Final Report Summary - DYNOTRAIN (Railway Vehicle Dynamics and Track Interactions Total Regulatory Acceptance for the Interoperable Network)
Executive Summary:
The certification of a rail vehicle according to European regulations, Technical Specifications for Interoperability, European Standards and national safety rules represents a significant element of both vehicle cost and time to market. Indeed, a large part of vehicle certification mandates testing for safety, performance and infrastructure compatibility. DynoTRAIN aimed to help meet the business scenarios listed in the ERRAC SRRA 2002 and 2007 by aiding the spread of European certification and acceptance procedures to speed up interoperable product approvals, thus making the process more efficient, while at least maintaining current levels of safety. In the field of vehicle dynamics, there was already a wide range of standards and regulatory requirements for the certification of rail vehicles. DynoTRAIN sought to promote interoperable rail traffic in the EU, partly by showing how virtual certification can be applied in the sector (and how this might be able to replace the existing ‘physical’ tests. A further important objective was to provide information to assist in the closing of some of the ‘open points’ in the relevant TSIs. In fact, DynoTRAIN was able to make a positive contribution to both the rolling stock (Loc & Pas) and infrastructure (INF) TSIs, which were revised during the course of the project. The project also contributed important data to the revision of one of the most relevant European standards – EN14363. The focus of the project was therefore on using the TSI route to consolidate the methodologies allowing the free exchange of certification data. The project was part of the TrioTRAIN cluster (Total Regulatory Acceptance for the Interoperable Network) which aimed, among other important objectives, to propose an innovative methodology via computer simulation/virtual certification to allow multi-system network and route approval in Europe to become a faster, cheaper and better process for all involved stakeholders. DynoTRAIN is the final project from this cluster of three to finish its work. This report explains the work carried out and main results achieved by the DynoTRAIN consortium.
Project Context and Objectives:
DynoTRAIN is a European research project funded under FP7. It is part of the TrioTRAIN cluster of projects, which aims at further promoting interoperability by increasing virtual certification, thus contributing to the competitiveness of rail. Through the DynoTRAIN project, TrioTRAIN addresses rail vehicle dynamics, which is one of the most relevant issues for a rail vehicle certification.
Background
Certification against EN standards on railway dynamics in particular, together with the relevant technical annexes of the TSIs, following various tests methods, extend train delivery times for months. Besides, the tests do not always capture all operating conditions. There is thus a risk of failure or unsafe approximation in such tests. In addition to this, some uncontrolled environmental and other boundary test conditions combined with restrictive operational limits can influence results. The costs and duration of tests performed in such conditions are also often increased by the need to do these tests several times so as to explore as much as possible all the range of environmental and boundary conditions and secure the results.
Project Objectives
The overall goal will be achieved by the following high level objectives:
• Address HS & CR TSI’s that effectively work to harmonise European and national standards on railway dynamics and track interaction to reduce costs and time of certification
• Reduce costs and time of certification by replacing existing tests
• Reduce costs of certification by introducing virtual testing
• Close “open points” in the HS and CR TSI’s
• Establish standardised conditions for derivation of results.
Project Results:
Introduction
WP1 –Measurements of Track Geometry, Contact Geometry and Vehicle Reactions.
This work package involved a test campaign that toured Europe, and visited:- Germany, Italy, Switzerland and France. As far as the project is concerned this is the largest test campaign ever undertaken anywhere in the world. The test covered 7500km of track and recorded 4.7 terabytes of data. The train incorporated four types of test vehicles (locomotive, coach and two freight wagons (that were tested in both tare and laden conditions) and included in the train was a track recording car so synchronised data of track input and vehicle reactions could be recorded simultaneously. In addition static tests were undertaken on all the test vehicles.
This work packaged has completed all objectives and has accumulated an enormous database that could be used for future railway research projects.
WP 2 – Track Quality Geometry
Current standards specify track quality in terms standard deviation and peaks values for short sections of track. Extensive measurement results were processed and reported. This analysis is the basis for Dynotrain making changes to EN14363, A new analysis method has been developed, the method “multi-regression “ as given a new understandings of the relationship between vehicle reactions and track quality. CEN.WG.10 have agreed to introduce multiple regression analysis in the next revision of EN 14363.
Analysis of the this vast track quality database identified many weakness in the current network measuring systems, the work package has made recommendations to improve the measured data by decolouring and improved filtering along with other quality checks.
The work package has studied a huge number of track quality assessment methods and surprisingly concluded that the current geometric methods are still superior.
Data was analysed to compare Network track quality, the analysis showed that on high speed lines the standard deviations were similar but there was considerable variation on slow speed lines.
WP3 – Contact Geometry
One of the most important objectives of the work package was to close “Open Point” in the TSI’s for all equivalent conicity issues.
Millions of wheel /rail combinations where considered in arriving at conicity recommendations at a European level. With the database available it was possible to make firm conicity recommendations to CEN for inclusion in the next issue of EN14363 which covers vehicle certification for running safety and recommendations at a system level to ERA for inclusion in both rolling stock and infrastructure TSI’s. It should be noted that the work package was not able to provide in-service limits for the TSI’s because an economic evaluation was required to apportion costs between rails and wheels to control conicity. This evaluation was outside the scope on Dynotrain. However, the work package was able to provide the following:-
• The methodology to be used for the economic evaluation. ERA.
• Equivalent conicity limits for vehicle certification, EN 14363
• Equivalent conicity limits for guidance of IM’s and RU’s when investigating in service instability, TSI CR INF. And TSI Loco and Pass.
The work package has also effectively undertaken an adhesion survey during the test campaign on small radius curves.
WP4 – Track Loading Limits Related to Network Access
The work package reviewed all national and international standards and the framework for cross acceptance. All the relevant parameters for track force assessment were reviewed along the national track construction and maintenance standards.
Sensitivities to track force inputs, relationships with damage mechanisms and practical operating controls were all considered.
For cross acceptance, national requirements can be assessed by use of existing data to compare with existing vehicles and multiple regression analysis can also be used with either test or simulation data. Also, the project has provided guidance on how to use the multiple regression analysis.
The results of WP4 have been presented to the ERA Cross Acceptance Unit and will be taken into account in future Guidelines.
WP5 – Model Building and Validation.
The work package has investigation and proposed procedure for validation of multi-body vehicle models to replace track testing for certification of vehicles running safety. This procedure has been recommended to CEN for inclusion in future issues of EN 14363. Validation methodology is based on:-
• Comparison of simulations data with vehicle tests data using force measuring wheel sets and body accelerations.
• A minimum of 12 test sections as required by EN 14363 with the four zones considered.
• A minimum of 24 data points comparing simulation to measurement for each of the 12 limit criteria being assessed.
• Difference limit criteria between simulation and test to be able to assessment each quantity in terms of standard deviation and mean difference. All must be satisfied for model validation.
Comparison of models with stationary tests was found to give mixed results, in some case it helped validation and others worsened to agreement.
Measured track data and contact geometry helped model validation but other factors are also found to be important.
WP6 –Virtual certification of modified vehicles running in other conditions.
Many aspects of railway safety cases used risk based criteria. Other industries such as the nuclear power generation use similar methods. This work package has investigated a probabilistic approach to running safety with convincing results.
Numerical simulations have been undertaken with a range of vehicle types on Virtual Track representing the various test track zones given in EN14363. This work provides a virtual solution for vehicle cross acceptance processes and certification. The method also allows the assessment of vehicle modifications and vehicles operating under different infrastructure conditions.
The methodology provides a solution to close an “open point” in the TSI on how to assess the importance of vehicle modifications.
WP7 –Regularity Acceptance
The work of regularity acceptance has been a continuous process with regular progress meetings with ERA, CEN and NSA’s. We have undertaken two formal progress meetings to update CEN WG 10, who are responsible for Vehicle Acceptance in respect of running safety. This process was also supplemented by the fact that four of our WP leaders are members of the CEN WG10.
This dissemination process has been very successful with all Dynotrain recommendations for changes to standards and TSI’s being accepted by the Regulation Bodies.
