Skip to main content

Virtual certification of acoustic performance for freight and passenger trains

Final Report Summary - ACOUTRAIN (Virtual certification of acoustic performance for freight and passenger trains)

Executive Summary:
Noise pollution is an important concern. At a European level the Environmental Noise Directive (END) requires noise mapping of large agglomerations and major routes and the development of Action Plans. It is estimated that 14 million people in Europe are exposed to levels of railway noise above 55 dB Lden, compared with 125 million people for road traffic [1]. In parallel, regulations controlling the noise emission from individual vehicles have been introduced. For the railway sector these were introduced from 2002 in the form of Technical Specifications for Interoperability (TSI) which limit the noise emission of new vehicles. The latest revision of the Noise Technical Specification for Interoperability (TSI) enters into force in January 2015.

Although the Noise TSI is strategically important in ensuring that the railway remains an attractive means of transport in the future and in allowing the sector to grow, it also means restrictions and increased costs to individual stakeholders. The conformity assessment required in the Noise TSI is mainly based on field tests, which is often a very expensive and time consuming process. Noise measurements are required under pass-by, stationary and starting conditions as well as inside the driver’s cab.

The ACOUTRAIN FP7 European Research project, coordinated by UNIFE – The European Rail Industry, aims to develop procedures and calculation tools to simplify the Noise TSI test procedures. From October 2011 to December 2014, the fifteen ACOUTRAIN partners have cooperated for introducing virtual certification for noise with a reliable simulation approach.

Moreover the ACOUTRAIN project has also contributed to the progress of other railway noise research areas such as:
- Methods for separation of infrastructure and rolling stock noise contributions: Three separation methods have been proposed and tested against measured data. These methods can be used either to transpose a vehicle to a different track or to transpose to another vehicle on the same track. The concept has been extended to a proposal for transposing measured results to a virtual reference track.
- Establishment of measurement procedures for new running conditions (e.g. braking and curving): A new procedure for type testing vehicles for curving noise has been proposed, including an on-board occurrence test procedure combined with trackside pass-by test.
- Development of procedures to obtain inputs for the European Noise Directive (END): A procedure to use ”virtual testing” models and data for the definition of equivalent noise sources for the END noise mapping has been established. This procedure leads to realistic vehicle source power allocation without costly testing for application of the END.

Project Context and Objectives:
The main goal of the ACOUTRAIN project is to speed up the rolling stock authorisation process by introducing some elements of virtual testing while retaining the same degree of reliability and accuracy. This general objective could be split in four scientific objectives:
1. Introduce virtual certification with a reliable simulation approach;
2. Establish a method for separation of infrastructure and rolling stock noise contributions;
3. Establish measurement procedures for new running conditions, specifically braking and curving;
4. Develop procedures to obtain inputs for the European Noise Directive.

In order to reach the four aforementioned objectives, each ACOUTRAIN Work Package had its own objectives:

WP 1: Procedures for a virtual certification of acoustic performance of freight and passenger trains
The basis for numerically assessing the noise emitted by a vehicle is to consider the rolling stock as a set of noise sources. To do so, the equipment contributing most in terms of noise is identified and turned into equivalent point noise sources. The same process is carried out to consider rolling noise when it is required (for pass-by simulation): wheels, rails and sleepers are represented as equivalent point noise sources.

In ACOUTRAIN, the rolling stock acoustically defined as a set of noise sources is called a Virtual Vehicle. This virtual vehicle is defined in a dedicated software called a simulation tool. Then, the overall noise level emitted by the rolling stock is generally assessed by energetic summation of the different equivalent noise source contributions, at the reception point. The simulation tool handles the propagation of noise contributions by taking into account environmental conditions such as ground effect and Doppler effect.

The use of such a virtual testing approach within the scope of a certification process requires:
o A full control of the virtual testing process (input - noise sources- characterization, calculation of the noise propagation in complex environment and computation of noise indices);
o A clear definition of “when and where” this process could be used (in which case).

This Work Package dealt with putting together the results of the other WPs to write a holistic procedure for the use of virtual testing in a Noise TSI certification process. Moreover, work was carried out in this Work Package for studying other scenarios of pass-by apart from the one covered by the TSI requirements: in particular curve and braking noise were studied. Finally, it was also proposed to check the potential of using the certification process output for supplying some inputs required by the noise mapping process such as noise source inputs.

WP 2: Noise sources: Rolling Noise
Rolling noise is usually the predominant source of noise from running trains at conventional speeds and remains important even at high speed. In order to implement virtual testing it is therefore paramount that reliable predictions of rolling noise can be achieved.

