ISSN: 1402-1757 ISBN 978-91-7583-XXX-X Se i listan och fyll i siffror där kryssen är

LICENTIATE T H E SIS

Department of Civil, Environmental and Natural Resources Engineering

Division of Operation, Maintenance and Acoustics Emilio Rodríguez Martínez Track Circuits’ Robustness

ISSN 1402-1757 Track Circuits’ Robustness ISBN 978-91-7583-045-2 (print) ISBN 978-91-7583-046-9 (pdf) Modeling, Measurement and Simulation Luleå University of Technology 2014

Modeling, Measurement and Simulation Emilio Rodríguez Martínez

TRACK CIRCUITS’ ROBUSTNESS Modelling, measurement and simulation

Emilio Rodríguez Martínez

Operation and Maintenance Engineering Luleå University of Technology

Printed by Luleå University of Technology, Graphic Production 2014

ISSN 1402-1757 ISBN 978-91-7583-045-2 (print) ISBN 978-91-7583-046-9 (pdf) Luleå 2014 www.ltu.se ACKNOWLEDGEMENTS

The research presented in this thesis has been carried out at the Operation and Maintenance division and funded by the European Community´s Framework Programme FP7/2007-2013 under grant agreement no. ”285259”, TREND project. I would like to thank them for providing the support to perform this licentiate, based on that research. The project was supervised by Prof. Diego Galar, Prof. Uday Kumar and Dr. Stefan Niska. They gave the support, guidance and valuable advice to help me to develop my ideas, allowing me to complete this licentiate. I would like to express also my sincere gratitude to the partners in the consortium, which consists of CEIT, CAF I+D, CEDEX, IFSTTAR, York EMC Services and, in special, to Trafikverket. I worked together with Dr. Stefan Niska from Trafikverket and his cooperation, kind personality and help made this journey much easier, making me feel like one more of their team members. I would also like to mention the coordinator of the project, Íñigo Adín from CEIT, who supported me and helped providing crucial information for the development of this research and Åke Wisten from LTU, whose experience and help in the tests are gratefully acknowledged. Thanks also to all my colleagues for their support and help. I am thinking in many names but among them Amparo Morant, Víctor Simón, Pablo Puñal and Iván Carabante stand out. Amparo Morant worked on TREND project and helped me every time I needed some assistance. I also have to thank her because of her contribution in my thesis and papers. Victor Simón came to the project later and it was also a pleasure to work with him on some parts of the project and papers used also in this thesis. Finally, although Pablo Puñal and Iván Carabante work also at LTU, they were not my direct colleagues. Even so, his advice, company and discussions were as important as any other technical contribution. And last, but not least, thanks to my family, who supported me every step of the way, to my friends (special mention to ‘the musketeers’) for the help in making the way easier and funnier and in general to every other one involved in the development of this research.

v

vi

ABSTRACT

In countries with rough weather conditions, frequent delays cause railway companies to waste time and money. Many of these delays are related to the train detection systems, as the old DC track circuits are still used in some countries, including Sweden, our case study. Since the most important factor in the railway system is safety, in some cases, the train detection system gets incorrect information and detects a non-existent train. The train slows down to avoid a problem in the track (with other trains or other faults), causing prolonged delays with cascading effects. The analysis in this licentiate contributes to the detection and reduction of TC failures; this, in turn, will save money for the railway community. A classification of the most probable causes of failures related to the train detection system was derived from the Swedish failures report database 0FELIA. After classifying failures, we focussed on the three most common worst case scenarios: low resistance between the rails, external interference such as a lightning strike, and iron-powder-bridges in the insulated joint. Electromagnetic interferences (EMI) are a problem for the railway system in general. One source of electromagnetic (EM) transients is the return current harmonic produced by the engine of the rolling stock itself. In the first stage of this licentiate, we implemented a Matlab model of the power supply system of the Swedish railway infrastructure, using the characteristics and previous measures of a real source. A model of a train as an active load validated by the manufacturer was integrated as a subsystem in different positions of the infrastructure. This method was used to study the behaviour of the low frequency system from an electrical point of view but it could also be used as input for an electromagnetic model using high frequencies. The model was validated through measurements taken in northern Sweden. In addition, a 3D model of the whole railway system was proposed. The simulation software was CST STUDIO SUITE® (Computer Simulation Technology Studio Suite), supported by real measurements on site and the lab to tune and validate the model. The results of the simulation show that the model fits with reality and is reliable for the study of track circuit sections. Some measurements followed the current standards, but we also analysed points not covered by them, allowing us to update the current standards. KEYWORDS: railway infrastructure, signalling, track circuit, relay, modelling, simulation, testing, robustness, interoperability, reliability, false positive signals.

vii

viii

LIST OF APPENDED PAPERS

PAPER A: Fault detection of Railway EMC problems using MATLAB models. E. Rodriguez, N. Raj Karki, D. Galar, D. Valderas, S. Niska. BINDT - The Tenth International Conference on Condition Monitoring and Machinery Failure Prevention Technologies, 18-20 June 2013, Kraków (Poland).

PAPER B: Simulation of electrical power supply system in railway infrastructure E. Rodriguez, D. Galar, S. Niska, N. Rai Karki. International Conference on Power & Energy Systems: Advances in Power Systems, 20-30 October 2013, Kathmandu, Nepal.

PAPER C: Safety issues of Track Circuits - A hybrid approach. E. Rodriguez, V. Simón, D. Galar, S. Niska. Published in the International Journal Communications in Dependability and Quality Management. Volume 17, Number 2, June 2014, ISSN 1450-7196.

PAPER D: El impacto de la complejidad de la electrónica en la seguridad del sistema ferroviario. E. Rodriguez, V. Simón, D. Galar, L. Berges, J. Tamarit. Accepted in the International Journal Mantenimiento, Asociación Española de Mantenimiento, Number 280, December 2014.

ix

x

TABLE OF CONTENTS

INTRODUCTION ...... 3 1.1 Problem definition and background ...... 3 1.2 Purpose and objectives ...... 5 1.3 Research questions ...... 5 1.4 Scope and limitations ...... 6 1.5 Structure of the thesis ...... 7 EMC IN THE RAILWAY SYSTEM ...... 9 2.1 EMI threats ...... 9 2.1.1 Rolling stock ...... 10 2.1.2 Power supply and electrification system ...... 11 2.1.3 Infrastructure defects and debris ...... 12 2.1.4 Signalling and Communication Systems ...... 12 2.2 EMC in the rolling stock...... 13 2.3 EMC in the infrastructure ...... 18 2.3.1 Power supply substation ...... 19 2.3.2 Booster transformer ...... 20 2.3.3 Autotransformer ...... 21 2.3.4 Design considerations affecting the electromagnetic characteristics of the infrastructure ...... 21 2.3.5 Overhead Catenary system ...... 22 2.3.6 Section breakers ...... 23 2.3.7 Neutral section (phase break) ...... 24 2.3.8 Grounding ...... 24 2.3.9 Layout ...... 25 2.3.10 External factors affecting the electromagnetic characteristics of the infrastructure ...... 28 CURRENT STANDARDS ...... 31 3.1 EMC on railways ...... 31 3.2 Summary of the standards ...... 33 3.3 EN 50121 –1: 2006: Railway applications - Electromagnetic compatibility - Part 1: General ...... 38

xi

3.3.1 Scope ...... 38 3.3.2 Annex A: the Railway System ...... 40 3.3.3 Useful additions to the EN50121-1 ...... 40 3.4 EN 50121 –2: 2006: Emissions of the whole railway system to the outside world …………………………………………………………………………………………………………………….. 43 3.4.1 Emission limits ...... 43 3.4.2 The measurement system ...... 45 3.5 EN 50121–3-1: 2006: Rolling stock – train and complete vehicle ...... 48 3.5.1 General issues of EN 50121-3-1 ...... 49 3.5.2 Compatibility with signalling and communication systems ...... 50 3.5.3 Interference on non-railway telecommunication lines ...... 50 3.5.4 Emissions ...... 51 3.5.5 Immunity ...... 53 3.5.6 Transients ...... 54 3.6 EN 50121 –3-2: 2006: Rolling stock – apparatus ...... 54 3.6.1 Emissions ...... 54 3.6.2 Immunity ...... 55 3.6.3 On-board Spot Signalling Systems ...... 55 3.7 EN 50121–4: 2006: emission and immunity of the signalling and telecommunications apparatus ...... 57 3.7.1 Emissions ...... 57 3.7.2 Immunity ...... 59 3.8 EN 50121–5: Railway applications - Electromagnetic compatibility - Part 5: Emission and immunity of fixed power supply installations and apparatus ...... 60 3.8.1 Emissions ...... 60 3.8.2 Immunity ...... 61 3.8.3 On-site testing ...... 61 3.9 EN 50122-1: Railway applications – Fixed installations. Part 1: protective provisions relating to electrical safety and earthing ...... 63 3.9.1 Scope ...... 63 3.9.2 Ground currents ...... 63 3.10 EN 50215: Railway applications – text of rolling stock on completion of construction and before entry into service ...... 64 3.10.1 Scope (related to EMC) ...... 64

xii

3.10.2 Supply issues and transients ...... 65 3.11 EN 50238: Railway applications – Compatibility between rolling stock and train detection systems ...... 65 3.11.1 Scope ...... 65 3.11.2 Neutral sections ...... 66 3.12 EN 50388:2005 Railway applications – Power supply and rolling stock – Technical criteria for the coordination between power supply (substation) and rolling stock to achieve interoperability ...... 66 3.12.1 Application procedure ...... 66 TRACK CIRCUITS ...... 69 4.1 Introduction ...... 69 4.2 CASE STUDY: Swedish DC track circuits ...... 74 4.2.1 Safety in the Swedish DC track circuit ...... 75 4.2.2 Main components of the Swedish DC track circuit ...... 76 4.3 Most common failures in track circuits in the Swedish railway infrastructure . 82 4.3.1 Scenario 1 – Low Resistance between rails ...... 82 4.3.2 Scenario 2 – External interference ...... 85 4.3.3 Scenario 3 – Iron-powder-bridge in the insulated joint ...... 90 4.4 Quality assurance of the DC Track circuits...... 91 RESEARCH METHODOLOGY ...... 93 5.1 Research strategy ...... 93 5.2 Data collection and analysis ...... 94 RESEARCH APPROACH ...... 95 6.1 Modelling ...... 95 6.1.1 Rolling stock ...... 95 6.1.2 Infrastructure ...... 97 6.1.3 Integration ...... 107 6.2 Hybrid model ...... 110 6.2.1 Solver ...... 110 6.2.2 Description of the scenarios ...... 115 6.2.3 Details of the components of the model...... 119 6.2.4 Validation test definition: ...... 126 RESULTS AND DISCUSSION ...... 145 7.1 Validation of the model ...... 145

xiii

7.1.1 Probes ...... 145 7.1.2 Procedure of the validation ...... 145 7.2 Analysis of the worst case scenarios ...... 146 7.2.1 Scenario 1 - Single track circuit section ...... 146 7.2.2 Scenario 2 – External interference ...... 157 7.2.3 Scenario 3 - Two track circuit sections ...... 172 7.3 Limitations of the current standards ...... 173 CONCLUSION ...... 175 8.1 Conclusions of the MATLAB model of the infrastructure ...... 175 8.2 Conclusions of the CST model simulations ...... 175 8.2.1 Scenario 1 - Single track circuit section ...... 175 8.2.2 Scenario 2 - External interference ...... 176 8.2.3 Scenario 3 - Two track circuit sections ...... 177 8.3 Conclusions of the validation experiments affecting the TCs ...... 177 8.3.1 Scenario 1 – Conclusions ...... 177 8.3.2 Scenario 2 – Conclusions 1 ...... 178 8.3.3 Scenario 2 – Conclusions 2 ...... 178 8.3.4 Scenario 3 – Conclusions ...... 179 FURTHER RESEARCH ...... 181 9.1 Improvement of the standards...... 181 9.2 Conclusions and further steps with the integrated MATLAB model ...... 183 REFERENCES ...... 185 APPENDED PAPERS ...... 189

xiv

1 INTRODUCTION

1.1 Problem definition and background

Robustness is defined as "the ability of a system to resist change without adapting its initial stable configuration" (Wieland et al., 2012). In this licentiate it was analysed the robustness of the DC Swedish track circuits from the fault-tolerance point of view. The performance of a railway measurement campaign is slow and expensive due to the continued traffic in the track. Furthermore, in order to ask for the permission is needed to contact all the different parts involved, as the government, the operator, the infrastructure administrator. The research is based on the FP7 project TREND, which main topic was EMC in railway, so the modelling was perform first in Matlab for the analysis of all the electrical infrastructure system context and then in CST for the electromagnetic one, particularly the DC Swedish track circuits. High powered electronic equipment, together with low power microcontrollers and other electronic devices, is being installed on trains in great numbers. Electromagnetic compatibility has therefore become a critical issue for the design of train related apparatus as well as of the train as a whole. Key information to understand the state of the art in EMC and interoperability in railway is provided by the European Railway Agency (Lloyds Register Group, 2010). This study is focused on the examination of the processes, procedures and methods ensuring electromagnetic compatibility between rolling stock and infrastructure in the 27 members of the European Railway Area. In particular, it analyses the demonstration of EMC of the rolling stock with the requirements of operating infrastructure. The document revealed the following issues:

3

- The two technical standards that should demonstrate electromagnetic compliance of rolling stocks do not ensure compatibility between the rolling stock and the signalling systems, neither with telecommunication services. They cannot ensure either the safe operation. - Telematics predominantly concerns the tracking and logistics associated with freight, particularly containerized freight. Cargo is identified by passive or active radio-frequency tags. These are considered to be a separate component of the railway and are assessed by a separate TSI (Commission Regulation, 2005). - In some countries (e.g. Austria, Belgium, etc.) additional requirements are set to limit radiated emissions that could affect telecommunication services (e.g. 80MHz, 160MHz, 450MHz, 900MHz, 2,4GHz, etc.) - In some countries (e.g. France, Greece, etc.) human exposure to electromagnetic fields (EMF) is considered part of the electromagnetic compatibility considerations for train acceptance. - Actual timescale of the testing and operational trials is between 3-4 months in some countries (e.g. Ireland, Germany) and 12 months for others (e.g. Czech Republic). Cost of the certification process varies between 25K€ (Poland) and 1.5M€ (Czech Republic). The conclusions derived from this exhaustive state of the art analysis can be summarized in the following points: - The certification process varies enormously in time and cost across Europe and extra requirements which are introduced in some countries. - Demonstration of permanent electromagnetic compatibility of the rolling stock in all the possible conditions is not feasible with the current available standards. - The RAILCOM project has made important research in the fields of EM emissions of the rolling stock that affect GSM-R, track circuits and BTMs. However this is only a starting point because only some issues of the overall problem have been partially addressed. - The research community has also made significant improvements in understanding the physical phenomena that lay behind the EM incompatibilities between the rolling stock and other systems (track circuit, signalling and telecommunications services). Unfortunately, the research areas have been isolated and a cross-domain approach has not been employed, yet.

4

1.2 Purpose and objectives

Our purpose in this licentiate is the modelling and simulation of the Swedish DC track circuits and their integration in the whole structure model, system of subsystems. The most common failures were analysed in order to design the worst case scenarios to study their robustness. The objectives followed can be divided in 3 groups: 1- Identification of the representative worst case conditions that includes not only EMC issues, but also other common failures in the infrastructure. This is achieved thanks to a process that includes data collection and data analysis, which is used for the modelling of the system. 2- Definition of a controlled test environment that comprises a test setup and a test site that will provide repeatability, reliability and accuracy. Moreover, the test environment will be able to reproduce infrastructure representative worst case condition. The design of a testing procedure that introduces stimulus to generate the representative worst case conditions will be able to validate our models. 3- Finally, the validated model can be used for simulations of these worst case scenarios saving money and time in comparison with measuring on the track and can be reused in the future. The results has been also used to propose an improvement to update the drawbacks detected in the current standards.

1.3 Research questions

The 4 research questions studied in this licentiate are: RQ 1: How can be modelled the electrical railway system and the track circuits to test them through simulations, cheaper and faster than measuring on the track? RQ 2: What measures are needed for the validation of the electrical model of the infrastructure, the electromagnetic model of the Swedish DC track circuits and their simulations? RQ 3: How the worst case scenarios, not covered in the standards, affect to the behaviour of the Swedish DC track circuits?

5

RQ 4: What are the improvements to the current standards with the drawbacks detected? The following table shows how the papers appended to this licentiate solve these questions and also one more field is added because not all the contents of the licentiate, as the improvement of the weak points of the standards, are published in papers:

RQ1 RQ2 RQ3 RQ4 PAPER A ݱ ݱ PAPER B ݱ ݱ PAPER C ݱ ݱ ݱ PAPER D ݱ ݱ ݱ Not published yet ݱ

Table 1: RQ answers in the appended papers.

PAPER A: Fault detection of Railway EMC problems using MATLAB models Study of the problems, first model of the infrastructure and objectives. PAPER B: Simulation of electrical power supply system in railway infrastructure Detailed methodology and modelling of the electrical railway system. PAPER C: Safety issues of Track Circuits - A hybrid approach State of art of track circuits, modelling, simulation and results. PAPER D: El impacto de la complejidad de la electrónica en la seguridad del sistema ferroviario General vision of the first period of TREND project.

1.4 Scope and limitations

The scope of this project is the analysis of the most common failures in the Swedish DC track circuits and their robustness to them.

6

The main characteristic of the safety in the track circuits is the failsafe principle. If anything does not work correctly, the signals turn red and the traffic slows down in this part of the line. The DC track circuits of Sweden are well designed robust devices, but some failures can occur. This project is studies this signalling system from an electrical point of view. The infrastructure is a wide system and we have to narrow it to focus in the track circuits and their context. Regarding the EMI threads, considering overall immunity levels for the railway infrastructure, it is not possible to define exact levels covering all cases of possible transients. If the infrastructure is divided into several parts, and each part is treated like a system of its own, it is easier to define an immunity level for that part. When dealing with electromagnetic interference in track circuits, the question of immunity has to be separated into at least two totally different cases, depending on the sources of the interference: internal sources (the source is related to a running train or its infrastructure) and external sources (the source is an external interference).

1.5 Structure of the thesis

I would like to stress that this licentiate explains the process followed by the FP7 project TREND.

Therefore, the licentiate starts with the introduction written before this section, where it is described the problem studied in the project and how it was faced.

It continues with the theoretical framework composed of EMC in the railway system (chapter 2), current standards (Chapter 3) and track circuits (chapter 4).

After this explanation of the context of the project it is explained the methodology (Chapter 5) and the research approach (Chapter 6), where a detailed analysis of the robustness of the Swedish DC track circuits is done.

Then, the results of the simulations are analysed (Chapter 7).

Finally, the closing of this licentiate is done with the conclusions (Chapter 8) and further research (Chapter 9).

7

8

2 EMC IN THE RAILWAY SYSTEM

2.1 EMI threats

The following list summarises the main EMI (Electromagnetic Interference) threats that can affect the signalling and communications system, considered in this licentiate as the EMI victim (Deliverable 2 TREND project, 2012): - Rolling stock. - Power supply & electrification systems. - Infrastructure defects & debris. - Signalling & communications systems. The coupling mechanisms can be separated into two classes with different characteristics. The first cause of interference is radiation, affecting the signal received by whichever receiver is in use. The second is interference by induction, capacitance or conduction that would affect either the transmitter or the receiver directly, but may not affect the signal integrity directly (although it is still plausible to disturb the signal by disturbing the transmitter, for example). In the first case, to cause a change in the signal levels on the victim signal, the emissions must be within the bandwidth of the receive antenna in use, and powerful enough to make a difference; however, the second type can cause problems at any frequency. The onus is, therefore, on reducing the radiated emissions within the sensitive bands while ensuring the lower frequency conductive, capacitive and inductive emissions are kept under control. This interoperability study will help, because if the same system is to be used throughout (for example, an RFID system), less variability will occur and, thus, the limits can potentially be reduced. The source of the interfering signal will dictate the type of coupling encountered; for instance, the /catenary interface will be a mainly radiated coupling method during arcing, but will have little close range effect on the broadcasting services when conducting ohmically.

9

2.1.1 Rolling stock

This section will explain possible problems in the rolling stock encountered by CAF, partner of the FP7 TREND consortium (Deliverable 2 TREND project, 2012). All of the various types of modern rail traction found throughout Europe use some form of power electronics. Diesel traction is less of an issue, as the threat emissions (most of which arise from the alternator) are likely to be relatively low frequency and, therefore, present little threat to broadcast services. Diesel traction also uses inverter drives, but these mobile sources of emission are not connected to the electrification system, they do not cause problems in the electrification system and are less of an issue. Rectifiers converting DC to AC may introduce higher harmonics from the alternator, but these are only likely to be a problem below a few MHz. After the alternator stage in a diesel , the electrical systems are quite similar, so some of the phenomena here can be applied to diesel traction, even though the return paths are different. As far as electric traction is concerned, semiconductor based converters and inverters condition the power from the sliding contact interface to provide suitable power for the on board systems, as well as the traction motors. An old fashioned but still used DC traction system will be fed through a thyristor chopper circuit operating at anything up to 400Hz and can radiate at this frequency plus harmonics. More modern traction circuits employ IGBT based inverter drives. This EM radiation may be propagated widely and affects all the systems considered in this study: - 9kHZ to 30MHz: Spot signalling systems and track circuits. - 30MHz to 530MHz: Broadcasting services. - 900MHz band of GSM and GSM-R frequency band: Broadcasting services and GSM-R. Motor noise for electric traction, alternator noise for diesel traction and the converting of the supplied power into usable power for both systems all have the capacity to interfere with the research areas of this project. For each dedicated frequency band, both AC and DC systems need to be considered. The other parameter that must be considered for rolling stock and is studied in this project is antenna positioning (for equipment using transceivers) of the victim systems. GSM-R, broadcasting or even some on-board signalling system antennas are affected by the rolling stock radiation. The correct positioning would require another parameter in the

10

EMC design of the rolling stock. The electromagnetic modelling of the whole system is crucial to minimise this type of problem. Note also that one of the problems with rolling stock EMC is that very often an EMC issue is not diagnosed as such, potentially meaning the ‘offending’ component may be exchanged for a new one; this may fix the problem, but it is an expensive fix for an intermittent problem caused by EMC. These situations can be addressed by means of a more complete and precise testing set-up and procedure which considers interoperability.

2.1.2 Power supply and electrification system

Another major source of radiated emissions is the power supply and electrification system. A main problem emanating from electric traction is the interface between the catenary and pantograph or the interface between the power rail and the sliding pickup shoe, together with the breaker operation after the power is received from the supply system. These systems are present in some form throughout Europe. The sliding nature of the contacts in these systems results in current arcing or ohmic conduction. The high current low voltage nature of the ohmic contact results in near field magnetic radiation and the low current high voltage arcing radiates predominantly in the electric field. The former can be considered a continuous threat, whereas the latter is a harmful transient signal. In any case, the entire spectrum is covered by these interferences. On the one side, the magnitude of the continuous supply signal, its frequency and the distance of the electrification system to outside world equipment has to have limits. On the other side, the magnitude and timed shape of the discontinuities of the transients must be modelled to give a better understanding of this malfunctioning and its effects within the railway on the different research areas. Both the transient and continuous effects should be considered in the measurement standards used for certification. DC railway power supplies are generally less of a threat to the broadcast services as the continuous emission is lower and the arcing is less. Interference from the electrification system may occur within the railway at lower frequencies and affects the railway signalling and communications systems due to the coupling between the S&T cabling and its parallelism to the power supply conductors. Another issue is environment. Rain or other moisture on the catenary or will reduce the arcing and, thus, reduce the emissions. However,

11

ice will cause large arcing problems due to the dielectric nature of the ice layer.

2.1.3 Infrastructure defects and debris

The power supply and electrification system, particularly when excited by the rolling stock, is a major EMI culprit, giving rise to transient emissions and their propagation. The discontinuities in the track caused by normal operation or by debris also produce arcing that radiates as a threat. When debris, such as the iron balls experienced by LTU, interfere with the communication paths of the spot signalling systems, the unwanted reflections could affect the normal functioning of these systems. A final parameter to consider is defective design and construction, poor maintenance, or damaging of the return current path, as this has proved to be harmful to the EM environment.

2.1.4 Signalling and Communication Systems

The signalling and communication systems used both on the train and off it must be considered as well. The railway uses various systems that operate in the radio frequency range; it is important to be aware that these have the potential to interfere with the four systems considered in this research. The signalling and communication systems are susceptible to external EMI and are able to interfere; this susceptibility is a major concern to the safe operation of the railway. When the catenary and pantograph are detached, significant current variations are created from both sides of contact. These fast current variations are broad-band transient conducted disturbances that run through all the metallic parts of the electrification system. The disturbances measured by the GSM-R antennas are the consequences of these conducted disturbances travelling along the different metallic elements, such as the catenary, the catenary supports, the pantograph, and the train roof, to name a few. Indeed, each metallic element included in the electrification system can behave as an antenna and radiate electromagnetic disturbances, which can then be received by the GSM-R antennas fixed on the train roof.

12

To illustrate this, Figure 1 presents a simulation result obtained with the WIPL-D software. In this simulation, a simplified pantograph and a λ/4 monopole antenna are placed on a ground plane. The dimension of the monopole antenna is fitted to 900 MHz. The pantograph and the antenna are illuminated by a 900 MHz plane wave (E field in vertical polarization). As the next figure shows, the induced densities of current on the pantograph are comparable to the induced current on the antenna (Slimen, 2009). Figure 1. WIPL-D modelling of a pantograph. This has not been studied in detail but it appears the main features that may impact the level of interference received by the GSM-R antennas are: - The dimensions and materials of the radiating elements. - The polarisation of the E field. - The distance between the antennas and the radiating elements.

2.2 EMC in the rolling stock

CAF also provided the next table 2 (Deliverable 2 TREND project, 2012). It is only applicable when the line voltage is 25 kV or 16 kV AC, but it summarises very well the different factors to be considered in the definition of the rolling stock model. The first column lists the variables influencing the electromagnetic emissions of the rolling stock identified in the sections above. The column indicates whether each variable corresponds to transient or continuous phenomena. The effects of the variables are classified in three groups in the adjacent three columns. The three groups are: - Harmonic currents returned to the rail by the rolling stock influencing the compatibility with track circuits and spot signalling system. - Magnetic field emissions created by the rolling stock, both in the near field (influencing the compatibility with signalling equipment using

13

inductive coupling between the source and the receiver) and the far field (influencing the compatibility with radio and broadcast services). - Electric field emission created by the rolling stock, both in the near field (influencing the compatibility with communication systems such as GSM- R) and the far field (influencing the compatibility with radio and broadcast services). This study used MATLAB to simulate the first effect. The second and the third effects were simulated with CST Microwave Studio.

Definition of the Harmonic Magnetic field Electric field variable currents emissions created emissions returned to by the rolling created by the the rail by stock, both in the rolling stock, the rolling near field both in the near stock (influencing the field influencing compatibility with (influencing the the signalling compatibility compatibility equipment using with with track inductive coupling communication circuits and between the source systems such as spot and the receiver) GSM-R) and signalling and the far field the far field system. (influencing the (influencing the compatibility with compatibility radio and with radio and broadcast services) broadcast services)

Discontinuities ݱ ݱ ݱ between the pantograph and the OCS – Transient phenomenon

Switching operations ݱ ݱ ݱ of the main circuit breaker. – Transient phenomenon

Impedance of the ݱ Transformer – Continuous phenomenon. Note 1

Switching frequency ݱ ݱ of the power electronic used for the rectifier of the High Voltage Converter – Continuous phenomenon. Note 1

14

Definition of the Harmonic Magnetic field Electric field variable currents emissions created by emissions created returned to the rolling stock, by the rolling the rail by both in the near stock, both in the the rolling field (influencing near field stock the compatibility (influencing the influencing with signalling compatibility with the equipment using communication compatibility inductive coupling systems such as with track between the source GSM-R) and the circuits. and the receiver) far field and the far field (influencing the (influencing the compatibility with compatibility with radio and radio and broadcast broadcast services) services)

Switching ݱ ݱ frequency of the power electronic used for the Traction Inverter – Continuous phenomenon

Switching ݱ ݱ frequency of the power electronic used for the Battery Charger – Continuous phenomenon

Rise time of the ݱ ݱ voltage transitions of the power electronic used for the rectifier of the High Voltage Converter – Continuous phenomenon. Note 1

Rise time of the ݱ ݱ voltage transitions of the power electronic used for the Traction Inverter – Continuous phenomenon

15

Definition of the Harmonic Magnetic field Electric field variable currents emissions created by emissions created returned to the rolling stock, by the rolling the rail by both in the near stock, both in the the rolling field (influencing the near field stock compatibility with (influencing the influencing signalling compatibility with the equipment using communication compatibility inductive coupling systems such as with track between the source GSM-R) and the circuits. and the receiver) far field and the far field (influencing the (influencing the compatibility with compatibility with radio and radio and broadcast broadcast services) services)

Rise time of the ݱ ݱ voltage transitions of the power electronic used for the Battery Charger – Continuous phenomenon

Cable definition ݱ ݱ – Continuous phenomenon

Grounding – ݱ ݱ Continuous phenomenon

Filtering – ݱ ݱ Continuous phenomenon

Layout of on- ݱ ݱ board equipment – Continuous phenomenon

Quality of the ݱ feeding system: harmonics introduced by the power supply – Continuous phenomenon

16

Definition of the Harmonic Magnetic field Electric field variable currents emissions created emissions created returned to the by the rolling stock, by the rolling rail by the both in the near stock, both in the rolling stock field (influencing near field influencing the the compatibility (influencing the compatibility with signalling compatibility with track equipment using with circuits. inductive coupling communication between the source systems such as and the receiver) GSM-R) and the and the far field far field (influencing the (influencing the compatibility with compatibility radio and broadcast with radio and services) broadcast services)

Quality of the ݱ ݱ ݱ feeding system: irregularities in the feeding system (broken rail, cross sections) – Transient phenomenon

Quality of the ݱ feeding system: variations in line voltage – Continuous phenomenon

Rain – Transient ݱ ݱ ݱ phenomenon

Ice/Snow – ݱ ݱ ݱ Transient phenomenon

Wind – ݱ ݱ ݱ Transient phenomenon Table 2: EMC in the rolling stock.

17

2.3 EMC in the infrastructure

A railway electrification system supplies electrical energy to railway and multiple units so they can operate without an on-board prime mover. There are several different electrification systems in use throughout Europe. Railway electrification has many advantages but requires significant capital expenditure for installation. Electricity enables faster acceleration and greater tractive effort on steep gradients. On locomotives equipped with regenerative brakes, descending gradients require little use of air brakes as the locomotive's traction motors either become generators sending current back into the supply system to be used by other vehicles, or energy is dissipated by on-board resistors, which convert the excess energy to heat. Other advantages include the lack of exhaust fumes at the point of use, less noise, and lower maintenance requirements of the traction units. The main disadvantage is the capital cost of the electrification equipment, especially for long distance lines which do not generate heavy traffic. Suburban railways with closely spaced stations and high traffic density are the most likely to be electrified; main lines carrying heavy and frequent traffic are also electrified in many countries. Another consideration is that if the overhead wiring breaks down, all trains can be brought to a standstill. There are many voltage systems used in railway electrification systems around Europe; these include both standard and non-standard voltage systems; see Figure 2.

18

Figure 2. Map of voltage systems used in Europe (ERRAC, 2012). The permissible range of standard voltage is stated in EN 50163 and IEC 60850. These take into account the number of trains drawing current and their distance from the substation.

2.3.1 Power supply substation

A substation is a part of an electrical generation, transmission, and distribution system. Substations have a number of important functions, including transforming voltage from high to low, or the reverse. Electric power may flow through several substations between generating plant and consumer, and voltage may change at several steps. A substation may include transformers to change voltage levels between high transmission voltage and lower distribution voltage, or to interconnect two different transmission voltages. Along with the transformers, substations generally have switching, protection, and control equipment. A large substation will use circuit breakers to interrupt any short circuits or overload currents that may occur on the network. Smaller distribution stations may use recloser circuit breakers or fuses to protect distribution circuits. Substations do not usually have generators, although a power plant may have a substation nearby. Other devices such as capacitors and voltage regulators may also be located at a substation.

19

Electrified railways use substations, often distribution substations. In some cases, a conversion of the current type takes place at the substation, commonly using rectifiers for direct current (DC) trains, and rotary converters or static converters for trains using alternating current (AC) other than that of the public grid. Sometimes there are transmission or collector substations if the railway network operates its own grid and generators. These substation systems can be used to convert three-phase 50Hz or 60Hz alternating current (AC) to supply AC railway electrification systems at a lower frequency and single phase, like 16.7 Hz, or to change AC into direct current (DC) for those systems (primarily public transit systems) using DC for traction power. As shown in Figure 2, European systems include 750 V DC, 1500 V DC, 3000 V DC, 15 kV 16.7 Hz AC and 25 kV 50 Hz AC. Sweden has two substation systems: a booster transformer (BT) system and an autotransformer (AT) system. Both operate at 16.7 Hz and 15 kV.