Test Vehicles
In the DoW three different vehicles are listed for the on-track tests, namely loco BR120 from DB, a passeger coach and a freight wagon with Y25 bogies. In addition Alstom suggested during the kick off meeting in August 2009 in Minden to include a 2-axle freight wagon with UIC double link suspensions in the test train because this type of wagon is frequently used in Europe by railway operators. Due to the friction elements in the Y 25 bogie and the progressive behaviour of the UIC double link suspension, the running behaviour of freight wagons is stongly dependent on the loading condition. Therefore the members decided to measure freight wagons in a fully loaded and in an empty condition.
Since it was not possible to provide a modern passenger coach with air suspensions by any partner, a passenger coach with steel spring suspensions form DB was used.
Stationary tests
The stationary tests were conducted before and after the on track tests. The main purpose of the tests is to provide data for the validation of the simulation models in WP5.
In order to get track access permission for the loco BR 120 in France and in Italy, a sway test was necessary to provide vehicle data for the determination of the kinematic envelope of the vehicle. In addition safety against derailment for running on twisted track has to be demonstrated.
The dynamic behaviour of the vehicle depends amongst others on the following parameters:
• The resistance against twist of the vehicle in transition curves.
• The lateral and rotational resistance of the bogie during the passage through curves, switches and crossings.
• The load distribution and the centre of gravity of the vehicle.
• The natural frequency of the eigenmodes of the car body and the bogie.
• The vehicle plays (wheelsets and bogie)
Most of these parameters were determined by the following stationary tests:
• Wheel unloading test and determination of Ct*on a test rig
• Running through a 150 m curve at very low speed
• Lateral bogie resistance test
• Rotational bogie resistance test
• Wedge test
• Measurement of the gaps of the wheelset and the bogie in lateral and longitudinal direction
The measurement results were provided by excel files in order to make the comparison between measurement and simulation easier. The excel files are stored on the server in Minden (www.dynotrain.de).
On track tests
The test vehicles were equipped with a lot of different sensors in order to measure the vehicle reaction to external excitations like track defects, alignment and changing contact conditions. Since the main objective of WP1 was to gather as much information as possible for the validation of vehicle models in WP5 many more sensors have been installed than would be necessary for an authorisation test. The control centre and the amplifiers of the vehicle reaction measurement were located in measuring coach of the train.
All of the measurement equipment was provided by DB except an angle of attack measurement system designed by IFSTTAR, which was installed in the first bogie of the passenger coach. The signals of the angle of attack measurement system will be sampled by DB and processed by IFSTTAR.
In total 222 parameters were measured for the running dynamics of the vehicles (55 parameters at loco BR120, 46 parameters at passenger coach Bim 547.5 44 parameters at empty and 29 parameters at loaded Sgns691 wagon, 34 parameters on empty and 14 parameters at loaded Laas wagon). In addition, the speed v and the unbalanced lateral acceleration aq, the pressure in the main brake pipe, the environmental temperature, the rail temperature, the air-pressure and the humidity have been recorded.
The track geometry was measured with RaiLab I, which is a regular track inspection vehicle from DB Netz AG. Lateral alignment and longitudinal level of the rail are measured with an optical laser system, whereas the lateral alignment is measured 14 mm under the top and the longitudinal level is measured at the top of the rail head. Both sensors are incorporated in a measurement frame which is mounted in the bogie. The measured signals refer to an inertia platform in the car body. The inertia platform represents a curve in space. Therefore the transfer function of the measurement systems is identical to 1.0 which means track defects are represented in their true amplitude and shape. Track defects are measured up to a wavelength of 70 m.
Rail profiles were measured with an optical laser system, which is mounted under the car body of the measuring coach. A separate computer controls the laser, cameras and stores the data. The laser flashes a beam over the cross section of the rail profile every 2 metres. A camera takes a photo from the laser beam and the pixels of the photo are analysed in order to digitize the rail profile.
The test train consisted of the 6 test vehicles mentioned above, 6 braking coaches, a measuring coach for sampling the vehicle reaction measurements and a track geometry measurement coach called RaiLab I. The on-track tests and their statistical evaluation were performed in October 2010. The on track tests were performed in Germany, France, Italy and Switzerland. In total, 4.7 Terabyte of data were sampled and are available on the web server in Minden (www.dynotrain.de).
WP2 Track geometry quality
Problem and Objectives
Rolling stock dynamic behaviour is a key safety requirement for the approval of rolling stock. This dynamic behaviour is strongly dependent on the track geometric quality as well as other parameters such as speed, cant deficiency, etc. Hence information on track geometric quality is required for the design and approval of rolling stock. Such track geometric conditions are defined in EN 14363 for approval of rolling stock as well as in EN 13848-5 for track geometric quality assessment.
Both standards use standard deviations and maximum values of alignment (rail coordinate in lateral direction) and longitudinal level (rail coordinate in vertical direction).
This approach involves the problem that the maximum values of track irregularities standard deviations do not correlate with the maximum values of the wheel-rail forces. We may find significantly different vehicle reactions at the same level of track geometry (standard deviation) as shown by many authors.
The poor correlation makes it very difficult to compare test results from different test tracks with differing track geometry.
This problem has been well known for a long time. There have been many attempts to define a method for describing track geometry or track geometry assessment method (“TGA”) which considers the typical vehicle behaviour.
One important objective of Dynotrain WP2 was to analyze, develop and assess different track geometry description methods. The aim was to find out, which track geometry description method gives the best correlation to the force reaction of “typical” vehicles and propose which one should be used in future.
Other objectives included proposing target values for a possibly proposed description method, to develop a method for transforming test results to other track geometry conditions, to develop a method to estimate maximum track geometry defects a vehicle is able to negotiate safely and to provide a track geometry data base as input for virtual homologation and for design of interoperable rolling stock.
Method and results for comparing track geometry assessment methods
This analysis was based on the results from on-track tests conducted in WP1, with synchronous measurement of track geometry, rail profiles and vehicle reaction. As this tests have been carried out in Germany, France, Italy and Switzerland and with six vehicles including a loco, a passenger coach and four freight wagons a wide range of conditions has been covered.
To assess and compare the effectiveness of different methods for the description of track geometry, a statistical approach was taken. It uses the vehicle assessment methods of EN 14363, where the vehicle behaviour is characterized by maximum, mean or rms values in track sections of 70-500 m length. On these statistical values a multiple regression model is applied with a defined set of input variables such as speed, curve radius, cant deficiency, equivalent conicity or the radial steering index. For dynamic vehicle assessment values, a Track Geometry Assessment result is included in the regression model as one or more additional input variables. The TGA method and its parameters are then varied. The quality of the regression model is characterised by its coefficient of determination R2, which is the ratio between explained variation and total variation. The TGA which explains the largest amount of variation or highest R2 is deemed to be the best one.
A complex software system had to be developed to do this analysis. It is completely open to new or changed TGAs or to new test results or even simulation results. It includes the full evaluation method from EN 14363 including signal processing, handling of track layout, statistical analysis on section values, etc.
A lot of different track geometry assessment methods have been applied on the measured track geometry and vehicle reaction data. Inputs to the TGAs are the measured track geometry parameters, but also derived signals like twist and differences left/right rail. Different wave length ranges have been studied as well as “chord transformations”, multi parameter assessments, derivatives, parameterization by triangles or wavelet transforms. Additionally a number of more complex algorithms (filters of MISO type) like PMA, WGB, Pupil, ETFE, adaptive filters have been studied.
First of all it became very clear that including track geometry in the statistical analysis reduces the unexplained variability significantly, in particular on straight track. Up to now, track geometric quality is not included in the statistical analysis of EN 14363, it is only prescribed to meet certain quality levels. Including track geometric quality by means of multiple regression gives a much lower unexplained variation and enables larger samples with a number of advantages. Track geometry as input parameter in a multiple regression allows also translating results obtained under certain track conditions to different track conditions simply by changing the target value of the TGA parameter at which the estimated maximum value is computed.
Many alternative proposals of track geometry description do not improve the regression model. Only in a few cases we find small improvements, e.g. if the wave length ranges are changed or if more than one track geometry parameter is used (alignment/longitudinal and twist or cross level).
More complex methods, like Pupil or adaptive filters, show potential improvements, but up to now not for all vehicle assessment parameters. In this field some further work is necessary.