Methods for predicting rolling noise, particularly the TWINS model, have been available since the 1990s. However, despite validation with field tests and widespread use throughout the industry, there remained aspects of the models and their use that required further refinement as well as more precise definition of the method of operation.

In addition, the wheel and rail roughness are important input parameters for predictions of rolling noise. The measurement of wheel roughness is not standardised in the same way that rail roughness is.

Separation of the contributions of wheel and track is a key issue, which is required if measurements in one situation are to be translated to another. For example, measurements made on a track which is non-compliant with TSI requirements could be used with a transposition procedure to determine the noise levels that would apply on a compliant track.

The objectives of WP2 were:
o To define more precisely methods to predict rolling noise reliably and consistently;
o To develop and define methods for source separation;
o To determine the limits of the accuracy and applicability of ‘translation’ procedures;
o To increase confidence in wheel roughness measurement so that it can be used in virtual testing;
o To quantify the variability in rolling noise due to variability in parameters describing a situation;
o To improve models for rolling noise.

WP 3: Characterisation of vehicle specific noise sources
The scope of this Work Package was to characterize vehicle specific noise sources relevant for the operational scenarios required for Noise TSI approval. The distinction “vehicle specific” means that all noise sources influenced by wheel-track interaction (i.e. rolling noise, curve squeal noise, braking noise) are excluded. More specifically, the following types of noise sources are addressed:
o Cooling and ventilation systems;
o Electric power and auxiliary systems (converters, transformers, inductors);
o Electric traction systems (motor and gearbox);
o Diesel engines and exhaust systems;
o Aero-acoustically generated noise for high train speeds.

Models for analyzing installation effects of screens and enclosures, including analytical diffraction models, energy boundary element methods and ray-tracing, are also applied and benchmarked against insertion loss test data for assessment of integration effects when the source is mounted on a vehicle.

The main challenges of this Work Package were related to the presence of a large variety of noise sources on a train leading to the need for different source rating methods. A major task was to recommend source assessment methodologies depending on source type as the ISO standards available for determining source power and directivity are frequently difficult to apply for real vehicle sources. It must be considered that the potential cost saving benefits of virtual testing may be jeopardized if standardized laboratory source characterization is required for all components on the trains. Therefore one main challenge was to develop and benchmark methodologies for source strength assessment that can be used within the industrial design process, including estimators of source variability and uncertainties in the source model.

WP 4: Methodology to certify tools for acoustic virtual testing
The scope of this Work Package was to provide a procedure to validate simulation tools for pass-by noise and standstill noise in order to apply them for future certification purposes.

The capabilities of such simulation tools have to be defined, based on:
o The input and output parameters needed;
o The accuracy expected;
o The validation process that has to be satisfied.

The first objective of this Work Package was to set the minimum performance requirements for such simulation tools. Within the project a new simulation tool was developed that satisfies these requirements and can eventually be used by third parties.

In order to be able to define criteria to certify the tools, one important part of this Work Package was to define reference cases mainly defined from analytical models.

WP 5: Validation of the noise virtual certification
The main objective of this Work Package was to validate if the requirements regarding applicability and reliability were fulfilled for the new virtual testing procedure proposed by ACOUTRAIN. The validation was performed by comparing results achieved with the new virtual test procedure to measurement results. Several application cases were defined that correspond to typical scenarios that a certification procedure has to cover and for which a virtual testing procedure would be of advantage. Two of these cases with the highest priority were chosen for the validation. It should demonstrate that the required accuracy is reached for all cases covered by the method and that significant sources of error and uncertainty are identified as well as the means to control these.

In order to perform the validation suitible data must be collected. One important goal of Wp5 was to perform a large measurment campaign with a test vehicle to assess pass-by noise, stationary noise as well as important parameters like rail- and wheel-roughness and track dynamic properties. Moreover, the noise sources of the test vehicle were to be charachterised to be used as input for a virtual vehicle of the test vehicle. Two different simulation tools were used to calculate the virtual vehicle and the results were compared to the measurement results.

Project Results:
A long term goal of ACOUTRAIN was to provide the foundation for the introduction of increased virtual testing in the Noise TSI. Virtual testing is less common in the acoustic field than other engineering fields but models for calculating the exterior noise emission based on sound power inputs of the relevant sources are already available and extensively used in decision support tools during the design phase of new rail vehicles.

However, for such models to complement real testing within a certification process, the tools themselves as well as the methods to assign acoustic source strengths, must be thoroughly scrutinized and a rigorous tool validation and verification procedure must be in place. Such procedures have been developed within the ACOUTRAIN project.

