Vous êtes sur la page 1sur 11

Edinburgh, Scotland

EURONOISE 2009
October 26-28

The equivalence of road traffic noise assessments using


XPS 31-133 Interim, RLS90, CRTN, RMW2006 and
Harmonoise
Simon J. Shilton1
Acustica Limited, Manchester, M22 5XB, United Kingdom.
Alan Stimac2
DARH2 Acoustics & Civil Eng. Ltd., HR-10430 Samobor, Croatia.
Renez Nota3
DGMR, Den Haag, The Netherlands.

ABSTRACT
Initial estimates indicate that over 115 million people dwell within approximately 160 major
cities across the EU for which noise maps and action plans have been prepared for the 1st
round of activities under Directive 2002/49/EC. The results derived from the maps, and
the summary action plans, have now largely been submitted to the Commission, and the
process of drawing comparisons between cities and Member States is being undertaken.
The Directive enabled Member States to use adapted national methods, either statutory or
de facto standards, or the EC recommended Interim methods, themselves adapted from
some existing national methods. The Directive contains a requirement for national
methods to demonstrate equivalence to the recommended Interim methods, and JRC
issued draft protocols designed to evaluate equivalence to the Interim methods. This paper
present results obtained from the use of these protocols in the evaluation of equivalence
between the recommended Interim method for road traffic noise, XPS 31-133 Interim and
compares them with the results obtained using three adapted national methods, and the
draft proposed Harmonoise method. These results may provide an initial estimate of the
uncertainty introduced into the mapping results through the selection of differing methods
of assessment, and draw comparisons with the extent of uncertainty introduced by
differences in the quality of input datasets and other factors.

1. INTRODUCTION
Directive 2002/49/EC of the European Parliament and of the Council relates to the
assessment and management of environmental noise, and is commonly referred to as the
Environmental Noise Directive or END1. The aim of the Directive is: to define a common
approach intended to avoid, prevent or reduce on a prioritised basis the harmful effects,
including annoyance, due to exposure to environmental noise. And to that end three
stages are set out:
Undertake strategic noise mapping to determine exposure to environmental noise;
1

Email address: simon.shilton@acustica.co.uk


Email address: alan@darh2.hr
3
Email address: nt@dgmr.nl
2

Ensure information on environmental noise and its effects is made available to the
public;
Adopt action plans, based upon the noise-mapping results, with a view to preventing
and reducing environmental noise where necessary and particularly where
exposure levels can induce harmful effects on human health and to preserving
environmental noise quality where it is good.
The END requires Member States to produce strategic noise maps for the main
sources of environmental noise, i.e. major roads, major railways, major airports and
agglomerations with a population of more than 250,000 persons in 2007 and those with a
population of more than 100,000 persons in 2012 and subsequent rounds. Estimates for
the 2007 mapping indicate that there were in the region of 160 cities to be mapped, and
over 85,000km of major roads.
The Directive sets out a requirement in paragraph 7 Data about environmental noise
levels should therefore be collected, collated or reported in accordance with comparable
criteria, and therefore proposes the development of a common methods of assessment
and recommends the use of EC recommended adopted Interim Methods to be used in the
meantime. Article 6 of the END does allow for the use of existing National Methods
providing that In such case, they must demonstrate that those methods give equivalent
results to the results obtained with the methods set out in paragraph 2.2 of Annex II. The
background to the basic requirement for an assessment of equivalence is discussed
further in the authors accompanying paper2.
In pursuance of an understanding of equivalence between National and Interim
Methods provision was made under a contractual arrangement from DG Environment3 to
DG Joint Research Centre (JRC) to prepare a set of technical protocols. These were
delivered to MS on 4 June 20084. Alongside the delivery of the protocols was an official
requirement to report back the results of an exercise on equivalency by the end of July
2008 either by implementing the JRC protocols or by provide their own proof of
equivalence by a means other than through using the JRC protocols.
In the Final Report on the project5, dated 18 December 2008, JRC noted that MS
showed to be reluctant in facing discussions concerning equivalency and indeed only
seven responses discussing the equivalency issue are provided in Annexes of the Final
Report, none of which have applied the JRC protocols. The authors had become aware of
this reluctance through contact with a number of MS national authorities, and in the
interest of widening the discussion on equivalence determined to undertake the application
of the JRC protocols as an independent undertaking.