2.3.2 Booster transformer

The BT is the most common system in Sweden. As the ratio in the transformers between the primary side and the secondary side is 1:1, the current is forced to use the path through the transformers to the converter. In this case, the path involves the S-rail and cables. In the S- rail, the rails are welded together to form one long rail, creating a long path for the current to return to the transformer. The transformers are spaced at intermediate intervals of 3 to 5 km, depending on the surrounding topography (Niska, 2008), see Figure 3.

Figure 3. Path of current in the BT system (Niska, 2008).

20

The benefit of the BT system is that it forces the current through the rail and transformer back to the converter.

2.3.3 Autotransformer

The AT is a 30 kV system that divides the voltage in half in the transformers. The overhead contact wire has a 15 kV power supply, and the return feeder, or negative feeder as it is called, has a different phase. The power of the return feeder is -15 kV, hence, the term negative feeder. The output current in this system can be higher than that in the BT system. Therefore, the AT system is used for tracks where more power, heavier loads, or faster accelerations are needed. The return current is sent back through the S-rail to the transformers. The transformers can be placed at an intermediate distance apart of 10 to 20 km depending on the surrounding topography. The AT system has a higher leakage to the ground of the return current (Niska, 2008) than the BT system; see Figure 4.

Figure 4. Path of current in the AT system (Niska, 2008).

2.3.4 Design considerations affecting the electromagnetic characteristics of the infrastructure

Electric trains receiving current from an system use a device such as a pantograph, bow collector, or . The device presses against the underside of the lowest wire of an overhead line system, the

21

contact wire. The current collectors are electrically conductive, allowing current to flow through to the train or and back to the feeder station through the steel wheels on one or both running rails. Non-electric trains (such as diesels) may pass along these tracks without affecting the overhead line, although there may be difficulties with overhead clearance. Alternative electrical power transmission schemes for trains include third rail, ground-level power supply, batteries, and electromagnetic induction. To achieve good high-speed current collection, it is necessary to keep the contact wire geometry within defined limits. This is usually achieved by supporting the contact wire from above by a second wire known as the messenger wire or catenary. This wire approximates the natural path of a wire strung between two points, a catenary curve, hence the use of catenary to describe this wire (sometimes called the whole system). This wire is attached to the contact wire at regular intervals by vertical wires known as droppers or drop wires. The messenger wire is supported at regular intervals by pulleys, links, or clamps. The whole system is then subjected to mechanical tension.

2.3.5 Overhead Catenary system

A catenary is a system of overhead wires used to supply electricity to a locomotive, streetcar, or light rail vehicle equipped with a pantograph. In the catenary system, the contact wire is one of the conductors of power to the electric train; the other conductor is the S-rail. The current is passed into the vehicle via a pantograph. Without significant voltage drops (losses), an IORE (iron locomotive) can pull 1000A and a double locomotive can pull a maximum of 840A. The contact wire must be at a constant height above the rails, normally 5.5 to 5.6 m from the running surface, so that collectors are required to move less. The work area is 4.8 to 6.1 m, however, as the height can be lower in tunnels. The use of contact wires requires the following considerations: - The contact thread must be suspended with carrier yarns to create a support line. - It must be tightened to create a straight line between the droppers. - It must allow an upward movement of 120 mm in response to dynamic forces from the .

22

The catenary system is mounted on a zigzag line between pylons so that the pantograph covers the entire width. The tension of the wires must not change when the temperature changes. The wire is sectioned into lengths of about 1200 m and stretched to 10 - 15 km. Weights are hung over the 2-wire-cut blocks to give 3 times the weight ~ 300-500 kg. The drive sections, often consisting of several electrically interconnected clamping sections, can be disconnected; therefore, an insulator is required between the drive sections to allow the pantographs to pass without arcing.

2.3.6 Section breakers

To allow maintenance on the overhead line without having to turn off the entire system, the overhead line system is broken into electrically separated portions known as sections. Sections often correspond to tension lengths as described above. The transition from section to section is known as a section break and is set up so that the locomotive's pantograph is in continuous contact with the wire. This continuous contact is achieved by having two contact wires run side by side over a length of about four wire supports: a new one drops down and the old one rises up, allowing the pantograph to smoothly transfer from one to the other. The two wires cannot touch each other, although the pantograph is briefly in contact with both wires. In normal service, the two sections are electrically connected to different substations, but this can be broken for servicing. On overhead wires, this is done by creating a neutral section between the wires, requiring an insulator. The driver must turn off the power when the train passes through the section insulator to prevent arc damage to the insulator; see Figure 5. Figure 5. Section insulator installed at a section break (Flury, 2012). Pantograph equipped locomotives may never run through a section break when one side is de-energized. If it does, however, the locomotive will become trapped. As it passes the section break, the pantograph will briefly short the two catenary lines together. If the opposite line is de-energized, this voltage loss may trip the supply breakers. If the line is under maintenance, personnel injury may

23

occur when the catenary is suddenly energized. Even if the catenary is properly grounded, the arc generated across the pantograph will likely cause damage to the pantograph, the catenary insulator, or both.

2.3.7 Neutral section (phase break)

Sometimes on a larger electrified railway, tramway, or trolleybus system, it is necessary to power different areas of the track from different power grids, the synchronisation of the phases of which cannot be guaranteed. (Sometimes the sections are powered with different voltages or frequencies.) There may be mechanisms for having the grids synchronised on a normal basis, but certain events may cause desynchronisation. This is no problem for DC systems, but in AC systems, it is highly undesirable to connect two unsynchronised grids. A simple section break is insufficient to guard against this, as the pantograph briefly connects both sections. Therefore, AC systems use a neutral section or phase break. This consists of two section breaks back-to-back, with a short section of overhead line that belongs to neither grid. If the two grids are synchronised, this stretch of line is energised (by either supply) and trains run through it normally. If the two supplies are not synchronised, the short isolating section is disconnected, leaving it electrically dead, ensuring that the two grids cannot be connected to each other. The sudden loss and subsequent reconnection to the supply over the neutral section might damage the locomotive if it is drawing power, so special signs are set up to warn the crew. When synchronisation is lost and the phase break is de-energised, the train's operator should put the controller (throttle) into neutral and coast through an isolated phase break section.

2.3.8 Grounding

The design and construction of the grounding systems have varied over time. The telecommunication system has a separate ground, consisting of a loop of copper wire in the earth around the building; an iron bar is connected to the loop to obtain a ground potential. Other systems use the S-rail as their ground potential. Former regulations dictated that only the S-rail was to be used as the ground. They also stated that the grounding

24

should be separated inside buildings but then reconnected when the S-rail was used as the ground. Return current in the rail or rails Both BT and AT systems use one or both of the rails as a return conductor for the power in the system. The conductor, the S-rail, is welded to make it a stable part of the system. Regenerative braking A regenerative brake is an energy recovery mechanism which slows a vehicle or object down by converting its kinetic energy into another form, which can be used immediately or stored until needed. This contrasts with conventional braking systems, where the excess kinetic energy is converted to heat by friction in the brake linings and therefore wasted. The most common form of regenerative brake involves using an electric motor as an electric generator. In electric railways, the generated electricity is fed back into the supply system. In battery electric and hybrid electric vehicles, energy is stored in a battery or bank of capacitors for later use, but this cannot be done in the railway: there must be another train in the right spot requiring power at precisely that moment.

2.3.9 Layout

Direct current Early electric systems used low-voltage DC. Electric motors were fed directly from the traction supply and controlled using a combination of resistors and relays connecting the motors in parallel or series. The most common DC voltages are 600 V and 750 V for and metros, 1,500 V and 650/750 V for third rail, and 3 kV for overhead. The lower voltages are often used with third or fourth rail systems, whereas voltages above 1 kV are normally limited to overhead wiring for safety reasons. The DC system is quite simple, but it requires thick cables and short distances between feeder stations because of the high currents involved. There are also significant resistive losses. In the United Kingdom, the maximum current that can be drawn by a train is 6,800 A at 750 V. The feeder stations require constant monitoring. The distance between two

25

feeder stations at 750 V on third-rail systems is about 2.5 km, and the distance between two feeder stations at 3 kV is about 25 km. Low-frequency alternating current Common DC commutating electric motors can also be fed with AC (universal motor), because reversing the current in both the stator and the rotor does not change the direction of torque. However, the inductance of the windings made early designs of large motors impractical at standard AC distribution frequencies. In addition, AC induces eddy currents, particularly in non-laminated field pole pieces, causing overheating and loss of efficiency. In such a system, the traction motors can be fed through a transformer with multiple taps. Changing the taps allows the motor voltage to be changed without requiring power-wasting resistors. Auxiliary machinery is driven by small commutating motors powered by a separate low-voltage winding of the main transformer. The use of low frequency requires that electricity be converted from utility power by motor-generators or static inverters at the feeding substations, or generated at altogether separate traction power stations. Polyphase alternating current systems This system provides regenerative braking with power fed back to the system, making it particularly suitable for mountain railways (provided another locomotive on the line can use the power). The locomotives use three-phase induction motors without commutators; these require less maintenance. The overhead wiring used in this system has two separate overhead lines, with the rail used for the third phase. This is quite complicated. In addition, the low frequency used requires a separate generation or conversion and distribution system. Train speed is restricted to one to four speeds, with two or four speeds only obtained by pole-changing or cascade operation or both. The system is only used today for four mountain (funicular or rack) railways using powered carriages; in these rail systems, the overhead wiring is less complicated and restrictions on speed are less important. As noted, two separate overhead wires are generally used, with the rail for the third phase, though three overhead wires are occasionally used. At junctions, crossovers, and crossings, the two lines must be kept apart, with a continuous supply to the locomotive. As the locomotive must have two live conductors wherever it stops, two collectors per overhead phase are used. The possibility of bridging a dead section and causing a short circuit

26

from the front collector of one phase to the back collector of the other phase must be avoided. The resistance of the rails used for the third phase or return is higher for AC than for DC because of the "skin" effect, but the frequency is lower than that used in industry. Losses also increase in this system, though not in the same proportion, as the impedance is largely reactive. Standard frequency alternating current This electrification system is ideal for railways covering long distances and/or carrying heavy traffic. Railway electrification using 25 kV, 50 Hz AC has become an international standard. The following two main standards define the system’s voltage: 1- BS EN 50163:2004 - "Railway applications. Supply voltages of traction systems”. 2- IEC 60850 - "Railway Applications. Supply voltages of traction systems" The permissible voltage range is stated in these standards; they take into account the number of trains drawing current and their distance from the substation. The Swedish system is now covered by the European Union's Trans- European railway interoperability standards 1996/48/EC "Interoperability of the Trans-European high-speed rail system" and 2001/16/EC "Interoperability of the Trans-European Conventional rail system". To prevent the risk of mixing out-of-phase supplies, sections of line fed from different feeder stations must be kept strictly isolated. This is achieved by Neutral Sections (also known as Phase Breaks), usually provided both at feeder stations and halfway between them. Typically, only half are in use at any time; the others simply allow a feeder station to be shut down and power provided from adjacent feeder stations. Neutral Sections usually consist of an earthed section of wire separated from the live wires on either side by insulating material, typically ceramic beads, designed so the pantograph will run smoothly from one section to the other. The earthed section prevents an arc being drawn from one live section to the other, as the voltage difference may be higher than the normal system voltage if the live sections are on different phases and the protective circuit breakers might not be able to safely interrupt the considerable current that would flow through them. To prevent the risk of an arc being drawn from one section of wire to the earth, when passing through the neutral section, the train must coast, and the circuit breakers must be open. In many cases, this is done manually by the driver. To help the drivers, a warning board is provided just before the neutral section,

27

with an advanced warning some distance before. Another board is set up after the neutral section to remind the driver to re-close the circuit breaker, although the driver must not do this until the rear pantograph has passed the board. In the UK, a system known as Automatic Power Control (APC) automatically opens and closes the circuit breaker; in this system, sets of permanent magnets alongside the track communicate with a detector on the train. Still, the driver needs to shut off power and coast; therefore, warning boards are provided at and on the approach to neutral sections.

2.3.10 External factors affecting the electromagnetic characteristics of the infrastructure

Quality of the feeding system The designs of the feeding systems are determined by the amount of traffic on the line. It is difficult to calculate the future design of engines or to know how they will work in a catenary system built decades ago. But the feeding system is only a cable delivering power to the trains; it is not critical in the power system. Weather conditions Frost creates an ice layer on the contact wire. The resulting distance between the pantographs and the contact wire creates an arc which has an impact on the EMI around the trains and infrastructure. Ballast resistance Ballast resistance is the resistance between the two rails of a track circuit. It comprises leakage between the rail fixings, sleepers, and earth. This resistance is dependent on the condition of any insulation, the cleanliness of the ballast, and the prevailing weather conditions. It is inversely proportional to the track circuit length and is expressed as ohm per kilometre, typical values being in the range of 2 to 10 Ω/km. Lower values may be obtained in wet conditions with bad drainage and/or contamination with conductive materials. However, according to UIC, that value must not fall below 1 Ω/km to keep the track circuit operational. Higher values may be obtained in dry/clean conditions or during frosty weather. A reliable track circuit must, therefore, be able to accommodate wide variations in ballast resistance.

28

Ground properties The ground potential differs around the globe, so geographical issues should be considered in any calculations. For example, the dampness in the ground and the specific soil have an influence. Train Engines, pantographs, and filters all affect the power supply system. This is an issue in itself.

29

30

3 CURRENT STANDARDS

3.1 EMC on railways

This chapter is based on the study performed by TREND consortium (Deliverable 3 TREND project, 2013) related to the current relevant railway EMC standards, including the main standards EN 50121, EN 50388, EN 50238 and EN 50122. The aim is to highlight areas that are currently unclear or require attention should the standards be modified or combined, or if new standards were to be written. Suggestions on how to address any identified issues are outside the scope of this licentiate. This section can be split into sub-sections. First, it details the parts of the EN 50121 standard It then moves on to include the other standards mentioned above. It includes points in the standards that are poorly defined or inaccurate or that introduce inaccuracies in both the test method and the results, points that result in excessive or unnecessary cost to the end user who commissions the testing, and points that impose design limitations on the various systems. Also included are some examples where the standards have proved to be insufficient. The EMC standards are based on the official Journal of the European Union, 2011/C 214/02 - Commission communication in the framework of the implementation of the Directive 2008/57/EC of the European Parliament and of the Council of 17 June 2008 on the interoperability of the rail system within the Community (recast). The European Union periodically publishes the titles and references of harmonised standards covered under the interoperability directive. The standards specifically written to deal with the electromagnetic environment in and around the railway environment are listed below: Cenelec EN 50121-1: Electromagnetic compatibility – Part 1: General Cenelec EN 50121-2: Electromagnetic compatibility – Part 2: Emission of the whole railway system to the outside world. Cenelec EN 50121-3-1: Electromagnetic compatibility – Part 3-1: Rolling stock – Train and complete vehicle

31

Cenelec EN 50121-3-2: Electromagnetic compatibility – Part 3-1: Rolling stock – Apparatus Cenelec EN 50121-4: Electromagnetic compatibility – Part 4: Emission and immunity of the signalling and telecommunications apparatus Cenelec EN 50121-5: Electromagnetic compatibility – Part 5: Emission and immunity of fixed power supply installations and apparatus. Cenelec EN 50122-1 Railway applications - Fixed installations - Electrical safety, earthing and the return circuit - Part 1 : protective provisions against electric shock Cenelec EN 50122-2: Railway applications Fixed installations - Electrical safety, earthing and the return circuit - Part 2: provisions against the effects of stray currents caused by DC traction systems Cenelec EN 50122-3: Railway applications Fixed installations - Electrical safety, earthing and the return circuit - Part 3: mutual interaction of AC and DC traction systems Cenelec EN 50125-1:1999 Railway applications - Environmental conditions for equipment - Part 1: Equipment on board rolling stock (EN 50125-1:1999/AC: 2010) Cenelec EN 50125-3:2003 Railway applications - Environmental conditions for equipment - Part 3: Equipment for signalling and telecommunications (EN 50125-3:2003/AC: 2010) Cenelec EN 50155:2007 Railway applications - Electronic equipment used on rolling stock (EN 50155:2007/AC: 2010) Cenelec EN 50159-1:2001 Railway applications - Communication, signalling and processing systems - Part 1: Safety-related communication in closed transmission systems (EN 50159-1:2001/AC: 2010) Cenelec EN 50215:2010 Railway applications - Rolling stock - Testing of rolling stock on completion of construction and before entry into service Cenelec EN 50238: Railway applications - Compatibility between rolling stock and train detection systems Cenelec EN 50388: Railway applications - Power supply and rolling stock - Technical criteria for the coordination between power supply (substation) and rolling stock to achieve interoperability

32

3.2 Summary of the standards

3.2.1 BS EN 50121 Railway applications. Electromagnetic compatibility.

Part 1: General The EN50121-X family is constantly referenced for all the systems and subsystems that are part of the railway, including spot signalling systems, track circuits, rolling stock, GSM-R and broadcasting services. Consequently, this set has to be analysed. The first part of this standard is a general introduction to the aspects of EMC in the railway system. It outlines the areas of EMC that are not included and introduces the parts of the EN 50121 standard as a whole. Part 2: Emission of the whole railway system to the outside world This standard details the emission limits of the whole railway system. It sets out the limits and the measurement system used and provides ideas for non-standard set-ups such as elevated railways. Part 3-1: Rolling stock - Train and complete vehicle This is the standard used by rolling stock manufacturers for the EMC assessment process of their new vehicles in most cases. This is true, not only in the EU, where adhering to the standard represents a way to demonstrate compliance with the requirements of the EMC Directive 2004/108/EC, but also in projects around the world where application of the EN 50121 is not compulsory. Part 3-2: Rolling stock – Apparatus This standard establishes the mandatory requirements the rolling stock must fulfil for the apparatus installed on the train. However, sometimes it is not enough to ensure the compatibility of the whole vehicle. This standard sets out the requirements for train-borne apparatus. It covers both emissions and immunity of apparatus. Apparatus installed in rolling stock using ‘good engineering’ practice is presumed to confer immunity for the whole vehicle; however, this is not always achieved. Part 4: Emission and immunity of the signalling and telecommunications apparatus. This standard concerns the emission and immunity levels for signalling and telecommunications (S&T) apparatus within the railway environment.

33

The EMC Directive states that any apparatus compliant with an OJEU listed standard is conferred with a presumption of conformity to the essential requirements of the EMC Directive, when that apparatus is installed in its intended environment. It is often used as the standard to test to for trackside equipment not covered in the other parts not designated as signalling or telecommunications, such as equipment installed on the platform. Part 5: Emission and immunity of fixed power supply installations and apparatus. This standard concerns the emission and immunity requirements for fixed power supply installations and apparatus within the railway environment, as defined in EN50121-1: 2006. The intention and purpose of the standard is that power supply installations and apparatus compliant with the standard will be conferred with a presumption of conformity to the essential requirements of the EMC Directive 2004/108/EC.

3.2.2 BS EN 50122 Railway applications. Fixed installations. Electrical safety, earthing and the return circuit.

Part 1: protective provisions against electric shock This standard specifies requirements for protective provisions relating to electrical safety in fixed installations associated with AC and DC traction systems and to any installations that may be endangered by the traction power supply system. It also applies to all fixed installations necessary to ensure electrical safety during maintenance work within electric traction systems. Part 2: provisions against the effects of stray currents caused by DC traction systems. This standard specifies requirements for protective provisions against the effects of stray currents resulting from the operation of DC traction systems. This standard applies to all metallic fixed installations forming part of the traction system and to any other unrelated metallic components located in any position in the earth, which may carry stray currents resulting from the operation of the railway system.

34

It also applies to all new electrification of DC railway systems. The principles may be applied to existing electrified systems where it is necessary to consider the effects of stray currents. Part 3: mutual interaction of AC and DC traction systems This European standard specifies requirements for the protective provisions relating to electrical safety in fixed installations, when it is reasonably likely that hazardous voltages or currents will arise for people or equipment, as a result of the mutual interaction of AC and DC electric traction systems. It also applies to all aspects of fixed installations necessary to ensure electrical safety during maintenance work within electric traction systems. Its scope is limited to basic frequency voltages and currents and their superposition. Note: this European standard does not cover radiated interferences.

3.2.3 BS EN 50125 Railway applications. Environmental conditions for equipment

Part 1: equipment on board rolling stock This standard defines the surrounding environmental conditions in 14 environmental parameters. The 12th of these parameters is the threat level of ambient electromagnetic noise to on-board equipment. It establishes EN 50121-3 as the reference to follow. Part 3: equipment for signalling and telecommunications This standard, approved by the technical committee TC 9X and closely related to EN 50125-1, specifies the environmental conditions for signalling and telecommunication electronic equipment. EN 50125-3 proposes to follow the testing requirements laid out in EN50121-4, but on-board equipment must comply with the surge requirements of EN 50155. The surge requirements of EN50121-4 were revised for the 2006 version of the standard, and these requirements are now more stringent than those required by EN50155.

35

3.2.4 BS EN 50155 Railway applications. Electronic equipment used on rolling stock

Subsection 5.5 explains equipment has to be protected against radiated and conducted emissions: the tests are defined in 12.2.8.1 and 12.2.8.2. Surges, electrostatic discharge and transient bursts are dealt with in 5.4 using the test specified in 12.2.7. All the specified tests in these two sections in this standard call on the EN50121-3-2 with no additional test information other than the requirement that equipment be set up as in the installed condition.

3.2.5 BS EN 50159 Railway applications. Communication, signalling and processing systems: Safety-related communication in closed transmission systems

This European standard is applied to communication between safety critical equipment. In the case of spot signalling systems like Eurobalise, on-board transmission equipment and the KER balises, these are included in closed transmissions systems. However, this standard is not interested in the physical layer and does not consider electromagnetic interferences in the receiving/transmitting path of the signal.

3.2.6 BS EN 50215 Railway applications. Rolling stock. Testing of rolling stock on completion of construction and before entry into service

As noted in the Introduction, this European standard specifies general criteria to demonstrate by testing that newly constructed complete railway vehicles conform with standards or other normative documents. In other words, the large list of tests the manufacturer must perform are related to the specific standard or norm to follow. Situated between static tests and dynamic tests, electromagnetic compatibility is in the latter list, in Section 9.15. Even so, “Interruption & voltage jump and short circuit tests,” as explained in Section 9.16, must be considered in this licentiate.

36

3.2.7 BS EN 50238 Railway applications. Compatibility between rolling stock and train detection systems

This European standard describes a procedure for mutual acceptance of rolling stock to run on specific routes. It describes how to measure interference currents and the susceptibility of train detection systems, explains how to charactise traction power supplies, and provides the procedure for acceptance. The result of the acceptance procedure is a structured justification document; this “compatibility case” document evidence that the conditions for compatibility have been satisfied. This European standard is not generally applicable to those combinations of rolling stock, traction power supply and train detection system accepted as compatible prior to its issue. However, as far as is reasonably practicable, this European standard may be applied to modifications of rolling stock, traction power supply or train detection systems which may affect compatibility. The scope of this European standard is restricted to the demonstration of compatibility of rolling stock with a train detection system’s characterisation (e.g. gabarit). Radio based signalling systems are not within its scope.

3.2.8 BS EN 50388 Railway applications. Power supply and rolling stock. Technical criteria for the coordination between power supply (substation) and rolling stock to achieve interoperability

This European standard sets out the requirements for the acceptance of rolling stock on infrastructure. It deals with the definition and quality requirements of the power supply at the interface between traction unit and fixed installations. The standard specifies the interface between rolling stock and electrical fixed installations for traction defined as the “supply system”.

37

3.3 EN 50121 –1: 2006: Railway applications - Electromagnetic compatibility - Part 1: General

This first part of EN 50121: 2006 is a general introduction to the aspects of EMC in the railway system. It outlines the areas of EMC that are not included and introduces the parts of the EN 50121 standard as a whole. This introduction to the standard set does not mention whether the standard should be updated with the use of more modern railway apparatus and systems. The disclaimer that the limits are set from measurements at the time when the EMC directive was developed is fair, but perhaps there needs to be some concession as to the changing of not only the railway system but the surrounding EM environment. One of the issues with railway EMC is introducing new systems such as rolling stock to older track and signalling systems, and it is probably necessary to include some time aspect in the standard. The EN 50121 –1: 2006 states adherence to both the immunity and emission limits in this standard does not “guarantee that the integration of the apparatus within the railway will necessarily be satisfactory”; in addition, “the immunity and emission levels do not of themselves guarantee that the railway will have compliance with its neighbours”. The standard also says “compliance with this Standard has been judged to give satisfactory compatibility”. As it is unreasonable to expect the standards to account for all eventualities in the extremely varied railway EM environment, the first two statements are sensible disclaimers. However, the third statement seems to contradict this, giving an impression that the compatibility is satisfactory overall, when in reality this may not be the case. Clarification of this point is required, or an expansion of the standard to be able to state that compatibility in the real world is likely to be achieved if the requirements of the EN 50121-x series of standards have been met. Additional requirements for the specific environment may also be necessary, for example, national technical rules for different track circuits in different countries.

3.3.1 Scope

This section provides a list of the aspects not covered by EN 50121: - Nuclear EM pulse - Abnormal operating conditions

38

- Direct lightning strike induction - Emissions from intentional transmitters, pacemakers, biological effects and safety. Not mentioned, but perhaps necessary, is intentional interference with the idea of deliberately causing a malfunction, for example, an act of terrorism, and the requirements to avoid this. Abnormal operating conditions are not defined, and it is unclear whether this simply refers to malfunctions of equipment that cannot realistically be accounted for in the standard. Medical awareness becomes more important, but since the EMC standards for industry etc. are clearly separated from medical aspects, it is reasonable the EMC standards for railways should also be separated from medical issues. If EMC standards for railways cover all frequency ranges between DC and a certain maximum frequency, this will give a good frame for medical expertise to work with – concerning exposure times etc. (In a railway environment, we can expect maximum electromagnetic fields of the following magnitudes, etc.). The levels of exposure are controlled in applying standard 50500, but in practice, this is applied only to DC-20kHz. So this Standard does not cover the frequencies considered in the present study. Note: EN 50500: 2008 covers measurement procedures of magnetic field levels generated by electronic and electrical apparatus in the railway environment with respect to human exposure. Since 2006, there has been more research on biological effects, driven by mobile telephony companies, among others, so a more informed view, if not an in-depth treatment, could perhaps be included. See Section 9 for a note on EM safety. The performance criteria outlined in EN 50121 –1: 2006 are mostly derived from the product description and documentation. The standard does not specifically mention what to do in the case of each type of failure, however, i.e. whether an immunity test is passed or not. It is given that if the apparatus becomes unsafe or dangerous, it will fail the test. This seems to imply that some knowledge of safety critical systems is required, but these systems are stated as not included. The addition of some safety information may aid in the definition of critical performance loss as a result of poor EMC. The safety issue may have resulted from the EMC Directive, 2004/108/EC, being defined by the EC as “not a safety directive”, whereas the interoperability directive, which requires conformance to be demonstrated by the same series of standards, is a safety directive.

39

3.3.2 Annex A: the Railway System

This annex includes a list of the uses of the on-board electrical supply on the train. Passenger services such as laptop charging points and on-board WIFI are not included. No mention is made of the effects of EMI on humans in the railway environment; as stated earlier, this is not part of the standard. Phenomena for immunity are listed as: conducted low frequency, radiated low frequency, conducted high frequency and radiated high frequency. This covers all potential incoming radiation to the railway system. Emissions only have limits applied to magnetic fields up to 9kHz, radio frequency fields (taken as between 9kHz and 2.5GHz) and voltage fluctuations. This misses a large portion of the EM spectrum, namely, any E fields below 9kHz and any conducted emissions that might occur. The voltage fluctuations mentioned are only those stemming from power frequency and harmonic currents, not from any other source. The radio frequencies are only “radio frequencies from trains” with no mention of radio emissions from other areas such as OHLE (Overhead Line Equipment). The test methods are only present up to 2.5GHz, and in the modern environment (including, for example, 4G networks operating at 2.6GHz), this upper limit is not high enough. The phrase “produce energy” used in this section should be reworded, perhaps to “create noise” or something similar. The overheads in an AC railway are recognised as acting as an antenna, with no mention of the main emission threat frequencies from OHLE. Trackside equipment effects on other aspects of the railway are given, but there is nothing about trackside immunity to the railway system. If this is not the place for such detail, there needs to be a reference to the area where this is given. In addition to the external sources of disturbance, there should probably be a victim circuit to highlight the issues met/ systems likely to be interfered with as a result of high train emissions.

3.3.3 Useful additions to the EN50121-1

This standard is understood as the main reference for establishing the framework to manage the EMC for railways, and, as such, should give the normative references needed to fulfil all the requirements for EMC in order to have a safe and functional railway system. Currently, it is possible to miss some standards during the validation process. Normative

40

references not present in Part 1 include EN 50122-1, EN 50122-2, EN 50122-3, EN 50215 and EN 50388. Some figures from the given normative references could be considered for inclusion in the general part of EN50121. In standard EN 50238, Figure 6 describes the various systems on the railway system and their influences on each other (see paragraph 4.9, EN 50238). This standard is meant to manage the influences between the rolling stock and the train detection systems, but it is also valid for any kind of auxiliary electronic equipment (signalling and telecommunication systems, rail switches detectors, etc.). It would be useful to include it in standard EN 50121-1 as a standard procedure to show the different interactions to consider when establishing the EMC framework in the railway environment.

Figure 6. Influences between systems (EN50388 Railway applications – Power supply and rolling stock). In EN 50388 there is also a flow diagram describing the process to follow when introducing a new element into the railway network. This applies to the power supply but the process shows the framework for adding any

41

new infrastructure component within the existing traction unit(s) and power supply network (see paragraph 10.2; EN 50388). In addition, it describes the acceptance procedure, compatibility study, tests and test methodology. It would be useful to include it in standard EN 50121-1 as the procedure to follow when introducing new electronic equipment to the infrastructure network.

Figure 7. Procedure for compatibility study of harmonics and dynamic effects when adding a new element to the railway system (1-222 A, 2007).

42

3.4 EN 50121 –2: 2006: Emissions of the whole railway system to the outside world

EN 50121-2:2006 details the emission limits of the whole railway system. It sets out the limits and the measurement system used, along with ideas for non-standard set-ups such as elevated railways.

3.4.1 Emission limits

EN 50121-2:2006 states that for a non-electrified line, the limits for a 750V DC railway system should be applied. Possibly a new set of limits for non-electrified lines could be established, as some limits may be too harsh for a diesel traction system, and, equally, some may be too lax, as diesel traction can emit at different frequencies and powers to DC traction. This may be required for both power and brake testing, as some diesel-electric traction units are fitted with regenerative braking. Since regenerative braking feeds current back to the catenary, additional magnetic fields are created in this process. There is no mention of regenerative braking in EN 50121-2:2006, but one area of the test requires emissions to be measured at 80% braking efficiency. The frequency limits for EN 50121-2:2006 are set between 9kHz and 1GHz. Apparatus limits go up to 2.5GHz. The standard mentions that sensitive frequencies should be tested, but for those below 9kHz or over 1GHz, no measurement or test method is given or inferred. Below 9kHz the EM environment may be too noisy to correctly isolate railway emissions. Therefore, these limits can leave testers in the dark when it comes to testing outside the frequencies, an issue of increasing size as, for example, on-board WIFI among other potentially high emitters (WIFI is not specifically mentioned, though presumably covered as an “intentional transmitter” in Part 1), working at frequencies in excess of 1GHz. Dependent on the equipment being used in on-board rolling stock or identified victims in the external environment, testing for emitted frequencies outside of the range specified may be required. Yet the standard provides no information on how this might be carried out. In practice, measurements have been conducted for EN 50121-2:2006 with the frequency range extended up to 6GHz using suitable receivers and horn antennas. Therefore, given the frequency limits, suitable measuring methods should be proposed.