Special attention has to be paid also to the uncertainties of the measurements. All measurements, track geometry and vehicle reactions are not perfect. Measurement errors may have a significant influence on the characteristics of the regression models. This has to be studied in future and may explain some of the differences between assessments based on measurements and on simulations.
Analysis of measured and provided track geometry data
Track geometry data has been made available by Deutsche Bahn (Germany), Network Rail (UK), Trafikverket (Sweden) (via KTH), Société Nationale de Chemins de Fer Francais SNCF, Réseau Ferré de France RFF (France), Ferrovie dello Stato, Rete Ferroviaria Italiana RFI, Trenitalia (Italy), Jernbaneverket (Norway). In total track geometry data for lines of 11000 km is available.
Additionally from WP1 track geometry data from four countries is available (approx. 3200 km).
It turned out, that many of the measurements show implausible characteristics. One measurement seems to be completely wrong; others contain parts with wrong measurement results. Speed category 80 km/h is especially affected from this
Different lines in the same network and the same speed category show in many cases large differences in the quality levels. These differences are generally larger than the differences between networks.
Besides a few exceptions, which can be explained, the 90%-tiles of standard deviations do not differ very much between the networks within a speed category. For lower speeds the collected data shows higher values than the test lines from WP1, at higher speeds (starting at 160 km/h) vice versa.
Looking into details we find in some cases significant differences in the spectral content. Especially in high speed lines we see very different shapes of the Power Spectral Density plot (PSD) of the longitudinal level.
The differences between the speed categories are significant and generally high. But we have to distinguish between parts of a line with locally reduced speed (mostly because of curves) and secondary lines with a general low speed.
On some networks and speed categories we find pronounced periodic deterministic processes. Often the reason are rail joints, especially at high speed lines other characteristics of these periodic deterministic processes are found which have to be originated by other features of the track.
A big part of the tracks show an unexpected shape of the PSD for longitudinal level. We found a significantly lower gradient between 8 and 20 m than at wave lengths below and above this range. Reason for this behaviour is unknown.
Besides the analysis of measured track geometry de- and recolouring methods have been developed and studied. Chord measurement systems exhibit a transfer function unequal 1, which may vary between 0 and 2. Decolouring is a process where chord measurement signals are transformed to signals equivalent to those from inertia based systems. This process can be applied also to the opposite direction.
Different methods for this process have been developed and applied.
The analysis showed clearly the problems associated with chord measuring systems. Special care has to be taken on dealing with the zeros of the measurement system’s zeros.
Method for estimating target values of track geometry and maximum defects
A method for defining a target system by means of multi body simulations. Since alternative characterisation parameters showed no significant improvement, the method utilizes existing assessment parameters of the present EN standards.
Five reference vehicles have been selected for the application of the method, representing different types of vehicles. These vehicles have been simulated on up to 400 track sections considering track layouts according to the test zones of EN 14363:2005, and track irregularities of Track Quality Class E of prEN 13848-6. The track geometry quality of these tracks has been assessed by means of standard deviations of track irregularities within wavelength range D1, which is 3m to 25m wavelength. The simulated vehicle responses to the track have been evaluated using maximum quantities for track loading and running safety as defined by EN 14363.
A statistical analysis is performed by calculating the regression and confidence intervals between vehicle response and track geometry quality. The intersection point of the upper confidence interval and the limits for the vehicle response according to EN 14363 is analysed.
The method can be used to yield a target system for track geometry quality and to determine at which types of isolated defects the reference vehicles reach limit values. However, due to simulation related issues, values for a target system have not been given.
Track geometry database
A European Virtual Test Track (EU VTT) toolbox built in Maltab has been constructed. The EU VTT allows measured track data to be read from various networks and for various line speeds, and recombine into a single Virtual Test Track of a more manageable size for virtual certification using vehicle dynamics simulation software. The toolbox includes signal analysis and plotting capabilities for viewing and checking input data as well as the produced EU VTT. Tests section based on the standard deviation of vertical and lateral irregularities are selected according to EN14363 (upcoming revision) test requirements.
Transforming test results to differing track geometry conditions
The current state of the art is to test and assess vehicles according to EN 14363 [5]. This is done for all new vehicles and if changes are applied to vehicles already in service. EN 14363:2005 allows two methods for statistical analysis: the one-dimensional method and a simple regression with cant deficiency as input variable. Therefore it is complicated or even impossible to transform these results to other conditions.
Multiple regression and a wider range of test conditions (input variables) allows transformation of the results.
Multiple regression methods have been applied to the test data from Dynotrain WP1 and the different vehicles. It has been shown what the requirements are to gather robust and statistically correct results. This affects also the standard methods within EN 14363 and especially the new method of multiple regression.
A number of conclusions have been drawn and some recommendations are given for an optimization of the analysis process in EN 14363. They promise to simplify the testing effort by a more sophisticated statistical analysis.
WP3 Contact geometry
Problem and Objectives:
The running stability of rolling stock is a key issue for the approval of rolling stock on the interoperable network.
The running stability is strongly dependent on the contact conditions between wheel and rail on straight track and in very large radius curves. A key parameter used to describe these contact conditions is the equivalent conicity.
If the equivalent conicity is high the natural wavelength of the wheelset will be short and the lateral acceleration value will be high. If the vehicle is not designed to withstand this motion this may lead to severe hunting with large forces between wheel and rail resulting in increased deterioration of both track and rolling stock and ultimately it may also be a safety matter.
The higher the running speed and the higher the equivalent conicity the larger is the risk of severe hunting. A common way of designing vehicles for high equivalent conicity values is to prevent the hunting motion itself. This is done by using a high horizontal stiffness of the wheelsets and high yaw damping of the bogie frame. These effective measures are often taken but they will have the effect that the horizontal forces between wheel and rail in curves and switches will increase and hence leading to a faster track deterioration.
A design for high equivalent conicity values will often have a detrimental effect on the track friendly properties of the Rolling Stock and will also lead to increased wheel wear.
In the case where the range of actual equivalent conicity is unknown, the Rolling Stock designer has no choice but to design for what he believes is the highest possible equivalent conicity to be expected.
The objectives of WP3 were the following:
- To close the open point equivalent concicity for in service in TSI Infrastructure and TSI Rolling Stock.
- To confirm the usage of radial steering index in order to avoid additional tests on other networks and define / confirm the range of application;
- To develop a method for the measurement of the wheel / rail friction coefficient during on track test;
- To set up a track data base (together with WP2) with:
o Rail profiles for the design of interoperable vehicles,
o Rail profiles for virtual homologation,
o A set of wheel / rail friction coefficients for virtual homologation.
Work carried out in this project:
The WP3 was divided in 7 tasks, namely
• Task 3.1: Collection of track data for the track conicity and radial steering capability
• Task 3.2: Collection of wheelset data for the wheelset conicity
• Task 3.3 Collection of wheel / rail friction coefficient data
• Task 3.4: Analysis for first selection of reference profiles and speed bands
• Task 3.5: Creation of Equivalent Conicity maps
• Task 3.6: Selection of limit values
• Task 3.7: Simulations considering the changing contact geometry over running distance
In task 3.1 worn rail profiles were collected for further analysis in task 3.5. Since the wear behaviour of the rail surface depends on the rail inclination, the curve radius, track construction and on the type of vehicles operating on the track, a big number of rail profiles were investigated in order to obtain a representative picture of the contact conditions on the TEN network. Therefore the rail profile measurements in WP1 were analysed in terms of equivalent conicity. In addition infrastructure managers, which are members in this project, provided a considerable number of rail profile data from different routes too. So in total rail profiles from UK, France, Switzerland, Italy and Germany were available for different speed categories.
In task 3.2 worn wheel profiles were collected for the same purpose as in task 3.1. Since the wear behaviour of the tread of the wheel might depend on the rail inclination, the bogie design and on the operating conditions a big number and variety of wheelsets were investigated in order to represent different kind of rolling stock operating on the TEN network. Therefore the partners provided a considerable number of wheel profile data.