2. EQUIVALENCE

The Compact Oxford English Dictionary6 defines equivalent as:


adjective (often equivalent to) 1 equal in value, amount, function, meaning, etc. 2
having the same or a similar effect.
noun a person or thing that is equivalent to another.
Within the Annexes of the JRC Final Report the MS take a number of different
approaches as to how equivalent is interpreted in the context of their response to the
Commission. These could probably be summarised as:
equivalent in use or application:
o it is possible to use the method;
o the methods had a similar original purpose or goal;
o the practitioners and legislators have experience and knowledge in its use
and interpretation;

the input datasets are similar and available;


the results generated are relevant in the context of the national legislation
and noise guidance, and will provide useful information within the Action
Planning phase of the END;
equivalent in elements of the computation methods:
o can the method be shown to have a similar approach to the adapted Interim
Methods?
source terms derived from similar factors, such as flows, speeds,
surface correction, gradient etc;
similar propagation factors taken into account, such as distance,
screening, air absorption, ground surface absorption, meteorological
factors etc;
adaptations have been added similar in approach to those added to
the Interim Methods;
equivalent in terms of agreement with measurements:
o studies undertaken to compare field measurements with calculations
undertaken using the national method or Interim Method, attempting to
demonstrate a closer agreement by the national method to the
measurements;
o
o

It is of note that none of the responses published to date take the one approach
which JRC had proposed through the development of the protocols, namely equivalent in
results generated. Indeed a couple of responses explicitly state that the results of
different national methods could not be equivalent. Even if one were to assess whether
results form a particular national method were indeed equivalent to an adapted Interim
Method, one would first need to determine the true result generated by the adapted
Interim Method for a given situation.
Defra research project NANR 937 developed the first map of uncertainty relating to
noise assessment, which was subsequently included within the WG-AEN Good Practice
Guide v28, p109. Whilst test protocols may seek to restrict the influence of some sources
of uncertainty, others will be almost unavoidable within a test procedure due to the
influence of the methodology, software, user settings and computational model datasets
on the final resultant level.
For example, the recommended interim methods are not accompanied by a set of
documented test cases or any type approval protocol; it is therefore considered likely that
different commercial software packages will produce varying results values when supplied
with the same input datasets. This is down to the influence of imprecise description within
current standards, and the variations between software tools as to how the procedure is
translated into the software tool, and how objects, geometries and assessments are
undertaken within each tool. This is an acknowledged area of uncertainty with a number of
existing national methods providing both test cases and acceptable tolerances for results9.
The JRC protocols attempt to reduce some of these influencing factors, and the approach
taken and results obtained are discussed below.

3. JRC PROTOCOLS
Four protocols were developed, one each for road, rail, industry and aircraft noise
assessments. The authors have restricted their investigations to the protocol for the
assessment of equivalency of road traffic noise methods at this stage.
The report describing the protocols is accompanied by a number of ESRI Shapefiles
providing the model data setting out the protocols for road, rail and industry. There were

two sets of delivered ESRI Shapefiles, for the road protocol these contained the following
seven model scenarios:
19 March 2008: City Depressed, City Embankment, City Flat, Open
Depressed, Open Embankment, Open Flat;
30 May 2008: Open Hills.
Table 1 sets out details of the model objects delivered for each scenario.
Table 1: ESRI Shapefiles delivered by JRC Protocols for each model