43

The substation emissions are well defined, but there is no mention of the environment in which the substation or the other fixed installations are present, for example, trackside or in a particularly harsh EM environment. These limits are applied to “other fixed installations” with no mention of what sort of fixed installation (trackside box, signalling unit or substation) and no reference to EN 50121-5:2006 which deals specifically with these power systems. As EN 50121-5:2006 will need to be followed in the presence of a substation or other fixed installation, a reference to EN 50121-5:2006 may be required rather than stating the same limits in the two different areas. Finally, there is no mention of conducted emissions from substations or directly from trains. A standardised hazard identification procedure, to be used whenever a change is made, may be one way of handling this issue. Usual practice in the UK is to perform hazard identification (HAZID) of the EM environment either before a new railway is constructed or before upgrading an existing section of railway. This information is used to inform both design and measurement requirements over and above EN 50121- 2 or –5 requirements. Sweden has an equivalent for implementing railways next to airports: according to a Swedish regulation, electrified railways cannot be built close to airports without special permission, and not without an analysis of potential electromagnetic interference. Studies in Norway and Germany have reached the conclusion that railways can be built as close as 300-500 metres to airports without electromagnetically affecting flight traffic. In Sweden, certain regulations, dating back to 1957, say overhead power lines for power distribution and rolling stock should not be built closer than 4 kilometres to airports (“Starkströmsförordningen” SFS 2009:22 published by ”Elsäkerhetsverket”). According to a document from International Civil Aviation Organization (ICAO), railways are considered physical obstacles, not a source of electromagnetic interference. Measurements taken by “Trafikverket”, verify there is no EM radiation risk from railway tunnels to instruments or people in terminals. EMC measurements in such areas will be very costly for the railway industry and rolling stock owners in Sweden. It would be much more cost-effective and eco-friendly for Sweden as a whole to run operations/services by having railways closer to airports, a situation achievable by proper EMC analysis. Such kind of analysis is currently not done in Sweden. Standard EN 50121-2 is the only EMC standard dealing with emissions of the railway system to the outside world. Despite this, EN 50121-2 has no

44

directives for infrastructure owners to build railways near the airports in Sweden, other than a clause mentioning special environments requiring more attention. This constitutes an issue for infrastructure; rolling stock owners are forced to carry out unnecessary and costly EMC measurements. A reference to the construction of the railway in sensitive areas like airports, or other areas sensitive to electromagnetic influences, would it make easier for infrastructure manufacturers to deal with possible limitations. One suggestion is to add all these cases to a guidance document setting out the need for hazard identification and demonstrating how the risks are mitigated. There is no mention of attempting to separate transient and continuous emissions by either measurement or analysis, largely because it is impossible to distinguish between these types of emission with a simple peak measurement. The 2006 update to EN50121-3-1 featuring the various moving test methods (slow speed, against mechanical brake etc.) attempts to distinguish continuous emissions from transient ones using the quasi-peak method (for stationary vehicles): it is designed so that a vehicle that passes Part 3-1 will also pass Part 2 (note: this is not necessarily achieved in practice due to resonance in the OHLE system).

3.4.2 The measurement system

EN 50121-2:2006 uses the peak measurement system with the measurement time as a function of the measuring set. A recommended value of the measurement time is given as 50ms in clause 5.1.1. The Standard also mentions correctly in 5.1.4 that “noise may not attain its maximum value as the traction vehicle passes the measurement point”. This suggests the measurement set should be “active for a sufficient duration” which would certainly be longer than 50ms. One of these systems should selected, preferably the latter. On-site line-side testing in UK shows interference from the OHLE can occur well before and after the train has passed the measurement point. Clause 5.1.6 in EN 50121-2:2006 states that the 10m measurements should be taken from the mechanical centre of the antenna, but the phase centre is potentially a more useful datum point (particularly for immunity measurements). Antennas are calibrated, and the standard mentions obtaining a background trace of EM noise when a train is not present to allow identification of environmental EM noise. It does not say what should be done when a large amount of EM noise is present in the

45

background scan from the railway system (this is mentioned in part 3-1, but the instructions say not to test if the noise is greater than 6dB below the limit, and there is no reference to this in part 1). This may be useful to obtain the worst-case scenario with emissions from OHLE or conducting rails when a train is not present. Standard practice is to run the test (over a particular frequency) band for an certain amount of time before and after the train passes. This automatically provides ambient measurements both before and after the train is present. The current EN 50121-2:2006 is not set up to obtain a worst-case scenario for the supply system. This is manifested by actively avoiding the areas that would cause the highest emissions, for example, stating that for an OHLE railway, measurements should be taken at mid-points between support masts away from the discontinuities, and for a trackside conductor, 100m from any gaps. This is done to “avoid inclusion of the transient fields associated with the make and break of collector contact”. However, in a real environment, these transients are present (for example, at sets of points or other crossover areas) and, as such, should be tested in the pursuit of a worst-case scenario and reasonable train immunity. Single side tests are suggested as “the majority of the emission is produced by the sliding contact if the train is moving”. This is fine if the system is symmetrical, as in an OHLE set-up, but in the case of a third rail system, symmetry is not present, so either both sides or, as suggested, the collector side should be measured. Finally, the majority of low frequency emissions do not stem from the sliding contact. A large issue is the factor used to modify the measured electric field to the 10 m electric field values. In the provided table no frequencies lower than 150 kHz are given; it suggests whether to carry on with the value for 150 kHz or extrapolate the factor for lower frequencies. This may cause a large problem when testing to this standard, as measurements are sometimes carried out down to 9 kHz. Resonance effects are “recognised” in clause 5.1.9, but there are no limits or specifications; if present, they are simply to be noted in the test report. Clause 5.3 states any transients resulting from switching (such as power circuit breakers) should be disregarded when selecting the maximum signal level. If this is a regular occurrence in real world usage, perhaps it should be included in the emission limits. A major problem with on-site testing of railways is weather, which can have a huge impact on the variation of the emissions. EN 50121-2:2006 says measurements should be taken “after 24 hours during which not more than 0.1mm rain has fallen”. However, as testing and/or track possession is often booked well in advance, this is very rarely met. Yet the

46

tests are still carried out, and weather conditions are recorded at the time of testing, as stated in EN 50121-2:2006. It would be beneficial to have a set of weather dependent emission limits, or a list of the changes in uncertainty/test levels caused by the weather conditions. This should include variables such as humidity, temperature, the dielectric constant of the ground and the state of the pantograph. The weather is one of the largest causes of different test-to-test emission values and must be addressed to reduce the uncertainties and tighten the limits. Testers need to know the potential effects of weather conditions. One problem of reducing the test uncertainty is alluded to in Annex B.12.3; other units operating in the area may currently be within the repeatability error of the current measurement system. If the repeatability error is reduced, this may become a problem. Since electromagnetic emissions from the arcs at the sliding contacts are due to the distance between the (carbon) contact and the (copper) wire, there is one way of standardising measuring procedures: a thin film of isolating material (plastic) can be attached to the carbon contact. This will give a continuous arcing, making it possible to standardise the measurements. In cold climates, with ice on the catenary, electric arcs can continue as long as the ice is there. Two test conditions are set out in EN 50121-2:2006, one where the traction is moving at over 90% of the maximum service speed at maximum rated power, and another at maximum rated power at a selected speed. There is no specific statement about regenerative braking, but if electric braking of any type is possible, it should be tested during the 80% brake efficiency test. Magnetic brakes/eddy current brakes are not mentioned Annex B states: “there is as yet no method by which this (the worst-case scenario for emissions under power) can be defined”. This will also need some work. It is not mentioned if the “outside world” referred to in this standard includes the inside of the train. If not, a test method should be designed to obtain the field strengths inside passenger carriages, for safety purposes and also to ensure the availability of passenger and train services. Annex B does not mention multiple power units / traction at both ends of the train, although multiple pantographs are mentioned. The section suggests higher noise emission may result from the trailing pantograph, which has a disturbed contact with the catenary in the double pantograph scenario. EN 50121-2:2006 seems to suggest emission limits can be relaxed for multiple pantograph arrangements, either a double pantograph on one unit or coupled unit of two or more single pantograph locomotives. If the emissions are higher than single pantograph limits, the

47

unit, as a whole, should fail the test, as the limit is set to stop interference irrespective of the pantograph system. There is no allusion to multiple units such as EMUs (electric multiple units) or trains with traction at both ends. If coupled units are to be used extensively, a fixed limit for the emissions is called for. An extreme effect can be seen in the case where the addition of multiple units under heavy load caused interference on telephone lines in Sweden. These traction units would have passed the EMC tests singularly, but obviously not as a group; hence, problems occurred when the ground return paths were overwhelmed. Both the pantograph effects and the increased return currents would have to be investigated in this modification to the standard to take into account multiple traction units or to reflect real world usage. There is no mention of testing in a different manner or calculations for either single track or more than dual tracks (for example, 4 tracks). It might be necessary for different emission limits to be applied to 4 track railways, or railways with 4 track sections. The current requirement is to test 10 m from the centre of the outer track: as long as any higher emissions resulting from multiple trains only appear on inner lines then that is reasonable. Levels on inner lines could be calculated and dealt with if too high. Calculations for the limits at 10m (Annex C) assume the radiation is being measured in the far field, which will not be the case for low frequency emissions from the overheads. This may cause an issue when the measurements are not representative of the actual emissions due to near field issues.

3.5 EN 50121–3-1: 2006: Rolling stock – train and complete vehicle

EN 50121–3-1: 2006 is the standard used by rolling stock manufacturers for the EMC assessment process of their new vehicles in most cases. This is true not only in the EU, where adhering to the standard represents a way to demonstrate compliance with the requirements of European Directive 2004/108/EC (the harmonised standard for the railway environment), but in projects around the world where application of EN 50121 is not compulsory. However, this standard is not enough to completely assess new rolling stock with respect to EMC, as many aspects are not covered. Various working groups are now focussed on this issue; it is expected that the Standard will be modified and new ones will appear.

48

The following sections describe the drawbacks of EN 50121-3-1 based on experience in recent projects.

3.5.1 General issues of EN 50121-3-1

First of all, the repeatability of the result of the measurements is not assured. The conditions for the test to be performed are not properly defined in the standard. Examples include the following statements in EN 50121-3-1: “Measurements shall be performed in well-defined and reproducible conditions. It is not possible to totally separate the effects of the railway system and the stock under test. Therefore the operator and the manufacturer have to define in the contract the test conditions and the test site for compatibility with signalling and communication systems and for interference on telecommunication lines”. For the radiated emission test, it says: “Since resonances may occur in the overhead line at radio-frequencies, it may be necessary to change the test site.” This means the train manufacturer could chose the most convenient emplacement, making it easier to pass the tests from the point of view of the manufacturer, but not necessarily ensuring EMC in all situations, for example, Alstom and the initial testing of the Class 390 Pendolinos. Train operators or infrastructure administrators could choose an area with the worst EM environment area in the infrastructure; this may be excessive, resulting in more cost and less efficiency. Apart from these two examples, other unspecified factors might contribute to the uncertainty of the results. These include: Weather conditions on the test day and for some time previously (and, thus, ground conditions) will affect the result of the test. EN 50121-2 defines some conditions for the radiated emission test described in this part of the Standard. However, EN 50121-3-1:2006 does not include any reference to weather conditions. During the winter and depending on the country where the new vehicles have to be validated, it might be difficult to find perfect weather conditions to perform the tests. Finally, if the same train is sold in two different countries to two different customers, the Standard does not make it clear if it is necessary to repeat all the tests according to country of supply. It is not clearly specified and the customer could ask for all the tests to be performed again, with all associated costs. The testing might not have to be repeated for CE marking purposes, but another operator might insist on tests being repeated to ensure that the

49

rolling stock will run satisfactorily on his line. This may result in vehicle(s) passing EN 50121-3-1, but when measurements are performed to EN50121-2 (railway emissions to the outside world), these limits may not be met. Hence, we should calculate worst-case resonance conditions and perform 50121-3-1 tests under these conditions.

3.5.2 Compatibility with signalling and communication systems

CENELEC Technical Specification CLC TS 50238 covers the requirements for track circuits and axle counters now in use. However, only certain preferred types of track circuits and axle counters are included, leaving several types of train detection systems used in the different infrastructures not covered. For cab signalling and ATP systems used in different infrastructures EN 50121-3-1:2006 does not specify any special requirements. This equipment shall fulfil the requirements of EN 50121-3-2:2006 and the train shall fulfil the requirements of EN 50121-3-1:2006. However, this may not be enough; problems can appear, affecting the availability and the safety of the vehicles, such as loss of cab signalling systems. The same situation occurs with the ERTMS (European Rail Traffic Management System) including the ETCS and GSM-R. It is not clear if the requirements of standards EN 50121-3-2:2006 and EN 50121-3- 1:2006 are enough to prevent EMC problems affecting the availability and the safety of the vehicles as a result of interference from GSM transmitters. GSM-R and other on-board transmission equipment are exempt from emission limits as they are designated “intentional transmitters”.

3.5.3 Interference on non-railway telecommunication lines

This section is divided into two parts: digital telecommunication lines and analogue telecommunication lines. For digital telecommunication lines, EN 50121-3-1:2006 does not define any test set-up, test procedure, or limit to be applied. Nor does it include a reference to other known standards. For analogue telecommunication lines, it does include a reference to the ITU-T (International Telecommunication Union – telecommunication Standardization sector), describing a method based on the measurement of the psophometric

50

current. However, no emission limit is defined for the different types of rolling stock or infrastructure to achieve compatibility with non-railway telecommunication systems. Limits, tests and details of and for railway signalling and telecommunication apparatus are provided in EN 50121- 4:2006. As a consequence, in most projects working on new rolling stock material, the purchaser does not know how to meet the standards. EN 50121-3-1 standard includes a reference to this test, in which the purchaser is asked to perform an analysis of compatibility with signalling and communication systems. As the method and the limits are unknown, test procedures and limit values taken from past projects are usually applied. This means the test lacks the certainty that it is either necessary or convenient for the project under development. This could create disputes between the purchasers and contractors, adding extra cost; it could also lead to non-compliances that delay the development of the project, add extra cost, and reduce the efficiency of the vehicles, as modifications such as extra filtering might need to be installed. Another issue to be considered is that most of the telecommunication lines, which run parallel to the railway tracks, are no longer conventional analogue lines. Most use digital transmission and, in some cases, an optical rather than an electrical signal. Therefore, the compatibility test described by ITU-T is not valid for those lines. In some cases, only the subscriber line between the commuter centre and the customer transmits analogue signals. This means only the urban vehicles are able to interfere with these communication lines due to their location. Therefore, different types of communication arrangements should be studied if this test is to be included in the standard. If necessary, there must be more details on test set-up and limits for the various types of rolling stock and infrastructure, for compatibility not only with analogue telecommunication lines, but also with digital telecommunication lines. Moreover, the applicability of this test must be different for the different types of rolling stock.

3.5.4 Emissions

The radiated emission requirements of EN50121-3-1 are divided into two parts: a stationary test and a slow moving test. Thus, two different sets of limits are described. For the stationary test, the limits are divided into two categories: trams/trolleybuses for use in city streets, and other rail vehicles. In the slow moving test, the limits are divided into three categories

51

depending on the supply voltage: a) 20/25 kV AC; b) 15 kV AC, 3 kV DC and 1.5 kV DC; c) 750 V/600 V DC, including trams/trolleybuses for use in city streets and diesel and diesel electric locomotives and multiple units. Moving at slow speeds should limit sliding conductor arcing and eliminate transients. The Quasi-Peak measurements are carried out with the vehicle stationary and were introduced in response to CISPR’s input when the EN50121-X standards were turned into IEC standards. The Quasi-Peak measurement aligns with, for example, CISPR 11 (EN55011), and is there to demonstrate that emissions will not affect radio and broadcast services. The division of the limits for the stationary test makes sense, as the street vehicles run closer to other victim equipment, making the application of a more restrictive limit necessary. It could be said (due to higher emissions and relaxed limits) that 25kV railways should have a greater distance to nearby infrastructure (domestic buildings) than 1.5kV DC. It could also be argued that for a city 750V DC might be the only appropriate electrification to minimise interference to the surrounding infrastructure. However, the division of the limits for the slow moving test is not reasonable. The limit line C, the most restrictive line, must be applied to all vehicles supplied by 750 and 600 V DC lines, as well as diesel and diesel electric vehicles. Electric vehicles supplied by 750 and 600 V DC lines include Metro vehicles which run exclusively underground, far from other victim devices. The application of such a restrictive limit to these vehicles could reduce their efficiency and increase their cost because of the necessity of stronger filtering. The same occurs for diesel and diesel electric vehicles. Although they are supposed to generate lower electromagnetic noise, they still include high power switching devices, which normally generate electromagnetic noise. Again the application of a more restrictive limit increases the cost of the vehicle and reduces its efficiency. A potential solution is to set a different limit for underground services or diesel services, but this could increase cost. There is no mention in the standard of whether the manufacturer of the rolling stock needs to define a worst-case scenario for the train; for example, maximum emissions may not be at 90% load but at some other traction load setting. EN50121-3-1 says the vehicle should be producing 1/3 of its total tractive effort in an attempt to maximise emissions by power electronics on “partial” conduction; however, it would be best to measure the worst- case scenario using modelling. EN50121-3-1: 2006 does not consider, or have any explicit requirement for, vehicles fitted with regenerative braking. The slow moving test permits emission measurements to be made under accelerating and

52

decelerating conditions. It should be made mandatory for vehicles fitted with regenerative braking to be tested for emissions when the vehicle is decelerating under the influence of the regenerative brake, in addition to the standard deceleration test. In the case of regenerative braking, the EN50121-3-1 option of replacing the slow moving test with a stationary test where brakes are applied is inadequate, as the regenerative brake will not be active and, therefore, not tested. Testing traction at 1/3 of tractive effort against mechanical brakes is specified in EN50121-3-1: to properly categorise this, there must be an indication of whether this is a realistic test environment. For example, testing a DC locomotive against a mechanical brake is unnecessary; the DC motors will not be turning so there will be no source of commutation emissions.

3.5.5 Immunity

Section 5 of EN50121-3-1 states “the vehicle can be deemed to be immune to a level of 20V/m over the frequency range 0.15MHz to 2GHz” if the individual apparatus is tested to EN50121-3-2: 2006. This is inaccurate. Inspection of EN50121-3-2: 2006 indicates the assumption can be applied only in the frequency range 80MHz to 1000MHz, meaning immunity cannot be assumed over 1000MHz. There is no requirement to perform immunity testing once the entire vehicle is assembled. There does not appear to be any formal requirement in EN50121-3-1 for immunity testing of traction motors or auxiliary motors and their associated circuitry. In particular, EN50121-3-1 references EN50121-3- 2, where there is no explicit requirement for a test involving a temporary interruption of the electric traction supply from, for example, a 25kV overhead power line. This has the potential to switch the motor into generator mode, which might inject high voltage onto, for example, a DC bus and stress semi-conductors with reverse voltages. This oversight is reinforced by Annex A of EN50121-3-2, which explicitly considers traction motors and auxiliary motors and simply states that there are “no test requirements” for immunity (or emissions).

53

3.5.6 Transients

Rare voltage peaks, with levels much higher than the average signal amplitudes, do occur in railway systems. Long cables can act as antennas and receive electromagnetic pulses from various sources. Thorough shielding and grounding can minimise the impact or damage caused by transient voltage peaks, but transients can still cause interruptions to services, black out video screens, give false alarms for hot bearings, etc. Transients of high magnitudes with very long time intervals have been reported. Transients may occur once an hour, once a week and, in some cases, once a month. Therefore, special measuring methods should be proposed and tested for future EMC standards.

3.6 EN 50121 –3-2: 2006: Rolling stock – apparatus

The main problem for a train manufacturer in this section of the standard is that its suppliers normally try to do only what is mandatory according to this standard. However, it may not be enough to ensure the compatibility of the vehicle.

3.6.1 Emissions

EN 50121-3-2 does not establish any requirements within the frequency band between 9kHz and 150kHz, either for conducted or radiated emissions. However, EN50121-3-1 considers this in the context of the whole vehicle by stating “emission requirements shall be specified according to the type of signalling and communication systems used” (see EN50238.) If the train manufacturer finds an EMC issue in the radiated emission test in this frequency band, it could be due to the emission of one of the pieces of equipment installed on-board. It is always more expensive and less efficient to solve problems in the late stages of the projects, making it sensible to reduce emissions in the frequency range below 150kHz before testing the whole vehicle. However, there might be problems determining the responsibility for this EMC problem, as the on- board equipment supplier will protest that its equipment fulfils the requirements of EN 50121-3-2 and, therefore, it is the train manufacturer’s responsibility. This, of course, will delay the solution of the problem.

54

The traction converters and auxiliary converters over 50 kV need not be tested individually but when the vehicle is tested as a whole in accordance with EN 50121-3-1. The problem here is that this type of equipment is deemed to be the main source of radiated emissions from the complete vehicle. Again, it will be more expensive and less efficient to solve problems in the late stages of the projects, so this must be tackled early in the design process. One of the objectives of EN 50121-3-1 is to separate the effects of the infrastructure when testing the new vehicles. The test campaign for the on-board equipment is the best chance to define repeatable conditions and eliminate any effect of the infrastructure. There must be no exceptions to the tests performed according to the standard EN 50121-3- 2, as this is the best scenario to ensure repeatability and independence from other external factors.

3.6.2 Immunity

No immunity requirement is given for rolling stock in the frequency range below 150kHz. This is a surprising anomaly, given that a train unit would normally be expected to be able to deal with high magnetic fields in this frequency range. In addition, there are no emission limits below 150kHz. At the very least, regulations should specify immunity at these frequencies to ensure EMC, even if it is still difficult in many cases to measure/limit emissions at these frequencies. Particular attention should be paid to sensitive areas such as 50Hz and 16 2/3 Hz as well as other frequencies of relevance to traction motors. Note: most equipment on board the train will fall within the 3m zone as in EN50121-4:2006. As such, special consideration may be appropriate through application of the relevant magnetic immunity tests given in EN50121-4:2006. This may give a base from which limits can be obtained.

3.6.3 On-board Spot Signalling Systems

For on-board spot signalling systems, EN50121-4 seems to be the standard to follow, as it is the most applicable to signalling and telecommunication (S&T) apparatus installed in the railway environment.

55

As the S&T equipment is physically on-board, however, it has to meet the EN50121-3-2 standard and, as such, should be tested to that standard. This means the main standard and testing procedure for on-board spot signalling systems is found in the EN50121-3-2 document. The application of tests depend on the particular apparatus, its configuration, its ports, its technology and its operating conditions. If a port is intended to transmit or receive for the purpose of radio communication, the emission and immunity limits in this standard at the communication frequency do not apply, i.e. the in-band frequencies. The disturbances radiated to the whole system are sufficient to distort the receive communication path of the spot signalling system are the RF electromagnetic fields. Following the generic EMC EN61000-4-3 standard (Testing and measurement techniques - Radiated, radio- frequency, electromagnetic field immunity test), an 80% AM (1KHz) signal from 80 MHz to 1000 MHz is radiated with an effective value of 20 V/m. It should be pointed out that no frequencies below 80 MHz are considered for this test. Annex B of EN50121-3-1 analyses conducted interference generated by power converters from 9 KHz to 30 MHz. This informative annex does not apply to the spot signalling system, and it is not considered. Unfortunately, the positioning of the antenna on the train body close to the motor cables could cause interference. Therefore, the antenna positioning for the spot signalling system should be considered for this frequency range (it also contains the working frequency of the Eurobalise on-board transmission equipment signalling system). This may require an additional section in the apparatus immunity section of EN50121-3-2. Another conclusion of the analysis of the current approval test norms is that the transients are not considered in EN50121-3-2 or EN50121-3-1; in the latter, the complete vehicle is considered as a whole. First, the transient signal captured during the rolling stock EMC approval tests defined in this document are hidden by the max-hold function of the capturing during the train passing. Second, Subsection 6.1 explains the requirements for the compatibility of the rolling stock with the signalling and communication systems should take into account other kinds of interferences, such as transients due to discontinuities, etc. Unfortunately, no limits or test procedures are defined for these interferences.

56

3.7 EN 50121–4: 2006: emission and immunity of the signalling and telecommunications apparatus

This standard concerns the emission and immunity levels for signalling and telecommunications (S&T) apparatus in the railway environment, as defined in EN50121-1: 2006. The purpose of the standard is that S&T apparatus compliant with the standard will be conferred with a presumption of conformity to the essential requirements of EMC Directive 2004/108/EC. This intention is made clear, as both EN50121- 4: 2006 and EN50121-1: 2006 are listed in the Official Journal of the European Union (OJEU). The EMC Directive says any apparatus compliant with an OJEU listed standard is conferred with a presumption of conformity to the essential requirements of the EMC Directive, when that apparatus is installed in its intended environment. Another aspect of this section of the standard is that it is often used as the standard to test the trackside equipment not covered in parts not designated as signalling or telecommunications, such as equipment installed on the platform. Examples of equipment tested in this way are ticket machines, station lifts and other trackside electronic apparatus such as axle counters. In order to test this equipment, specific limits have to be incorporated into the 50121 standard, in an added section. It would be helpful to have these limits within the 50121 standard, even though this equipment has passed industrial standards, to avoid unnecessary gap analysis. Drawbacks to and limitations of EN50121-4: 2006 are summarised in its Sections 3.7.1 (emissions) and 3.7.2 (immunity).

3.7.1 Emissions

EN50121-4:2006 states: “Apparatus which complies with the emission levels of EN61000-6-4 will meet the emission requirements of this standard providing that emissions from any DC power port are within the emissions limits for AC power ports”. (Note: EN61000-6–4 is the generic industrial emissions standard, currently listed in the OJEU as EN61000-6- 4:2007+A1:2011). In other words, it is recognised that the railway environment is electromagnetically harsh. Apparatus conforming with the generic industrial emissions standard is likely to provide levels of emissions which are somewhat below those expected in the railway environment; thus, apparatus compliant with EN61000-6-4 can be presumed to be benign in the railway environment.

57

Unfortunately, the nature of most signalling systems (in the UK at least) is that their associated track circuitry operates at relatively low frequencies (less than 10kHz), outside the range of emission frequencies considered by EN61000-6-4 (the latter standard does not concern itself with emission frequencies below 150kHz). Although EN50121-2:2006 does require magnetic field measurements to be made from the railway as a whole down to a frequency of 9kHz, many common track circuits use frequencies that are even lower than this. There is, therefore, a significant region of the frequency spectrum up to 150kHz which is highly relevant to signalling and telecommunications apparatus, and for which there is no regulation on emissions within EN50121-4:2006. Note that this spectrum of frequencies encompasses base band audio frequency telecommunications, including base band voice transmissions on telephone lines. Significant audio frequency (psophometric) noise emitted by rolling stock and its apparatus has the potential to render voice communications unintelligible. This oversight in the EN50121-4:2006 standard is reinforced by the statement: “No measurements need to be performed at frequencies where no requirement is specified.” In view of the preceding paragraphs, this statement is clearly reinforcing the poor practice of ignoring the potential interference of S&T apparatus to other existing S&T apparatus, the very type of apparatus for which the standard was written. With regard to intentionally radiating apparatus in the railway environment (e.g. a transponder system), EN50121-4:2006 states: “The emission and immunity limits in this standard at the communication frequency do not apply”. This statement does not take cognisance of the fact that an intentional transmitter sited too closely to other S&T or other generic industrial apparatus may cause interference if the impinging level of irradiation is above that to which the victim apparatus has been tested. It would be prudent to address this problem with a supplementary statement in the standard regarding the necessity to maintain an adequate protection distance from other victim apparatus, according to the immunity level of the victim apparatus at the transmission frequency. This could be easily done by referring to Annex E of EN61000-4-3:2006, where an explicit method is provided to establish an appropriate protection distance.

58

3.7.2 Immunity

EN50121-4:2006 states: ‘The immunity levels of EN61000-6-2 will also be adequate except for the special case of apparatus as defined in note 1. Note 1 appears beside a number of immunity tests and says: ‘The tests given apply to apparatus inside (sic) 3m - zone and vital equipment such as interlocking or command and control which are mounted in areas where a high risk of interference from mobile radio telephones has been identified”. This wording concurs with the zoning scheme adopted by a number of UK railway projects (e.g. Crossrail, Thameslink Programme), whereby the first 3m from the centre line of an electrified railway are considered to be particularly electromagnetically harsh. At distances beyond 3m, the level of immunity required for apparatus is identical to that of the generic industrial immunity standard EN61000-6-2:2005. EN50121-4:2006 makes special provisions for safety critical apparatus through application of additional radiated immunity tests at VHF frequencies between 800MHz and 2.5GHz, encompassing hand-held mobile phone frequencies. The immunity levels required by EN50121- 4:2006 in this frequency range are more stringent than those required by the generic industrial standard EN61000-6-2, resulting in the need for a gap analysis. Section 6.2 of EN50121-4:2006 states that “voltages induced by traction currents are not treated here. They have to be covered by the functional specification”. It may be prudent to pre-empt mitigation of EMC problems for S&T apparatus connected to significant lengths of metallic cabling running parallel to electrified rails, by requiring conducted immunity testing in the frequency range 0-150kHz as covered in EN61000-4-16. For example, there is already such a requirement in London Underground standard G-222, Manual of EMC best practice. Note: radiated immunity at 4G mobile phone frequencies (2.6GHz band) are not considered by EN50121-4:2006. The standard only requires testing up to 2.5GHz.

59

3.8 EN 50121–5: Railway applications - Electromagnetic compatibility - Part 5: Emission and immunity of fixed power supply installations and apparatus

This standard concerns the emission and immunity requirements for fixed power supply installations and apparatus within the railway environment, as defined in EN50121-1: 2006. The intention and purpose of the standard is that power supply installations and apparatus compliant with the standard will be conferred with a presumption of conformity to the essential requirements of the EMC Directive 2004/108/EC. Drawbacks and limitations to the EN50121-5: 2006 standard are summarised in Sections 3.8.1 (emissions) and 3.8.2 (immunity).

3.8.1 Emissions

For emissions from the substation to the outside world (section 5.1), there is a significant region of the frequency spectrum up to 9kHz which is highly relevant to signalling and telecommunications apparatus, and for which there is no regulation on emissions in EN50121-5:2006. Note that this spectrum of frequencies encompasses base band audio frequency telecommunications, including base band voice transmissions on telephone lines. There should be a requirement during testing to measure the magnetic emissions below 9kHz (e.g. down to 5Hz) and DC track circuits (in some counties like Sweden), even if no explicit limits for these emissions are given. This will assist in the necessary studies to establish that EMC will exist between local signalling systems and the substation. The measurements will also assist in any necessary psophometric noise studies to establish that induced psophometric noise into telephone wires will be below the ITU-T recommended level of 1mVp. It is worth noting that most psophometric noise will be excited by the rolling stock, with negligible noise from the substation. For intentional radio transmitters, EN50121-5: 2006 states, “The limits in this standard do not apply to intentional communication signals”. This does not take cognisance of the fact that an intentional transmitter sited too closely to a power supply or other apparatus may cause interference if the impinging level of irradiation is above that for which the victim apparatus has been designed and tested. It would be wise to address this problem with a supplementary statement in the standard on the necessity to maintain an adequate protection distance from other victim apparatus,

60

according to its immunity level at the transmission frequency. This could be easily done by referring to Annex E of EN61000-4-3: 2006, which provides a method to establish an appropriate protection distance. Emissions from substations are also within the scope of EN50121-2: 2006 using a measurement distance of 10m.

3.8.2 Immunity

Section 6 of EN50121-5: 2006 states “voltages induced by traction currents are not treated here. They have to be covered by the functional specification. It may be prudent to pre-empt mitigation against EMC problems for power supply related apparatus connected to significant lengths of metallic cabling running parallel to electrified rails, by requiring conducted immunity testing in the frequency range 0-150kHz as covered in EN61000-4-16. For example, there is already such a requirement in London Underground standard G-222, Manual of EMC best practice”. Nor are radiated immunity at 4G mobile phone frequencies (2.6GHz band) considered by EN50121-5: 2006; the standard only requires testing up to 2.5GHz.