In task 3.3 measured lateral and vertical wheel-rail forces were compared with simulated ones. After certain basic corrections, the input friction for the simulations was iteratively modified so as to improve the agreement in measured and simulated forces. The final input friction value was then considered as the friction estimate for the test runs. In addition an algorithm has been proposed for estimating the wheel-rail friction based on lateral/vertical/longitudinal wheel-rail forces, wheel contact position and wheelset angle of attack. The three forces and the contact position were used to obtain the friction estimate. The angle of attack was introduced to improve the estimated total creep which, together with the estimated spin, was used for quality assurance of the friction estimate. The algorithm was evaluated by simulations only since measurements of all five quantities for the same vehicle were not available.
In Task 3.4 reference profiles for wheels and rails were defined for the calculation of conicity maps. The reference wheel and rail profiles act as a sort of co-ordinate (scaling) system for the conicity maps. It is important to note that the reference rail profile is determined by the frequency distribution of worn wheel profiles and vice versa. The reference profiles for wheels and rails should cover a wide range of equivalent conicity values.
In Task 3.5 conicity maps were calculated from selected wheel and rail profiles which have the same frequency distribution as the whole sample. The conicity maps were calculated for different speed categories and for wheels operating on networks with a rail inclination of 1/20 and 1/40 respectively. In addition non-linear parameters of worn rail profiles were calculated. According to [20] the non linear parameter is a performance parameter which describes the vehicle behaviour at the stability limit.
From the conicity maps calculated in task 3.5 limit values of equivalent concity tan e for the homologation of vehicles and in-service limits for tracks were derived in task 3.6. Thanks to the investigations of WP3 the open point “in-service conicity” in the TSI Infrastructure [13] and TSI Loc&Pas [12] has been closed.
Last but not least, simulations carried out in task 3.7 confirmed that the methodology proposed by the EN15302 standard (mean value over 100m) is accurate enough for the calculation of equivalent conicity for varying rail profiles. Therefore there is no need to calculate this parameter taking into account the varying profiles in a more accurate manner. Concerning the curve radius it is shown that the conicity function can be significantly influenced by curve radius. By direct analytical calculation of a minimum radius based on flangeway clearance, it was shown that sufficient clearance between the flange and the high rail is available in radii below the current limit of 2500m down to around 1500-1000m. Verification of stability behaviour using vehicle dynamics simulation showed that for freight traffic running below 120km/h, hunting is generally not an issue below 2500m.
WP4 Track loading limits related to network access
The focus of WP4 was on cross acceptance related to track loading limits and therefore this work has concentrated on the vehicle assessment parameters related to track forces rather than those considering derailment risk or other safety performance. There is a concern that the requirements for cross acceptance that are used today are leading to excessive cost and time delay in obtaining approval in an additional member state for vehicles which are already operating successfully in one or more networks.
The framework for vehicle authorisation in Europe and the role of the European Railway Agency in cross acceptance work has been reviewed together with some experience of cross acceptance on a bi- and multilateral basis. Specific national requirements may result from national legal requirements and/or may reflect special conditions of track construction or maintenance. In most cases no scientific justification is given in the available documents.
A range of track force assessment parameters are applied for vehicle acceptance including cross acceptance and for technical compatibility checks. These parameters have been investigated and found to be useful for assessing the influence of a vehicle on track forces and the deterioration of track and track components. The limit values related to these various parameters have been developed over many years and, in some cases, the background is not clear today. Some limit values are based on scientific studies while others are based on experience and comparison with existing vehicles which were at that time believed to be at the limit of acceptability for the track.
The objectives of DynoTRAIN WP4 were:
• To improve the process of cross-acceptance between different countries
• To reduce the number of additional tests required for approval on non-TSI compliant infrastructure
• To develop limit values related to infrastructure construction and maintenance
• To clearly identify specific local requirements
• To provide proposals for cross-acceptance processes including use of simulations
• To provide proposals for operating limits dependent on infrastructure conditions.
The current requirements across a range of different countries have been identified and listed with the relevant conditions of application and limit values. The requirements are divided according to whether they are referenced in:
- International documents (TSIs, ENs and UIC leaflets)
- National documents (European) – including Specific Cases or National Technical Rules mentioned in the TSIs.
The relationship between the existing limit values and the track construction was then investigated by simulation technique using available vehicle data to simulate the vehicle and track interaction for a series of vehicle and track conditions. In this study the track construction is regarded as the infrastructure type including the maintenance condition.
The investigations were complemented by a second study using simulations of the relationship between vehicle service condition and the dynamic forces in track elements, assuming that the vehicle type of interest has heavy axle loads (locomotive and freight) and track conditions were large (R>600m) radius and very small (250≤R<400) radius curves.
In a questionnaire it was found that most networks use a track categorisation system based on speed and service density. For line speeds ≥ 160 km/h the current standards for track construction across the member states appear to be similar. Differences between construction and maintenance standards are, as expected, more significant for lower speed lines. On lower speed lines in some countries a ‘weaker’ track condition may require a lower limit for one of the vehicle assessment parameters.
For each of the identified track force criteria the sensitivity to different input parameters was investigated by multiple regressions on test data from WP1 and simulation results from locomotives and a loaded freight wagon. The most relevant parameters were identified. Summary tables indicate the relationships between:
• Vehicle assessment parameters and track deterioration effects,
• Track installation and maintenance conditions and different track damage mechanisms,
• Vehicle assessment parameters and track / operating conditions.
These tables allow potential operating controls to be considered that may control the track deterioration if the track is weak or if certain vehicle performance parameters have high values. In combination with the developed methods for extended evaluation of data from testing and from simulation, additional testing for track loading is no longer necessary in most cases.
Possible tools to handle compatibility between vehicle and infrastructure have been identified including the use of multiple regression of test- and simulation data to describe the track loading under different operating conditions related to track layout, track maintenance and track design parameters. There should be no special requirements related to the design rail inclination on different networks if a suitable range of equivalent conicity is covered for stability assessment.
The use of multiple regression analysis allows the estimated maximum value for relevant parameters to be evaluated for different target conditions and then compared with the appropriate limit value, or with values for existing, comparable vehicles. This method can be applied to either test data or results from a validated dynamic model. This will enable the amount of testing needed to demonstrate vehicle compatibility with different infrastructure conditions to be reduced. Some recommendations for the application of multiple regression have been developed.
WP5 Model building and validation
The objectives of DynoTRAIN WP5 were:
• To review the state-of-the-art of building and validation of multi-body railway vehicle models
• To test these models by comparisons between simulations and measurements
• To specify the requirements for agreed process of validation of vehicle models for virtual certification.
The DynoTRAIN WP5 investigations were structured into 5 Tasks. The review of state-of-the-art of modelling and validation conducted in Task 5.1 resulted to a paper presented on the IAVSD Symposium in Manchester 2011 and published in [1]. A questionnaire and presentations by project partners about their experience with model validation carried out in the framework of Task 5.1 showed rather small experience due to limited availability of measured data. The missing track irregularities or/and rail profile data were usually mentioned as the reason for the deviations between simulations and measurements. Task 2 was dedicated to investigations about suspension modelling experiences, which was then used for the modelling of vehicles used in the validation investigations. The preparation of vehicle models and the identification of uncertain or unknown parameters by comparisons with stationary tests was the topic of Task 3. Tasks 4 and 5 were dedicated to the validation studies and analyses, which resulted in the proposed new approach for the validation of vehicle models with regard to the vehicle acceptance.
The simulations, comparisons with measurements and evaluations carried out in DynoTRAIN WP5 were performed using 10 vehicle models, built with the use of different simulation tools by different partners; using over 1,000 simulations of validation exercises. For all vehicles tested in WP1, the comparisons between simulation – measurement used the same selected test sections. The comparisons between simulation and measurement using values based on EN 14363 represented more than 50,000 single comparisons. The evaluation of time or distance as well as power spectral density diagrams compared the simulation and measurement in more than 21,000 plots. The subjective validation assessments were accomplished by a workshop during which 26 recognised European running dynamics experts and project partners assessed a set of 120 selected diagrams.