Case
City Depressed
City Embankment
City Flat

Barr

Bldg

Gabs

Hlin

Open Depressed
Open Embankment
Open Flat
Open Hills

Recv

Road

A. General description
The road protocol is a test environment for the assessment of road traffic noise. The
Interim Method is the French NMPB-routes-96. There are three main configurations for the
test environment: a motorway in an open situation, a main street representing a city and an
area of hilly terrain with short noise barriers.
The tested cross sections then include a case with a road on flat terrain, one with the
road on an embankment, and a third with the road in a cutting. There are additional
variations within this cases a there may be none, one single or two absorbing or reflecting
noise barriers alongside the road. There is also variation due to the presence of buildings
either parallel or perpendicular to the road.
Finally, the receptor locations may be free field, or 2m from a building faade. The
receptors may have an unscreened view of the road, be behind one barrier, or multiple
barriers.
B. LAeq values to be reported
The protocol requires the calculated LAeq level to be reported. The specific configuration
tables then set out traffic flow per hour, so one is to assume that the specific configuration
calculations will be for a 1 hour period.
C. Shape of the test environment
The majority of situations have a predominantly flat terrain profile apart from the two hilly
scenarios. Buildings are located within the model, and free field receptors are located at
25, 50, 100, 200 and 500m from the road.
D. Noise source descriptions
For the City models, the roads are to be standard two lane highways, with each centreline
in the model representing on carriageway carrying one-way traffic as per the appropriate
direction of flow. The traffic flow is assessed as pulsed and continuous. The composition of
the traffic is 300 vehicles per hour per lane, including 10% heavy vehicles. The vehicle

speed is set at 50 km/h for both light and heavy vehicles, and the road pavement is dense
asphalt concrete.
For the Open models, the roads are to be four lane highways, with each centreline
representing two one-way carriageways plus a narrow emergency lane without traffic.
There is a 2m separation between the directions of flow. The traffic flow is assessed as
steady and continuous. The composition of the traffic is 1000 vehicles per hour per lane,
including 20% heavy vehicles. The vehicle speed is set at 110 km/h for light vehicles and
90 km/h for heavy vehicles, and the road pavement is dense asphalt concrete.
E. Meteorological situations
There are two situations specified: (1) 100% favourable, which is to correspond to a nighttime assessment; and (2) 50% favourable which is to correspond to a daytime or evening
assessment.
F. Cross Sections
There are some 15 different cross sections identified, labelled A through Q. There are
receptors located along each of these cross sections.
G. Configurations
A table sets out a list of global calculation parameters to be used for setting up the method,
where applicable. These include temperature, relative humidity, wind speed, number of
reflections, source search radius distance, and building and barrier absorption coefficients.
H. Specific Configurations
There are some 20 Specific Configurations listed in Appendix 1 which are to be calculated.
For each Specific Condition a number of factors are listed which determines what has to
be assessed. The factors listed are: cross sections to be tested; road height; total number
of lanes; width of lanes; vehicle type; vehicle speed; vehicle flow; steady or pulsed traffic
flow; ground factor; and meteorological conditions.
For each of the 20 Specific Configurations it is necessary to cross reference the
cross sections to be tested against the seven model scenarios in order to determine
which models are to be run, then determine by means of the meteorological correction
whether the Specific Configuration is a 1 hour Lnight or Lday assessment, then the number of
lanes to determine the total flow per emission line.

4. UNCERTAINTIES ASSOCIATED WITH THE JRC PROTOCOLS


During the application of the JRC road protocol in two different software tools, using four
different methods of assessment, the authors noted a number of areas where the JRC
protocols suffered from imprecise descriptions, aspects without guidance and uncertainty
in application. A number of these will be discussed below to provide an overview. Whilst it
may be argued that each may only have a minor impact upon the results obtained,
collectively the impact could be of significance, but more importantly they highlight the
complexity of designing and documenting a clear unambiguous test protocol which
provides a precision, repeatable and accurate test environment.
A. General description
The Interim method is described as the French computation method NMPB-routes96.
However there is no specific mention of the EC recommended adaptations, the method is
not described as XPS 31-133 Interim Method as per the NANR 93 research and WGAEN GPGv2, and there is no confirmation that the software tools tested by JRC were
using the adapted Interim Method version, rather than the original French method.

The receptor locations for all protocol cross sections are located 2m from the building
facades, which is at odds with WG-AEN GPGv2 which recommends 0.1m from building
facades, apart from the assessment of a quiet faade which may be at 2.0m. Receptor
points are said to be located at a height of 4m, however it is not expressly stated that this
is 4m above local ground level, and this becomes relevant when some of the receptors
closest to the emission line are within the footprint of the modelled cutting or embankment,
here a 4m absolute level, and 4m relative level will result in quite different assessment
locations, as illustrated by Figure 1. The END and WG-AEN GPGv2 recommend 4m
relative to local ground, however the protocol test models have 4m absolute height
assigned, even for those within the extent of the cuttings and embankments. Should the
user follow the recommended approach, or the protocol test file?