The Work Program of the project was organised around five technical Work Packages:

• WP 1: ‘Procedures for a virtual certification of acoustic performance of freight and passenger trains’;
• WP 2: ‘Noise sources: Rolling Noise’;
• WP 3: ‘Characterisation of vehicle specific noise sources’;
• WP 4: ‘Methodology to certify tools for acoustic virtual testing’;
• WP 5: ‘Validation of the noise virtual certification’.

The main scientific and technologic results and foregrounds developed during the project are described in the following sections dedicated to each WP.

WP 1: Procedures for a virtual certification of acoustic performance of freight and passenger trains
The preliminary basis of the use of virtual testing in the TSI certification process have been estbalished by two means:

o The first achievements was to give more clarity to the so-called “simplified method” already mentioned in the current Noise TSI (ref: 2011/229/EU). This method has been proposed as an alternative process, for specific cases, compared to the full measurements required in the Noise TSI. ACOUTRAIN has proposed 9 flowcharts (see public deliverable D1.1 Clarification of the simplified method in the partial revision of the TSI) that clarify the use of a simplified method, again for specific cases, and therefore support its use within the scope of a TSI certification process;
o Secondly, it was defined the first recommendations for the use of virtual testing (VT) within the scope of an acoustic certification process (see public deliverable D1.2 Recommendations for a future certification process). These recommendations give the framework of a VT process:
- The specific vocabulary for acoustic VT;
- The “when and how” applying VT for different possible scenarios (distinction is made between full virtual testing, hybrid virtual testing for which measurements are still used at least to validate the calculations; and extension of approval case, for which the studied vehicle is designed according to an already certified rolling stock);
- The documents that have to be provided: the way information is shared with the Notified Body in charge of the certification;
- The validation process that explains how evidence will be brought to prove the VT reliability, compared to a full measurement process.

Then the ACOUTRAIN project delivered the final recommendations for using virtual testing in an acoustic certification process, within the scope of Noise TSI process (see public deliverable D1.8 a virtual certification process). They are based on the different results and conclusions gathered in the other Work Packages during the whole project.

Moreover a numerical methodology has been developed to compute, from the tools developed for virtual testing such as virtual vehicle, the input required for noise mappings, according to the CNOSSOS method (noise mapping computation as required by the European Noise Directive) – see public deliverable D1.9 Transposition procedure for END noise sources.

Also measurement procedures for curve squeal and braking noise have been developed. These procedures are based on existing ones (proposed in ISO 3095 (Railway applications — Acoustics — Measurement of noise emitted by railbound vehicles) or as final report of the Curve Squeal project) with improvements defined according to the feedback of measurement teams. The curve squeal measurement procedure (that proposes an on-board measurement with microphone in the bogie area, to quantify the propensity of the rolling stock to squeal in various curve conditions) has been tested by ACOUTRAIN.

WP 2: Noise sources: Rolling Noise
It is important, especially in the context of virtual testing, that the results from calculation models are independent of the user. Experience shows that significant differences can be obtained by different users due to the large number of input parameters and sub-models available in a package such as TWINS. Comparisons were therefore made between a set of benchmark calculations carried out by four partners, three of them using the TWINS software. The results have helped to define values for sensitive parameters and give recommendations for good practice.

Results have been published as a ‘user guide’ describing procedures for modelling wheels and tracks (public deliverable D2.1 User guide describing procedures for modelling wheels and tracks). Recommendations have been made in public deliverable D2.6 (Proposed acceptance procedures for different wheel and track types) about how to use calculations for the acoustic acceptance of new wheel designs, extending the existing procedure in EN13979- 1 (Railway applications — Wheelsets and bogies - Monobloc wheels - Technical approval procedure — Part 1: Forged and rolled wheels).

Sources of uncertainty in rolling noise predictions are often associated with input parameters such as wheel and rail roughness and track decay rates. A study has been performed to quantify the effect of these uncertainties on the overall rolling noise (public deliverable D2.3 Report quantifying the sources of uncertainty and their effect on the outcomes of virtual testing). Rail roughness is routinely measured and is the subject of an international standard (EN15610, Railway applications — Noise emission — Rail roughness measurement related to rolling noise generation). Limits for rail roughness are included in ISO3095 (Railway applications — Acoustics — Measurement of noise emitted by railbound vehicles) and the Noise TSI. The wheel roughness is often just as important in determining noise levels but there are no standards for its measurement. A measurement and analysis protocol has therefore been proposed (public deliverable 2.4 Proposed analysis method for wheel roughness) and tested by a number of partners within the project.