The following models of vehicles tested in DynoTRAIN WP1 were developed and compared with measurements:
• Locomotive DB BR 120 – model by Siemens in simulation tool Simpack
• Locomotive DB BR 120 – model by IFSTTAR in simulation tool VOCO
• DB passenger coach Bim – model by Bombardier Transportation in Simpack
• DB passenger coach Bim – model by IFSTTAR in VOCO
• Empty freight wagon Sgns with Y25 bogies – model by Technical University Berlin in Simpack
• Empty freight wagon Sgns with Y25 bogies – model by IFSTTAR in VOCO
• Laden freight wagon Sgns with Y25 bogies – model by Technical University Berlin using Simpack
• Laas 2-axle flatbed wagon unit modelled by Alstom in Simpack.
Furthermore, other models of two recently developed vehicle types were assessed using measurement results provided by the suppliers of the vehicles. A high speed train delivered to Turkey by CAF was modelled by this company using the simulation programme SIDIVE, and the train DMU IC4 delivered by Ansaldobreda to Denmark was modelled by the vehicle supplier in Simpack.
The investigated validation assessments by comparisons between the simulations and the measurement results consisted of:
• Assessment based on measured quantities, filtered and processed by analogy with EN 14363.
• Subjective engineering assessment using a simple “Yes/No” method by project partners as well as during a project workshop with invited experts.
• Validation metrics, i.e. quantitative measures comparing simulation and measurement in the time histories with the aim to maintain agreement with engineering judgement.
The correlations between the different groups of assessment as well as the relationship between the assessments and the achieved results were investigated. Neither the subjective engineering assessment not the validation metrics concluded to be suited for an objective validation method.
The project investigations showed that to approve the model validity, it is more important to assess an overall agreement between simulation and measurement, than to concentrate on maximum differences in a few individual comparisons. An assessment of a large set of comparisons between simulation and measurement values evaluated by analogy with EN 14363 was finally selected as the best way to an objective and reliable model validation. The final method, criteria and validation limits were derived based on the proposals of project partners and also on the results achieved in WP5 to ensure the feasibility of the proposed validation method.
The proposed criteria and validation limits are based on 12 quantities covering the quasi-static and dynamic wheel/rail force measurements and vertical as well as lateral vehicle body accelerations. For each quantity a set of at least 24 comparisons between simulations and measurements is evaluated using values based on EN 14363 from at least 12 sections, which represent all 4 test zones according to EN 14363.
A difference between the simulated value and the corresponding measured value is evaluated for each value and each quantity. The whole set of differences simulation – measurement per each quantity is then assessed statistically. The following values of the whole set of differences per quantity are calculated and compared with the corresponding validation limits:
• Mean value
• Standard deviation.
The proposed validation method represents an overall assessment of a large number of data which is not practical to carry out by using engineering judgement of the plots, as it would involve having to display, check and document the approval of such a large number of plots. The specified set of 12 quantities to be evaluated covers the quasi-static as well as dynamic behaviour of the vehicle in regard to the vehicle acceptance, which is the intended range of the application for a validated model. The vehicle’s safety relevant behaviour and the track loading results are validated by comparing quantities measured using force measuring wheelsets. The representation of vehicle ride is validated by comparing the rms values and maximum values of the car body accelerations.
The proposed validation method, criteria and limits were applied to all 78 model configurations evaluated in this project. Only 20 of 78 model configurations fulfilled the proposed model validation limits. The validated models were only those vehicle models, which were tested in the framework of DynoTRAIN WP1, where the measured track layout and track irregularity data were available and could be used for the model validation. The models successfully validated were only the models of locomotive DB BR 120 and DB passenger coach Bim. Neither a model configuration of the freight vehicles nor a model configuration without usage of measured track irregularities could be validated. This result demonstrates the difficulties of modelling freight vehicles with large uncertainties of friction suspension parameters, and stresses the importance of actual infrastructure data for a successful model validation.
The application results demonstrate, that the proposed validation limits are sufficiently strict, and it shows, that the proposed method represents a robust validation approval. Various investigations into adjusting the validation limits demonstrated that a slight increase of the validation limit of a single quantity did not change the number of successfully validated model configurations.
The identification of model parameters can be supported using comparisons of stationary tests measurements with simulation of those tests. However, the DynoTRAIN investigations showed that an improvement of the overall model quality regarding the simulations of on-track tests using comparisons with stationary tests is often marginal.
References:
[1] S. Bruni, J. Vinolas, M. Berg, O. Polach and S. Stichel: Modelling of suspension components in rail vehicle dynamics context. Vehicle System Dynamics 49 (2011), pp. 1021-1072
WP6 Virtual certification of modified vehicles and vehicles running in other conditions
The vehicle/track system is complex and non-linear containing a number of sources of variability. The track geometry, its quality and the track stiffness change from one location to another. The contact conditions vary greatly along the track, for example the friction coefficient depends on local environmental conditions, rail profiles wear at different rates and to different shapes depending on age, traffic and curve radius and this results in varying contact conditions. The mechanical properties of a vehicle are also variable, for example their mass and inertia depend on the number of passengers, or distribution of goods. In the same manner, there is variability in the suspension characteristics, as their design and manufacturing processes are numerous (for example there is a large range for the elastomers), as they stem from different suppliers, and as they change during the service life of the trains.
The current approach to characterization of the system non-linearities is to achieve a number of measurements in different locations and to post-process the output using statistical processing.
In the automotive field, the nuclear field and in the off-shore energy fields, complex systems are already certified using experiments on sub-systems and simulation on the whole system. When there are high safety concerns the system is often re-certified regularly during its life cycle. In the energy field, the state moreover asks the operators to certify that the probability of incident is lower than a given threshold. A full probabilistic approach is then required. The nuclear field has thus developed open source numerical tools to handle the variability in complex, non-linear systems.
In the work achieved in WP6, some of these methods have been applied to the virtual certification of different track/vehicle systems. These are always based on a three step approach: the description of the mechanical problem, the identification of the uncertainties and the propagation through the modelling.
In this context, it has been shown that the track geometry, the stiffness and the contact conditions play a key role on the extreme values studied during the certification. In order to accurately take into account these effects, a method to generate representative virtual tracks has been proposed.
These tracks are built from the concatenation of measured track sections, according to the standard requirements. The number of sections used, their origin and their specificities (e.g. curve radius, quality, etc…) make the method highly versatile and capable of taking the testing conditions beyond those possible during on-track tests. The work shows that a larger number of track sections would enhance the precision of the estimated values.
To build these virtual tracks an EU VTT toolkit compliant with EN14363 has been developed. A library of Virtual Test Tracks, which have been built based on WP1 measurements, using a combination of input data from all four networks: Germany, France, Italy and Switzerland has been set up. It has moreover been shown that it can be difficult to match the EN14363 requirement in terms of the mean radius of all curves for testing in zone 3. But using virtual test tracks, it is easily conceivable that the simulation environment could be slightly modified so that the representative spread of test curves fit the requirement without changing the quality of the estimated certification output.
Variability of rail profiles and of the friction coefficient also has to be introduced. What has been proposed in this work is to make a random choice of friction coefficient and of a rail profile from the current track section for each concatenated track..
Taking into account the variability of vehicle parameters would also be interesting. For practical purposes, the proposed procedure is only based on the assumption that a normalized and validated model for the train is available, for which mechanical properties have been accurately identified from experimental data. Nevertheless, in order to verify the robustness of the vehicle model and also to be sure that the parameters important for the certification criteria are well modelled, it is important to achieve a sensitivity analysis before the whole virtual certification procedure. In this regard, several sensitivity procedures have been described and tested.
Finally, it has been pointed out that wear and changes in the mechanical characteristics of the train can significantly modify its dynamic response during its life cycle. Taking advantage of the numerical possibilities, it could be interesting in the near future that the certification not only focuses on the response of new trains but also tries to integrate the behaviour of this train over its whole lifecycle. However in this case the limit values should be changed since they include a margin to allow for change in behaviour over the normal maintenance cycle
In order to validate the proposed numerical procedure, the complete virtual certification of two trains tested in WP1 has been achieved in this work. The certification results computed from the simulated results have then been compared to the criteria obtained from measurement. Even if the mean values are quite well reproduced, differences have been noticed when analyzing the extreme quantiles. When validating the model for certification purposes, special attention therefore has to be paid not only on the mean response of the sub-systems of the train but also to the rare events.