Figure 1: Receptors close to the road, absolute or relative heights?


It is also relevant to mention that in many cases the receptors located behind multiple
barriers or buildings will not be testing the variation between methods of assessment;
rather they will be testing the programmers interpretation and solution to the generally undescribed situation on multiple barriers. These test situations are largely outside the
description of the documented standards, therefore they will be open to variance not
attributable to the methods, rather the software designers chosen solution.
B. LAeq values to be reported
The protocols are to be assessed to provide a 1 hour LAeq result. Appendix 1 states that
Due to the fact that eventual differences in dB between the LAeq values calculated
according to the interim method and those according to the national method will be
identical to those obtained if the Lden and Lnight formulas are used, only LAeq shall be
required. This may be the case for many national methods; however it is not the case for
the CRTN method used in the UK and Ireland. CRTN produces results in terms of the L10
indicator. The LAeq is then derived by applying a back end correction factor. The exact
correction factor to be applied is different depending upon whether the traffic data input is
for 1 hour, a complete day or evening or night period, or an 18 hour day 06:00 24:00 hrs.
the L10 to LAeq correction factor therefore varies between day, evening, night and Lden
periods, and between different types of traffic flow input data. The assumption set out
above therefore no longer holds and a source of uncertainty in application of the protocol is
introduced.

C. Shape of the test environment


The description mentions that the buildings are shaped to prevent any lateral diffraction,
however there is no mention of an appropriate setting to use with a software tool or
method of assessment which would normally use lateral diffraction.
The closest free field receptors are set 25m from the road emission line, however this
can mean that the receptor is within the region of the slope of the embankment of cutting,
as shown in Figure 4. In these cases it is important that receptors have relative height, and
not absolute height.
D. Noise source descriptions
The description of the traffic flow parameters includes a number of aspects which are
specific to the Interim Method. The traffic flow type can be difficult to map across into other
methods, and may not be possible in some methods. In several national methods, the
effects of acceleration and deceleration are assessed by the use of junction corrections.
The input files accompanying the test protocols did not provide any of such objects
although they might be the cause of the pulsed traffic flow in the urban test cases.
Rather than testing the assessment of different traffic flow types, which is rather
uncommon across the different national methods, it would have been useful to assess the
effects of road gradients, since correction factors for this phenomenon are part of many of
the national methods as well as the Interim Method.
The different speeds for light and heavy vehicles are also not supported in a number
of methods, but there is no guidance as to whether a single speed approach should use
the light vehicle speed, heavy vehicle speed, average or weighted average speed,
therefore introducing inconsistency in approach. The pavement type is specified as dense
concrete asphalt however this does not appear within the GPGv2 Toolkit 5, so the
selection of the correct correction factor is open to interpretation.
A question that may be raised is whether the objective of equivalence should apply to
the noise source description or not. Since the requirements for and the characteristics of
the traffic fleet vary among the EU MS as well as the appearance of winter tyres and
studded tyres, it would be rather surprising to find equivalent results from the different
noise source descriptions.
E. Meteorological situations
The JRC report setting out the instructions for the use of the test protocols states explicitly
that when a national method assigns specific values, those values should be used instead
of those set out in the JRC test protocols. As a basis of comparison, it is understandable
that equivalence of the national methods should be tested using the most straightforward
implementation of the methods, in order to minimise sensitivity for the chosen software
tools, flexibility introduced by software developers and reduce the effects of extra options
as much as possible. However, as the average meteorological conditions vary among the
different MS, input values for atmospheric temperature and relative humidity should merely
be considered as default input values to the calculations and not as a part of the method.
By using different values for the meteorological input, as suggested, the test protocols
introduce non-equivalence that may not be inherent to the calculation methods.
F. Cross Sections
The cross sections provided by the test protocols consist of a variety of free field receptor
locations and, in the urban cases, receptors behind the first or second row of buildings. In
many European cities, significant noise levels from mayor roads or railways penetrate
much further into the built-up area. For such situations, a few national methods provide
merely a macroscopic approach instead of calculations with discrete buildings. The Dutch