3.8.3 On-site testing

On-site testing is covered in Part 2 (Emissions of the Whole Railway System to the Outside World) and Part 3.1 (Rolling Stock – Train and Complete Vehicle). Testing to Part 3.2 is carried out in a laboratory. On-site testing is usually performed: - To verify the EMC Management plan for a system has been adhered to, and EMC has been achieved between the whole railway and the outside world. - In the case of Part 3-1, to ensure rolling stock satisfies the prescribed emission limits to meet: interoperability requirements, safety requirements or CE marking requirements under the EMC Directive. This is needed for the rolling stock to be used on the railway system. The following points follow real on-site test measurement sessions. The 10m antenna distance is rarely practicable due to constraints on the test site, resulting in reliance on normalising measurements to 10m as cited in Part 2. Normally, measurements have to be taken at 3m. The far field approximation becomes unreasonable at lower frequencies for larger sized emitters at this distance. The suggestion that both sides of the train do not need to be

61

independently tested is often followed, not because it is stated in the standard, but because it is not possible for various reasons (including health and safety) to test both sides of the track. For a pantograph OHLE railway, the symmetrical aspect is important, but for a third rail arrangement with the contact off centre, it is not. Some of the requirements in the standard have to be ignored; an example is the requirement that other locomotives be at least 20km away. This is normally unrealistic while testing on a real railway unless operating under a track possession with a requirement that all vehicles within a prescribed area have their pantographs down, and this is often only possible during a night-time possession. The substation testing procedure is particularly long, especially if the substation is large, necessitating a measurement from each side and from each corner. Again this can be limited by the available space around the substation, which is often less than 10m. It can take up to three days to test a large substation. Often the customer will ask for more tests than called for in the standard. The most requested extra test is emission measurement at over the stated 1GHz limit. This is usually carried out to 3GHz to encompass the current spectrum usage. Immunity measurements on the actual train may prove too challenging to include in a measurement campaign or test method, as the normal way of testing immunity is not immediately obvious (no ports to induce transients on, ESD to train body does not make sense etc.). Immunity is conferred on rolling stock by meeting EN50121-3-2: 2006 and by maintaining the integrity of the immunity of the individual on-board systems through “best practice”. Whole vehicle immunity testing is not possible without access to an extremely large anechoic chamber (e.g. the BAE systems chamber at Warton used for Eurofighter/Typhoon aircraft) with rail access (and including a dynamometer); otherwise, radiated immunity testing would extensively disrupt broadcast services and the test set-up would represent an illegal generator. On-board immunity testing can be devised, for example, to test second-hand rail grinders imported for use on Network Rail, but must be used with caution. Deserving mention is the particularly harsh environment encountered on the roof of a train, particularly on an OHLE railway. Different immunity limits need to be applied for any equipment to be situated in this area. For the substation tests in EN50121-2, the test method given in clauses a) to e) in Annex A involves finding the frequency of the maximum emission in the range 0.12MHz to 2MHz and treating the quasi-peak measurement as the “maximum radio emission at a frequency in the neighbourhood of 1MHz” required by the standard. A similar routine can be carried out for the “maximum radio emission at a frequency in the neighbourhood of 350MHz” with the frequency band of 300MHz to 400MHz. In the case of 9kHz to 150kHz, one minute’s worth of measurements is taken with a peak detector then compared to the limit line. Any emissions above the line are treated with a quasi-peak measurement after discarding bona fide radio

62

transmissions. This is repeated for 150MHz to 30MHz, 30MHz to 300MHz and 300MHz to 1GHz. This procedure is more detailed than is given in the standard, and results in a more representative emission profile for substations.

3.9 EN 50122-1: Railway applications – Fixed installations. Part 1: protective provisions relating to electrical safety and earthing

This section details the drawbacks, provides missing information and suggests some improvements to the EN 50122 standard.

3.9.1 Scope

The EN 50122 standard primarily concerns provisions for electrical safety with respect to people rather than electronic equipment. The EN 50124:2005 standard concerns clearances and creep age distances for electrical and electronic equipment. The title of the standard specifies its scope as safety and earthing, but the document focuses on human safety and ignores earthing and protection of the equipment. The latter can cause malfunctions of the different apparatus that can be sensitive to EMI, even when they are of no harm to people.

3.9.2 Ground currents Annex C of EN 50122-1 provides a test for the voltage in the case of double track sections (see Figure 8), but this Annex is informative, not normative.

Figure 8. Potential decrease test for double track sections and average soil earthing resistance (EN50238 Railway applications – Compatibility between Rolling Stock and train detection systems).

63

This kind of test would prevent transients that could provoke a malfunction in equipment because of the surroundings. EN 50122-1 only references the double track section scenario; no reference is made to other scenarios, for example, single track, the most common situation in Sweden. The most suitable change to this standard would be revising the Annex to make it normative to help reduce transients by ground currents due to lack of isolation and take into account all the scenarios possible when performing the tests. Another point to mention is that maximum limits for ground current return are not specified; these would help diagnose malfunctions due to earthing that can trigger equipment failures or cause safety issues.

3.10 EN 50215: Railway applications – text of rolling stock on completion of construction and before entry into service

In this generalist standard, two sections are directly related to electromagnetic compatibility and issues with signalling systems: Sections 3.10.1 and 3.10.2. These points will be covered in this section.

3.10.1 Scope (related to EMC)

The first subsection of EN 50215 states that to verify that all equipment functions correctly after installation, without interference effects, the equipment should have passed the EN 50121-3-2 standard with a sufficient margin between the electromagnetic emissions and their immunity levels. This margin is not specified, however, and no measurement of the radiated emissions is required inside the vehicle. The second and third subsections explain that the external interferences produced by the vehicle, including radio frequency interferences (a safety related issue) must follow the EN 50121-3-1 standards, as expected. The fourth and fifth subsections deal with external interferences to the vehicle and a voluntary test of electrostatic discharge. Both address the standard EN50 121-3- 2 to show the immunity of the complete vehicle; again, problems may arise because no tests are performed inside the vehicle.

64

3.10.2 Supply issues and transients

The objective of this section is to verify voltage changes in the external supply do not adversely affect the performance of the vehicle from the perspective of the traction system of the train (EN 61377-1, -2 and -3). In addition, where applicable, the co-ordination of protective systems between the traction vehicle and the power supply infrastructure should be checked as per EN 50388. The four subsections of 3.10.2 specify the tests to be applied to the train: - Voltage jumps based on EN 50163. - Interruptions, with total time of interruptions from disconnection to reconnection of 10ms to 10s. - Voltage variations from a full range of operating point of view. - Short circuits with an unlimited time. None of these supply issues considers the transients that affect the signalling systems tackled by this project (magnitude, frequency and time duration). Once again, special measuring methods should be proposed and tested for future EMC standards.

3.11 EN 50238: Railway applications – Compatibility between rolling stock and train detection systems

This section details the drawbacks of, supplies missing information for, and suggests improvements to EN 50238. Further information will also be provided where the EMC issues related to track circuits are linked to the drawbacks of the current tests.

3.11.1 Scope

EN 50238 lacks a reference to all apparatus or equipment not directly involved with train detection (equipment of this nature would come under 50121-4), and there is no reference to specific differences in power supply systems in the section titled “Characteristics of traction power supplies”. An example of a piece of equipment not covered is a rail-switch detector. Some railway equipment (like detectors and other electronic apparatus) has the same specifications as train detection systems; therefore, it may be appropriate to include them in the scope of the various standards for train detection systems.

65

Some electronic equipment is based line-side, like the rail switches detectors that work on DC. This apparatus can be influenced by the low frequency emissions from the rolling stock, signalling systems, etc. and this should be checked. This apparatus should also pass the immunity requirements in EN 50121-4.

3.11.2 Neutral sections

The influence of the neutral sections is treated in EN 50388, but only for the relationship between the rolling stock and the power supply installations. This can influence the electronic equipment installed in the surroundings of the track. Annex A of EN 50238 provides various scenarios used to determine the susceptibility of train detection systems. Transitions between neutral and powered sections should be described in this section and in EN 50388. A way to avoid influences on signalling equipment located in the surroundings of neutral sections could be testing not only the behaviour of the train in the neutral section, but also the transition between sections, studying the peaks produced when the train enters and exits a neutral section and the influences of these maximum peaks on the electronic equipment in the surroundings.

3.12 EN 50388:2005 Railway applications – Power supply and rolling stock – Technical criteria for the coordination between power supply (substation) and rolling stock to achieve interoperability

This section details the drawbacks of, provides missing information for, and suggests improvements to the EN 50388 standard.

3.12.1 Application procedure

EN 50388 has a diagram showing the process to follow when introducing a new element to the railway network. The standard describes the acceptance procedure, compatibility study, tests and test methodology. This may be more suitable for inclusion in the standard EN 50121-1 or in the EN 50388, as a procedure to follow when introducing new electronic equipment to the infrastructure network.

66

3.12.2 Neutral sections

EN 50388 specifies that a train should bring the power consumption to zero when entering the phase separation section (paragraph 5.1; EN 50388). When entering the powered section, a voltage peak can be produced. This may produce electromagnetic emissions that will influence the power supply installations as well as any other electronic equipment (signalling, telecommunication, detectors, etc.). A way to avoid influences on signalling equipment located in the surroundings of neutral sections could be testing not only the behaviour of the train in the neutral section, but also the transition between sections, studying the peaks produced when the train enters and exits a neutral section and the influences of maximum peaks on the electronic equipment in the surroundings. Annex A of EN 50238 shows the different scenarios used to determine the susceptibility of train detection systems. Transitions between neutral and powered sections should be described in this section and in EN 50388.

67

68

4 TRACK CIRCUITS

4.1 Introduction

A track circuit is an electrical system used to detect the location of a train or other obstacle on the rail track. Due to its failure protection mode, it cannot absolutely prove the presence of a train, but when a track circuit section is marked as clear track, a safety route can be set through this section (General Information on Track Circuits, 1998). There are different kinds of track circuits on the market, depending on the return current, frequency, etc. They include DC, AC, high frequency, audio- frequency etc. We develop DC models as they were analysed in the Swedish measurement campaign.

4.1.1 Origin

The first prototype was based on an open circuit system. Several circuit- instruments in close proximity to the railroad were activated by the wheels of the rolling stock. Through a relay action, the circuit was closed when the wheels activated a lever at one point of the track. Magnet relay attracted its armature and kept its own circuit closed. The action of opening or closing the circuit was controlled by the magnet relay. When the rolling stock passed a certain point on the track, this activated another lever, which, in turn, activated a reverse signal, opening the relay circuit by cutting off the battery. Such a system is extremely limited and may show a safety signal when a danger exists. One problem is caused by the detachment of one part of the train. When the rolling stock enters a new track section, it activates the track occupied signal. If the train stops between two sections, something quite likely to happen in sharp curves and grades, the forward section activates the clear track signal in the previous one, even though it is still occupied. Another problem occurs when a train enters a section from the opposite end or from a siding, thus blocking the track, but the track is shown as clear. A final problem is a false positive signal that occurs when the relay battery fails for any reason. If a line wire breaks or there is a break in a connection, the

69

signal will invariably show a clear track. The problems listed above are linked to open circuits. Therefore, changing from an open circuit system to a closed circuit system is necessary. The Robinson closed rail circuit forms the basis of every efficient automatic electrical system and is illustrated in its simplest form in the next figure.

Figure 9. Closed track circuit scheme (American Railway Association, 1922). Figure 9 shows the railroad divided into sections. The rails in each section are insulated from adjacent sections. A light battery has its terminals connected to the opposite rails at one end of the section. A magnet relay is placed at the opposite end of the section. The current passes through the whole length of the circuit, keeping the relay on the continuous closed circuit magnetized as its normal condition. The relay keeps the secondary circuit normal closed. This circuit controls the signal which is normally held in a safety position. The discontinuity in the rails means each section of railway track is electrically defined by the provision of insulated rail joints (American Railway Association, 1922).

4.1.2 Common components and behaviour

A section of a railway track is electrically limited by insulating joints in the rails at either ends. A general track circuit section includes: a source of electrical energy, connected across the rails via series impedances, and a detector. These are placed at opposite ends of the rails. The detector senses the transmitted electrical energy when no train is within its boundaries, energizing the repeater circuit. Otherwise, a train will cause the rails to be short-circuited; the detector no longer sees enough electrical energy and the state switches to track circuit occupied. A leakage of electrical energy through the relay, for any reason, will cause the track circuit to fail and change its state. As discussed above, such a circuit configuration requires a high degree of failsafe measures. At the same time, a good electrical contact between the wheel set of the train and the rails upon which they run is key, as is a low impedance path between the steel tires of each wheel, produced via the connecting axle. Both requirements will be

70

analysed later (see train shunt imperfection). When a train enters a section, the wheels and axles short out the current from the relay. An obstacle can also produce a short circuit between the rails. When no obstacle is present, for security reasons, the relay is energised by the current flowing from the power source through the rails; if the track is considered occupied, the current to the track relay coil drops, and it is de-energised. The failsafe principles dictate that the relay interprets the presence of the signal as unoccupied track, whereas a lack of a signal indicates the presence of a train. Circuits through the relay contacts therefore report whether or not the track is occupied. Next, the relay releases its armature, opening the signal circuit. The signal immediately turns to the danger position. When no obstacle is present, the relay is energized by the current flowing from the power source through the rails. When an obstacle produces a short circuit between the rails, the current to the track relay coil drops, and it is de- energised. Circuits through the relay contacts therefore report whether or not the track is occupied. The failsafe principle dictates that the relay interprets the presence of the signal as unoccupied track, whereas a lack of a signal indicates the presence of a train. This basic principle governing the operation of the system is performed in a variety of ways depending on the track circuit type. There are several kinds of sources which imply different types or receivers. For instance, an electrical DC energy source needs a simple relay; an electrical AC energy source needs a more sophisticated AC vane relay; a receiver tuned to a particular frequency is used with AC source at power frequencies, etc.

4.1.3 Types

Track circuits apply this basic principle in a variety of ways for various reasons. A short classification of them is listed below. - We can have AC or DC sources; in the latter group, we can find different work frequencies (50 Hz or 831/3 Hz). - The type of functioning, as mentioned, can refer to an open or closed track circuit. - The traction return current can be a single or double rail track circuit. - The type of regulating device (only in AC track circuits) can be: resistance, reactance or condenser fed track circuit. - The type of wiring (only in case of point & crossing tracks) can be series or multiple types.

71

- The type of Independence Bond (only with AC double rail track circuits) can be resonated or auto coupled track circuit. - When immunity from the DC current is required, 50 Hz AC track circuit can be used. Other operating frequencies at 75 Hz and 831/3 Hz can be used by the track circuit in the case of immunity from both DC and 50 Hz AC. - Finally, 50 Hz AC track circuits can be used in both single and double rail configurations, depending on the track layout and traction return arrangements. The principle feature of the 50Hz AC track circuit is the double element vane relay. This relay is composed by a thin aluminium vane which is pivoted on an axle and able to rotate. Buffer stops are provided to limit the motion in either direction. A mechanical linkage enables the electrical contacts to operate in sympathy with the vane rotation. The counterbalance action ensures that the vane and the contacts assume the track circuit in occupied position in the absence of any torque on the vane. The interaction of the two coils, whose AC magnetic fields induce eddy current in the vane, produce enough torque to rotate the vane to the energized position. There are two types of coil: the local coil which is permanently connected to a 110 V AC supply and the control coil which receives its energy from the track circuit itself. Eddy currents in the vane are continuously induced by the local coil, but no mechanical torque results if the control coil remains without energy. The counterbalance action of the vane ensures that the relay remains on its backstop. A steady mechanical torque is applied to the vane, if the frequency is identical to that on the local coil and there is a suitable phase difference between the two, when the second AC voltage is applied to the control coil by the track circuit Consequently the torque gets its maximum value when there is a phase difference of 90° between the two voltages and falls to its minimum, equal to zero, when the voltages are in-phase or anti-phase, i.e. 0° or 180°, respectively. The direction of the torque’s motion depends upon whether the control voltage lags or leads the local voltage. In real life, it is not possible to achieve 90° phase angle due to leakage of current through the ballast and the inductance of trails. To mitigate this leakage, an internal tuning capacitor is connected across the control coil, or across a third (tertiary) winding to partially compensate for this effect. The value of the capacitor must be appropriate to achieve parallel resonance with the control or tertiary coil inductance at the track circuit operating frequency. A phase angle of 70°-75° is more achievable, what means that a 25% above nominal control voltage must be applied. The maximum applied voltage must not exceed 60% above the nominal value. In a single rail configuration, a feed transformer isolates the track circuit and provides an initial voltage step up and adjustment. The supply is fed to the rails via a traction fuse and an adjustable feed capacitor. The function of this capacitor is to ensure that the control current in the rails leads the 110 V local supply voltage by approximately 90°. The voltage from the rails is out of phase

72

with the local supply voltage applied to the local coil by approximately 90o. This effect is due to the voltage from the rails is fed to the control coil via another traction fuse and therefore, torque is created in the vane and the relay energizes if the control coil voltage is high enough. An important thing to be considered is that the connections to the rails at both the feed and relay end must be in the same phase relationship. The connections may be reversed to fix the electrical stagger of the track circuit. An important point to consider is that the reversal of connections must be carried out at both ends of the track circuit in order to maintain the phase relationship between the rail voltage and the supply to the local coil of the relay. The wiring should be reversed in the secondary of the transformer and at the control coil of the relay. If a reversal at only one end of the rail is done it will cause the relay vane to attempt rotation in the wrong direction, forcing it against its back stop. The double rail configuration is quite similar to the single rail one. The main difference is that the rails are isolated with joint bars and there is no common rail. Both configurations are arranged so that trains usually run on to the relay end first. On bi-directional lines, any predominance of direction should be taken into account.

4.1.4 Operation and adjustment of a simple Track Circuit

The following DC track circuit scheme can help give a better understanding of the track circuit operation.

Figure 10. Track circuit layout including Ballast Resistance and Train Shunt. As mentioned before there are two states: track circuit clear and track circuit occupied. To explain these two situations more thoroughly, we use the above configuration. The real voltage source (including the feed resistance and the cable resistance) is placed at the left end of the track circuit, limited by insulated rail joints. The relay is at the opposite end. Ballast resistance and train shunt are modelled with real impedances. The ballast resistance forms an additional load in parallel with the relay. Because the ballast resistance falls in wet weather, the current drawn from the feed rises. This situation will augment the voltage across the feed resistor, decreasing the

73

rail and relay voltages. A further reduction of the relay voltage to below the relay changing state value will indicate the track is occupied without the presence of a train. A measure to increase the current fed into the rails and raise the voltage in the rail and relay is reducing the value of feed resistance. It should be noted that the end length has significance. A long feed end produces an immovable resistive value that has to be added to the feed resistance, thereby reducing the effectiveness of the adjustable feed resistance. However, a long relay end reduces the ratio of relay voltage to rail voltages by potential divider action. Consequently, the track circuit could indicate the track is occupied at a higher ballast resistance. It therefore imposes a shorter maximum workable length. A short circuit current will flow from the feed end of the rolling stock, and this closes the circuit. The train shunt resistance is in parallel with the ballast resistance. The relay will operate at particular values of combined ballast and train shunt resistances, regardless of the value of the feed resistance. Because both resistances have a direct interaction with the relay, a higher ballast resistance will require a lower value of train shunt resistance to operate the relay within its banks. Thus, a higher value of the train shunt will require a lower value of the ballast resistance. As previously mentioned, the track relay is cut off by short circuits rather than disconnections. Therefore, the drop-away time of the relay is increased because the inductive circuit prolongs the decay of the coil current.

4.2 CASE STUDY: Swedish DC track circuits

In this licentiate the case study is a Swedish DC track circuit section. It was the location chosen to perform the first Swedish measurement campaign to obtain all the values needed to reproduce different simulation scenarios. Each track circuit detects a defined section of track, such as a block. In terms of boundary conditions, track sections are separated by means of an 6 mm thick, rubber, insulated joint in the I-Rail. Close to each insulated joint, there is a cabin with all the components required to feed, monitor and control the track circuit section. The other rail, denoted the S-rail, is unbroken and carries the return current from the engine, back to the feeding transformer. This allows the negative traction return to the propulsion system to remain intact while providing individual track circuits separated by insulated joints. The lengths of the I-rail sections vary from 200 m to 2500 m. The shortest sections are located at railway stations. Normally, in a station area,

74

there is one I-rail-section between two adjacent switches. Each one of the I-rail-sections is connected to a separate voltage source delivering 7 V DC to one of the ends. At both ends of each I-rail-section there is also a relay, which is pulled when the DC voltage is above 1.7 V, and not pulled when the voltage is below 1.7 V. These relays give information to the automatic train control, and to the train control centre; they also control the red and green signals. When a train enters a certain section of the I-rail, the voltage between the S-rail and the I-rail drops from 7 V to zero, because the wheels and the axles make a short circuit between the rails. This makes the relays change position and green lights are switched to red. The automatic train control, ATC, system then makes it impossible for another train to enter the section. In case the 7 V-voltage-source should fail, the voltage of that section also drops to zero, and the relays switch the signals to red, etc. The limits for immunity against failures of the track system should be defined by voltage levels of about 1.7 V DC. The safety checking routines of the Swedish railways are based on a voltage interval around 1.7 V, separating high voltage from low voltage. The schema of the DC track circuit with the framework deployed in Sweden is shown in Figure 11.

R1

R2 R2 15Ω 15Ω

Figure 11. DC track circuit schema.

4.2.1 Safety in the Swedish DC track circuit

As a crucial security part of the railway infrastructure, the DC track circuit design includes many safety actions: - Relay operating follows the failsafe principle functioning. If the current through the relay drops down for any reason, the system will show an emergency signal.

75

- Relays and power supply are attached to opposite ends to ensure the whole track circuit section is checked; this allows the system to detect rail breaks. - For lengths up to 1800 metres, most of the Swedish DC TCs consist of two relays placed at the ends of the track circuit section with the power source in one end. Having two relays is a safety method providing a double check against failure. If the relays are contradictory, a danger signal will be activated and sent to the TCC (Train Control Central) to activate an emergency brake call. - This power supply in most stations is a rectifier with a DC battery as backup. Furthermore, circuits are commonly battery-powered at low voltages to protect against line power failures. - In the event of insulation failure between track circuit sections, one circuit could falsely power the next one. Reversing the electrical polarity from section to section, in the case of a short-circuit of the IRJ the consecutive relays will show an obstacle on the railroads.

4.2.2 Main components of the Swedish DC track circuit

Battery The battery gives a 6 DC V power supply to feed the whole track circuit section.

Figure 12. Track circuit battery.

Relays As explained in Section 5.2, for security reasons, there are two relays per track circuit section, one in each end. They receive the input of the circuit and indicate whether the track section is occupied or not. The relay is only activated if the current through the coil is in one direction. The model used in our case study is the JRK 10470, a measuring polarized relay.

76

Figure 13. JRK 10470 relay. Relays are located in the cable between S-rail and the 15 ohms resistance and choke. Its characteristics are: Vnom = 1.4 V, Von = 1.5 V, Voff = 0.70 V and R (XL) = 30 Ω.

Figure 14. Characteristics of JRK 10470 relay. Chokes There are 3 chokes per track circuit section, one in serial with the battery and a 0-6 ohms resistance and one in serial with each relay and a resistance 0-15 Ω. All are represented as Dr, or drossel, a Swedish word for high inductance. They are made up of two coils of insulated wire of 2.4 H each to protect the DC equipment against AC currents from the track and train as transients, conducted and radiated emissions, or external, such as lightning.

77

These two coils can be connected in series or parallel depending on the configuration. For the relays, the coils are connected in series and for the power supply the coils are in parallel. Their characteristics, depending if the coils’ configuration is in series or in parallel, are respectively: Z = 500 ohms, R = 2.6 ohms, V(AC)max = 100 V, Freq = 16.67 Hz and I(DC)max = 0.25 A Z = 125 ohms; R = 0.65 ohms; V(AC)max = 50 V, Freq = 16.67 Hz and I(DC)max = 0.50 A Figure 15. Choke.

Resistances Two resistances, R1 and R2, must be calibrated to ensure the correct behaviour in the track circuit section. R1: A 0-6 ohms resistance in series with the battery to limit the current when the track circuit is short-circuited, saving battery power. R2: A 0-15 ohms resistance in series with each relay to ensure that impedance value of the relay circuit is higher than the impedance value of the train circuit. R: A 15 ohms resistance in parallel with the relay. As seen in Figure 16, mechanical movable pieces can be screwed in to create the corresponding resistance.

Figure 16. Resistance.

Resistance was introduced in the simulation (at the beginning and the end of the rails in the track circuit section limits). To model the scenario of a short-circuit between two I-rail track sections, we introduced a variable resistance between them as a broken or shorted insulated joint.

78

Cables Cables connect the battery, relays, resistances, chokes and rails. The cable used for the catenary is 100% copper, but it was necessary to define two other cables, the BV RQ 1x10 mm2 and the BV RQ 2x10 mm2, made of copper and protected by black polyethylene. The cable goes under the rail parallel to the sleeper, in a corrugated tube for protection. Both cables that go to the relay/power source from the 2 rails are in the same corrugated tube. After passing the rails, the cables go underground (below the ballast) to the relay / power source. The depth is approximately 600 mm; it does not go straight, from the rail to the cabin, as shown in Figure 17.

Figure 17. Cable route. Cable connections There are two kinds of connections depending on where the connection is located, in the rail (Figure 18) or in the cabin (Figure 19). Both can be considered ideal in the model. The cables are connected to the rail in two positions as a security measure; the 2 connections make the same cable.

79

Figure 18: Connections to the rail.

Figure 19. Connections inside the cabin.

Rails The S-Rail is the rail through which the return current circulates; the length of the I-Rail is approximately 1000 m. The rails are the interface between the train and the track circuits. In our case study, the standard of the profile of our rails is U.I.C. 60 with a conductivity of 4.4e+06 S/m. The measures are detailed in Figure 20. It can be added that the distance between the middle point of the rails is 1507 mm (1435 mm. between rails + 72 mm. width) and from the middle point of the pole to the middle point of the rails is 3000 mm.

80

Figure 20. Rail U.I.C. 60 profile.

Electrical insulated joint These joints are at the limit of the I-rail between track circuit sections. Their function is to isolate both track circuit sections to let different voltages in the rails detect obstacles in the track. They are 6 mm thick and made of rubber in our case study. Figure 21. insulated joint.

81

4.3 Most common failures in track circuits in the Swedish railway infrastructure

The first step to determine the worst case scenarios is to identify the main problem faced (Deliverable 8.1 TREND project, 2014). Since it is hard to define a global immunity level, specially an accurate limit covering all cases of possible transients, for the whole system, it is easier to divide the system into subsystems and study them separately. To study the robustness of the Swedish DC track circuits, the criterion is the source of the problem (0FELIA, 2013) the library where all failures in the Swedish railway system are reported. The resulting three most common worst case scenarios are: - Short-circuit between rails: Low ballast resistance, failures in crossings, switches, obstacles, etc.

- External interference: In the extreme case, a lightning strike.

- Short-circuit between track circuit sections: Failures in the insulated rail joint.

4.3.1 Scenario 1 – Low Resistance between rails

Due to the crucial need for safety in the railway system, in one of the most commonly reported failures, the train detection system warns that the track is occupied when there is no train. This is often caused by unwanted grounding of the I-rail, making the voltage drop below 1.7 V, without the presence of a train, mainly at railway stations. The opposite failure is extremely dangerous, but seldom occurs. In 30 years, only one case of an undetected train has been reported. The signals were indicating “No train on the section” (voltage between the rails above 1.7 V DC), but there was a train standing at the station. The case occurred at Boden Central station and was due to massive snowfall. All the wheels of the train were floating on an isolating layer of snow. This was a very rare case, and has only been reported once. From the EMC point of view, it is important to follow the common rules of EMC and make proper groundings. The cabinets in the technical houses must be designed for EMC and components and apparatus must be mounted according to EMC rules (Niska, 2009). Track circuits restrictions, causes of leakage and other details are explained in PAPER C.

82

Ballast resistance The resistance between the two rails of a track circuit is called ballast resistance. This resistance comprises leakage between the rail fixings, sleepers and earth; its value is dependent upon several factors, including insulation conditions, cleanliness of the ballast and weather conditions, among others (Patra, A. P. et al., 2009). The ballast resistance values vary between 2 and 10 Ω/km. These values are inversely proportional to track circuit length, although lower values can be reached in some special situations as in wet conditions with bad drainage or contamination with conductive materials. Therefore, a reliable track circuit must accept a wide range of ballast resistance variation. In a simpler explanation, the ballast resistance can be considered as a single resistance connected between the rails. This representation is useful in explaining the simple behaviour of DC track circuits. However, it is important to understand that the limitations of the model may not explain many of the more complex phenomena demonstrated by track circuits. A more complex and accurate model represents the ballast resistance as a series of resistances between earth and each rail. There is a further component of resistance between the rails independent of earth, but it is high compared to the rail-earth resistance and is not taken into account in the calculations. An obstacle across the track connecting the two rails is rare and easily detected; this event can be caused by flooding, massive snowfall or train freight drops. The minimum resistance value of the ballast is discussed in PAPER C. The term ballast denotes the layer of crushed stone (and, in exceptional cases, of gravel) on which the sleepers rest. The ballast fills the space between sleepers as well as at some distance (called ballast shoulder) beyond the sleeper ends. The railway ballast performs several functions: - Further distributing stresses transmitted by the sleepers. - Attenuating the greatest part of train vibrations. - Resisting track shifting (transverse and longitudinal). - Facilitating rainwater drainage, allowing track geometry to be restored and correcting track defects. The above functions are clearly contradictory in some aspects; thus, the ballast cannot completely fulfil all of them. For good load bearing

83

characteristics and added track stability, the ballast needs to be well graded and compact; but this makes dispersal of water and associated maintenance more difficult. A balance among the various functions ballast is required to perform is aimed at, and this depends on the material. For example, in the Malmbanan line case, north of Sweden, the thickness of the ballast is upgraded to 60 cm to handle the 30 tonne loads; two types of ballast material are used: - Luleå – Gällivare (206 km): Gabbro. - Gällivare – Kiruna (100 km): Porphyry. - Kiruna – Riksgränsen (130 km): Porphyry. The tests for low resistance between rails are performed in the track, but the data come from the track circuits cabin (Fellman, 2004). Rail impedance The DC resistance of a rail forms the real part of the rail impedance and is around 0.035 Ω/km. This value could increase up to 0.25 Ω/km if the insulated joints are of galvanized iron. The inductance of rail can raise the overall impedance per rail from approximately 0.3 Ω/km (50 Hz) to 10 Ω/km (2 kHz). Although in DC track circuits, the rail inductance does not have an effect, it has to be considered in AC track circuits. Furthermore, rail inductance in audio frequency track circuits produces a steep decline in rail voltage as distance from the transmitter increases. However, it is of little consequence when considering the operation of AC power frequency track circuits, where the rail voltage can be expected to decline very little between the feed and relay ends. All such alterations compromise the length of the rails. The workable length of a track circuit is influenced by: the declining value of ballast resistance, the increasing value of rail impedance and the electrical requirements. Train Shunt imperfections The resistive value of the train shunt determines the energy experienced by the relay; in fact, it can produce an undesirable switch in the track circuit state. The energy seen by the relay will be null only in the case that the train shunt is zero. Whilst the resistance of an axle and wheels is virtually zero, there are a number of factors that can increase the effective train shunt value. The factors applied to a particular vehicle in a particular place can be very variable, since some factors affect the track, whilst others are vehicle specific (Zerbst, U. et al., 2012).

84

Rust films on the rail head or tyre tend to act as a semi-conductor, thus implying a high resistance until the voltage exceeds a particular threshold value and breaks down completely. The breakdown voltage is directly proportional to the level of contamination. The biggest problem caused by heavy rust films is the inability to detect trains, something usual with prolonged disuse. The mechanical strength of rust films is considerably reduced by the presence of humidity. Therefore, lightly rusted rails will only be a problem when dry (Zerbst, U. et al., 2012). This problem is most severe when showery weather is accompanied by dry wind or in prolonged periods without trains. Leaf residue in areas where the extent of line side afforestation is significant or problems with the coal dust on the rail head tend to be confined to colliery areas, but these have the same repercussions as heavy rust films. In addition, sand contamination is associated with slow moving locomotives. With the intention of improving brake performance, certain types of rolling stock are fitted with a composite type of reading brake block. This implies the appearance of a contaminant layer on the traditional steel tyre which tends to insulate the train from the rails. In the context of braking, treads and disc brakes can be analysed. Disc brakes have a thin film of insulation between two pieces of metal. Given the electrical contact with these types of brakes, we can expect that the surface roughness of the metal will permit high spots to penetrate the film, producing an ineffective insulation. Train shunting is improved when the vehicles are travelling on curved tracks, since the vehicle guidance force and wheel rail slippage are increased.