In acoustic certification the aim is to quantify the noise production of the train but in practice the track plays at least an equal role in the noise generation. It is therefore essential to be able to separate the contributions of wheel and track. This then allows the results obtained at one location to be ‘transposed’ to another. Work on this topic of separation and transposition has compared a number of approaches and tested them on data from several different cases. In reality it is not straightforward as large changes cannot be predicted reliably and small changes fall within the measurement accuracy so it is difficult to demonstrate that they can be predicted reliably. Nevertheless this remains a very important topic. The results of this study are issued in public deliverable 2.5 (Source separation and transposition techniques).

A number of aspects of the rolling noise models were being addressed in fundamental research. This has result in recommendations for improved model components (public deliverable 2.7 Improved model components) which can form part of future revisions of models such as TWINS.

WP 3: Characterisation of vehicle specific noise sources
Computationally efficient source models of aerodynamic noise from pantographs have been developed and validated using wind tunnel experimental data. For fan sources a methodology where the source is modelled as a small number of equivalent monopole sources has been successfully applied using in situ HVAC cooling fan data. Also it has been shown that fairly simple analytical models of screens can be used to determine installation effects of roof mounted sources.

In addition significant progress has been made regarding: (i) design parameters for the sound power generation of transformers; (ii) modelling of diesel engine encapsulations; (iii) wind-tunnel validation of component based models for aero-acoustic noise generation of pantographs and bogies; (iv) transfer functions for bogie and roof mounted sources based on loudspeaker tests; (v) modelling of installation effects of bogie and roof mounted sources; (vi); validation of a sound intensity method for power determination in a typical industrial test environment; (vii) source power and directivity of traction motors and HVAC systems at different rpms and loads.

The main output was a methodology for determining source descriptors for significant vehicle sources. The source descriptors can be in the form of, (i) source power and directivity or (ii) as a number of equivalent monopole/dipole sources, and should be in such a format that they can be used as input for the simulation tools used for virtual certification, e.g. the ACOUTRAIN tool developed in Work Package 4. For each type of system studied, an important task was to determine an appropriate level of complexity for the source model needed, ranging from a single uniform point source to a more general case built up of several equivalent sources with a certain geometrical distribution and complex radiation directivity patterns. Estimators of the accuracy of the source models in terms of variability based on the source testing methodology applied are also addressed.

WP 4: Methodology to certify tools for acoustic virtual testing
The capabilities of a simulation tool have been defined in terms of the definition of the input/output data. These have been reported in the public deliverable D4.1 (Report with definition of input/output data for each global model). This phase highlighted the differences between the tools available in the consortium (Sitare from Alstom and Vamppass from SNCF). An ACOUTRAIN simulation tool dedicated to the certification process that can also be used by third parties has been developed, see public deliverable 4.2 (Basic global prediction tool and user manual). This tool allows computations to be made with different ground impedance models.

To be able to certify a tool, it was necessary to build reference data based on typical situations. More than thirty test cases have been defined. These cases were based on sources defined analytically like monopoles and dipoles. The aim of those cases was to be able to test tools on their ability to take into account a propagation model depending on the type of ground (grass, ballast and reflecting ground). Some cases with a moving source are also included. It has provided the opportunity to clearly analyze the influence of the main parameters as the input data but also the influence of the theoretical models used.

The results of the study show that, besides the effect of the ground impedance, height of the source and receiver position, also the directivity of the source has a big influence on the final result. The influence of the ground topography has been also studied. Overall, the influence of a complex topography compared to a situation for which the ground between sources and receiver is defined as a completely flat ballast bed is rather low and limited to low frequencies.

To have also experimental data, a flat ground with straight ballasted track and grass field adjacent was characterised. Those tests have produced experimental results for similar cases to the numerical ones. It was also measured the impedance of a grassy ground. The comparison with numerical cases allows the project to consolidate the choice of the numerical formulation. The experimental cases were also interesting to verify the choice of the model used for grassy ground. Nevertheless, it was noticed that the case of ballasted groundwas not so well mastered: it was difficult to measure the impedance of ballast and to have a perfect model for all the frequency range. The project noticed that models of ballast used up to now give rather lower values for frequencies below 400 Hz compared to experiments.

Among all the numerical test cases, the most representative cases of the phenomena to be modelled were selected. Combinations of several cases among 26 test cases were selected (24 static cases and 2 cases with moving source). All the selected cases have been computed with the available tools for the static configurations. Globally, good results have been obtained.