To add consistency to this work, the full proposed procedure has also been applied to a case that would have met the required conditions of the virtual certification. The Turkish EMU train used in this example was first certified in Turkey. It has then been slightly modified to run in Arabia. First, it has been verified that the two trains showed very similar behaviour when tested on the same virtual track, which is the first requirement for virtual certification. Then, its dynamic behaviour in the new conditions has been shown to be better than the first one with respect to the certification criteria. Under the restriction that it can be justified that the mathematical models of the two trains are validated, then this slightly modified train could have been numerically certified. The same procedure has been applied to the EMU IC4 train.
WP7 Quality assurance & regulatory acceptance
Objectives
WP7 is transverse to all the other ones. In the general project framework it addresses all the high-level objectives of DynoTRAIN (harmonisation of standards, cost reductions, closure of “open points”) in that none of these would be possible without the acceptance of the results of the project on behalf the relevant stakeholders. "Acceptance" of course does not imply an automatic uptake of results, rather the acknowledgement by the main stakeholders, that are responsible for TSIs and standards (i.e. ERA and CEN), that the results are worthy of consideration for input to their work.
The purpose of the WP is to ensure such an "acceptance" in three main ways:
• By striving to represent information on the accuracy of existing and proposed assessment methods as exhaustively as possible; it was acknowledged in fact at the start of DynoTRAIN that accuracy, or rather its quantitative counter-part, uncertainty, is a key scientific aspect that needs to be carefully considered, particularly since expressing uncertainty is a way to express the degree of confidence in one's results; this was addressed in WP7's Task 1 "uncertainty analysis";
• By ensuring personal involvement and implication of the stakeholders in an early stage of the project; it was realised that the only way for the project to generate short-medium term impacts was to keep the stakeholders informed throughout the project and obtain feedback on the usability of results, so that they knew what to expect at the end of DynoTRAIN; this was addressed in WP7 Task 2 "liaison and dissemination";
• By identifying and structuring the project's outputs suitably, and subjecting it to thorough analysis and review in a transparent way to all the project's Partners, thus arriving at quality-assured output that could be called with confidence "DynoTRAIN output"; this was addressed in WP7 Task 3 "proposals for integrations to standards and authorisation processes".
Activities
In the first part of the project:
• The existing framework of standards was analysed through a formal logical analysis, i.e. an analysis that identifies logical elements (for this specific case, provisions in the standards that were categorised – e.g. assessment tool, accuracy, validation conditions, assessment conditions, limit criterion etc.) and uses a formal method to represent them; the Partners were asked at an early stage to state which standards / TSIs they were expected to target with their work;
• A guideline for the drafting of output documents was developed based on the structure of the existing framework as identified by the formal logical analysis; the Partners were asked to provide, as output, proposals for provisions (requirements and recommendations) that would, in their view, usefully replace or integrate existing provisions;
• A workshop was held to discuss the issue of accuracy related to vehicle assessment; the discussion addressed how the accuracy of a vehicle assessment depends on measurement uncertainty, eventually simulation uncertainty, but also the inevitable uncertainties associated with the way standards are written (e.g. processing requirements, safety margins associated with limit values etc.); correspondingly guidelines on the representation of uncertainty were drafted; awareness was raised on the issue, so that all project Partners would be more eager to compile uncertainty information for their reports;
• An Advisory Council was set up, through which input was given to and received from the Target Group.
In the second part of the project:
• WP7 collected the proposals of the technical Work Packages in the form of “output documents”, that responded to a series of requirements (e.g. only information that may become public, output possibly in the form of proposed provisions);
• And analysed them generally in five different ways: 1. a formal logical analysis to identify the provisions of the existing framework that corresponded to proposed project provisions, 2. a NSA/NoBo perspective analysis to verify potential difficulties in third-party assessment, 3. a feasibility check to identify eventual practical issues that could make the proposal difficult to apply (e.g. data retrieval), 4. an internal “2nd party” peer review to which all project Partners were invited, and 5. an uncertainty analysis, to establish the entity of assessment uncertainty of the proposed process with respect to that of the existing process;
• The output of the projects was structured with a public “top-document” (the main WPX deliverable) describing each proposal with the process that it had undergone within WPX, with the output documents as annexes (versions before and after peer review, with the corresponding comments resolution document).
Results
Task 1
An uncertainty framework was set up incorporating where possible the ISO GUM framework for measurements. The framework was then populated with the quantitative results obtained through experimental and simulation work in the project.
The main results of this work are figures for indicators of uncertainty, allowing comparison of the accuracy of existing and proposed assessment methods. The framework itself is a result with potential long-term impacts on the sector. In fact it addresses for the first time an issue that has long been neglected although it is often the matter of discussions and delays. One impact could be on future standardisation, e.g. for guidelines on how to express assessment results, and the methodology is a way to evaluate any future proposal for reducing authorisation costs through virtual or other alternative tools.
It was shown that the accuracy of a virtual assessment according to the process defined in DynoTRAIN (WP6 procedure and WP5 validation) in overall terms can be of the same order of magnitude as that of an experimental assessment. The possibility to achieve this depends on:
• The assessment quantities and their level;
• The techniques and modelling effort used: the Partners in DynoTRAIN implemented state-of-the-art models, using measured track geometry and wheel-rail profiles, and particular care when assembling the virtual test track;
• How the "shift" from validation conditions (in which measurements are available for comparison) to assessment conditions (after having modified vehicle or track for recalculation of the assessment quantities) is controlled.
It was possible to draw a clear picture on the relevance of the influence quantities in contributing to intra-assessment variability − i.e. the one used for the assessment itself in the calculation of statistical estimated values − and inter-assessment variability − i.e. the variability observed if we could ideally replicate assessments on the same type of vehicle many times and that we have to neglect (consequently generating uncertainty) when only one assessment is performed.
Out of the many contributing influence factors, only few stand out as particularly relevant:
• Intra-assessment variability is mostly generated by:
o Firstly, friction variations, track irregularities (+varying stiffnesses to an unknown degree), cant deficiency and curve radius;
o Secondly, contact geometry, for which however the currently used indicators (equivalent conicity, radial steering index) show experimentally a correlation with the varying assessment quantities that is either very vehicle-specific (equivalent conicity) or very low (radial steering index);
o Thirdly, vehicle speed; and finally other influences, such as the mostly unknown contribution of environmental conditions.
• Inter-assessment variability is dominated in both the experimental and virtual assessment by the freedom of choice of samples for the statistical evaluation; the results suggest that the variability associated with different choices of EN-compliant samples of track sections is higher than that associated with the other main sources − vehicle-to-vehicle variability in experimental assessments and modelling choices in virtual assessments; this cannot be affirmed with full confidence since further substantial research effort is needed for a thorough understanding of the latter variabilities.
All these sources of variability can be reduced as a consequence of the increase of the level of knowledge on their contributions. The consequent improvement of assessment repeatability would contribute to reduce the costs due to repetition of tests. The DynoTRAIN project has provided contributions in this sense. However, some important aspects are still open. Among these, the fact that continuous on-track wheel-rail friction and stiffness measurements are still not practical or widespread, leaving these important influence quantities uncontrolled in vehicle assessments, is certainly one of the most important.
Task 2
All liaisons between the projects and the Target Group occurred under the auspices of the projects’ Advisory Council. This body was set up when TrioTRAIN was still being created by inviting ERA, CEN, CENELEC and any interested NSA to participate. The final composition saw:
• The Head of the Cross-Acceptance of ERA chairing the council;
• CEN represented by the Chair of Technical Committee TC 256 “Railway Applications”;
• CENELEC represented by the Chair of Technical Committee TC9X “Electrical and electronic applications for railways”;
• The participation of the NSAs of France, Germany and the United Kingdom.
The main result was a mutual understanding and acknowledgement of roles and responsibilities of researcher groups and standardisation/regulatory bodies. A way of working was pioneered that is already being exploited and refined in other projects. The Target Group has been well aware of intended project output for a long time.