4.3.2 Scenario 2 – External interference

The most harmful scenario related to the natural forces that can interfere with the railway system is lightning. A cloud-to-ground lightning strike can cause damage in two different ways: by a direct strike or by induction effects while striking somewhere nearby. The largest currents are produced by return strikes. The currents produced by these peak return strokes are significant where the struck object presents a resistive load. The typical value for the peak current is

85

about 30 kA and the wave is conducted to the object by other means, for instance, conducting systems and power lines. The most relevant properties of lightning that cause damage are the current peak and the maximum rate of current change. The maximum rate of current change appears on objects that present inductive impedance such as wires, earth leads etc. Assuming that 10% of the peak current value with 0.3 μs duration finds its way to the wiring of an electronic device, for an inductive load of 1 mH per metre, the inductive voltage produced in a 10 cm long wire could reach 1000 V. Indirect strikes can affect into the train detection system but obviously they are less severe than direct strikes. The severity depends on the stroke currents involved in the return stroke phase of the lightning flash. There are some differences between the first return stroke and the subsequent return strokes. The first return stroke is slow, with a low frequency, in the range of tens of kHz to few hundreds of kHz. Subsequent return strokes could be up to a few MHz. The train inrush current has to follow a specific limit so it does not disturb the TC. The inrush current of 45 Amp in 1.5 seconds or 25 Amp in 2.5 seconds must not be exceeded (Kraftförsörjningsanläggningar, 2013). This inrush current is mainly produced when the train is stopped and has to start moving. At this precise moment, the track circuit receives a peak of current and cannot detect any obstacle in the track, a well-known issue in other locations in Europe. A cloud-to-ground (CG) lightning can cause damage to an object on the ground directly by hitting it or indirectly by induction effects while striking somewhere near the object. The extent of the damage depends on the characteristics of the lightning and the characteristics of the object. The most important properties of lightning current that cause damage are peak current, maximum rate of change of current and the integral of the current over time. In a CG lightning strike, the largest currents are produced by the return strokes. Peak return stroke are important in cases where the struck object presents a resistive load, e.g. the surge impedance of a long power line. As an example, when a lightning return stroke with a peak current of 30 kA strikes a power line with a surge impedance of 400 Ω, it can produce an over voltage of 1200 kV. This large voltage can cause flashover across the insulator, from line to ground, to adjacent lines and to other objects nearby. The magnetic forces produced by the peak current can cause wires to be pulled out of the walls and electrical machines, and the metal tubes to be crushed. A 30 kA current entering earth through a grounding

86

impedance of 10 Ω causes a potential rise of 300 kV and also cause surface arcing. In objects that present inductive impedance, such as wires in electronic systems, earth leads etc., the maximum over voltage is proportional to the maximum rate of change of current. Maximum di/dt occurs at the return stroke current wave front. Assume 10% of the 30 kA peak current with front time 0.3 μs finds its way to the wiring of an electronic apparatus. For an inductance of 1mH per metre, the inductive voltage produced in a 10cm long wire is 1000 V, enough to destroy most electronics unless there is adequate protection. In negative return strokes, the average value of di/dt is 110 kA/μs. In positive return strokes, these values are much smaller. The most important field parameters are the peak electric field and the maximum time rate of change of electric or magnetic fields. Peak voltages on exposed metallic surfaces in lightning fields are proportional to peak electric fields, and peak voltages produced in a loop of wire are proportional to the rate of change of magnetic fields. For instance, a typical return stroke striking 100 m away may induce an over voltage in excess of 200 V/m2 of the loop area formed by the equipment and its cables, for a certain orientation of the loop. The degree of penetration of fields inside shielded enclosures through apertures is largely proportional to the rate of change of magnetic and electric fields. The magnitude of peak fields and the rate of change of fields are important parameters in over voltages caused in above ground wires and underground cables. The finite conductivity of the ground creates a horizontal component of an electric field on the surface of the earth. This component of the field is large if soil conductivity is low. This field is oriented radially from the lightning channel and induces over voltages in overhead lines and cables on the ground. The effect of this horizontal field may be seen as series voltage sources distributed along the conductors, each source turned on in sequence as the field sweeps along the conductor. These series voltage sources will drive a common-mode current in the conductors.

87

Figure 22. Current waveforms measured at tower bottom for a flash with two subsequent strokes by Eriksson (Nelson, 2005).

Indirect strikes are not severe when compared to direct strikes. However, severity depends on the stroke currents involved in the return stroke phase. The first return strokes and the subsequent return strokes have different properties. First return strokes are slow and have lower frequencies than subsequent strokes. A typical example of return stroke current waveforms measured by Eriksson for a flash is shown in Figure 22. During the return stroke phase, the resulting electromagnetic fields illuminate structures and conductor systems whether above the ground, on the ground or under the ground. For the present discussion and for the sake of explanation only, consider a return stroke channel illuminating a system of three conductors at location (P1, P2 and P3) as shown in Figure 23. Depending on the distance of the stroke from the object struck and magnitude of the return stroke, two components of the electric fields (horizontal and vertical) contribute to what are known as fictitious sources (distributed sources appear along the length of the conductor as long as the fields are illuminating it, a classical way of imagining and realizing field to wire coupling phenomena) on the wires as shown in the lower sketch in the figure. The vertical component of the electric field will create distributed voltage sources appearing all along between the line and reference, proportional to the height or depth of the conductor. If the conductor is on the ground, there is no contribution from this component. The vertical component of the electric field has negligible effects because of the ground medium. The vertical component only contributes to the total voltage on the line at a given point. The most important distributed voltage sources that will always exist are from the horizontal component of the electric field. The distributed voltage sources from the horizontal field are proportional to the height or depth of the conductor and also on the finite ground medium.

88

Figure 23. Sketch showing field illumination phenomenon due to return strokes and the consequent representation of field to wire coupling phenomena by means of distributed voltage sources (Nelson, 2005).

89

The return stroke phase of the lightning determines the extent of damage to systems, while the stepped leader phase of the lightning determines whether the stroke will terminate on the system to be protected, i.e., direct strikes, or on the flat ground, i.e., indirect strikes. The magnitude of the currents and their corresponding wave shapes decide the severity of the damage; hence, they should be included in a lightning protection study.

4.3.3 Scenario 3 – Iron-powder-bridge in the insulated joint

The insulated rail joint (IRJ) is a simple but important component of the train detection system. It electrically insulates the two consecutive rails, and also acts as a bond between the rails to form a long track distance. The bond malfunctions when it electrically connects both consecutive track sections. The physical gap between the two rails becomes smaller each time a train passes because of wearing, possibly ending in a signalling failsafe fault with a potential electrical connection. The wheels on the rolling stock roll over the IRJ and this may mill the rail over the IRJ magnetic fields, making iron particles stick together and building conducting bridges over IRJ. The electrical polarity from section to section is reversed, so if the IRJ is shortcut by an iron-powder-bridge, both sections will show an occupied track state following the failsafe principle. Therefore, the system will activate an emergency brake call and a non-existent obstacle will be detected, producing unnecessary delays and costs This may happen in locations where the traffic is heavy and the wearing of wheels is extreme, as in main stations. Typically this issue does not imply a problem on long distance tracks, but is a problem for crucial railroads with many trains. In these kinds of situations, extra maintenance is necessary, for example, to clean the IRJ of metallic particles regularly to avoid short-circuits between adjacent I-rail sections. The short-circuit between track circuit sections is emulated in the track and measured in the track circuits cabin. The insulated joint in the I-rail electrically insulates the two rails and connects them to create one long track distance. At times, the failsafe system will show red, indicating something is not correct in the system. The wheels from the trains and wagons roll over the insulated joint; this mills the rail ends over the insulated joint and over time the gap between

90

the two rails diminishes. The problem is a wear out problem but the outcome is a signalling fault. If an isolated joint between two sections of the I-rail is bridged by a piece of metal, the voltage in both sections can drop enough to generate “train presence” in both sections. If a road crossing is near, the traffic lights turn red and the barriers go down. This is how the system should work and can be used as a simple test that the system is working. The Swedish system of train detection by means of the I-rail normally works very well, and train collisions have never happened where the ATC system has been used. However, in some locations where the traffic is heavy and the wearing of wheels is extreme, magnetic fields make iron particles stick together, building conducting bridges over isolated junctions. The rails always become magnetic because of magnetic fields from railway currents. This is normally not a problem on long distance tracks, but can be a problem at points on some lines where there are a great many trains, as in Stockholm Central Station, where it is necessary to clean the isolated junctions from metallic particles to avoid short circuits between adjacent I-rail-sections. Induced voltage peaks can interfere with sensitive strain gauges. Detectors for flat wheels may be disturbed by passing trains, if an isolated I-rail junction is very close. The detectors for flat wheels consist of strain gauges glued to the rail. If a train enters or leaves an I-rail section, a transient E- field peak may be induced, due to sudden changes of current-paths. Sudden changes of a current through an inductive circuit will generate a voltage peak and a corresponding electromagnetic peak (E-peak), strong enough to affect the sensitive strain gauge sensors. Such an E-peak can make a measuring sequence useless, if it occurs in the middle of the measuring sequence. This kind of disturbance has been reported only in a location where the flat wheel detector is very close to the I-rail junction. It not occur if the isolated junctions are far away from the flat wheel detector. The detectors for hot bearings are much more insensitive to electromagnetic pulses, but are not affected by closeness to I-junctions.

4.4 Quality assurance of the DC Track circuits.

The quality of the behaviour of the track circuit is assured by means of its correct calibration. After introducing the battery, the resistances R1 and R2 must be calibrated following this method:

1- R1 = 0-6 Ohms resistance in series with the battery. Its value is the one providing the corresponding current in the source cable with the

91

train in the track. This value is taken from a graph depending on the soil and length of the track circuit section. In our case, the minimum is 2 Amps.

Figure 24. Current graph to the R1 calibration.

2- R2 = 0 - 15 Ohms resistance in series between the choke and the relay. The first step is tuning the resistance to 0 ohms, without a train in the track. Then, a variable resistance must be introduced between the rails to provides a voltage difference of 4 volts. In this situation, the R2 is increased until it reaches 1.4 volts between the terminals of the relay.

92

5 RESEARCH METHODOLOGY

5.1 Research strategy The research methodology can be classified in two styles, quantitative and qualitative (Neuman, 2003). In most of the cases the right combination of them is the key for the optimal research. In this licentiate it was also studied a specific case, the DC Swedish Track Circuits. The use of case studies helps to get real conclusions from the particular to the general (Yin, 2009). The next table shows the 3 types of research goals followed and their characteristics. Exploratory Descriptive Explanatory Become familiar with Test a theory´s the basic facts, setting prediction or Provide a detailed, and concerns. principles. highly accurate picture.

Create a general mental Elaborate and enrich Locate new data that picture of conditions. a theory´s contradict past data. explanation.

Formulate and focus Create a set of questions for future Extend a theory to categories or classify research. new issues or topics. types.

Generate new ideas, Support or refuse an Clarify a sequence of conjectures or explanation or steps or stages. hypotheses. prediction.

Document a casual Determine the Link issues or topics process or mechanism. feasibility of with a general

conducting research. principle. Report on the

background or context Develop techniques for Determine which of of a situation measuring and locating several explanations is future data. best.

Table 3. Type of research goals. (Neuman, 2003).

93

5.2 Data collection and analysis

The collection of data used in this project comes from different sources:

- 0felia: Trafikverket database of the records of all the failure data of the railway infrastructure and work orders. - Measurement campaigns on the track. - Experiments in the lab.

0felia can be filtered by different fields. Using type and teknik (technology) filtering the most common failures in the track circuits were detected looking at the failure mode (symptom) frequency. The maintenance personal of the DC Swedish track circuits agreed with our choice.

On the other hand, in the next section are explained the specific tests for the measurement campaigns and the lab experiments designed for the validation of the modelling and simulations.

There were also analysed the weak points of the related current standards to propose an improvement of these drawbacks.

94

6 RESEARCH APPROACH

6.1 Modelling

The main objective in this research methodology is the design and verification of a railway system model that estimates electromagnetic interference levels affecting the functionality of signalling and train detection systems. Specifically, rolling stock and infrastructure models should help to identify the low frequency interfering signals affecting track circuits by creating harmonic currents on the track. The low frequency model will be implemented using the following strategy: Rolling stock model: considering an ideal power supply source, the rolling stock model is implemented and compared to results of the measurements. This model only covers stationary conditions, not transients. Infrastructure model: considering a lineal power load, the infrastructure model is implemented and compared to results of the measurements. This model only covers stationary conditions, not transients. Integrated model of rolling stock and infrastructure: with the already validated models of rolling stock and infrastructure, an integrated model is implemented and compared to results of the measurements. In a first approach, the model is only intended to cover stationary conditions, not transients.

6.1.1 Rolling stock

Rolling stock was provided and validated by the manufacturer, CAF (Deliverable 4.1 TREND, 2013). The aim of the rolling stock simulation model is to predict the harmonic currents generated by the train and affecting train detection systems such as track circuits. The model to be developed will focus on AC trains (the type of rolling stock to be used in the test campaigns). Only stationary conditions will be studied by the rolling stock model to more easily validate its results. Once the simplified model is validated some transient events will also be studied. The primary requirements of the train model are:

95

The model should be able to represent the harmonic current content generated by the train as closely as possible. The model should be easily adaptable to different type of trains (for which the available information is minimal), thus requiring it to be a general and high level model. The main problem of the model development is obtaining all the necessary parameters because of inaccessible information. The test campaigns will use trains not manufactured by CAF, and their characteristics will not be fully accessible. Basic components can affect the electromagnetic characteristics of the rolling stock model. The aim of the model is to study the influence of the different rolling stock components in the whole system. The typical structure of a train is shown in Figure 25.

Pantograph

DC Link Filter

Converter Inverter

Motor

Rail Speed APS sensor

Converter control Inverter control

Figure 25. AC train block diagram.

In general, the model behaves correctly in the frequency bands of interest up to 50 kHz. The key frequency ranges of the harmonic spectrum correspond to the principal emission bands due to the PWM converter switching (3800-4600Hz and 7600-9200Hz). The simulation is only valid for the prediction of harmonics multiples of the synchronization signal, which in this case is the 50Hz AC power source. The model considers the ideal performance of the components and systems, thus, only harmonic of frequency n×50, with n=1, 2, 3, etc. Moreover, this simulation model considers an ideal power source, which in the real situation, includes voltage source harmonics at low frequencies, at odd frequencies of the source fundamental frequency. Real converter control strategies include thermal control and power efficiency and consumption control to improve the performance of the converter, and this may influence the switching pattern of the PWM converter. Some additional features which are caused by the infrastructure or the presence of other train or equipment have a slight influence in the model as well.

96

Basically, the train behaves like a harmonic current source, closing the circuit defined by the infrastructure substation, the catenary and the rails as the return conductor.

6.1.2 Infrastructure

As also shown in PAPER B, a generic simplified scheme of the rail infrastructure appears below.

Figure 26. Generic scheme of a railway infrastructure.

This model is integrated with another Matlab model of the train to run the simulation and get the signals. As seen in the figure, the model corresponds to a track section from the converter to the end of the line. The main blocks in all infrastructure are the converters, the transformers (AT or BT) and the transmission lines with their couplings. To get the right configuration of the system, it is necessary to have data on the following: Converter: Input and output frequencies and voltages and the R and L in series. Transformer: Operating power, voltages and frequency. For losses, it is necessary to specify the power losses obtained from the open circuit test (no load test) and short circuit test, the voltage drop obtained from short- circuit test and the magnetization current from the open circuit test. It is important to obtain the contact impedances and impedances of the AT. Transmission lines: Voltage levels, the distances between the conductors and their sections, and material conductivity to get the couplings between them.

97

Poles: Distance between poles and their configuration. Other distances: distance between transformers (d), converters and connections of the rail to ground.

Swedish railway supply system

A detailed scheme of the Swedish infrastructure is shown in Figure 27.

Figure 17. Detailed Scheme of the Swedish Railway infrastructure.

Also taken from PAPER B, Figure 28 shows the track section to be modelled, from the exit of the converter to the end of the line.

98

Figure 28. Detailed scheme of one section of the Swedish Railway infrastructure.

The following data about the infrastructure were provided by Trafikverket and used in the modelling: AT system’s impedance (Banverket, 2002): Z=R+jωL=(0.189+0.0335×length(km))+j(0.293+0.031×length(km))Ω, where f=16.7Hz (Note: the length of the cables connected to the AT and Z is included in the AT, not in the cables.)

Mutual inductances: M12= k ඥଵଶ End of circuit with an AT.  End of circuit of cables sources: with 3 loads of 100 kΩ in star configuration and connected to ground with a load of 1 MΩ. Source cables of 50 Hz connected to ground with R=1MΩ. AT-transformers: TYP: KYRU 36 NC 3250 Rated Power: 5000 kVA Rated Voltages: 33000/16500 V Rated Frequency: 16.7Hz No Load Power Loss: 2600 W Full Load Power Loss: ½: 11000 W Short Circuit impedance: ½: 0.4 % Transformer for the power supply system of 132 kV TYP: KYRU 145 NC 16000 Rated Power: 16000 kVA Rated Voltage: 132000/16500 V Rated Frequency: 16.7 Hz No Load Power Loss: 7270 W

99

Full Load Power Loss: 57470 W Short Circuit impedance: 4.77 % Transformer from the local source and to the backup system (for the 50Hz system) TYP: KYRU 36 NC 2500 Rated Power: 2500 kVA Rated Voltages: 22000/11000 V Rated Frequency: 50 Hz No Load Power Loss: 3300 W Full Load Power Loss: 19500 W Short Circuit impedance: 6.5 % Power Line Pole configuration: it is based on “The Bothnia Track” and follows the structure used to calculate the electrical parameters.

Figure 29 shows the wire coordinates, diameters and electrical data for the train power supply lines of the Bothnia track.

Figure 29. Swedish Power line pole.

Table 4 includes the characteristics and position of each of the cables numbered in the previous figure. Note: the last three columns were not included in PAPER B.

100

Conductor X, Y, Area, Rad Material R, L, C, Number/ mm mm mm2 ius, /conduc Ω/km mH/ nF/km Voltage mm tivity, km (To frequency 106 S/m ground) (theoreti cal values) 1 Catenary 300 560 120 6.2 Cu +16.5kV, 0 0 57 16.7 Hz 2 Carrier 300 730 70 4.7 Bronze line 0 0 28 +16.5kV, 16.7 Hz 3 Neg feeder - 685 212 8.2 Al -16.5kV, 600 0 28.5 16.7 Hz 4 Ground - 630 212 8.2 Al return line 210 0 28.5 5 “single 300 0 6648 46 Fe(steel) 6 Neg feeder - 785 212 8.2 Al -16.5kV, 600 0 28.5 16.7 Hz 7 Support 600 785 212 8.2 Al Feeder 0 28.5 +16.5kV, 16.7 Hz 8 Top feeder 500 925 99 5.6 FeAl 0.371 1.05 11.014 22kV, 50 Hz 0 31 1 19 9 Top feeder 0 101 99 5.6 FeAl 0.371 1.05 11.014 22 kV 16 31 1 19 50 Hz 10 Top - 925 99 5.6 FeAl 0.371 1.05 11.014 feeder 500 0 31 1 19 22 kV Ground - Table 4. Characteristics and position of each cable.

Design and simulation of the main blocks To make the simulation more realistic, instead of an ideal source, as a linear transformer for the power supply we used, the voltage and current measured before the converter and entered the information into MATLAB.

101

Figure 30. Current measured in Sweden before the converter.

Figure 31. Voltage measured in Swedish railway system before the converter.

As shown in PAPER B, we considered the first 8 harmonics. In the next figures, we show the differences between the original sources and the simulated sources, for current and voltage respectively.

102

Figure 32. Original sources and simulated sources.

We have grouped the information on amplitude and phase of each harmonic in the tables 5 and 6, again for the current and the voltage respectively.

Current 1x 2x 3x 4x 5x 6x 7x 8x (16.7 Hz) Amplitude 427.91 0.2428 11.4845 0.3090 42.17 0.4919 25.53 0.6126 (mA) Phase 249.45 297.95 77.39 166.48 209.13 260.18 11.1 337.6 (Degrees) Table 5. Amplitude and phase of the harmonics in current.

Voltage 1x 2x 3x 4x 5x 6x 7x 8x (16.7 Hz) Amplitude 2.8909e+4 23.7876 2e+3 14 874.12 2.46 156.5 3.46 (V) Phase 94.78 148.58 133.32 316.82 221.14 52.27 334 337.6 (Degrees) Table 6. Amplitude and phase of the harmonics in voltage.

In the next section, we use the mirror method to calculate the parameters for the coupling, assuming a distance h below the rails' level of 1 metre for the electrical ground zero. With that assumption, as well as the necessary characteristics and the distances between the cables, we calculate the coupling values as follows.

103

Figure 33. Description of the pole.

The rail resistance is R = 4.976Ω/km and the self-inductance L = 0.964mH/km. The catenary with AT system has an impedance: 0.125 + j0.343Ω/km (Banberket, 1999). Assume the frequency 16.7Hz. The self- inductance is 3.3 mH/km.

The distance from the power source to the converter is around 400-500 km. The positive line of the 3-phase lines is assumed to be 0.1025+j 0.126Ω/km. The equivalent inductance is then 1.2mH/km. The 400- 500km with impedance will have a resistance of 0.1025 × 450 Ω = 46.125Ω and the inductance of L = 0.126 × 450 / (2 × 3.1416 × 16.7) = 540mH.

As for the transformer, we wrote a Matlab program to calculate using the multitransmission lines method (Lee et al., 2006) the mutual inductances and capacitances. A screen shot of both applications and the used formulas are provided in PAPER B.

The results for the Swedish railway infrastructure are detailed in Tables 7, 8 and 9.

104

Table 7.Transmission lines impedances.

Table 8. Mutual inductance.

Table 9. Mutual capacitance.

For a 10 km long line, the voltage drop is as shown in Figure 34.

Figure 34. Voltage drop.

105

Finally, by merging all this information, we designed the final Matlab model of the Swedish railway infrastructure, shown in Figure 35.

Figure 35. Matlab model of the Swedish railway infrastructure.

106

6.1.3 Integration

To integrate the Matlab model to the Oaris train, we adapt the infrastructure using the model designed from the data provided by Trafikverket. In this integration, the train model is added to the infrastructure model as a subsystem with two ports. One of the ports is connected to our catenary and the other one to our rail as a real train in a line. First, the Powergui configuration of the Matlab Simulink model is changed to “discrete” to enable us to run the simulation of the integrated model. The Spanish railway infrastructure uses the same power supply to feed two tracks, which should be in parallel. The power supply in Toledo – La Sagra is: 1x25 kV, 50Hz, so there is no negative feeder. The source used is an ideal sinusoidal input of 400 kV amplitude and 50 Hz frequency instead of the real measured signal. This signal is used as input of a transformer to the 25 kV, 50 Hz. The transformer is a BT type, with rate 1:1 between 25 kV and ground. The electrical characteristics of the substation of this infrastructure, where the transformers are placed, are: -Rated Power: Sr = 20 MVA -Rated voltage of the secondary: U r = 27.5 kV -Short Circuit Impedance: 8.0% The characteristics of the cables are shown in the next table.

Table 10. Characteristics of the cables.

Figure 36 is a scheme of the power pole in the Spanish infrastructure.

107

Figure 36. Power supply pole of the Spanish railway infrastructure.

And the resulting Matlab model is given in Figure 37.

108

Figure 37. Matlab model of the Spanish infrastructure.

109

6.2 Hybrid model

This section discusses the work required to design high frequency infrastructure models permitting us to study the EM interferences produced and received by another subsystems that are able to interfere with communication and signalling equipment (Deliverable 4.2 TREND project, 2013). The objective is to develop a model that includes the major interference sources and reflects the main couplings that lead to the action of radiated interference on sensitive systems. The final model should allow us to perform any kind of test, especially interesting for the worst case conditions. This hybrid model is based on the known characteristics and observations from measurement campaigns. Two measurement campaigns were carried out, one in Spain and one in Sweden. In this case, the model and simulation can be very complex, so the strategy of our modelling methodology is to obtain a realistic model in terms of coupling between the main EM noise sources and victim systems, but with reasonable computation time, looking carefully into the solver and meshing impacts and the definition of the elements included in the model.

6.2.1 Solver

Several very powerful numerical tools are available on the market. After identifying our needs in terms of the frequency range to study, dimensions of the model structure and nature of the interference signals, as introduced in PAPER A, we determined the most appropriate software. The structure to model is large and the frequencies vary between 20 Hz (conducted interferences) and 1.8 GHz. The nature of the disturbance signals can also significantly vary: they can be permanent or transient with very variable time characteristics. We selected CST software because it offers a wide range of solvers that can adapt to the different configurations that must be addressed and it is particularly suitable for modelling transient signals. All models of the various parts of the project were designed using the same platform, CST STUDIO SUITE 2012, to have the possibility of integrating them. CST microwave studio (MWS) is a software package for electromagnetic analysis and design. It provides a modelling graphical tool shown in Figure 38. It allows many applications to be handled, including antennas, filters, transmission lines, couplers, connectors, printed circuits boards, resonators

110

and many more. It also allows the integration of technical systems design and permits electromagnetic analysis of large complex systems like cars, trains and aircraft.

Figure 38. Graphical CST interface (Microwave Studio, 2013).

The elements of the structure to model are successively defined in the interface. The simplest elements can be defined by planes, wires and perfect electrical conductors (PEC). To form more complex elements, libraries of "solid" and “materials” are available; they allow the modelling of different types of objects. After the model has been constructed, a fully automatic meshing approach can be applied before the simulation is started. The most salient feature of CST is that it represents a complete technology approach; it offers the possibility of adapting the simulation solver computing and the mesh approach to the user’s problem. This property is particularly advantageous in view of the different issues addressed in the licentiate. As noted above, one of the advantages of CST MWS is the “Complete Technology Approach” which allows choosing the best solver for the electromagnetic problem from among many possibilities. This advantage gives the user flexibility to choose the best method while maintaining a high level of accuracy. Figure 39 presents the different solvers and their EM fields applications and can be used as a guideline rather than a rule for which solver to use.

111

Figure 39. Comparison of CST solvers applications (Microwave Studio, 2013).

As introduced in PAPER A, this software is one of the most accurate and efficient computational solutions for electromagnetic designs, such as circuit simulation in a wide range of frequencies. CST offers a wide range of EM simulation software to address design challenges across the electromagnetic spectrum, from static and low frequency to microwave and RF, for a range of applications, including EDA & electronics, EMC & EMI and charged particle dynamics. The main piece of CST’s product range is CST STUDIO SUITE®, which provides a complete set of 3D electromagnetic simulation tools, along with a number of related products dedicated to more specific design areas, such as cable harnesses, PCBs and EM/circuit co-simulation. These functionalities fit perfectly into our requirements. In our work, we use the tool CST CABLE STUDIO. This simulation software is dedicated to the three-dimensional analysis of signal integrity (SI), conducted emission (CE), radiated emission (RE), and electromagnetic susceptibility (EMS) of complex cable structures in electrically large systems. It provides powerful import filters from popular MCAD and ECAD tools for smooth integration into the industrial workflow.¨ It is equipped with enhanced visualization capabilities to interactively highlight selected signals or cables in both the 3D graphic view and the 2D schematic view, as shown in Figures 40 and 41 respectively.

112

Figure 40. CST CABLE STUDIO 3D model.

Figure 41. Circuital 2D schematic view.

Information about which cable carries which signal or which wires are connected to which connector pins can be directly accessed in the CST CS cable navigation tree, as shown in Figure 42. It is also possible to define different tasks to show the values measured with the probes or perform a parameter sweep in one or more component.

113

Figure 42. Cable navigation tree menu and navigation tree menu.

Meshing parameters

To get correct results, it is critical to choose the right meshing parameters. In this model, we use a coupling distance of different routes of 5700.0 mm. This distance takes into account the coupling contribution of the both rails instead of a separated contribution. The simulation analyses the ohmic and dielectric losses. With the uni-directional coupling to the 3D field solver, we have the option of integrating the train in the model and connecting the chassis to the ground. In these simulations, we have to introduce the train, so there is no AC/transient simulation. The final meshing is shown in Figure 43.

Figure 43. Final meshing used.

114

6.2.2 Description of the scenarios

Scenario 1 - Single track circuit section

This scenario has one track circuit section and a variable resistance between the rails with a sweep of values. The scenario represents a typical Swedish track circuit section. It introduces a variable resistance R3 between the I-rail and the S-rail as the axis of the train. This variable resistance is also used for the calibration of the resistance in series with the choke and the relay. There are two possible commutations of the relay ON to OFF and OFF to ON in different voltage, due to the hysteresis window of the relay. We will check this, sweeping a range of resistances, decreasing with initial condition ON and increasing with initial condition OFF.

Figure 44. Sweep of the 11 resistances from 4 to 5 Ohms (step: 0.1 Ohms).

Figure 45. Sweep of the 21 resistances from 2 to 0 Ohms (step: 0.1 Ohms).

Figure 46. 3D model of the scenario.

115

Figure 47. Schema of the scenario.

Scenario 2 - External interference

In this scenario, we have one track circuit section with an external interference to check the immunity of the DC track circuit. We use an external port connected to the track as the wheels of the train. This scenario removes the variable resistance R3 and introduces a lightning strike as an external interference to the track by means of external port 1. The lightning strike produces a 30 kA current entering the railway infrastructure through the grounding impedance during 0.3 μs. This is implemented using a CST task called “Transient” related to this port. Because of its properties, it is possible to import an ASCII file as a 30 kA current signal of 300 ns duration, as shown in Figure 48.

116

Figure 48. Import of the lightning signal to port 1.

The 3D model of the scenario is equal to the previous one, but the schema of the scenario is modified as shown in Figure 49.

Figure 49. Schema of the scenario.

117

Scenario 3 - Two track circuit sections

The third scenario has two successive track circuit section with a variable resistance with a sweep of values between the I-rails sections as a broken or shorted insulated joint. The scenario represents two successive track circuit sections. To prevent one circuit from falsely powering another in the event of insulation failure, the electrical polarity is usually reversed from section to section. In this configuration, if there is a short circuit between them, both relays will show an object on the tracks, and the train will activate the emergency brake. To model this, we add a second track circuit section changing the poles of the battery and the connections of the relays. The I-rail sections are separated by means of a variable resistance modelling the 6mm rubber isolation joint.

+ - + - + - + - Relay I-rail Relay

S-rail

Figure50. Inverse polarity in successive track sections.

Figure 51. 3D model of the scenario.

As can be seen in Figure 52, to get the inversion of the polarity, we reverse the connections of the battery and relays lines to the opposite rails.

118

Figure 52. Schema of the scenario.

6.2.3 Details of the components of the model.

Battery The battery is an ideal DC source which gives 6 DC power supply to feed the whole track circuit section.

Figure 53. Battery.

119

Relays The CST block called “Voltage controlled switch” represents the same behaviour as a relay and is used to model the relays. As can be seen in the characteristics, it is possible to change the initial condition, the functionality used to validate both possible commutations of the relay: ON to OFF and OFF to ON.

Figure 54. Relay characteristics. Figure 55. CST relay block.

Chokes Each choke consists of two huge coils of 2.4 H, which can be connected in series or in parallel depending on how it is configured. For the relays, the coils are connected in series and for the power supply, the coils are in parallel.

Figure 56. Coils in series. Figure 57. Coils in parallel.

Resistances All the resistances in the circuit are considered ideal. Some are variable and to model them, we assign their resistance value as a parameter. This method allows us to perform a sweep of values or easily change them in the whole model.

Figure 58. Resistance with a value R1.

120

Cables BV RQ 1x10 mm2

Figure 59. Cable BV RQ 1x10 mm2 definition.

Profile:

Figure 60. Cable BV RQ 1x10 mm2 profile.

Materials:

Figure 61. Characteristics of its materials.

121

BV RQ 2x10 mm2

Figure 62. Cable BV RQ 2x10 mm2 definition.

This cable has the same shape, proportions and materials as the previous one but is defined as two cables together, because there are two cables in the same corrugated tube and there is a coupling effect between them.

Figure 63. Cable BV RQ 2x10 mm2 profile.

In this case, the cable goes underground (below the ballast) to the relay or to the power source, but is not straight, so it is represented as a 600 mm depth spline, as shown in previous sections.

Catenary

122

Figure 64. Definition of the catenary cable.

Figure 65. Catenary profile.

Cable connections The connections are defined by the pins or nodes and the junctions between them.

123

Figure 66. Characteristics of the connection.

Rails As in the case of cables, the rails must be modelled according to their profile and material.

Figure 67. Rail definition and list of points of the polygonal profile.

Profile:

Figure 68. Rail profile.

124

Figure 69. Material characteristics of the rail.

Electrical insulated joint The 6 mm rubber insulated joint between I-rail sections is represented by means of a resistance. A working insulated joint is modelled with a 1 MOhm resistance. In the case of scenario 3, the broken insulated joint is modelled with a variable resistance.

Secondary of the relay circuit To check the right behaviour of the track circuit we introduce a simple circuit in the secondary composed of a 2 volts DC battery and a resistance to the ground. With a probe, we can measure the voltage or the current in the resistance. If these values are different from zero, it can be said that the relay is de-energised (there is an obstacle between the rails) and the low resistance in the secondary of the relay closes the circuit. Figure 70. Secondary of the relay circuit.

External ports Scenario 2 integrates an external port into the model to introduce an interference to the track circuit.

Figure 71. External port.

125

Medium It must be stressed that all around our model, the material is the air, with characteristics as indicated in Figure 72.

Figure 72. Material characteristics of the air.

6.2.4 Validation test definition:

Scenario 1 – Test site The test site must meet as far as possible the “free space” requirements within the existing restrictions of the railway environment. The location of the track for the measurement must be close to the cabin with the supply, chokes and relays and as isolated as possible from any population or main road (the closest must be a minimum of 10 km away). The track must not have any trains passing the measuring track circuit section at the moment the test is performed. To ensure the least possible interference to the tests, the meteorological conditions must be within a reasonable range. Specifically, the humidity must be low enough to prevent condensation on the power supply conductors (after 24 hours during when no more than 0.1 mm rain has fallen). Since it is necessary to plan the tests before the weather conditions can be known, tests could be made in weather conditions which do not meet the target conditions. In these circumstances, the actual weather conditions must be recorded with the test results. Finally, all similar tests should be carried out on the same working day to ensure the same test site measuring conditions (temperature, humidity, etc.).

Scenario 1 – Test setup This scenario is related to the precise moment when the relay switches from ON to OFF and vice versa. Therefore, a voltage sweep has to be performed with a variable resistance to determine when the relay switches from one state to another. The instrumentation required to perform the measurement campaign is: - Variable resistance (R Ballast): At least from 0 to 8 Ω and 0.1 Ω accuracy.