ministry of environment even designed such a solution explicitly for the purpose of noise
mapping. Whether such a macroscopic approach is equivalent to noise calculations by the
Interim Method, is a question that will not be answered by using the JRC test protocols.
G. Configurations
It is noted that in the global parameters the use of two reflections, rather than first order
reflections, is at odds with the recommendation in 2.43 of GPGv2. Evidence from across
the EC appears to suggest that first order reflections were used almost universally for
strategic noise maps.
The way in which noise calculation software packages deal with reflections against
vertical surfaces is perhaps the most diffuse part of software implementation and a major
source of implementation uncertainty. Calculation methods often give a lot of freedom to
interpretation of how to deal with ray curvature for reflected propagation paths, reflections
from small surfaces etc. In light of this information, the choice of second order reflections is
awkward since it will not lead to testing the variations between methods of assessment,
rather the programmers interpretation and chosen solution will be tested.
H. Specific Configurations
The requirement to cross reference the cross sections with the model scenarios, the
metrology with the day / night assessment period, and the number of lanes with the flow
per lane, can be deciphered with care; however it does not provide a clear and transparent
description of the testing to be undertaken. It may provide for a brief description; however it
does open up a potential for misapplication between testers, thus reducing repeatability.

5. TRUE VALUE FOR INTERIM METHOD


In their Final Report, JRC reported the results of an assessment of the road protocol they
undertook using three different commercial software packages, running the same models,
with the same Interim Methods, by the same user. Analysis of these results across the
1741 receptor locations indicates an average 95% CI of 4.6dB, and a range of 95% CI of
0.2dB to 33.3dB.
The authors ran the road protocols through two other noise mapping software
packages not tested by JRC. These were undertaken by highly experienced users of the
software tools, using the documents and files available from JRC, and setting the models
up to use the EC adapted XPS 31-133 Interim Method. The users discussed interpretation
of the JRC document and models at some length to agree a common approach to the
modelling. The results from these assessments where then placed alongside the JRC
results, and statistical analysis undertaken on the same basis. Across the five different
software implementations of the EC adapted XPS 31-133 Interim Method the average
uncertainty was 7.8dB 95% CI, with a range of 95% CI of 0.4dB to 28.1dB. Figure 2
presents the results of the calculations, with the five sets of results from the software
packages compared with the linear average result from the five results.
The 95% CI is based upon a fully expanded standard deviation, i.e. 4 x s, as there
were not enough samples to construct a full probability distribution curve. In line with other
uncertainty assessment work the 95% CI is expressed, as there is no information as the
relationship between the mean of the results and the true result.
The extents of these uncertainties clearly indicate the challenge associated with
transposing a documented standard into a software tool in the absence of clear and
unambiguous text, relevant test cases and validation protocols.
Following this assessment, for each of the receptor points within the protocol we now
have a mean value and 95% CI associated with the mean (2s). In line with general
statistical approach, should the resultant from an assessment of the same receptor, for the

same protocol, using a different method of assessment, fall within this range the results
could be said to be equivalent.

Figure 2: Five software implementations of the Interim Method

6. RESULTS FROM NATIONAL METHODS


Following the assessment using the EC adapted Interim Method, versions of the road
protocol scenario models were then created for the CRTN, RLS90, RMW2006 and
Harmonoise methods of assessment. The calculations were run in line with the instructions
within the JRC report, where possible, given the restrictions noted above. The results for
each of the national methods were generated in one software tool, by one user, and the
results for each receptor point compared with the mean and 95% CI value of the Interim
Method from the five software tools.
The results for the individual points are presented in Figure 3. Whilst complex analysis of
the results would probably constitute a further paper, the summary results may be
presented. When running analysis across the five methods, four national methods and
the mean value of the Interim Method, the average uncertainty was 16.4dB 95% CI, with a
range of 95% CI of 0.5dB to 58.6dB. Whilst Figure 3 may show CRTN displaying the
widest range of scatter, when compared with the mean Interim Method, it has the lowest
average deviation, of -0.29dB. The trend for RLS90 appears to have less scatter than
CRTN, whilst generally calculating lower than the Interim Method at low noise levels, and
higher than the Interim Method at high noise levels. The average difference was -1.4dB.
RMW2006 also appears to have a restricted scatter, however it also generally appears to
calculate levels approximately 5dB below the Interim Method, illustrated by an average
difference of -5.3dB. Finally, the Harmonoise results display a comparatively wide degree

of scatter, but again a general trend towards calculating lower levels, with the average
difference being -4.9dB.

