility and
reliability of calculated noise levels. Often the measurement results have to be combined with calculations to
correct for operating or propagation conditions different from those during the actual measurement. The
IMAGINE method for determining the L
den
and L
night
by measurement is described in ISO standard form in
ref. [27].

Databases
Source databases have been set up, containing average, default (averaged) or (representative) example
data for different source types and vehicle classes. With the data from the databases typical European
noise situations can be calculated, but also regional corrections can be made for road, noise spectra of
typical types of trains and tracks can be used for railway and many different examples are available for
industry noise. Only the aircraft noise database still needs to be filled with flight data of aircraft/engine types.
The databases for rail and industry are fit for further development by easy implementation of new data.

IMAGINE and GIS
GIS systems are an important tool for the production of strategic noise maps. In order to enhance the quality
and to achieve reproducibility of the maps a set of GIS specifications has been put together, which are
Page 71

suitable for noise mapping purposes. On the other hand the best practice of using available geospatial data
has been investigated.

Propagation method
The propagation of the sound emitted by the noise sources to the receiver is described in the
HARMONOISE/IMAGINE propagation method. This method describes direct, reflected and refracted ray
paths and it includes reflections from buildings, ground reflection, relief of surroundings, barriers, etc. under
variable meteorological conditions.
The HARMONOISE project provided two kinds of propagation methods: a theoretical reference method and
a more practical engineering method. Both models were validated against experimental data. In IMAGINE
both methods were developed for higher sources (for aircraft) and for the detection of direct, reflected and
diffracted ray paths in a complex three dimensional scene (industry).

IMAGINE software
The P2P software for the propagation method, built and improved during HARMONOISE and IMAGINE,
was originally meant to be used inside the projects for case testing. The software turned out to be a
practical tool for understanding and testing of noise propagation situations. Therefore it has been made
freely available and is used by many technicians and software developers. Several software developers
have started integrating the HARMONOISE/IMAGINE modules in their commercial noise mapping products.
For the railway database software is available for automatic uploading of new data in the database.

Propagation classes
The IMAGINE propagation method takes into account the influence of the meteorological circumstances on
the sound propagation. Four propagation classes are defined based on local data on wind speed, wind
direction and thermal stability of the atmosphere. These data can be easily collected from normal weather
stations. This classification of propagation conditions applies to calculations as well as to measurements.
The Member States define their situation by considering the frequency of occurrence of each propagation
class over a relevant year. See ref. [78].

The IMAGINE website
The IMAGINE website contains an overview of the Project by the description of the IMAGINE events during
and after the project and the many documents and presentations which are downloadable. Events, news
items and links with other relevant projects and European developments are described on the website.
www.IMAGINE-project.org

Você também pode gostar