126

- Oscilloscope: Must be portable and configured as a multimetre. A DC current and a DC voltage probe for that model of oscilloscope are also required. - DC multimetre to measure voltage between rails. - Cables and clamps for connections.

Figure 73. Test setup scenario 1.

Scenario 1 – Test procedure There are two relay switching possibilities: ON-OFF and OFF-ON, and they appear in at different voltage due to the hysteresis window of the relay. Two different sets of measurements must be performed for this scenario, increasing and decreasing a variable resistance between rails for the calculation of the current in the source cable, voltage in the primary of the relay and voltage between rails. The tests are conducted as follows: 1- Connect the current probe and the voltage clamp of the oscilloscope to the primary of the relay. 2- Measure voltage and current with no obstacle between rails. 3- Connect the variable resistance by attaching the two clamps to the rails. 4- Connect the multimetre between rails. 5- Set the variable resistance to the maximum value (around 8 Ω) and check values for voltages and current. 6- Decrease the resistance in as many steps as needed to check the behaviour of the relay until reaching 0 Ω. Continue checking the values for voltages and current, as well as the relay switches.

127

7- Increase from the resistance from 0 Ω to the maximum resistance value, using as many steps as required to check the behaviour of the relay. For each change, check the values for voltages and current and also the relay switch. 8- Save the results. 9- Disconnect the oscilloscope and variable resistance. Scenario 2 The tests of the worst case scenario due to an external interference is done by emulating a lightning strike in a laboratory. The track circuit is tested in two different scenarios: 1- Behaviour of the isolated chokes. 2- Behaviour of the relays with the introduction of the lightning into the I-rail in the whole track circuit section. Scenario 2 – Test site 1 The goal of this test is to check the immunity of the chokes to different emulations of the lightning strike. This validation is done in a laboratory with conditions inserted into the standard laboratory conditions. Because many definitions of standard temperature and pressure differ significantly from standard laboratory temperatures (e.g., 0 °C vs. ~25 °C), reference is often made to "standard laboratory conditions" (a term deliberately chosen to differ from the term "standard conditions for temperature and pressure", despite its semantic near identity when interpreted literally). However, a "standard" laboratory temperature and pressure is inevitably culture-bound, as different parts of the world differ in climate, altitude and the degree of use of heat/cooling in the workplace. For example, schools in New South Wales, Australia, use 25 °C at 100 kPa for standard laboratory conditions, while in Sweden, the temperature is approximately 5 degrees less. Scenario 2 – Test setup 1 The required equipment for this test consists of: - High voltage oscilloscope connected to a laptop with a software data collector. - 30kV generator. - A photodiode, optical fibre and a 9 V battery to trigger the oscilloscope and measure the spark duration.

128

- Rolled cable used in different lengths to check the desired resistances with a multimetre. - Big capacitors supporting till 40 kV. - Choke. - 10 MΩ ceramic resistance to connect to ground. - Connections and cables.

Figure 74. Test setup 1 scenario 2.

After the oscilloscope is calibrated, it is connected to a laptop with the software data collector Octave to take data and plot them in graphs.

Figure 75. Calibrated oscilloscope connected to the laptop.

129

To get the right signal, the oscilloscope must be triggered in the exact moment the spark arrives, something that cannot be done using a manual actuator.

Figure 76. Actuator to start the oscilloscope measure manually.

Because it is not possible to know the duration of the spark from the graphs because they are resonant graphs, we tried calculating the duration of the spark by means of the number of frames of a video of a predetermined high fps (frames per second) rate, but this method was not accurate enough. Ultimately, we solved the problems of the triggering and duration of the spark by using a photodiode in the second bunch of tests.

130

Figure 77. Photodiode pointing to the spark. When the photodiode is supplied by a power source, this introduces a great deal of noise, Figure 78.

Figure 78. Power source to supply the photodiode.

131

We finally decided to use an optical fibre connected to the photodiode, which was now supplied by an ordinary 9 V battery instead of a power source to avoid any noise, Figures 79 and 80.

Figure 79. Optical fibre pointing to the spark.

Figure 80. 9 V battery to feed the photodiode and photodiode at the end of the optical fibre.

132

The resistance of the serial RC circuit is created by a rolled cable, as shown in Figure 81.

Figure 81. Multimetre measuring a rolled cable resistance of 33.4 Ω.

The two biggest capacitors available on the market that support 30 kV are 3.5 nF and 2.7 nF; they can be configured in parallel to get a total of 6.2 nF, Figure 82.

Figure 82. Both 3.5 nF and 2.7 nF connected in parallel.

The choke consists of two huge coils of 2.4 H, which can be connected in parallel to protect the relay, and in series to protect the power supply. The electric characteristics of the chokes, as explained in previous sections, depend on the configuration: Series: Z = 500 Ω; R = 2.6 Ω; V (AC) max = 100 V; F = 16.67 Hz and I (DC) max = 0.25A. Parallel: Z = 125 Ω; R = 0.65 Ω; V (AC) max = 50 V; F = 16.67 Hz and I (DC) max = 0.50 A. In this experiment, we test both configurations, connecting them to a RC series circuit and to the ground through a safety ceramic resistance of 10 MΩ, Figure 83.

133

Figure 83. 10 MΩ ceramic resistance connecting the choke to the ground.

Scenario 2 – Test procedure 1 These are the test steps for scenario 2: 1- Connect the wanted capacitors and rolled resistance for the serial RC circuit. 2- Configure the choke in series or parallel. 3- Connect the capacitors to the power supply circuit (switch 1 closed and switch 2 open), Figure 84. 4- Start the power supply in the wanted voltage and wait some seconds to have the capacitors charged. 5- Connect the capacitor to the choke (switch 1 open and switch 2 closed), producing the spark, Figure 85. 6- Connect the capacitor to the ground of the circuit to ensure the discharge, Figure 86. 7- Turn the power supply off. 8- Save the measure of the voltage between the resistance and ground to check how the insulation is provided by the choke.

134

Figure 84. The figure shows the starting position, where the capacitor is charging.

Once it is charged, the red cable must be connected to the choke- resistance circuit to emulate the lightning going into the circuit, Figure 85.

Figure 85. Spark produced in the connection between the capacitor and the choke.

At this point, the capacitor is discharged again, Figure 86.

135

Figure 86. Connection of the capacitor to ground to ensure the discharge.

Scenario 2 – Test site 2 This test is performed in the same laboratory conditions as the previous one. In this case, it the lighting going into the I-rail of a whole track circuit section is emulated, as shown in the test setup. Scenario 2 – Test setup 2 For this circuit, the following must be added: Ceramic resistance to emulate the train. 8 x 1.5 V batteries in parallel (12 V in total) and a 10 kΩ resistance for the circuit on the secondary of the relay.

136

2 variable ceramic resistances divided into R = 15 Ω and R2. 6 V DC battery and a 3 Ω serial resistance for the power supply. 2 additional chokes, as there are 3 chokes per track circuit section to give protection to each relay and the battery. 2 metal bars to emulate the rails. 4 channel high voltage oscilloscope, with at least one high voltage probe, connected to a laptop with a software data collector.

Figure 87. Test setup 2 scenario 2.

The train is emulated by a variable ceramic resistance between rails, which can take values from 0 to 10 Ω. As checked in advance in the measurement campaign in the track, the resistances R1 and R2 are tuned to make the relay switch at 1 and 6.2 Ω, Figure 88.

Figure 88. Ceramic resistance at 6.2 Ω, the one which makes the relay switch.

137

The measures in the secondary of the relay are made in the resistance of a simple circuit composed of 8 x 1.5 V batteries in parallel (12 V in total) and a 10 kΩ resistance, Figure 89.

Figure 89. Secondary of the relay circuit.

The resistances of each relay circuit are obtained from two variable ceramic resistances divided into R = 15 Ω and R2, Figure 90.

Figure 90. Ceramic resistance in the relay circuit.

138

The battery is a 6 V DC car battery, Figure 91.

Figure 91. 6 V DC battery.

The two chokes are connected as shown in Figure 92.

Figure 92. Both chokes connection.

139

Finally, Figure 93 shows the definitive mounting. The figure also shows the two metal bars that emulate the S-rail and the I-rail, where the spark that emulates the lightning strike is introduced.

Figure 93. Final mounting.

Scenario 2 – Test procedure 2 The steps to follow in this test procedure are: 1- Connect the wanted capacitors and rolled resistance for the serial RC circuit. 2- Configure the choke in series or parallel. 3- Connect the capacitors to the power supply circuit. Switch 1 is closed and switch 2 is open to charge the capacitors. 4- Start the power supply in the wanted voltage and wait some seconds to have the capacitors charged. 5- Connect the capacitor to the choke, producing the spark. Switch 1 is open and switch 2 is closed, so the peak of current will enter the track circuit as a lightning strike. 6- Connect the capacitor to the ground of the circuit to ensure the discharge. 7- Turn the power supply off. 8- Save the measure of the voltage between the resistance and ground to check how the insulation is provided by the choke.

140

Figure 94. Introduction of the lightning strike as a spark in the circuit.

Figure 95. Discharge of the capacitor.

A four channel oscilloscope may be used as follows, Figure 96: high voltage probe, photodiode, secondary of the first relay and secondary of the second relay.

141

Figure 96: 4 channels oscilloscope.

The high voltage probe may now be connected to the I-rail, Figure 97, and/or to the exit of the choke (input of the relay), Figure 98.

Figure 97. High voltage probe connected to the I-rail.

142

Figure 98. High voltage probe connected to the exit of the choke (input of the relay). In order to be saved, the data may be recorded in graphs, with voltage in volts and time in seconds, connecting the oscilloscope to a laptop with a software data collector. The insulated rail joint (IRJ) is a simple but important component of the train detection system. It electrically insulates the two consecutive rails, but also it acts as a bond between the rails to form a single long track. The bond malfunctions when it connects the consecutive track sections electrically. The physical gap between the two rails becomes smaller each time a train passes because of wearing. This can end in a signalling failsafe fault from an electrical connection if the wheels from the rolling stock roll over the IRJ, milling the rail over the IRJ magnetic fields. This can cause the iron particles to stick together and build conducting bridges over the IRJ. The electrical polarity from section to section is reversed, so if the IRJ is shortcut by an iron-powder-bridge, both sections will show an occupied track, following the failsafe principle. The system would activate an emergency brake call and a non-existent obstacle would be detected on the rails, producing unnecessary delays and costs This may occur in locations where the traffic is heavy and the wearing of wheels is extreme, as in main stations. Typically, this issue does not imply a problem on long distance tracks, but is reported on crucial railroads with many trains. In such situations, extra maintenance is necessary, for example, regularly cleaning metallic particles from the IRJ to avoid short- circuits between adjacent I-rail sections. The short-circuit between track circuit sections is emulated in the track and measured in the track circuit cabin.

143

144

7 RESULTS AND DISCUSSION

7.1 Validation of the model

7.1.1 Probes

First we decided to introduce the following probes in the simulation to check the model’s behaviour: - In series with the battery. - Primary of the relay to measure the voltage value which commutates the relay. - Secondary of the relay to check the commutation in the relay’s exit. - In the I-rail. As the S-rail is connected to ground, the voltage in the I-rail is very close and can be approximated as the voltage between rails. Location Probe Battery P1 Primary of the relay 1 P5 Primary of the relay 2 P6 Secondary of the relay 1 P7 Secondary of the relay 2 P8 I-rail measured in the relay 1 P11 I-rail measured in the relay 2 P12 Table 12. List of probes used for the validation of the model.

7.1.2 Procedure of the validation

As expected, the results after calibration are not equal to the real measured ones, so we needed to tune some values of our model. We changed the Von to Voff values in the relay (1.5 V y 0.7 V respectively). The first measurements in the track showed this switching: - ON to OFF: In the change from 1.5 Ohms to 1.4 Ohms.

145

- OFF to ON: In the change from 4.8 Ohms to 4.9 Ohms. The new voltage values for an experimental resistance, R1 = 2.99 Ohms and R2 = 7.50 Ohms, are: - Von = 2.125 Volts. The relay has initial condition OFF because there is an obstacle of a resistance less than 4.8 Ohms on the track. If we increase it, the voltage in P5/P6 will be over 2.125 Volts and the relay commutates to ON, the working default. - Voff = 1.2 Volts. The relay has initial condition ON, the working default, because there is no obstacle or its resistance is more than 1.5 Ohms. If we decrease it, the voltage in P5/P6 will be below 1.2 Volts and the relay commutates to OFF, the behaviour when an obstacle or train is detected on the track. With these new voltages, as shown in the next section, the behaviour of the simulated track circuit is extremely close to the real measured track circuits.

Figure 99. New characteristics of the relay.

7.2 Analysis of the worst case scenarios

We tested a single rail track circuit fed by a DC source, a model deployed throughout Sweden, in several worst case scenarios.

7.2.1 Scenario 1 - Single track circuit section

Simulation results The simulation currents measured with the probe P1 for the same states of the track as those measured in the field are shown in the next table:

146

Resistance between rails Current in P1 0 Ohms between the rails 2 Amps 0.5 Ohms between the rails 1.725 Amps 1 Ohms between the rails 1.53 Amps 1.4 Ohms between the rails 1.415 Amps 1.5 Ohms between the rails 1.385 Amps 1.6 Ohms between the rails 1.365 Amps 1.8 Ohms between the rails 1.32 Amps 2 Ohms between the rails 1.275 Amps 4 Ohms between the rails 1.01 Amps 4.4 Ohms between the rails 0.975 Amps 4.6 Ohms between the rails 0.96 Amps 4.8 Ohms between the rails 0.945 Amps 4.9 Ohms between the rails 0.935 Amps No connection between the rails 0.42 Amps Table 13. List of simulated currents measured in the probe P1.

The simulated voltages measured with the probes P11 or P12 (same result) for the same states of the track as those measured in the field appear in the following table: Resistance between rails Voltage in P11 or P12 0 Ohms between the rails 0.005 Volts 0.5 Ohms between the rails 0.83 Volts 1 Ohms between the rails 1.415 Volts 1.4 Ohms between the rails 1.765 Volts 1.5 Ohms between the rails 1.84 Volts 1.6 Ohms between the rails 1.915 Volts 1.8 Ohms between the rails 2.05 Volts 2 Ohms between the rails 2.175 Volts 4 Ohms between the rails 2.97 Volts 4.4 Ohms between the rails 3.08 Volts 4.6 Ohms between the rails 3.13 Volts 4.8 Ohms between the rails 3.17 Volts 4.9 Ohms between the rails 3.2 Volts No connection between the rails 4.5 Volts Table 14. List of simulated voltages measured in the probes P11 or P12.

147

Verification on the track: Some measures must be taken outside and others inside the cabin on the relays.

Figure 100. Monitoring panel and relays inside the track circuits cabin.

Equipment Figure 101. Portable digital oscilloscope Scope III - OX7042. The equipment used to measure the voltage and current in the measurement campaign was the portable digital oscilloscope Scope III - OX7042. It was configured as a multimetre; it used a DC current probe, with a clamp for the DC voltage. Its characteristics are: - General data: 8000 digit + MIN/MAX bar graph; TRMS; graphical memory with date/time. - AC-, DC-, AC+DC voltage: 600 mVolts to 600 Volts TRMS; 800 mVolts to 800 Volts DC; accuracy Volts DC 0.5 % + 5 digs; bandwidth 200 kHz.

148

- Error detection and measurement window error: error duration adjustable; memory from up to 100 mistakes with date/time in text file. - Resistance: 80 Ohms until 32 MOhms; accuracy 0.5 % display + 25 D. - Further measurement types: temperature; capacity (5 nF to 5 mF); frequency (0 … 200 kHz), diode test 3.3 Volts. Figure 102. Variable resistance between rails.

We also used a variable resistance between the rails with a 0.1 Ohms accuracy.

Location and conditions of the experiments Both campaigns were conducted in stations in the north of Sweden, one in Niemisel and the other in Gammlestad (Luleå).

Niemisel The campaign described in PAPER A was down in the old station building in Niemisel, just 20 m from the track.

Figure 103. Old Niemisel station.

Measurements were taken from 22 to 24 January 2013, between 22.00 and 04.00, without rainfall or snowfall and with temperatures around -17 Celsius degrees every night.

149

Results in Niemisel The measured currents in the source cable are: Resistance between rails Current en P1 0 Ohms between the rails 1.84 Amps 0.5 Ohms between the rails 1.61 Amps 1 Ohms between the rails 1.46 Amps 1.4 Ohms between the rails 1.33 Amps 1.5 Ohms between the rails 1.31 Amps 1.6 Ohms between the rails 1.28 Amps 1.8 Ohms between the rails 1.23 Amps 2 Ohms between the rails 1.22 Amps 4 Ohms between the rails 0.93 Amps 4.4 Ohms between the rails 0.91 Amps 4.6 Ohms between the rails 0.88 Amps 4.8 Ohms between the rails 0.86 Amps 4.9 Ohms between the rails 0.85 Amps With the train in the track 2.6 Amps No connection between the rails 0.9 Amps Table 15. List of currents in the source cable measured on the track.

The measured voltages between rails are: Resistance between rails Voltage in P11 or P12 0 Ohms between the rails 0.7 Volts 0.5 Ohms between the rails 1.52 Volts 1 Ohms between the rails 2.13 Volts 1.4 Ohms between the rails 2.55 Volts 1.5 Ohms between the rails 2.58 Volts 1.6 Ohms between the rails 2.66 Volts 1.8 Ohms between the rails 2.81 Volts 2 Ohms between the rails 2.95 Volts 4 Ohms between the rails 3.89 Volts 4.4 Ohms between the rails 4.02 Volts 4.6 Ohms between the rails 4.09 Volts 4.8 Ohms between the rails 4.15 Volts 4.9 Ohms between the rails 4.14 Volts With the train in the track 0.52 Volts No connection between the rails 4.5 Volts Table 16. List of voltages between rails measured on the track.

150

As checked in the campaign, the commutations of the relays occur as: - ON to OFF: In the change from 1.5 Ohms to 1.4 Ohms. - OFF to ON: In the change from 4.8 Ohms to 4.9 Ohms.

Comparison between experimental and numerical results Also provided in PAPER C, the next tables and figures compare the simulations and the tests. State of the track Measured Simulated current current (P1) 0 Ohms between the rails 1.84 Amps 2 Amps 0.5 Ohms between the rails 1.61 Amps 1.725 Amps 1 Ohms between the rails 1.46 Amps 1.53 Amps 1.4 Ohms between the rails 1.33 Amps 1.415 Amps 1.5 Ohms between the rails 1.31 Amps 1.385 Amps 1.6 Ohms between the rails 1.28 Amps 1.365 Amps 1.8 Ohms between the rails 1.23 Amps 1.32 Amps 2 Ohms between the rails 1.22 Amps 1.275 Amps 4 Ohms between the rails 0.93 Amps 1.01 Amps 4.4 Ohms between the rails 0.91 Amps 0.975 Amps 4.6 Ohms between the rails 0.88 Amps 0.96 Amps 4.8 Ohms between the rails 0.86 Amps 0.945 Amps 4.9 Ohms between the rails 0.85 Amps 0.935 Amps No connection between the 0.9 Amps 0.42 Amps rails Table 17. Comparison of currents measured in the source cable and those measured with the P1 probe in the CST model.

151

2,5

2

1,5

Simulated Current 1

Amperes (A) Series1 0,5

0 0 0,5 1 1,4 1,5 1,6 1,8 2 4 4,4 4,6 4,8 4,9 Ohms between lines (Ω)

Figure 104. Graphical comparison of currents measured in the source cable and those measured with the P1 probe of the CST model. State of the track Measured Simulated voltage voltage (P11/P12) 0 Ohms between the rails 0.7 Volts 0.005 Volts 0.5 Ohms between the rails 1.52 Volts 0.83 Volts 1 Ohms between the rails 2.13 Volts 1.415 Volts 1.4 Ohms between the rails 2.55 Volts 1.765 Volts 1.5 Ohms between the rails 2.58 Volts 1.84 Volts 1.6 Ohms between the rails 2.66 Volts 1.915 Volts 1.8 Ohms between the rails 2.81 Volts 2.05 Volts 2 Ohms between the rails 2.95 Volts 2.175 Volts 4 Ohms between the rails 3.89 Volts 2.97 Volts 4.4 Ohms between the rails 4.02 Volts 3.08 Volts 4.6 Ohms between the rails 4.09 Volts 3.13 Volts 4.8 Ohms between the rails 4.15 Volts 3.17 Volts 4.9 Ohms between the rails 4.14 Volts 3.2 Volts No connection between the 4.5 Volts 4.5 Volts rails Table 18. Comparison of voltages measured between the rails and voltages measured with the P11 or P12 probes (same electrical point) of the CST model.

152

4,5 4 3,5 3 2,5 2 Measured Voltage

Voltage (V) 1,5 Simulated Voltage 1 0,5 0 0 0,5 1 1,4 1,5 1,6 1,8 2 4 4,4 4,6 4,8 4,9 Ohms between lines (Ω)

Figure 105. Graphical comparison of voltages measured between the lines and voltages measured with the P11 or P12 probes (same electrical point) of the CST model.

The comparison of the measures and the simulations shows the values of “No connection between the rails” but this was not taken into account in the graphs because it is not possible to show an infinite R3. The higher the resistance between the rails, the closer the error between the simulated and real voltages; this means we chose the correct supply value and resistance R2. Gammlestad The second validation test was performed in Luleå Gammlestad, in the north of Sweden, on 26 February 2014, with a temperature of -10 °C outside and 15°C inside the cabin.

Figure 106. View from the exit of the track circuit cabin.

153

The next figures show the oscilloscope and how it was connected to the relay by a DC current probe and DC voltage clamp.

Figure 107. Portable oscilloscope and connections of the probes to the primary of the relay, pins 21 and 22, the ones connected directly to the coil of the relay, and to the rail (clamp). On the other side, the variable resistance was connected to the rails by special clamps adapted to the rail shape. The voltage between the rails is checked directly in the variable resistance connected between the rails.

Figure 108. Measure of the voltage between rails.

154

Figure 109. Equipment used to measure inside the cabin.

Results in Gammlestad The tests in this validation correspond with the next resistance steps: - Test 1: No resistance between the tracks: Current relay: 120 mA Voltage relay: 1,6 V - Test 2: Resistance between the tracks R = 8 Ω: Current: 25 mA Voltage: 1.4 V Voltage track: 3.8 V - Test 3: Resistance between the tracks R = 4 Ω: Current: 25 mA Voltage: 1.2 V Voltage track: 3.2 V - Test 4: Resistance between the tracks R = 2 Ω: Current: 25 mA Voltage: 0.9 V Voltage track: 2.5 V

155

- Test 5: Resistance between the tracks R = 1.1 Ω: Current: 30 mA Voltage: 0.8 V Voltage track: 2.0 V - Test 6: Resistance between the tracks R = 1 Ω: Current: 25 mA Voltage: 0.7 V Voltage track: 1.95 V The relay switched. - Test 7: Resistance between the tracks R = 0.5 Ω: Current: 25mA Voltage: 0.6 V Voltage track: 1.5 V - Test 8: Resistance between the tracks R = 0 Ω: Current: 25mA Voltage: 0.8 V Voltage track: 0.8 V - Test 9: Resistance between the tracks R = 3 Ω: Current: 25 mA Voltage: 1.1 V Voltage track: 2.9 V - Test 10: Resistance between the tracks R = 6 Ω: Current: 25 mA Voltage: 1.3 V Voltage track: 3.6 V - Test 11: Resistance between the tracks R = 6.1 Ω: Current: 25 mA Voltage: 1.3 V

156

Voltage track: 3.65 V - Test 12: Resistance between the tracks R = 6.2 Ω: Current: 25 mA Voltage: 1.3 V Voltage track: 3.7 V The relay switched - Test 13: Resistance between the tracks R = 8 Ω: Current: 25 mA Voltage: 1.4 V Voltage track: 3.75 V

The switch from energized (not obstacle in the track) to not energized (obstacle in the track) occurs with resistances lower than 1.1 Ω, corresponding with voltage between the rails lower than 2 V. The switch from not energized (obstacle in the track) to energized (no obstacle in the track) occurs with resistances higher than 6.1 Ω, corresponding with voltage between the rails higher than 3.65 V. This measure validates the previous ones in a new location.

7.2.2 Scenario 2 – External interference

Simulation results The following table shows the simulated currents measured with the P1 probes, with the influence of the lightning for the same states taken in the field. Resistance between rails Current en P1 0 Ohms between the rails 1.84 Amps 0.5 Ohms between the rails 1.61 Amps 1 Ohms between the rails 1.46 Amps 1.4 Ohms between the rails 1.33 Amps 1.5 Ohms between the rails 1.31 Amps 1.6 Ohms between the rails 1.28 Amps 1.8 Ohms between the rails 1.23 Amps 2 Ohms between the rails 1.22 Amps

157

4 Ohms between the rails 0.93 Amps 4.4 Ohms between the rails 0.91 Amps 4.6 Ohms between the rails 0.88 Amps 4.8 Ohms between the rails 0.86 Amps 4.9 Ohms between the rails 0.85 Amps With the train in the track 2.6 Amps No connection between the rails 0.9 Amps Table 19. List of currents measured in the probe P1 with the influence of the lightning.

In the same way, the simulated voltages measured with the P11 and P12 probes with the influence of lightning for the same states taken in the field are shown in the following table. Resistance between rails Current en P1 0 Ohms between the rails 0.7 Volts 0.5 Ohms between the rails 1.52 Volts 1 Ohms between the rails 2.13 Volts 1.4 Ohms between the rails 2.55 Volts 1.5 Ohms between the rails 2.58 Volts 1.6 Ohms between the rails 2.66 Volts 1.8 Ohms between the rails 2.81 Volts 2 Ohms between the rails 2.95 Volts 4 Ohms between the rails 3.89 Volts 4.4 Ohms between the rails 4.02 Volts 4.6 Ohms between the rails 4.09 Volts 4.8 Ohms between the rails 4.15 Volts 4.9 Ohms between the rails 4.14 Volts With the train in the track 0.52 Volts No connection between the rails 4.5 Volts Table 20. List of voltages measured by the P11 or P12 probes with the influence of the lightning.

Location and conditions of the lab experiments

In this case study, we conducted two different experiments to validate the robustness against lightning strike failures: 1. Behaviour of the isolated chokes.

158

2. Behaviour of the relays with the introduction of the lightning into the I-rail in the whole track circuit section.

Results of experiment 1 These are the tests used to validate this experiment. Test 1. Choke configured in series, power supply 2 kV and rolled resistance 33 Ω. The peak to peak voltage is 1.98 V (multiplied by a factor x1000 in the probe), close to the 2kV emulated lightning strike, and the duration from zero to peak (rise peak duration) is 110 ns.

Figure 110. Test 1.

Test 2. Choke configured in series, power supply 3 kV and rolled resistance 33 Ω. The first peak is 2.92 V (multiplied by a factor x1000 in the probe), once again close to the 3kV emulated lightning strike, and the rise peak duration is 130 ns. In this case, the second peak is later in time because the resonation produces a virtual peak to peak voltage value of 3.8 V.

159

Figure 111. Test 2.

Test 3. Choke configured in series, power supply 5 kV and rolled resistance 3 Ω. The peak voltage value is 1.58 V and the peak to peak voltage is 2.3 V. With this lower resistance, the rise in peak duration is 5.3 ns.

Figure 112. New rolled resistance with resistance value of 3 Ω.

160

Figure 113. Test 3.

Test 4. Choke configured in series, power supply 10 kV and rolled resistance 3 Ω. The first peak is 1.16 V, the peak to peak voltage is 2.12 V and the rise in peak duration is 4.8 ns.

Figure 114. Test 4.

Test 5. Choke configured in series, power supply 15 kV and rolled resistance 3 Ω.

161

The voltage peak is 1.2 V, peak to peak voltage is 1.86 V and the rise in peak duration is 5.4 ns.

Figure 115. Test 5.

Test 6. Choke configured in series, power supply 20 kV and rolled resistance 3 Ω. The voltage peak is 4.46 V, the peak to peak voltage is the same since it starts at zero, and the rise in peak duration is 4.8 ns.

Figure 116. Test 6.

162

Test 7. Choke configured in series, power supply 30 kV and rolled resistance 3 Ω. The voltage peak is 5.62 V, the peak to peak voltage is the same since it starts at zero, and the rise in peak duration is 20.4 ns.

Figure 117. Test 7.

Test 8. Choke configured in series, power supply 30 kV and rolled resistance 0.8 Ω. The voltage peak is 1.54 V, the peak to peak voltage is the same since it starts at zero, and the rise in peak duration is 11.2 ns.

Figure 118. New rolled resistance with a resistance value of 0.8 Ω.

163

Figure 119. Test 8.

Test 9. Choke configured in parallel, power supply 10 kV and rolled resistance 0.8 Ω. The voltage peak is 1.64 V, the peak to peak voltage is the same since it starts at zero, and the rise in peak duration is 18.4 ns.

Figure 120. Test 9.

164

Test 10. Choke configured in parallel, power supply 30 kV and rolled resistance 0.8 Ω. The voltage peak is 2.14 V, the peak to peak voltage is the same since it starts at zero, and the rise in peak duration is 12.8 ns.

Figure 121. Test 10.

Test 11. Two chokes in parallel, both in parallel configuration (i.e. 4 coils in parallel), power supply 10 kV and rolled resistance 0.8 Ω. The voltage peak is 1.96 V, the peak to peak voltage is the same since it starts at zero, and the rise in peak duration is 11.2 ns.

165

Figure 122. Adding a second choke, also in parallel configuration (4 coils in parallel)

Figure 123. Test 11.

Test 12. Two chokes in parallel, both in parallel configuration (i.e., 4 coils in parallel), power supply 20 kV and rolled resistance 0.8 Ω. The voltage peak is 1.72 V, the peak to peak voltage is the same since it starts at zero, and the rise in peak duration is 9.4 ns.

166

Figure 124. Test 12.

Test 13. Two chokes in parallel, both in parallel configuration (i.e., 4 coils in parallel), power supply 30 kV and rolled resistance 0.8 Ω. The voltage peak is 2.48 V, the peak to peak voltage is the same since it starts at zero, and the rise in peak duration is 9.4 ns.

Figure 125. Test 13.

167

Figure 126. Test 13 with a longer time scale.

Results of experiment 2 In these tests, the 4 channel high voltage oscilloscope used was the Agilent Technologies InfiniiVision MSO-X 2024A Mixing Signal Oscilloscope (200 MHz, 2GSa/s). It was possible to get digital graphs by connecting to a laptop with the software data collector Octave and using a simple code: plot((0:size(d,1)-1) × xinc, O.d(:,1)) where xinc is the time per sample. Not all are shown because some are irrelevant. Because we needed to have the correct resolution to observe the behaviour of each channel, we selected the following time scales for the various channels: 20 and 500 μs/div for CH1 (high voltage probe), 2 μs/div for CH2 (photodiode) and 50 μs/div for CH3 and CH4 (secondary of the relays). There was only one high voltage probe available so we used it to perform the experiments twice, first to measure the I-rail and second to measure the exit of the choke (input of the relay). We tried using it another high voltage probe, but its AC accuracy was only 60 Hz, too slow to measure the spark (around 3 μs). We took measurements from 5 kV to 30 kV, but this licentiate restricts itself to the most representative: 30 kV emulating the worst case scenario and 15 kV where the lightning is either weaker or not directly hitting the system. Test 1. 15 kV, high voltage probe connected to the I-rail and DC battery circuit open.

168

Figure 127. CH1 (high voltage probe) 20 μs/div.

Figure 128. CH1 (high voltage probe) 500 μs/div.

169

Figure 129. CH2 (photodiode) 2 μs/div.

Figure 130. CH3 (secondary of the relay 1) 50 μs/div.

170

Figure 131. CH4 (secondary of the relay 1) 50 μs/div.

Last two Figures 130 and 131 are saturated, but still useful to see the reaction of the relays. Similar figures were obtained from the next tests, so there won’t be in this licentiate because they are not relevant enough. Test 2. 15 kV, high voltage probe connected to exit of the choke (input of the relay) and DC battery circuit open. Test 3. 15 kV, high voltage probe connected to the I-rail and DC battery circuit closed. Test 4. 15 kV, high voltage probe connected to the exit of the choke (input of the relay) and DC battery circuit closed. Test 5. 30 kV, high voltage probe connected to the I-rail and DC battery circuit open. Test 6. 30 kV, high voltage probe connected to the exit of the choke (input of the relay) and DC battery circuit open. Test 7. 30 kV, high voltage probe connected to the I-rail and DC battery circuit closed.

171

Test 8. 30 kV, high voltage probe connected to the exit of the choke (input of the relay) and DC battery circuit closed.