Task 3
The main result of this task is the structured collection of documentation called "DynoTRAIN output" that has been analysed and reviewed thoroughly within the project itself, with a process that is usually not performed in research projects. This output contains proposals written in text form with precise reference to the documents addressed (TSI, standard, other). It is hoped that in this way regulatory/standardisation bodies can save effort in having to "translate" scientific results into text for their documents. The text is already there to be reviewed, revised and eventually incorporated into regulatory/standardisation documents.
Conclusion
The project gained benefit from being part of Triotrain, both Aerotrain and Pantotrain were 3 year projects, Dynotrain was a four year project and was therefore it was possible to learn lessons from the other two projects. The project held regular meetings with stake holds in ERA and CEN which brought significant progress towards Regulatory Acceptance. Dynotrain started with 34 objectives and achieved 31 completely, of the other three,one objective became obsolete and two relating to equivalent conicity limits were partially completed by producing processes to devise in service limits but need economic evaluation to decided costs splits between infrastructure managers and rolling stock operators for rail and wheel maintenance costs, This work was outside the scope of Dynotrain and is normally undertaken by ERA.
Technical success:
• Measurement campaign toured Europe, visiting:- Germany, Italy, Switzerland and France. As far as the project is concerned this is the largest test campaign ever undertaken anywhere in the world. The tests covered 7500km of track and recorded 4.7 terabytes of data. The train incorporated four types of test vehicles (locomotive, coach and two freight wagons that were tested in both tare and laden conditions) and included in the train was a track recording car so synchronised data of track input and vehicle reactions was recorded. In addition static tests were undertaken on all the test vehicles. This work packaged has completed all objectives and has accumulated an enormous database that could be used for future railway research projects.
• Current standards specify track quality by geometric descriptors, sections of track are measured for quality and defined in terms of standard deviation and peaks values. Extensive measurement results were processed and reported. Recommendations have been made to CEN WG 10 for inclusion in the next issue of EN 14363. New analysis methods are being developed; multi-regression techniques are giving new understandings of the relationship between vehicle reactions and track quality. CEN WG 10 have agreed to introduce multiple regression analysis in the next revision of EN 14363.
Analysis of the this vast track quality database identified many weakness in the current network measuring systems, the work package has made recommendations to improve the measured data by decolouring and improved filtering along with other quality checks.
The work package has studied 11 other track quality assessment methods and surprisingly concluded that the current geometric methods are still superior for determining the relationship between vehicle reaction and track quality.
Data was analysed to compare Networks, the analysis showed that on high speed lines the standard deviations were similar but there was considerable variation on slow speed lines.
• Millions of wheel /rail combinations where considered in arriving at conicity recommendations at a European level. With the database available it was possible to make firm conicity recommendations to CEN for inclusion in the next issue of EN14363 which covers vehicle certification for running safety and recommendations at a system level to ERA for inclusion in both rolling stock and infrastructure TSI’s. The TSI in service limits will be defined after the ERA economic evaluation to consider the cost share between rails and wheels.
The work package has also effectively undertaken an adhesion survey during the test campaign on small radius curves.
• All the relevant parameters for track force assessment were reviewed at both international and national levels and the national track construction and maintenance standards.
Sensitivities to track force inputs, relationships with damage mechanisms and practical operating controls were all considered.
For cross acceptance, national requirements can be assessed by use of existing data to compare with existing vehicles and multiple regression analysis can also be used to again with either test or simulation data. Also the project has provided guidance on how to use the multiple regression analysis can be used for cross acceptance.
• The work on virtual certification has resulted in a proposed procedure for validation of multi-body vehicle models to replace track testing for certification of vehicles running safety. This procedure has been recommended to CEN for inclusion in future issues of EN 14363. Validation methodology is based on:-
• Comparison of simulations with vehicle tests using force measuring wheel sets and body accelerations.
• A minimum of 12 test sections as required by EN 14363 with the four zones considered.
• A minimum of 24 data points comparing simulation to measurement for each of the 12 limit criteria being assessed.
• Limit criteria is given for each assessment quantity in terms of standard deviation and mean difference. All must be satisfied for validation.
Comparison of models with stationary tests was found to give mixed results, in some case it helped validation and others worsened to agreement.
Measured track data and contact geometry helped model validation but other factors are also important.
The change will allow track tests to be replaced by simulation as part of the vehicle certification and will have economic impact on new vehicle certification costs. The cost savings for new designs will be limited but larger savings are possible for repeat builds and the management of changes to existing vehicles.
Potential Impact:
Potential impact of the project
Development and implementation of Technical Specifications for Interoperability and EN norms
In DynoTRAIN one important goal was to help close ‘open points’ in the TSIs as well as, where possible, improve the vehicle certification process in order to make it more efficient in terms of time and cost. To ensure the implementation into the TSIs of the project outputs, a specific Work Package (WP7) was dedicated to ensuring that the project’s proposals were of a high quality and directly applicable. WP7 was also designed to ensure that the proposals were of a sufficient standard that it would be possible to convince the relevant stakeholders of their safety, effectiveness, efficiency and feasibility as elements of a proposed certification process. To this end, there was a structured communication between the project and the various stakeholders.
The DynoTRAIN partners ensured that the project had a regular and intensive interaction with the standards and regulatory bodies. Within the Technical Management Team, the majority of members were also members of CEN TC256 Working Group 10, Vehicle/Track Interaction. This meant that the CEN Working Group was constantly well informed on the status and progress of the project. Work undertaken and results achieved were discussed at meetings of the CEN Working Group by the TMT members, who discussed their project work with their CEN WG10 colleagues. The Technical Management Team was also invited to two meetings of the working group. At both of these plenary meetings of CEN TC256 WG10, the TMT presented the work undertaken and results achieved within the various Work Packages in great detail. This interaction gave the project an excellent understanding of the context and requirements of standardisation. It also gave the consortium direct access to, and intimate knowledge of, the relevant CEN Working Group. This interaction presented an extremely important opportunity for the project to present and justify its results. This has played an important role in facilitating the uptake of the results, with some of the content of EN14363 having been informed by the work within DynoTRAIN.
The project partners have had regular contact with the standards and regulation bodies through the twice-yearly meetings of the TrioTRAIN Advisory Council. This saw the Technical Leader, Co-ordinator and some TMT members meet face-to-face with the Chairman of TC256 of CEN and the Head of Unit for Cross Acceptance at the European Railway Agency, ERA. During these meetings, the project partners had the opportunity to understand the regulatory context, which was important to them as a group seeking to make proposals for TSIs. Between meetings of the Advisory Council, meetings were organised between the project and ERA, for example in October 2012 during which the TMT presented an update on the results to date to the ERA officers following the Loc & Pas and Infrastructure TSIs. In the same way that many of the Work Package leaders were members of the CEN group, many were also members of the ERA Working Party for dynamics issues. Results from DynoTRAIN are incorporated in both the recently revised Loc & Pas and Infrastructure TSIs, and this constitutes the achievement of an important objective for the DynoTRAIN project.
Reduction of the migration time for the implementation of new interoperable solutions
One of the main objectives of DynoTRAIN was to reduce the time and cost for certification of new interoperable rail vehicles. This was to be achieved by:
- Harmonising the national requirements through focusing on the TSI’s, making it possible to transfer certification data from one country to another, obviating repetition in the certification process;
- Replacing existing test configurations with new procedures, such as the introduction of virtual certification. Since simulation models are already commonly used in the industry, the possibility to use them for certification as well as design validation will save cost and time for the required tests. In addition it gives the opportunity to have better controlled environmental testing conditions in the future by introducing new scenarios to cover. This will also help to test in certain boundary conditions which cannot always be reproduced at the time of undertaking physical on-track tests. This could therefore have the effect of increase safety levels or confidence in tests.
Impact on competitiveness
DynoTRAIN has contributed to establishing and strengthening common technical standards and compatibility between the various European systems, by proposing input for TSI and EN norms. These proposals go in the direction of the simplification of certification procedures. This contributes to a reduction in the time and cost of certification and therefore of the putting into service of a rail vehicle. These anticipated changes in the process can save hundreds of millions of Euros for the European rolling stock manufacturing industry, making it more competitive abroad and, assuming that the savings are passed to Railway Undertakings and then to passengers in turn, will make the rail transport mode in Europe more competitive.