Figure 3: Four national methods compared to the average Interim Method

7. CONCLUSIONS
The END requires Member states to obtain comparable results, either by the use of the
recommended EC Interim Methods, including the necessary adaptations, or by using
Identified National Methods, which have been adapted as necessary and demonstrated to
produce equivalent results to the recommended adapted EC Interim Methods. In light of
the general reluctance of MS to provide proof of equivalence between their selected
National Methods and the Interim Methods, DG Environment commissioned DG JRC to
develop test protocols for the assessment of equivalence.
The final report published by JRC acknowledges that the MS authorities were then
reluctant to apply the protocols as a means of testing equivalence, and due to the lack of
published information the authors determined to apply the JRC protocols as an
independent undertaking in order to widen the discussion on equivalence, software
standards, and the use of test models.
The results obtained through the application of the JRC protocols for road
methodologies indicate that it in practice it is somewhat challenging to determine quite
what the true result is for the EC recommended Interim Method, which in turn makes
comparison with other national methods a complex issue. When comparing the average
result for the Interim Method obtained, with four results for CRTN, RLS90, RMW2006 and
Harmonoise the average difference for RLS90 and CRTN of less than 1.5dB could suggest
a form of equivalence, although the point for point scatter of results presented in Figure 3
would probably suggest otherwise. The results obtained for RMW2006 and Harmonoise

indicate a general bias of lower calculated levels by around 5dB, compared to the average
value of the Interim Method, whilst RMW2006 resulted in less scatter of individual results
than Harmonoise.
Overall, the attempt at undertaking a test on equivalence is to be lauded, and the
process has the potential for producing results which may be able to inform ongoing
discussions regarding comparability of results, and the cross MS analysis being
undertaken by EEA on behalf of the Commission. Unfortunately there have been found to
be unclear descriptions, bugs and inconsistencies within the JRC protocols and the
accompanying reports which makes consistent, repeatable application of the protocols
somewhat challenging.

REFERENCES
1

Official Journal of the European Union, L 189, 12-25, 18 July 2002.


Equivalence within noise mapping projects, Shilton et al., Proceedings of Euronoise 2009,
Edinburgh, Scotland, 26-28 October 2009.
3
Administrative Arrangement No. 07-0307/2007/477794/MAR/C3 on Technical advice on the
equivalence of the national assessment methods used by the EU Member States for strategic noise
mapping specified in Annex II of the Directive 2002/49/EC and the EC guidelines adopted on 6
August 2003 stipulated in October 2007.
4
Implementation of Directive 2002/49/EC on Environmental Noise Protocols for checking the
equivalence of national noise mapping methods against the interim methods. 31 May 2008.
5
Administrative arrangement between DG ENV and JRC on Services to support the
implementation of Directive 2002/49/EC on Environmental Noise (Contract no. 070303/2007/477794/MAR/C3) FINAL REPORT (TECHNICAL REPORT no 2, 22nd October 2008 to
22nd December 2008) on Assessment of the equivalence of national noise mapping methods
against the interim methods, 18 December 2008.
6
Compact Oxford English Dictionary, Oxford University Press, 2009.
7
DEFRA Research Project NANR 93: WG-AENs Good Practice Guide and the Implications for
Acoustic Accuracy, Final Report: Sensitivity Analysis for Noise Mapping, HAL 3188.3/3/2, May 2005.
8
European Commission Working Group Assessment of Exposure to Noise (WG-AEN), Position
Paper, Good Practice Guide for Strategic Noise Mapping and the Production of Associated Data on
Noise exposure, Version 2, 13 August 2007.
9
E. H. van Banda, H. Stapelfeldt, Software implementation of the Harmonoise/Imagine
method, the various sources of uncertainty, proceedings of INTERNOISE 2007, Istanbul, 2007
2

Vous aimerez peut-être aussi