7.2.3 Scenario 3 - Two track circuit sections

Simulation results In this model, composed of two successive track circuit sections, the resistance R3 is introduced between the rails of section 1 and a new resistance R4 isolating the I-rail section from section 2 is added as a problematic joint. We used two more probes, P15 and P16, in the secondary of both relays of section 2 to determine when the commutation of these relays is produced. The first test used a huge R4 as a working insulated joint and R3 lower than 1.4 Ohms. In this test, only section 1 detects an obstacle on the track. The second test used a huge R4 as a working insulated joint and R3 bigger than 4.9 Ohms. In this test, neither section detects an obstacle on the track. The third test reduced the R4 to zero as a short-circuit between the I-rails sections as a broken insulated joint and R3 lower than 1.4 Ohms. In this test, both track circuit sections detect an obstacle on the track. The fourth test reduced the R4 to zero as a short-circuit between the I-rails sections as a broken insulated joint and R3 bigger than 4.9 Ohms. In this test, both sections detect an obstacle on the track because of the safety configuration inverting the polarization of successive track circuit sections. Finally, using our model we calculated the minimum resistance that ensures the correct insulation. The result was 2 Ohms. Thus, every insulation joint with an equivalent resistance over 2 Ohms will ensure the correct behaviour of the track circuits, in terms of insulation.

Experiment results The results of the 4 possible scenarios were expected and are summarized below. First test: Well working IRJ and obstacle between rails. Result: Only the first section shows occupied track section state. Second test: Well working IRJ and no obstacle between rails. Result: None of the sections detects anything.

172

Third test: Broken IRJ and obstacle between rails. Result: Due to safety reasons, both sections detect an obstacle on the track because of the configuration inverting the polarization of two successive track circuit sections. Fourth test: Broken IRJ and no obstacle between rails. Result: Due to safety reasons both sections detect an obstacle on the track because of the configuration inverting the polarization of two successive track circuit sections. Finally, the minimum resistance that ensures the correct insulation between track circuit sections is calculated at 2 Ω.

7.3 Limitations of the current standards

The flexibility of the EN50121-X: 2006 series of standards could be seen as an advantage, as it is easier to obtain the conditions for the tests to be done. However, new vehicles tested according to EN 50121 may have EMC problems when introduced onto existing networks. It is also possible, due to the requirements of this standard that some equipment (or element of the new rolling stock) is over-engineered unnecessarily. It may be necessary to define a unique test site, test procedure and limits for the different types of rolling stock and various infrastructure arrangements, taking into consideration national rules and differences. This would be best served by an overarching standard coupled with more specific country standards, as is currently the case. The drawbacks to all standards mentioned in this licentiate would need to be addressed, however. The objective must be, as with other product families, that two laboratories “A” and “B” would test to different parts of the EN 50121 standard at separate times and separate locations and obtain the same result. Yet the uncertainties in testing, particularly the on-site testing of the train as a whole, may make this difficult. As regards on-site testing, more effort is needed to separate transient and continuous emissions; also the emission limits perhaps need re-visiting. There should be some attempts to make the standards more future-proof, as a small change in this stage of development is much less costly than at a later date. An example is the frequency range: the extra effort involved in increasing the range for emission measurements to 4 or 5GHz (to catch future higher frequency technologies) instead of 2.6GHz (which encompasses all current technology) is small. Particular attention should be paid to the situation where new rolling stock is introduced to an older network, as this is extremely common. In this, there should be provision for the more up to date signalling systems and telecommunication designs such as digital broadcasting. Safety should be

173

referenced but kept as a separate standard, and an additional test of the fields inside the train should be investigated. This list highlights the issues with the standards that can be addressed in future work. 1. On board emissions testing: An addition to EN50121-3-1 involving on- board emissions would be very useful to help EMC of GSM and GSM-R and to inform medical and EMF standards. 2. Categorising and measuring transient emissions: Neither conducted nor radiated transient EM energy is accounted for in any EMC standards. Additions to the various standards need to be investigated to reduce transient production and boost immunity to transients. 3. Neutral sections: These have been identified as an area not treated correctly by the current EMC standards, along with other infrastructure design aspects that can cause issues. 4. Radiated emissions limits: As identified in case studies and examination of the standards, the emission limits stated in EN50121-X can be interpreted as too high, allowing interference on radio services at long distances. 5. Frequency limits: The frequency range specified by EN50121-2 in the tests to be carried out is insufficient at both the upper and lower ends. 6. Conducted emissions: These emissions are mainly from substations or induced by passing trains; their categorisation and measurement should be included in EN50121-2. 7. Overhead line resonances: These are not covered but it would be useful to categorise them in a measurement campaign. 8. Capacity to handle future issues: One of the drawbacks identified in this licentiate is to the lack of aid in testing up to date systems (e.g. immunity of the GSM-R system). This should be investigated to ensure the standards are as up to date as possible. 9. Specific typical worst case scenarios. As a result of addressing the points in both sections of this conclusion, once a train has been assessed and has passed the EMC tests, different customers, independent of their country, should all accept the item has been tested to a level that ensures compatibility with their railway system.

174

8 CONCLUSION

8.1 Conclusions of the MATLAB model of the infrastructure

The infrastructure model has been tested in the presence and the absence of a train, and logical results seem achievable. Idle currents when the track is empty and reactive loads when a train is present may validate the assumptions made in the model. External factors like meteorological events may behave differently and produce anomalies in the model so system robustness should be tested for these issues. For integration with the train model, the Spanish infrastructure is designed as an ideal power source. However, the train is a harmonic current source which connects the catenary with the rails, so the integration of the train model introduces new harmonics to this feeding frequency. This influence is more relevant in the starting of the train, with the presence of another train in the same section (and a different track, something not possible in the Swedish infrastructure, a single track system) or with the occurrence of undesired events which produce external signals as a lightning strike close to the track. These events seldom occur, but their impact on the performance of the infrastructure is relevant.

8.2 Conclusions of the CST model simulations

8.2.1 Scenario 1 - Single track circuit section

The train cannot be modelled as a simple resistance between the rails; it is even more restrictive than a 0 Ohms resistance between the rails. As a result, the track circuit system has high immunity against the non-detection of the train in terms of resistance between the rails.

175

8.2.2 Scenario 2 - External interference

The two tables below compare the simulated currents and voltage values with and without lightning. State of the track Simulated Simulated current current without with lightning lightning 0 Ohms between the rails 1.84 Amps 1.84 Amps 0.5 Ohms between the rails 1.61 Amps 1.61 Amps 1 Ohms between the rails 1.46 Amps 1.46 Amps 1.4 Ohms between the rails 1.33 Amps 1.33 Amps 1.5 Ohms between the rails 1.31 Amps 1.31 Amps 1.6 Ohms between the rails 1.28 Amps 1.28 Amps 1.8 Ohms between the rails 1.23 Amps 1.23 Amps 2 Ohms between the rails 1.22 Amps 1.22 Amps 4 Ohms between the rails 0.93 Amps 0.93 Amps 4.4 Ohms between the rails 0.91 Amps 0.91 Amps 4.6 Ohms between the rails 0.88 Amps 0.88 Amps 4.8 Ohms between the rails 0.86 Amps 0.86 Amps 4.9 Ohms between the rails 0.85 Amps 0.85 Amps No connection between the 0.9 Amps 0.9 Amps rails Table 21. Comparison of currents with and without the influence of lightning measured with the P1 of the CST model.

State of the track Simulated Simulated voltage voltage without with lightning lightning 0 Ohms between the rails 0.7 Volts 0.7 Volts 0.5 Ohms between the rails 1.52 Volts 1.52 Volts 1 Ohms between the rails 2.13 Volts 2.13 Volts 1.4 Ohms between the rails 2.55 Volts 2.55 Volts 1.5 Ohms between the rails 2.58 Volts 2.58 Volts 1.6 Ohms between the rails 2.66 Volts 2.66 Volts 1.8 Ohms between the rails 2.81 Volts 2.81 Volts 2 Ohms between the rails 2.95 Volts 2.95 Volts 4 Ohms between the rails 3.89 Volts 3.89 Volts 4.4 Ohms between the rails 4.02 Volts 4.02 Volts

176

4.6 Ohms between the rails 4.09 Volts 4.09 Volts 4.8 Ohms between the rails 4.15 Volts 4.15 Volts 4.9 Ohms between the rails 4.14 Volts 4.14 Volts No connection between the 4.5 Volts 4.5 Volts rails Table 22. Comparison of voltages with and without the influence of lightning measured with the P1 of the CST model.

As seen in the tables, the results are exactly the same. Therefore, using the chokes, our simulation checked for total immunity to a lightning strike of 30 kA during 300 ns. But in the simulation, the components cannot suffer real effects like be burnt or broken and it is also contrary to what happens in other systems in the railway infrastructure (Morant, 2012).

8.2.3 Scenario 3 - Two track circuit sections

One of the simulation tests reduced the equivalent resistance of the insulated joint to zero as a short-circuit between the I-rails sections and used a resistance between rails in the first section lower than 1.4 Ohms, as a detectable obstacle in the track. The result was as expected. Even without any obstacle in the track, if there is not a good insulation between track circuit sections, both detect an obstacle because of the safety configuration inverting the polarization of successive track circuit sections. Regarding the limits, using our model we calculated the minimum resistance to ensure the correct insulation; this turned out to be 2 Ohms. That means that in terms of insulation, the immunity limit will be provided by an insulation joint with an equivalent resistance over 2 Ohms.

8.3 Conclusions of the validation experiments affecting the TCs

8.3.1 Scenario 1 – Conclusions

In this configuration, the current is 120 mA without an obstacle between the tracks and very close to 25 mA for the various tests with resistance between rails, but the voltage between the rails changes with the t voltage values. This

177

controls the relay switching with resistances lower than 1.1 Ω and higher than 6.1 Ω. On the one hand, we find there cannot be an obstacle or any deteriorated ballast between rails with a resistance lower than 1.1 Ω or a non-existent train will be detected. On the other hand, there cannot be any isolation between the track and the train that increases the resistance to over 6.1 Ω between the rails when the wheels of the train run over them, or the train wouldn’t be detected. These limits have to be taken into account in the design and maintenance of the infrastructure in terms of minimum ballast resistance allowed and the maximum resistance value of the dirt on the rails.

8.3.2 Scenario 2 – Conclusions 1

The representative value is the first peak voltage, since the negative voltage and following peaks are non-relevant due to the resonation. This effect can produce a virtual peak to peak voltage even higher than the power source. Normally, the rise peak duration increases with the resistance and the voltage, but there are some exceptions. The attenuation increases with a smaller rolled resistance, which also produces a shorter duration spark. In addition, with smaller resistances the signal oscillates more. Configuring the choke in parallel, the equivalent resistance is 4 times less, so the voltage peak is significantly higher and the duration is slightly longer. We could check this by comparing tests 8 and 10. In the extreme case (not even in the track) of two chokes in parallel, both in parallel configuration (i.e., 4 coils in parallel), the voltage peak is slightly higher, but shorter in duration, making the interference less harmful for the system. These values lead us to conclude that a high inductance (see Henrio) is needed to attenuate the effect of the lightning to protect the battery and the relay circuits.

8.3.3 Scenario 2 – Conclusions 2

In order to get the switches of the relay to the same resistance values as in the measurement campaign, 1 Ω and 6.2 Ω, the resistances R1 and R2 must be

178

tuned. In our case, the track circuit was designed with a R1 = 3 Ω series resistance for the power supply and R2 = 14 Ω series resistance for the relays. The total equivalent resistance of the power supply is around 0.2 Ω of the battery in series with a 3 Ω resistance. The maximum current peak tested was 30 kA, producing more than 5 cm sparks, but the resulting power was not high enough to break any of the components. Even if it is supposed to be a symmetrical structure, one side had less resistance and it was possible to see the spark through the configuration connection of one of the chokes. As shown in the graphs, there is a different spark size for the same experiment. It is very difficult to control, but normally it is bigger with a higher power supply. In the case of the smaller sparks, we encountered more noise in the resulting measured signals. The highest was the emulated lightning strike; its measurements were also more reliable. Our main conclusion is that in the Swedish DC track circuits, the relay is slow; none of the tested sparks could make it switch, making it a very robust device against external interferences. An average lightning strike is 300 ns long, but even testing much bigger sparks (around 3 μs, 10 times more) proved insignificant when compared to the delay of the relay (0.8 s to switch to close and 0.3 s to open). As expected, we saw no switching of any of the relays in the various lightning test cases. Further research can build on these experiments by using a known electrical data source to model the track circuits with a simulated lightning strike.

8.3.4 Scenario 3 – Conclusions

To ensure the correct performance of the track circuits, in terms of insulation, an IRJ with an equivalent resistance over 2 Ω is required. If there is no obstacle/train or there is an obstacle/train in only one of the two consecutive track circuit sections, but the IRJ is broken, both track circuit sections will be shown as occupied. Otherwise, only the occupied one will be shown. Finally, because IRJ is a crucial component of the railway system, it is very important to perform good maintenance on this part of the infrastructure to avoid any situation which could lead to train delays, especially as delays have important economic repercussions.

179

180

9 FURTHER RESEARCH

9.1 Improvement of the standards

Further changes to the EN50121 standard, parts 2 and 3-1, and possibly to the EN50238 standard (recently, a great deal of work has been done on the EN 50238 family) may well be necessary. As explained in previous sections, changes need to consider specific typical worst case scenarios, as well as the differences between transient and continuous emissions, and to include other signalling systems such as ERTMS. The standards do not distinguish between transient and continuous emissions. The current measurement technique needs to be improved upon to reduce both types. The result of this failure of the standards is a higher than expected emission from a moving train system. Our main contribution to the improvement of the standards includes a detailed description of the procedure to build the test site, test setup and test procedure for the previously mentioned three worst case scenarios, not covered by the current standards. Frequency range The frequency ranges stated throughout the EN50121-x: 2006 series of standards are inadequate for the full protection of broadcasting services. The emissions from the whole train include test limits from 9kHz up to 1GHz; this is not high enough to protect more recently used services. The immunity limits given in part 4, to protect the signalling and telecommunication systems, go up to 2.5GHz, which is insufficient to encompass the newer 4G networks at 2.6GHz, but sufficient to encompass the lower GSM and radio frequency bands. Transient and Continuous Emissions One of the main outcomes of the 2002 report is the conclusion that the limits given in the EN50121-X: 2000 standard are insufficient to ensure compatibility with existing radio services. Most influences are covered, but the limits are deemed to be insufficient. Since the introduction of the 2006 version of the EN-50121 series of standards, the emission limits stated in part 2 have not changed.

181

The limits given in EN50121-2 are derived from trackside measurements of emissions from moving trains and, as such, are a combination of transient and continuous emissions; the transient emissions (for example, due to pantograph bounce) may be significantly larger in amplitude than the continuous emissions. However, the standards specified in EN50121-2 actually mean that rolling stock manufacturers can legitimately allow continuous emissions to emanate from rolling stock up to these limits. If rolling stock does emit continuously at the appropriate limit, the chance of interference to radio services is greatly increased. The rolling stock rarely emits continuously at a level that causes severe interference to broadcasting services; if that were the case, the number of reported interference cases would be much higher. The fact that reported issues are low suggests emissions interfering with broadcasting services are transient in nature. A problem arises if the rolling stock emits continuously below the limit but high enough to interfere with broadcasting services, something quite conceivable with the current limits. Accordingly, the 2006 standard introduced a stationary emissions requirement to separate the transient and continuous emissions; the stationary test allows the use of the quasi-peak measurement system, which is much more useful for measuring disturbance to radio systems. However this does not solve the issue of separating the transient and continuous emissions for the moving train test. On-board systems, RFID, WIFI and GSM On-board systems operating in the radio frequency region (such as freight RFID and WIFI) come under the heading of “intentional transmitters” and, as such, are not required to meet emissions limits set out in EN-50121. However, there is no mention in the EN50121 series of standards of the interference effects when using these systems. Guidelines on interference from broadcasting services affecting the railway system (i.e. rolling stock immunity) on board the train are severely lacking in the EN-50121 set of standards. There is no provision for protection from mobile phones within the passenger compartment, other than the implicit assumption that any passenger borne equipment will have passed the relevant emission requirements. On-board WIFI is not mentioned at all in the European standards, although in some cases the WIFI frequencies are within an OFCOM (Office of communications) designated unrestricted band, meaning there is no limit on emissions in these bands (OFCOM UK, 2010). However, there is still potential of interference at these unrestricted frequencies and therefore a need for on-board train immunity testing; this is missing from the current standard. There are immunity limits for mobile telephony with respect to interference to signalling and telecommunication systems (in EN50121-4) but not regarding on-board immunity.

182

The signalling and telecommunication standard (EN-50121-4) is set up to allow immunity testing at mobile telephony frequencies, but the standard only requires testing up to 2.5GHz, which does not encompass the new 4G mobile networks at 2.6GHz or 802.11a at 5.2 GHz. Emission limits from the railway system as a whole (EN-50121-2) are applied from 9kHz to 1GHz; there are no guidelines on limits above 1GHz. Therefore, there is a potential for interference onto any system using higher frequencies, such as mobile telephony upper bands, 3G and 4G, and both on-board and station-based wireless systems. Throughout the EN50121 series of standards, emission limits are defined for lower top frequencies than immunity levels, except in the case where immunity limits are not present, for example, immunity of the whole train. Safety Considerations Another aspect related to broadcasting services is the potential interference to safety and emergency services, many of which use radio frequencies to coordinate and organise responses or operations. A clause in the EN50121 –x series states that “sensitive frequencies must be tested”, but it does not give different (lower) limits for these sensitive frequencies. There is no mention of safety in any of the parts of the EN50121. Relation to current EMC standards As the interference to radio services are purely from radiated emissions from the railway environment, there are no other applicable standards available to reduce the interference, and, as such, the EN-50121-x series require the capacity to reduce interference. Equally, immunity to potential field strengths present within the carriages is not included in any part of EN-50121 for the train unit as a whole. EN50121-3-2 applies immunity limits to the apparatus within the rolling stock, referencing the EN 61000 set of standards (BS EN 61000-4- 3:2006 +A 2:2010 Electromagnetic compatibility — Part 4-3: Testing and measurement techniques — Radiated, radio-frequency, electromagnetic field immunity test). To solve these problems, changes to, or updating of, the EN50121-x: 2006 series of standards now in use is needed.

9.2 Conclusions and further steps with the integrated MATLAB model

During the validation of the models, we faced difficulties getting access to the information because of conflicts of interest with third parties not included in the project. For the Spanish case, there was enough information available from the rolling stock (as it was CAF’s design) but limited information from the infrastructure. For the Swedish case, there was enough information available from the infrastructure (as Trafikverket was providing these data) but limited

183

information from the rolling stock (the train used in the test campaigns was Alstom’s design). Therefore, the implementation of the models of the rolling stock and the infrastructure were independently implemented and validated. However, the integrated model of both rolling stock and infrastructure did not succeed because of the problems getting access to the information necessary to implement the models. Considering the difficulties encountered in the first period, we decided to leave the low frequency model and to focus efforts on the high frequency model with CST and compatibility issues. In further research, this model can still be used; if it is possible to get the pending data, the results could be used to improve the high frequency model.

184

REFERENCES

0FELIA, 2013, Trafikverket's Swedish railway failure and work orders database.

1-222 A, 2007, Electromagnetic Compatibility, London Underground.

American Railway Association, 1922. The history of Dr. William Robinson's invention of the track circuit, the fundamental unit which made possible our present automatic block signalling and interlocking systems. Signal Section, the invention of the track circuit, New York.

Banverket, 1999, Dokument Impedanser för KTL 132 kV, 30kV och 15 kV ML.

Banverket, 2002, TV doc. BKE 02/28.

Commission Regulation EC No 62/2006, December 2005.

Deliverable 2 TREND project: Collection of EMI Experiences and establishment of qualitative relationships, 2012.

Deliverable 3 TREND project: Drawbacks of Current Harmonised EMC Approval Tests & Other electromagnetic influences, 2013.

Deliverable 4.1 TREND project: Characteristics of the rolling stock and infrastructure, 2013.

Deliverable 4.2 TREND project: Identification of the rolling stock and the infrastructure electromagnetic characteristics, 2013.

Deliverable 8.1 TREND project: Cross acceptance EMC test setup, test site and test procedure for Swedish DC Track Circuits, 2014.

ERRAC Roadmap WP 01, The Greening of Surface Transport, http://www.errac.org/IMG/pdf/errac_roadmap_energy_deliverable_v10.pdf , 2012 Fellman, C., 2004, Banverket document BT 95088: “Signalteknisk information”, chapter 7.1

Flury, A., 2012, Electrification, http://www.railway-technology.com

185

General Information on Track Circuits, December 1998. Safety & Standards Directorate, Railtrack, PLC. GK/RC0752.

Kraftförsörjningsanläggningar Elektriska krav på fordon med avseende på kompatibilitet med infrastrukturen och andra fordon, http://www.transportstyrelsen.se/Global/Jarnvag/Vagledning/Godkanna nde/BVS_543_19300(2010-03).pdf, 2013-04-11

Lee, H., Lee, C., Jang, G. and S. Kwon, 2006, “Harmonic Analysis of the Korean High-Speed Railway Using the Eight-Port Representation Model”, IEEE Transactions on power delivery, Vol. 21, No. 2.

Microwave Studio, 2013, Computer Simulation Technology (CST), Online: www.cst.com.

Lloyds Register Group, June 2010. “EMC for European Railways”.

Morant, A., Wisten, Å., Galar, D. et al., 2012, “Railway EMI impact on train operation and environment”, Rome EMC Europe seminar.

Nelson, T., 2005, “Electromagnetic interference in Distributed Outdoor Electrical Systems, with an Emphasis on Lightning Interaction with Electrified Railway Network”. Uppsala University thesis.

Neuman, W. L., 2003, “Social research method: Qualitative and Quantitative Approaches”, 5th edition, USA.

Niska S., 2008, “Measurements and Analysis of Electromagnetic Interference in the Swedish Railway Systems”, Doctoral thesis, Luleå University of Technology.

Niska, S., January 2009, “Electromagnetic Interference: A major source of Faults in Swedish Railway”, Journal of Performability Engineering (IJPE)”, Vol. 5, No. 2, pp. 187-196.

OFCOM UK radio frequency allocation chart, 2010, http://stakeholders.ofcom.org.uk/spectrum/spectrum-management/UK- FAT-Table-2010

Patra, A. P., Kumar, U., 2010, “Availability analysis of railway track circuits”, Institution of Mechanical Engineers, Proceedings, Part F: Journal of Rail and Rapid Transit.

186

Slimen, B., 18 December 2009, “Search Procedures Characterization Environment EM Rail Adapted to Context Communications Embedded Systems”, Doctoral Thesis at the University of Science and Technology of Lille (USTL).

Wieland, A., Wallenburg, C.M., 2012. “Dealing with supply chain risks: Linking risk management practices and strategies to performance.”, International Journal of Physical Distribution & Logistics Management, 42(10).

Yin, R. K., 2009, “Case Study Research: Design and Methods (Forth ed.)”, SAGE Publications, London.

Zerbst, U., Beretta, S., Köhler G., et al., 2012, “Safe life and damage tolerance aspects of railway axles – A review”, Engineering Fracture Mechanics.

187

188

APPENDED PAPERS

189

Paper A

Fault detection of Railway EMC problems using MATLAB models

Emilio Rodriguez, Nava Raj Karki and Diego Galar Luleå University of Technology, Division of Operation and Maintenance Engineering Luleå, 97187, Sweden, telephone: +46(0)920491937 [email protected], [email protected], [email protected]

Daniel Valderas CIET and Tecnum, University of Navarra, 20018 San Sebastián, Spain [email protected]

Stefan Niska Trafikverket, Luleå, Norrbotten, Sweden [email protected]

Abstract

The complexity of the railway system increases when more electronics are used. When dealing with railway infrastructure that has not been renewed over the last several years like in Sweden and with new trains equipped with significant electronic components, it can find itself in one of the most relevant railway problems: the intensive emergency stop can be brake triggered, causing prolonged delays with cascading effects. The reason is that the train control central (TCC) could find unexpected signals from the track due to transient electromagnetic fields that interfere in the train control system circuits.

In this paper a Matlab model of the power supply system of the Swedish railway infrastructure is proposed in which it is possible to integrate a train as an active load. The source of these transient electromagnetic (EM) fields is the engine which can be integrated in different position of the track to study the behaviour of the low frequency system from an electrical point of view. The output of the model can be used as an input for a design of an electromagnetic model in higher frequencies. After the design of the model, a measurement campaign in the north of Sweden to validate the model was carried out.

1. Introduction

These days more and more electronics is used in the railway system with the ultimate aim of improving the reliability, availability, maintainability and safety of the system. The increased use of electronics enables the system operators including the locomotive driver, control centre operators, maintenance crews, planning division to obtain the required data and information on the state and location of train, state of railway tracks, presence of maintenance activities on the track, and so on. As the safe and smooth operation of rolling stock depends upon various factors including the reliable functioning of electronics components and systems, there are possible adverse effects also. The signalling system may produce erroneous signal prompting the control and command centre to fail. The electromagnetic emissions may interfere with each other resulting in erroneous outcome. The rolling stock electromagnetic emissions are a major concern for train manufacturers and railway infrastructure operators [1] in Europe and elsewhere. Available harmonized electromagnetic compatibility (EMC) standards (EN50121-2 [2], EN50121-3-1 [3] and EN50121-3-2 [4]) do not completely address interoperability issues caused by rolling stock interferences with signalling systems (GSM-R, BTM, LTM and STM). Moreover, these standards do not cover representative worst-case conditions derived by transients in the rolling stock behaviour typically generated by feeding and track circuits' discontinuities.

This situation causes waste of time and resources for train manufacturers when integrating rolling stocks and signalling systems. And with tested trains in use, problems may still arise occasionally. Then, the technical solutions are not straight forward. The duration of the field testing employed to solve this kind of problems and to go through the certification process may vary between 3 months and 12 months costing between 25 k€ to 1,5 M€ [1]. Further, the railway infrastructure operators suffer the railway infrastructure availability reduction caused by the rolling stock electromagnetic incompatibility with the safety critical signalling systems and may cause an estimated reduction of 10% of the availability in the most crowded lines.

In this context, TREND (Test of Rolling Stock Electromagnetic Compatibility for cross- Domain Interoperability) project has been initiated with the objective of addressing this situation by means of the design of a test setup to enable the harmonization of freight and passengers rolling stock approval tests for EMC focusing not only on interferences with broadcasting services but also on railway signalling systems. It aims to identify and design the cross acceptance test sites on electrified and non-electrified lines that reproduce representative worst case conditions for steady state and transient behaviours. These worst case conditions will be obtained thanks to the modelling of the rolling stock and the rail and feeding infrastructure. The thorough analysis comprises measurement, modelling and safety and availability analysis of the effect of rolling stock’s electromagnetic interferences (EMIs) on the neighbouring systems. The system potentially affected by these EMIs will be completely covered. These are classified in four research areas: spot signalling system (which includes BTM, LTM and STM), track circuit, GSM-R and broadcasting services (which include TV, radio, Freight RFID, WIFI and GSM). This complete physical environment will permit a precise analysis of the EMI coupling model affecting the whole communication systems. The successful design of a test procedure recreates representative worst-case for the rolling stock electromagnetic emissions that could affect interoperability including transient phenomena.

2. The complexity of EMC model in railway

Directive 2008/57/EC on the interoperability of the railway system within the European Community [5] defines the railway as a series of subsystems: · Infrastructure · Control command and signalling · Power

2 · Rolling stock · Operation and traffic management · Maintenance · Telematics.

However, the national railway system in every country consists of two physical parts; the mobile part (rolling stock, telematics) and the static part (infrastructure, control command and signalling and power). The mobile part can be further sub-divided into two categories defined by its power source; electric or non-electric. The electrically powered mobile part must comply with all the physical requirements of the non-electric mobile part e.g. gauge, loadings, platform height and so on but must also be compatible with the electrical systems. Unlike the physical aspects, the electrical aspects do not have an easily defined or constrained interface with the rest of the world and hence they are potentially more difficult to assess.

There are three potential modes of interaction between all electrical systems; these are conduction, induction and radiation. Although, in theory, all three modes take some part in every interaction most interactions are dominated by a single mechanism. However it would not be practical to define compatibility in terms of the pure interactions by asking general questions e.g. “How are induced effects considered in your safety management system?”. Rather the compatibility demonstration is specified between defined parts of the system e.g. between train return current and corrosion of bridge supports. This reduction to specific systems, subsystems and interactions makes a generic definition for cross acceptance extremely problematic even if limited, as in this case, to compatibility between rolling stock and infrastructure (or neighbouring systems).

The international railway community always aspires to explore any possible common consensus between the requirements of individual country and the wider generic phenomena capable of causing similar interactions in all the member countries. Existing assessments may be narrowly defined or even be specific to a single train or infrastructure component.

2.1 Infrastructure

The power supply system is a subsystem of the overall railway system and cannot be designed in vacuum. Therefore, to specify the requirements of the power supply, the starting point is the physical layout of the route, the location of the passenger stations, the topography of the track route including the curves to be encountered and their physical characteristics. This information together with the required acceleration and speed of the train as a function of position along the track can be used, to determine the tractive effort requirement of the train as a function of position.

Different operating scenarios are assumed starting with the number of trains per hour to run, in what station a train should stop, what are the speed limits at various lengths of the route, the required acceleration and deceleration, and so on. All these acts are to determine location of the trains at any instant of time. Having those snapshots, the concerned engineer determines the power requirements at the different points on the tracks (where the trains are). This information provides the interface between the overall

3 system and the power supply system energizing the trains. The specifications of the electrification system constitute a subset of the overall specifications, for example, voltage specifications are specified by international standards such as BS EN 50163 [6] and IEC 60850 [7].

There are two basic components that influence the performance of the railway power supply system:

 Capacity and number of Auto Transformers (AT) and converters: The power requirement of each train on each track and the location of each train is calculated using the speed and the traction effort and therefore capacity and number of the power supply units is established in order to fulfil the demand of the vehicles.

 Power lines. The power lines (wires) are the messenger, the catenary, the static, and the feeder wires; one for each track. In few literatures, the messenger wire and the catenary wire are called the catenary system or the Overhead Catenary System (OCS). The rails are used as return conductors also. The configuration of the system specifies the coordinates, material and diameter of each conductor (horizontal and vertical distances, usually from the centre of the track) and also what type of conductors are to be used.

2.2 Rolling stock

Various types of rail traction used in Europe use some form of power electronics. Diesel engines are less of an issue in that they do not use electric traction and as such the threat emissions (most of which will arise from an alternator are likely to be relatively low frequency. Rectifiers converting DC to AC may introduce higher harmonics from the alternator but these are only likely to be a problem below a few MHzs.

However, from both the radiated emissions and EMI with broadcasting services perspectives, electric traction is more of an issue. Electric traction units receive power through conductors either above or below the train parallel to the track through either a conducting shoe on a separate power rail; or, more commonly, an overhead catenary/pantograph system. In Sweden the overhead arrangement is run at 15 kV AC at 16.7 Hz.

3. Matlab model for the infrastructure

A typical AC railway power feeding system receives the electricity supply at the substation. For technical reasons, like feeding reliability, protection, rotation of phases, and so on, any feeding section is isolated from the others and supplied with only one power substation/converter. A feeding section is typically about 100 kilometres long. Conventionally, the power substation/converter is connected to the feeder transformers 10 kilometres apart from each other.

The detailed scheme of the power supply system in the Swedish railway infrastructure is shown in figure 1.

4

Figure 1. Detailed scheme of the power supply system in the Swedish railway infrastructure after the converter.

There are two major components to be modelled in the infrastructure. The cables feed the system and run along the tracks hanged from the poles. The transformers are located at every certain number of kilometres and assure the power supply to the transmission lines.

3.1 Cable system

The catenary system (C) in the figure below represents the messenger wire and the contact wire. The contact wire (the wire touching the pantograph) should be as horizontal as possible so that the contact pressure between the contact wire and the pantograph is more or less uniform. However, a wire supported at the two ends on poles will assume a catenary shape represented by hyperbolic function. Hence, a wire, called the messenger wire is supported at the two ends at the poles with the help of insulators and the contact wire is attached to the messenger wire by hangers of varying lengths to keep the contact wire as horizontal as possible. Figure 2 shows a schematic diagram of a Swedish pole with ten conductors in a single track configuration. For greater accuracy, each wire is modelled separately and no bundling of conductors was performed. The mutual impedances have been calculated using readily available formula and presented in the tables below. To model the line in SimPower, the mutual inductance element is used. Since there are 10 wires, there are large number of mutual impedances and are connected as shown in figure 3.

This single track has been modelled as a black box that comprises of all conductors between two poles (60 meters distance) and represented as a 10 inputs / 10 outputs system where the lines are interacting according to coupled inductions and mutual capacitances.

5

Figure 2. Swedish Power line pole.

Figure 3. Cable coupling between transformers.