Global strategic impact
By strengthening interoperability within the EU, DynoTRAIN is helping to further consolidate the rail industry and to promote European standards and practice outside Europe. This is very important for the competitiveness of the European industry. During the life of the project and thanks partly to the co-ordination of ERA, a relationship has developed between the DynoTRAIN project and the United States Department of Transport and Federal Railroad Administration. These bodies are extremely interested in the European approaches towards managing dynamics issues, so much so that representatives of these bodies have come from the United States on three separate occasions to meet representatives of the DynoTRAIN project. It is likely that, with continued co-ordination of ERA, this relationship will continue. This presents an important opportunity for Europe to have an influence on standards and practice outside of Europe. Within Europe in general, efficiency gains and improvements in quality will deliver substantial benefits for both users and suppliers in general. Within Europe this means benefits for Railway Undertakings, Infrastructure Managers and the manufacturing industry.
By advancing interoperability, as well as by showing how the cost and time of putting vehicles into service can be reduced, the results of this project will contribute to the required increase in competitiveness of the rail transport mode within Europe. It is necessary to increase the attractiveness of the rail transport mode within Europe. By passing savings from simplified certification test configurations all the way through Railway Undertakings to passengers (assuming that these savings are indeed passed on in this way), the results of this project are able to contribute to this. DynoTRAIN is helping to create the conditions of possibility for this to be accomplished.
Implementation of greater interoperability will also reduce the range of goods and services required and generate economies of scale. Both of these measures will lead to reduce the cost and the time of development and manufacture. This last point also shows how the project contributes to the overall competitiveness of the European industries. This is especially important at a time of stiff competition in the global market for rolling stock and sub-system components.
By producing results that work towards the implementation of efficient interoperability and introduction of certifications based on simulation rather than on physical track tests (or an optimised mix of simulation and track tests or a simplification of the current track tests) DynoTRAIN should bring the following benefits:
- Increased competitiveness of the railway mode;
- Increased interoperability;
- Reduction in the cost of train ownership;
- Improved availability and reliability of rolling stock.
Strategic impact of DynoTRAIN on the rolling stock manufacturing industry
The results of DynoTRAIN will strongly impact the European rolling stock manufacturing industry. The main benefits expected are:
- Reduction of the duration and the costs of certification procedures thanks to less costly test configurations and through the part-transfer of expensive and time-consuming physical track tests to numerical simulation, as described above;
- Reduction of the time-to-market for new products, this has also been described in the preceding sections of this report;
- Harmonisation of certification procedures and mutual recognition of certificates across Europe;
- Optimisation of R&D resources and identification of priorities for future innovation in the field of certification;
- Development of the state-of-the-art in railway dynamics (track and vehicle interaction). This is an important element of the project, as we can say that several Work Packages have advanced the existing knowledge in railway dynamics. Much more is known about many of the key topics than was the case before the project. This is the case for each and every technical Work Package.
Strategic impact of DynoTRAIN on railway undertakings
The project results also bring benefits to railway undertakings:
- Reductions in the cost of vehicle procurement and ownership;
- Improved reliability of rolling stock thanks to improvement in safety standards and in interoperability;
- Reduction in time and costs for further certification of replacement materials: not all the testing will have to be done again; instead only some parameters might be changed in the simulation.
Furthermore, by stimulating a collaborative approach among the European system integrators, the DynoTRAIN project has the potential to generate competitive gains on a worldwide basis. This contributes to making European railway products attractive to emerging or established markets (for example Asia and North America) and developing economies in serious need of affordable rolling stock (such as Asia, Africa and Latin America).
Community societal needs
The results obtained during the project will facilitate the certification of railway rolling stock against EN standards and TSIs. The introduction and increased used of simulation-based certification tests will contribute to improve the attractiveness of rail vehicles, and therefore of rail-based transport as a whole. This will serve community economic needs, as the high numbers of people employed by the rail sector can be maintained or even grow. Moreover, cost reductions and efficiencies are essential, as there is more and more pressure on public finances. Rail services are very often part-financed by the public purse, and must be provided at not only a cost which passengers can afford, but at one which provides value for money for the taxpayer.
This will contribute to the promotion of rail transport both for passengers and freight purposes as an alternative to other modes, thus fostering the environmental and economic sustainability of transport. It is already known that rail is the most environmentally-friendly form of public transport, and by helping to enhance the attractiveness of rail, DynoTRAIN can bring a massive potential environmental benefit. This is very important, as modal shift away from roads and onto rail is essential if the European Union’s transport and climate goals are to be met.
The main expected benefits of this for European citizens are:
- Reliable and safe services through increased availability and reduced time-to-market of high-performance rolling stock. In particular, changes will come about in the expectations of passengers regarding the quality of their journeys due to various factors as forecast by ERRAC in the SRRA II;
- Better quality of scheduled services by avoiding operational disruption caused by physical track tests;
- Enhanced utilisation of rail-based traffic capacity on the basis of free choice of transport mode and thanks to the quality/ price ratio increase;
Moreover, the rail transport mode is now becoming the natural ‘partner’ for international flight connections: high-speed lines are gradually replacing many short airline journeys (under 500 km). Reinforcing the competitiveness of rail transport could and should in this case bring a considerable contribution to the reduction of overall noise levels for neighbouring communities by replacing the excessive use of flight, in particular night flights.
Security and safety
Safety and security are major societal needs. In our developed countries, as certain other major needs have been more or less met, safety is often considered as the number one issue. As far as mass transit and public transport networks are concerned, the safety and security of passengers and staff is a key factor of the “competitive and sustainable growth” for the various operating companies.
The propositions made by DynoTRAIN for improvement in European certification standards and the submission of resolutions to close uncertainties in the Technical Specifications for Interoperability and safety have the potential to lead to a significant improvement in safety and security. By contributing to the closing of open points in TSIs, DynoTRAIN is certainly contributing to the improvement in safety and security on the European rail network. The advances in knowledge and state-of-the-art in railway dynamics achieved under this project will certainly help to make the rail transport mode even safer than it currently is.
In terms of general safety, it is recognised that rail is safer than road and air transportation. AeroTRAIN can contribute to helping reduce the number of overall transport accidents, by playing a part in encouraging modal shift towards rail.
Environmental issues
The integration of DynoTRAIN results into European certification standards will ensure at least a neutral impact on climate change. Indeed, the introduction of a new certification process based on computer simulation will reduce the need for physical on-track tests:
- Numerical models will allow a large range of conditions (climatic, geometric, etc.) to be taken into account, thus reducing the need of repetitive on-track tests;
- By improving the TSIs and cross-acceptance, the number of on-track tests to be performed in each country will be decreased.
Perhaps this has the additional benefit of helping to reduce the noise emissions to lineside neighbours, as there could well be fewer numbers of trains running for the purposes of certification.
At the EU level, the shift for both passengers and freight transport from road to railways will have an important impact on the environment through:
- Reduction of greenhouse gas emissions: NOx, SOx, and particles emissions from trucks and cars. Road transport was responsible for 40 % of GHG emissions in Europe at the time of the submission of the TrioTRAIN project proposal, and this is without speaking about the contribution of air transportation;
- Reduction of energy consumption: freight trains consume less energy than the equivalent number of trucks for a same mass transported. The same statement can be made for passenger trains and cars.
Furthermore, in the framework of short and middle range distances journeys, the competitiveness of transport by train compared to air transportation must be highlighted, knowing that commercial air transportation produced, at the time of the submission of the TrioTRAIN project proposal, 2.5 % of the global CO2 emissions due to human activities.
DynoTRAIN will play its part in facilitating modal shift of passengers and freight to rail, which is a clear contribution to improving the environment in Europe. Through better rolling stock, which comes more quickly to the market, it will also play a role in ensuring that the rail transport mode is equipped to take the increasing number of passengers coming to rail from road and air transport for other reasons.
List of Websites:
www.triotrain.eu
Coordinator’s Contact detail:
Ross Hanley
UNIFE - The European Rail Industry
Avenue Louise 221
B-1050 Brussels
Office: +32 2 642 23 21
ross.hanley@unife.org