Propagation is similar to ISO 9613 (3) and takes into consideration geometrical divergence,
air absorption in addition to ground, barrier, reflection and meteorological
effects. However, the fact that railways are dipole sources is also taken into
account in the method.
3.
IMPLEMENTING The Dutch Railway Noise
Model
RLM2 has been
implemented in Brüel & Kjær’s Predictor Type 7810 software based on
previous experience gained from implementation of the 1987 version, user
feedback and authority guidelines. While the new method improved the modelling
of the source, the main aims of the software implementation were to reduce
errors in interpretation of the method in the software to a minimum and reduce
user error by providing an intuitive user interface. Thus, close cooperation
between users, developers of the method and the developers of the
software was essential.
Previously, calculation methods were developed and described for manual
calculation with calculators or spreadsheets. A software developer interprets
the text and implements it. This gives rise to possible errors due to
imprecision in the text and small mathematical differences between the text and
the software. The individual’s use of the software then adds an extra source of
error. In order to improve the quality of results, the following steps can be
taken:
·
Provide authorized test cases for the method
·
Provide uncertainty limits in the method description in
general or for these test cases
·
Standardized implementation of parts of the method
·
Consultation between the developers of the method and
the developers of the software
·
Intuitive
implementation
of the method using the software by users
·
Extensive testing
RLM2 is no exception. However, there are no uncertainty
limits given in the RLM2 method – neither for the input
nor the result. And, unlike the CRTN method (4), the RLM2 method includes no test cases for developers to validate their software. On the other hand, the
angle of sight, number of reflections and source modelling (source height and
railway profile) were standardised. The weaknesses in the method description
were countered by intensively pursuing the other aspects:
·
Consultation between the developers of the method and the
developers of the software to clarify implementation doubts
·
Extensive testing of individual factors in the method
(e.g. divergence, barrier effect) against results calculated in spreadsheets by
method experts, external to the development group
·
Extensive testing by several users experienced in the
method. Complex models (e.g concerning reflections or
screening) were compared to expected results
·
Intuitive
user interface of the software by user involvement and beta releases
·
Information to users on how to implement the method in
help, manuals and at seminars
The verification method used for Predictor
follows the Nordic method for verification of environmental calculation
software (5) where the software’s calculation result in several
test cases are compared to result limits (in this case internal limits) for
each software release version.
4.
Application outside THE NETHERLANDS
In Holland, the traffic flow of all trains
is standardised in one database (the “Acoustisch Spoorboekje”) which everybody
can buy. This is maintained by the Dutch Ministry of the Environment and
contains the number and types of trains, the speed, the track construction,
etc. for all railways in Holland at a resolution of 100m, valid for the current
year and for a future year. This further reduces discrepancies in results
between different software and different users.
The RLM2 method is suitable for use
internationally. This is a major reason for its selection by the European
Commission as the de-facto standard for use in the proposed Environmental Noise
Directive for noise mapping. The method is one of the newest railway noise
calculation methods available. The propagation part of the method is close to
ISO 9613, purely physical and thus international. The source model is based on
Dutch trains, rolling stock and track. These are similar to other European
countries and therefore can be used internationally. As the RLM2 source model
produces a sound power/metre, alternative source models based on national databases
can easily be implemented.
Outside Holland, the following factors are
probably not standardised and will reduce the accuracy of the method:
·
standardised calculation factors such
as angle of sight and number of reflections used
·
standardised source model (source
height and railway profile)
·
a national traffic flow database
Thus, the European Commission is urged to
take these factors into account. In addition, implementation of the method in
software packages without detailed knowledge and experience of the RLM2 method
may contribute to reduced accuracy.
5.
Application to noise mapping
The RLM2
method is an octave-band method. This is a slower calculation method than a
broad-band LAeq-based method as 8 octave bands have to be calculated
for each source-receiver combination before the total A-weighted result can be
seen. Compared to LAeq-based methods, this reduces speed by a
comparable factor. However, the accuracies of LAeq-based methods are
poor in complex situations where several attenuation factors (e.g. reflections
and screening) are present. The improvements in accuracy greatly outweigh the
reduction in calculation speed.
The RLM2
method has already been successfully used to map railway noise over large areas
in the Netherlands. For example, the entire length of the railway line from
Rotterdam Harbour in the West of the Netherlands to the German border in the
East was mapped to determine whether the 57 dB(A) limit at dwellings would be
adhered to. In total, an corridor of 1 km or so was mapped over the entire 150
km long line. When this was first done some 5 years ago, the calculations of 20
to 30 km long stretches of the track were performed overnight. Today, with
modern computers and calculation software, calculation speeds have increased
tenfold, resulting calculation times of an hour or so.
In urban
areas, calculations can often be limited to shorter propagation distances, in
particular in areas where other sources of noise (e.g. road traffic) dominate.
This further increases calculation speed. On the other hand, more closely
spaced receiver points are often required to cover the noisy and quiet facades
of residences with the necessary resolution. In built up areas, it may be
beneficial to increase the number of reflections to improve accuracy at the cost
of extra calculation time. For these quiet facades, the effects of reflections
are particularly important. However, software implementations differ in their
treatment of multiple reflections and results from different users and software
will have larger discrepancies.
6.
CONCLUSION
Implementations of the Dutch Railway Noise
Model have been developed for over 20 years. During this, the model itself, its
implementation in software and its use have improved and now give good results
in the Netherlands through the use of standardised input parameters.
Implementation of the model in Brüel & Kjær’s Predictor Type 7810 software is based on
experience gained from implementation of the 1987 version, user feedback and
the guidelines that have been developed. The main aims of the software
implementation were to reduce errors in interpretation of the method in the
software to a minimum and reduce user error by providing an intuitive user
interface. This involved cooperation between users, developers of
the method and the developers of the software. Despite (or perhaps because of)
a lack of authorized test cases and uncertainty limits for the method,
intensive validation was undertaken.
The RLM2 method is suitable for use
internationally. However, outside Holland, some non-standardised factors will
reduce the accuracy of the method. The European Commission is urged to take
these factors into account. In addition, implementation of the method in
software packages without detailed knowledge and experience of the RLM2 method
may contribute to reduced accuracy.
Compared to the world of sound level meters, calculation software methods
in general still lack accuracy requirements and test procedures in standardized
quality assurance regimes. Adding this to new versions of current methods or to
the new common EU method should be a high priority for the controlling body
(the European Commission). This will result in better methds, better software
implementations and better results.
7.
REFERENCES
1. The Dutch Railway Noise Model Reken en Meetvoorschriften Railverkeerslawaai '96, Centrale Directie Voorlichting en Externe Betrekkingen, 14/1997, 1996
2. Proposal for a Directive of the European Parliament and of the Council relating to the Assessment and Management of Environmental Noise European Commission, COM(2000) 468 final 2000/0194(COD), 2000
3. ISO 9613 Part 2 Acoustics – Attenuation of Sound During Propagation Outdoors – Part 2: General Method of Calculation, 1996
4. Calculation of Road Traffic Noise, HMSO, ISBN 0 11 550847 3, 1988
5. Acoustics. Framework for the Verification of Environmental Calculation Software, Nordtest Remiss 1455-99, 1999