A guideline defining criteria to decide if a tool can be certified or not for virtual certification of trains was produced.

WP 5: Validation of the noise virtual certification
A measurement campaign on a new electric multiple unit train (EMU) was performed, applying the new methods that were developed as well as the established ones. This campaign gave valuable data for the validation of the application cases and demonstrated the typical use of the new procedure. Each noise source was analysed in detail and it resulted in proposals for improvements both for source characterisation and the simulation tool calculation.

In Wp5, the first recommendations for the use of virtual testing as proposed in Wp1 were further elaborated and procedures were developed to ensure the reproducibility of the virtual testing method. The methods that had been developed separately in the more technical work packages were integrated in a complete procedure that should be reliable and applicable after the finalisation of the project. The proposed procedure covers each step of the acoustic vehicle certification including methods for choosing the suitible virtual testing approach in a certain case and the validation of the virtual vehicle. Besides the technical aspects, the validation in WP5 also evaluated the feasibility of the new procedure from a user’s point of view. The purpose was to give recommendations regarding which cases fulfil the requirements for the process and which weak points still have to be resolved for the final procedure that was proposed in WP1.

It will take time to introduce virtual testing into the Noise TSI procedure and for it to gain wide acceptance but the concept of virtual testing has great potential to achieve significant cost and time savings in the certification of new, quieter railway vehicles.

ACOUTRAIN has delivered a framework for noise virtual certification including definition of three different modes of application. Dedicated tools for virtual modelling of TSI Tests have been developed. Moreover testing and modelling procedures for characterization of significant vehicle and track sources have been developed and evaluated. However more work is needed to validate and refine the virtual certification concept and to detail the validation of reference vehicles, including modelling updating procedures and best-practice for source representations. Moreover a cost-benefit analysis of different virtual testing scenarios needs to be produces in the future.

Potential Impact:
One of the main objectives of ACOUTRAIN is to reduce the time and cost for the conformity assessment against the Noise TSI certification of new interoperable rail vehicles, and by doing so reduce the time to have better performing railway vehicles operating on the railway network.

The ACOUTRAIN project further promotes the operational and technical integration of the different national railway systems in the European Union and accession countries. A TSI conformity assessment based on virtual testing will reduce the need for testing on dedicated tracks and test sites, for which access is rather limited today.

The ACOUTRAIN project:
• is dedicated to both freight and passenger trains and has the objective to reduce cost and time to market by simplifying the acoustic certification. The simplification of the conformity assessment procedure is one element for removing obstacles to the introduction of new vehicles and it will stimulate the expansion of the railway market.
• aims at facilitating the certification of railway rolling stock against EN standards and TSI. This will contribute to the promotion of rail transport both for passenger and freight purposes as an alternative to other modes, thus fostering environmental and economic sustainability of transport.
• leads to a better understanding of noise sources which will enable more effective control of noise emissions and assessment of the impact of quieter technologies.

In order to maximise the communication and exploitation of project outputs, the partners of project used several effective communication systems: participation to congresses, technical fairs, publications in scientific journals, design and operation of a public area in the web site.

A public website was created and maintained, and information relevant to the project was made available through this. The website was updated as the project progressed. 1flyer and a Newsletter were produced and distributed at the numerous events at which ACOUTRAIN was presented.

Articles on the project were published in mainstream railway sector publications (European Railway Review, EURAIL mag).

Examples of the fora in which ACOUTRAIN was presented include:
• UNIFE General Assembly 2012, 2013 and 2014;
• InnoTrans 2012 and 2014;
• Transport Research Arena 2014;
• International Workshops/events on Railway Noise;

In order to ensure that project outcomes are well known and reach targeted decision makers, the project made sure that the whole railway industry and the railway operators (including also the ones not involved in the consortium) are aware of the virtual certification process defined by ACOUTRAIN through association internal committees and ACOUTRAIN public website. Moreover the ACOUTRAIN results were presented during CEN Working Group 3 (acoustic) meeting and shared with theEuropean Railway Agency Noise Working Party o link the ACOUTRAIN project with the review of the Noise Technical Specification for interoperability (TSI).

In order to ensure that the project outputs reach targeted decision makers who will implement them and use ACOUTRAIN results for integrations to existing standards and processes for virtual certification a liaison with ERA, NSAs, NoBos and CEN/CENELEC was implemented.

List of Websites:
More information is available on the project’s website:

Coordinator’s Contact detail:

Nicolas FURIO
UNIFE - The European Rail Industry
Avenue Louise 221
B-1050 Brussels
Office: +32 2 626 12 62