3.2 Transformer

As the autotransformers are used in the Swedish railway infrastructure, it is relevant to present the equivalent electrical circuit used to model the auto transformer as shown in figure 4:

6 Figure 4. Equivalent circuit diagram of the autotransformer.

The whole infrastructure system can now be represented by a single model taking into consideration that series configuration of the blocks corresponding to the cable will come out as an accurate version of the real catenary system.

Figure 5. Matlab model of the power supply system in the Swedish railway infrastructure.

Figure 5 shows the model developed to simulate the railway infrastructure in terms of power supply. Distance between the converter and end of the line is 100 kilometres (section isolated). In between, several auto transformers can be found in the interval of ten kilometres to feed the corresponding sections. These track sections have cable sets (10 input/outputs boxes) connected in series configuration (each box corresponds to 60 meters as a distance between poles).

4. Integration with rolling stock: System of systems approach

The proposed railway infrastructure system modelled with Simulink offers the flexibility integration of train anywhere along the track. This enables the system to check different scenarios when the train is close to the transformer or far and the power transferred is minimum. It also allows to simulate the worst scenarios for transmission lines in terms of signal propagation.

7 The accuracy of the integration is such that the train can be allocated anywhere between two poles, but the train being an active load will unfortunately contaminate the power line with harmonics. One very important verification test was to check the validity of the model for the idle currents in absence of train. See figure 6.

Amps Volts

Secs Secs Amps Volts

Secs Secs

Figure 6. Original and simulated current and voltage between the converter and the AT.

The similarities in both signals partially verifies that the model reproduces real condition in absence of active load what creates optimum initial condition to introduce the rolling stock as a subsystem of the overall railway system.

Once the train load is connected, the signals (power supplied by the converter) are entirely different and there is an obvious phase shift between the voltage and current. Measured and simulated signals don’t differ significantly and, therefore, the model can be assumed as valid under certain boundary conditions.

Figure 7. Schematic diagram of the train.

The outcomes of this integration will be the different power signals produced by the presence of the train in different position along the track. This dynamic response containing harmonics and high power consumption will produce some undesired effects in the surrounding areas affecting electronics systems and components and other stuffs as a consequence of radiated emissions produced by these conducted signals.

Therefore, the previously described model exhibits the capability to replicate the conducted emission along the track in the power systems. The conducted emission may

8 be potentially dangerous for the safety of signalling systems and other electromagnetic sensitive stuff.

5. From conduction to radiation model: Fault detection stage

Computed power signals must be converted in radiated signals in order to check their effects on the neighbouring electronic devices and systems. Matlab/Simulink is limited for that purpose and there are several commercial tools and software available which can provide a high frequency view of radiation.

The software chosen was the CST STUDIO SUITE [8], a 3D design software. Our case of study was the Niemisel Station (Northern Sweden), when we performed a measurement campaign to obtain all the required values to reproduce the same scenario in the model.

This software has demonstrated its ability to provide the most accurate and efficient computational solutions for electromagnetic designs as circuit simulation in a wide range of frequencies. It offers a wide range of EM simulation platform to address design challenges across the electromagnetic spectrum, from static and low frequency to microwave and RF, for a range of applications, including EDA & electronics, EMC & EMI and charged particle dynamics. The main piece of this software’s product range is CST STUDIO SUITE®, capable of providing a complete set of 3D electromagnetic simulation tools, along with a number of related products dedicated to more specific design areas such as cable harnesses, PCBs and EM/circuit co-simulation. These functionalities fit perfectly into the requirements of the work being carried out as part of the consortium. For this purpose, CST CABLE STUDIO™ will be used. This simulation software is dedicated to the three-dimensional analysis of signal integrity (SI), conducted emission (CE), radiated emission (RE), and electromagnetic susceptibility (EMS) of complex cable structures in large electrical systems. It provides powerful import filters from popular MCAD and ECAD tools for smooth integration into the industrial workflow. CST CABLE STUDIO™ (CST CS) is equipped with enhanced visualization capabilities in order to interactively highlight the selected signals or cables in both the 3D graphic view, as it can be shown in the figure 8, and the 2D schematic view.

Figure 8. CST CABLE STUDIO 3D model.

The goal of the model is the study of the radiated emissions, mainly produced by the train and in the connections between the cables. See figure 9 in the next page.

9

Figure 9. Radiated emissions.

6. Conclusions

The paper discusses about the possible erroneous operation in railway system because of EMI of different electrical and electronics systems and has proposed the model of the power supply system in the Swedish railway infrastructure to integrate a train as an active load to carry out simulations. The goal of this fault detection technique is the simulation of interaction between complex subsystems, which comprise the railway system. These systems of systems approaches aim to reproduce faulty scenarios, which are not dependent on the individual functioning of the subsystems but appear when subsystems interact. EMC issues in railway seem to be a feasible technique for NFF (Non Fault Found) techniques where traditional condition monitoring fails due to the lack of knowledge regarding the multivariable interdependence among the systems.

Acknowledgements

The research has received funding from the European Community´s Framework Programme FP7/2007-2013 under grant agreement no. ”285259”. The Consortium consists of CEIT, CAF I+D, CEDEX, IFSTTAR, York EMC Services, Trafikverket and Luleå Tekniska Universitet.

References

1. ERA EMC Report 2010. 2. BSI/CENELEC, Railway applications – Electromagnetic compatibility Part.2: Emission of the whole railway system to the outside world, BS EN 50121-2: 2006. 3. BSI/CENELEC, Railway applications – Electromagnetic compatibility Part 3-1: Rolling stock - train and complete vehicle, EN 50121-3-1: 2006. 4. BSI/CENELEC, Railway applications – Electromagnetic compatibility Part 3-2: Rolling stock - rolling stock apparatus, BS EN 50121-3-2: 2006. 5. Directive 2008/57/EC of the European Parliament and of the Council of 17 June 2008 on the Interoperability of the Rail System within the Community, European Commission Publication L, 2008. 6. BS EN 50163:2004: “Railway applications. Supply voltages of traction systems”. 7. IEC 60850, Railway applications – Supply voltages of traction systems, 2007-2. 8. www.cst.com.

10 Paper B

Simulation of electrical power supply system in railway infrastructure An integration with rolling stock in Sweden

Emilio Rodríguez Martínez, Diego Galar Pascual Division of Operation, Maintenance and Acoustics Nava Raj Karki LTU (Luleå University of Technology) Department of Electrical Engineering Luleå, Sweden T.U. (Tribhuvan University) [email protected], [email protected] Kathmandu, Nepal [email protected] Stefan Niska Trafikverket Luleå, Sweden [email protected]

Abstract—A Matlab model of the electrical power supply system energizing the trains. The specifications of the system of a 15 kV AC and 16.7 Hz, the most common in the electrification system constitute a subset of the overall Swedish railway infrastructure, is proposed. Within the validated specifications, for example, voltage specifications are specified models of rolling stock and infrastructure an integrated model is by international standards such as BS EN 50163 [1] and IEC implemented. The train can be integrated in different position of 60850 [2]. the track to verify the behaviour of the supply system from an electrical point of view. Its output can be also used as an input for The international railway community always aspires to a design of an electromagnetic model in high frequencies. explore any possible common consensus between the Specifically, our aim is to identify how the rolling stock and requirements of individual country and the wider generic infrastructure, mainly in the low frequency, affects to other phenomena capable of causing similar interactions in all the components due to the harmonic currents on the track. In a first member countries. Existing assessments may be narrowly approach the model is only intended to cover stationary defined or even be specific to a single train or infrastructure conditions, not transients. After the design of the model, a component. measurement campaign in the north of Sweden to validate the Directive 2008/57/EC on the interoperability of the railway model was carried out and compared with our results. These measures were also used in the implementation of a real source. system within the European Community [3] defines the railway as a series of subsystems: Index Terms— Railway, power supply, EMI impact, train • Infrastructure operation. • Control command and signaling • Power I. INTRODUCTION • Rolling stock • Operation and traffic management The specification of power supply characteristics is based • Maintenance on the physical layout of the route, the location of the passenger stations, the topography of the track route including • Telematics. the curves to be encountered and their physical characteristics. However, the national railway system in every country This information together with the required acceleration and consists of two physical parts: the static part (infrastructure, speed of the train as a function of position along the track can control command and signaling and power) and the mobile part be used to determine the tractor effort requirement of the train (rolling stock, telematics). as a function of position. All this subsystems must be independent on the power Different operating scenarios are assumed starting with the supply and can be tested in our model. number of trains per hour to run, in what station a train should Regarding the first one, it is needed to study the capacity stop, what are the speed limits at various lengths of the route, and number of Auto Transformers (AT) and converters. The the required acceleration and deceleration, and so on. All these power requirement of each train on each track and the location acts are to determine location of the trains at any instant of of each train is calculated using the speed and the traction time. Having those snapshots, the concerned engineer effort and therefore capacity and number of the power supply determines the power requirements at the different points on units is established in order to fulfil the demand of the vehicles. the tracks (where the trains are). This information provides the The power lines are the ten wires detailed in the next interface between the overall system and the power supply sections. In few literatures, the messenger wire and the catenary wire are called the catenary system or the Overhead Catenary System (OCS). The rails are used as return conductors also. The configuration of the system specifies the coordinates, material and diameter of each conductor (horizontal and vertical distances, usually from the centre of the track) and also what type of conductors are to be used. On the other hand, the mobile part can be further sub- divided into two categories defined by its power source; electric or non-electric. The electrically powered mobile part must comply with all the physical requirements of the non- electric mobile part e.g. gauge, loadings, platform height and so on but must also be compatible with the electrical systems. Unlike the physical aspects, the electrical aspects do not have an easily defined or constrained interface with the rest of the world and hence they are potentially more difficult to assess. Figure 2. Generic schematic diagram valid for most of the The rolling stock electromagnetic emissions are a major power supply system railway infrastructures. concern for train manufacturers and railway infrastructure operators [4] in Europe and elsewhere. A typical AC railway power feeding system receives the This situation causes waste of time and resources for train electricity supply at the substation. For technical reasons, like manufacturers when integrating rolling stocks and signalling feeding reliability, protection, rotation of phases, and so on, systems. And with tested trains in use, problems may still arise any feeding section is isolated from the others and supplied occasionally. Then, the technical solutions are not straight with only one power substation/converter. A feeding section is forward. The duration of the field testing employed to solve 100 kilometres long maximum to ensure its correct this kind of problems and to go through the certification performance. Conventionally, the power substation/converter is process may vary between 3 months and 12 months costing connected to the feeder transformers 10 kilometres apart from between 25 k€ to 1.5 M€ [4]. Further, the railway infrastructure each other to ensure the power supply to the transmission lines. operators suffer the railway infrastructure availability reduction These cables feed the system and run along the tracks hanged caused by the rolling stock electromagnetic incompatibility from the poles. with the safety critical signalling systems and may cause an A generic simplified schematic diagram valid for most of estimated reduction of 10% of the availability in the most the power supply system railway infrastructures is shown crowded lines. [5] below. The main blocks in all infrastructures are: the converters, II. GENERIC SCHEMATIC DIAGRAM the transformers (AT or BT) and the transmission lines with Our model corresponds to a track section from the exit of their couplings. the converter to the end of one single track line, the most The Swedish railway supply is a catenary/pantograph typical case in Sweden. See Fig. 1 and Fig. 2. system at 15 kV AC and 16.7 Hz and the rail is connected to ground. The transformer mostly used is the BT (Booster- Transformer), but in all new lines that are building in Sweden it is only the AT (Auto-Transformer) that is used, with a distance around 10 km between them. The required data for this model is: Distances: Between poles, transformers, converters and connections of the rail to ground. Converter: Input and output frequencies and voltages as well as the R and L in series with this source. Transformer: Operating power, voltages and frequency. Regarding the losses, it is necessary to specify the power losses obtained from the open circuit test and short circuit test, the voltage drop obtained from short-circuit test and the magnetization current from open circuit test. It is important to obtain the contact impedances and impedances of the AT. Transmission lines: Voltage levels, coordinates in the pole, shapes/diameters and electrical data, as material conductivity. As commented in the introduction, a Matlab model of the train, Fig. 3, can be integrated in any point of the track between Figure 1. Power supply system from the exit of the converter to rail and catenary to run the simulation and get the resulting the end of the track. signals of the integration. Table 1. Amplitude and phase of the harmonics in current.

Harmonic (16.7 Hz) Amplitude (mA) Phase(Degree) Overtones 1x 427.91 249.45 2x 0.24 297.95

Figure 3. Schematic diagram of the train. 3x 11.48 77.39 4x 0.31 166.48 III. MODELLING OF THE DIFFERENT MAIN BLOCKS. 5x 42.17 209.13 A. Converter 6x 0.49 260.18 The function of the converter it to transform the domestic 7x 25.53 11.10 supply (3 phases, 50 Hz, 70 - 220 kV) to the railways supply (1 8x 0.61 337.60 phase, 16.7 Hz, 15 kV).

As it will be shown in the description of the pole and the Table 2. Amplitude and phase of the harmonics in voltage. transmission lines in the Table 3, there is a positive and a negative feeder and the electric potential is 30 kV voltage. A Harmonic (16.7 Hz) Amplitude (V) Phase(Degree) first good approximation of the converter was a linear Overtones transformer with -15 kV and 15 kV, but after the measurement 1x 28900.00 94.78 campaign we could reproduce a customized source. This customized source was produced by means of the Fourier 2x 23.79 148.58 series of the superposition of the 8 first harmonics of those 3x 2000.00 133.32 signals, given in Tables 1 and 2. The Fig. 4 and Fig. 5 show the 4x 14.00 316.82 commented measures performed in the Swedish railway infrastructure. 5x 874.12 221.14 6x 2.46 52.27

7x 156.50 334.00 8x 3.46 337.60

Fig. 6 and Fig. 7 show the differences between the original sources and the simulated sources, for current and voltage respectively with the train in the track.

Amps

Figure 4. Phase shift between the voltage and the current measured in Swedish railway power supply system before the converter with the train in the track. Time (s) Amps

Time (s)

Figure 6. Measured and simulated current comparison.

Figure 5. Phase shift between the voltage and the current measured in Swedish railway power system before the converter without train in the track.

Volts This application performs the computing necessary to calculate the equivalent parameters of the Auto-Transformer. The inputs for this code are the next characteristics of the Auto-Transformer: power, voltage in the primary, frequency, losses without power, losses in short circuit, the voltage loss and the current Loss. And the outputs are the equivalent Time (s) parameters for a two-coil transformer within the next Volts characteristics:

(1)

Time (s) (2)

Figure 7. Measured and simulated voltage comparison.

(3) B. Transformers They are used to assure the power supply to the transmission lines. As the Auto-Transformers are used in the Swedish railway infrastructure, it is relevant to present the (4) equivalent electrical circuit used to model the auto transformer as shown in the Fig. 8. C. Coupling between transmission lines The catenary system represents the messenger wire and the contact wire. The contact wire (the wire touching the pantograph) should be as horizontal as possible so that the contact pressure between the contact wire and the pantograph is more or less uniform. However, a wire supported at the two ends on poles will assume a catenary shape represented by hyperbolic function. Hence, a wire, called the messenger wire is supported at the two ends at the poles with the help of Figure 8. Electric circuit diagram of the autotransformer. insulators and the contact wire is attached to the messenger wire by hangers of varying lengths to keep the contact wire as For the calculation of its values it was designed the Matlab horizontal as possible. Figure 11 shows a schematic diagram of application Auto-Transformer Parameter Calculator, Fig. 9. a Swedish pole with ten conductors in a single track configuration. For greater accuracy, each wire is modelled separately and no bundling of conductors was performed. This single track has been modelled as a black box that comprises of all conductors between two poles (60 m. distance) and represented as a 10 inputs / 10 outputs system where the lines are interacting according to coupled inductions and mutual capacitances. In the case of the transmission lines, it is needed to model the coupling between them. Fig. 10 shows the block which represents this coupling between the transmission lines from one auto-transformer to the next one. The mutual impedances have been calculated using readily available formula and presented in the tables below. To model the line in SimPower, the mutual inductance element is used. The image method was used to calculate these parameters with a the image at a distance h below the rails' level of 1 meter, so it can be assumed that under h the electrical ground is

really zero. Figure 9. Matlab implementation for the calculation of the autotransformer parameters. 5 Fe(steel) 3000 0 6648 46 “Single rail” 8.5 6 Neg. feeder Al -600 7850 212 8.2 -15kV, 28.5 16.7 Hz 7 Support Al Feeder 600 7850 212 8.2 28.5 +15kV, 16.7 Hz

8 Top feeder FeAl 500 9250 99 5.6 Figure 10. Detail of the coupling block. 22kV, 31 50 Hz On the other hand, proposed design is based on 'The 9 Top feeder FeAl Bothnia Track' design [6], with the characteristics of the cables 0 10116 99 5.6 22 kV 31 and rails numbered in the Fig. 11 are detailed in the Table 3. 50 Hz

10 Top feeder FeAl -500 9250 99 5.6 22 kV 31 50 Hz

Finally we need the impedance of each transmission line to build the resistance matrix directly with the real part of the impedance and the inductance matrix as follows. The catenary has impedance: 0.125 + j0.343 Ω/km [7]. The resistance is directly 0.125 Ω/km and assuming the frequency 16.7 Hz it is possible to obtain an inductance of 3.3 mH/km by means of the following formula:

    where Im(z) represents the electrical inductive reactance and f the frequency 16.7 Hz. Figure 11. Matlab application for the calculation of the In the case of the rail the resistance is R = 4.976 Ω/km and coupling between transmission lines. the self-inductance L = 0.964 mH/km and so on.

It is taken a distance of 10 km between Auto-Transformers Table 3. Characteristics of the transmission lines. and a distance of 450 km from the power source to the

converter. Material/ Conductor conductivity, Using the multi-transmission line (MTL) method, the Number/ X, Y, Area, Radius, 106 S/m mutual inductance for any pair of transmission lines i and j can Voltage mm Mm mm2 mm (theoretical frequency be calculated with the next formula: values) 1   catenary Cu +15kV, 3000 5600 120 6.2  57 16.7 Hz -4 where µ0 = 4 × π × 10 H / km, Dij the distance from the 2 line i to the image of j and dij the distance from the line i to the Carrier line Bronze 3000 7300 70 4.7 line j. The mutual capacitance can be obtained from (6) as: +15kV, 28 16.7 Hz     3  Neg. feeder Al -600 6850 212 8.2 -15kV, 28.5 where c = 3 × 108 m/s and L the mutual inductance. 16.7 Hz ij As in the case of the Auto-Transformer, a Matlab program 4 Al Return line -210 6300 212 8.2 was designed for the calculation of the mutual inductances and 28.5 Ground capacitances. A screen shot of the application is shown in the Figure 11. IV. COMPLETE MATLAB MODEL in the power systems. The conducted emission may be The whole infrastructure system can now be represented by potentially dangerous for the safety of signalling systems and a single model taking into consideration that series other electromagnetically sensitive components, apertures and configuration of the blocks corresponding to the cable will systems. come out as an accurate version of the real catenary system. V. CONCLUSIONS The distance between the converter and the end of the line depends on the number of 10 km long coupling blocks The paper proposes a model of the power supply system in integrated in the model. In between, several auto transformers the Swedish railway infrastructure to integrate a train as an can be found to feed the corresponding sections. These track active load to carry out simulations. The implementation of the sections have cable sets (10 input/outputs boxes) connected in models of the rolling stock and the infrastructure were series configuration (each box corresponds to 60 meters as a independently implemented and validated. distance between poles). The infrastructure model has been tested in presence and The train model is added to the infrastructure model as a absence of train and logical results seem to be achievable. Idle subsystem with two ports. One of the ports is connected to our currents when empty track and reactive loads when train is catenary and the other one to our rail as a real train in the line. present may validate the assumptions made in the model. The proposed railway infrastructure system modelled with External factors like meteorological conditions may produce Simulink offers the flexibility of integration of train anywhere anomalies in the model; therefore, system robustness should be along the track. The train is a harmonic current source which tested regarding these issues in a future work. connects the catenary with the rails. So, the integration of the This influence is more relevant in the starting of the train in train model introduces new harmonics to this feeding the presence of another train in the same section and different frequency. This enables the system to check different scenarios track (uncommon in the Swedish infrastructure which is when the train is close or far to the transformer with the typically a single track system) or with the occurrence of minimum transferred power. It also allows simulating the worst undesired events which produce external signals, as the strike scenarios for transmission lines in terms of signal propagation. of a lightning close to the track. These events are very seldom, The accuracy of the integrated model is such that the train but the impact on the performance of the infrastructure is can be allocated anywhere between two poles, but the train relevant. being an active load will unfortunately contaminate the power ACKNOWLEDGMENT line with harmonics. One very important verification test was to check the validity of the model for the idle currents in the The authors acknowledge and appreciate Trafikverket, the absence of the train. Swedish Transport Administration, for its support and As it can be seen in Fig. 6 and Fig. 7, the similarities in collaboration during the research project especially in the both signals partially verifies that the model reproduces real experimental analysis. condition in absence of active load what creates optimum REFERENCES initial condition to introduce the rolling stock as a subsystem of the overall railway system. [1] BS EN 50163:2004: “Railway applications. Supply voltages of Once the train load is connected, the signals (power traction systems”. supplied by the converter) are entirely different and there is an [2] IEC 60850, Railway applications – Supply voltages of traction obvious phase shift of 180 degrees between the voltage and systems, 2007-2. current. Measured and simulated signals don’t differ [3] Directive 2008/57/EC of the European Parliament and of the significantly and, therefore, the model can be assumed to be Council of 17 June 2008 on the Interoperability of the Rail System within the Community, European Commission valid under certain boundary conditions. Publication L, 2008. The outcomes of this integration will be the different power [4] ERA EMC Report 2010. signals produced by the presence of the train in different position along the track. This dynamic response containing [5] Rodríguez, E., Karki N.R., Galar D., Valderas D., Niska S., “Fault detection of Railway EMC problems using MATLAB harmonics and high power consumption will produce some models”. Conference Paper, CM 2013 and MFPT 2013, Kraków. undesired effects in the surrounding areas affecting electronics [6] Banverket-Projektering, BB10-725 065-01, 2006-06-30. systems and components that can be affected from [7] Banverket-Dokument, Impedanser för KTL 132 kV, 30kV och electromagnetic field as a consequence of radiated emissions 15 kV ML. produced by these conducted signals. Therefore, the previously described model exhibits the capability to replicate the conducted emission along the track

Paper C

                CDQM, An Int. J., Volume 17, Number 2, 2014, pp. 15-26

UDC 625.143:624.04 ID NUMBER 209636620

Safety Issues of Track Circuits – A Hybrid Approach

Emilio Rodríguez1, Víctor Simón1, Diego Galar1 and Stefan Niska2

1 Division of Operation, Maintenance and Acoustics, Luleå University of Technology, University Campus, Porsön, 97187, Luleå, Sweden E-mail: [email protected], [email protected], [email protected] 2 Trafikverket, Swedish Transport Administration, EMC group, Sundsbacken 2-4, 97242, Luleå, Sweden E-Mail: [email protected]

accepted May 23, 2014 Summary The study of railway electromagnetic interference (EMI) seeks to determine the source of the interference or to ensure the correct operation of the equipment within adverse conditions. The complexity of railway system increases when more electronics are used. However a simple DC track circuit is still used in train detection systems in many countries, including Sweden, our case study. Most of the failures reported in the Swedish railway infrastructure are related to the detection system, making this research of interest to the railway community. By searching the Swedish failures report database, 0FELIA, for the most repetitive and probable causes of failures, they were identified three worst case scenarios: low resistance between the rails, external interference as a lightning and iron-powder-bridges in the insulated joint. They were simulated using the software CST STUDIO SUITE® (Computer Simulation Technology Studio Suite), supported by real measurements on site. Measurements followed the current EMC standards and were used to tune and validate the models, resulting in simulations very close to the real measures. Key words: Signalling, track circuit, EMC, EMI impact, train operation, interoperability and reliability.

1. INTRODUCTION

Several European countries already have high-speed trains running across their borders, e. g. Spain, France, Italy and Germany. On certain routes, the trains can reach 350 km/h [7]. In Sweden, however, many trains run at a maximum of 200 km/h even though hundreds of kilometres of track are ready for at least 250 km/h operation [7, 4].

15 A system upgrade to solve the compatibility problems between new trains and old infrastructure is desperately required. For example, railway infrastructure operators suffer from railway infrastructure availability reduction caused by the rolling stock electromagnetic incompatibility with the safety critical signalling systems. But neither the signalling system, the catenary, nor the trains themselves are prepared for this updating. In railway systems, one of the simplest but biggest problems is to identify when a rail track is occupied. Train detections systems, such as the track circuit are used to solve this problem. A track circuit is an electronic/electric device which detects whether there is a train on the track. If so, it warns the rest of the system that the track is occupied. This simple action has a crucial responsibility for travellers' security. If a track circuit fails, a train crash could occur [8]. Although this seldom occurs, some false positives have caused injuries and fatalities; in the Cowan rail accident on 6 May 1990, 6 people died and 99 were injured. In short, a good performance of the track circuit is essential for safe transportation [13]. Unfortunately, track circuit malfunctions cause many delays. A track circuit may detect the presence of a train when the track is empty. Trains do not have access to the empty rail track, and this situation may not be detected for a long time. Even though this situation is not as much a priority as one described above, it can cause a significant loss of time and money. In Sweden, the DC track circuit is most commonly used. Each line is divided into track circuit sections about 1 km long separated by 6 mm thick rubber insulated rail joints (Isolerade Rail, I-rail). The principle of the DC track circuit is the connection of the two rails by the wheels and axle of the railway vehicle, also known as rolling stock, to close the open circuit [2]. The systems used to avoid failures are strengthened with backup supplies, fail-safe relays (Figure 1), opposite supplies in consecutive track circuit sections following the fail-safe principle, chokes, double relay circuits, etc.

Figure 1. Detail of the most used relay in Sweden, model JRK 10470 In the context of failure avoidance, the FP7 project TREND (Test of Rolling Stock Electromagnetic Compatibility for cross-Domain Interoperability) designs a test setup for harmonising freight and passenger rolling stock approval tests for electromagnetic compatibility focusing not only on interferences with broadcasting services but also on railway signalling systems [5]. To identify interferences affecting the track circuit, it was performed a failure review using data from the Trafikverket's 0FELIA database. After the identification of the three most representative worst case scenarios, these are analysed. A thorough study of cross-talk causes of failure in the train

16 detection system revealed the need for a virtual scenario; to produce a wider variety of scenarios that are difficult to measure on site, it was modelled a simulated framework in CST STUDIO SUITE®. Finally, the results were cross-referenced with the simulations to validate the virtual environment. Eventually some of the input parameters in the simulation are dependent on the measurement performed.

2. SWEDISH DC TRACK CIRCUIT WORST CASE SCENARIOS 2.1 Introduction In our case study, a single rail [12] track circuit model is fed by a DC source. The chosen measurements were from a 575 m track circuit section at the Niemisel station, in the north of Sweden. With the collected data it was possible to reproduce the scenarios for the simulations, showed in the next sections. The whole track circuit is divided in sections, each considered a unique block. Track sections are separated by an insulated joint at the rail end. In the Swedish case, a 6 mm thick rubber insulated rail (Isolerade Rail, I-rail) transports information on the track circuit status [11]. The other rail (Svetsade rail, S-rail) is welded without insulated joints and carries the return current of the railway infrastructure. Figure 2 shows the general scheme of the track circuit.

Figure 2. DC Track circuit general scheme

Extra fail-safe actions in the Swedish track circuit include the following: • Relay operating follows the fail-safe principle. If the current through the relay drops for any reason, the system displays an emergency signal. • Relays and power supply are attached to opposite ends to ensure the whole track circuit section is checked; this allows the system to detect rail breaks. • For lengths up to 1800 metres, most of the Swedish DC TCs consist of two relays placed at the ends of the track circuit section, with the power source in one end. Having two relays provides a double check against failure. If there is a contradiction between relays, a danger signal is activated and sent to the TCC (Train Control Central) to activate an emergency brake call. • This power supply in most stations is a rectifier with a DC battery as backup. Circuits are commonly battery-powered at low voltages to protect against line power failures. • If there is an insulation failure between track circuit sections, one circuit could falsely power the next one, reversing the electrical polarity from section to section. In the case of a short circuit of the insulated rail joint, both consecutive relays will show an obstacle on the railroad.

17 2.2 Features 2.2.2 Low resistance between rails In the first case, they were considered two scenarios. As explained, the train detection system may warn that a train is occupying the track when there is no train. The opposite may occur as well, but the possibility that a train is not detected is negligible [9]. In the case of the detection of a non-existent train most failures occur at railway stations. The Swedish DC Track circuit is limited to 1500 metres between the feeding and relay. This limit is set so that diversion and reduction in track voltage between sections should not cause faulty signalling. Some factors influence the diversion and track voltage, causing low resistance between rails and triggering a faulty signal. These influences are: • ballast resistance, • grounding of the I-rail, making the voltage drop below 1.7 V DC (voltage state change), even in the absence of a train, • rail structure, • switches and crossings (Figure 3), • obstacles on the track.

Figure 3. Switch on the track The differences in maximum distance between the feeding and the relay in the Swedish DC track circuits occur because the resistance of the ballast varies at different climatic conditions; thus, the extent of the leakage current differs on track sections (i.e., between the I- and S-rail, over each sleeper). The size of the diversion and track voltage are influenced by several factors: • material of the ballast and its consistency, • ballast humidity, • temperature, • type of sleeper and rail fastenings, • switches and crossings over the distance of the track circuit, • pollution in the ballast, • ballast going up against the rail foot or alternatively not going up. When the track is frozen or completely dry, the dissipation is practically zero. On the other hand, in warmer autumn rain, the ballast resistance will fall to its minimum value and the diversion is therefore at its maximum.

18 For the purposes of our study, it was estimated the minimum resistance value of the ballast as follows: • ballast of gravel, 2 /km track circuit, • ballast of macadam and wooden sleepers, 4 /km track circuit, • ballast of macadam and concrete sleepers, > 7 /km track circuit [6]. An obstacle across the track connecting the two rails can be caused by flooding, massive snowfalls or dropped freight, but this seldom occurs and is easily detected. On the other hand, the undetected real train is a dangerous failure that cannot be allowed. During the last 30 years, this happened only once in Sweden. This rare and special event occurred at Boden Central Station because of massive snowfall. Even with the train occupying the track section, the automatic train detection system showed no presence of a train; the voltage did not drop below the voltage state change because all the wheels of the train were floating over a layer of snow [1]. 2.2.3 External interference The biggest natural problem that can interfere with the railway system is lightning. Cloud-to-ground lightning can cause damage in two ways: by a direct strike or by induction effects resulting from a nearby strike. Occasionally lightning can strike far from an object and still affect it. Indirect strikes can also affect the train detection system, but are not as severe as direct strikes. In this case, the wave is conducted to the object by other means, for instance, conducting systems and power lines. The most relevant properties of the lightning that cause damage are peak current and maximum rate of current change. The largest currents are produced by return peak currents when the struck object presents a resistive load. The typical value for a peak current is about 30 kA. A current with this magnitude entering the earth with a grounding impedance of 10 causes a potential rise of 300 kV and may cause surface arcing. Another lightning property that affects the railway system is the maximum rate of change of current in objects that present inductive impedance, such as wires, earth leads etc. For instance, assuming that 10% of the peak current value with front time 0.3 µs finds its way to the wiring of an electronic device, for an inductive load of 1 mH per metre, the inductive voltage produced in a 10 cm long wire could reach 1000 V. The safety devices used to protect against this kind of interference are the chokes (Figure 4), placed in series with relays to protect them from any disturbance.

Figure 4. Choke with possible double configuration, in series or parallel

19 2.2.4 Iron-powder-bridge in the insulated joint

The insulated rail joint (Figure 5) is a simple but important component of the train detection system. It electrically insulates two consecutive rails and acts as a bond between to form a long track distance. The wheels roll over the insulated joint, milling the rail ends; the physical gap between the two rails becomes smaller each time a train passes, and may create an electrical connection. The problem is a wear out one, but the outcome is a signalling fault, as dictated by the fail-safe principle.

Figure 5. Insulated rail joint Conducting iron-powder-bridges might electrically connect two consecutive track sections. If this is bridged by a metal piece, the voltage in both sections could drop enough to generate an occupied track state on both sections. Magnetic fields make iron particles stick together, building conducting bridges over insulated joints, especially in areas where the traffic is heavy and the wearing of wheels is extreme. Typically this issue does not lead to problems on long distance tracks, but is reported in high volume areas. In such situations, extra maintenance is necessary, as for example, cleaning metallic particles from insulated junctions to avoid short circuits between adjacent I-rail sections. Detectors for flat wheels or hot bearings may be considered as well; these are more insensitive to electromagnetic pulses and unaffected by the closeness of the junctions on the I-rails [3]. 2.3 CST Model 2.3.1 Short circuit between rails As all the worst case scenarios are related to the precise moment when the relay switches its state, a voltage and current sweep is done to determine when the relay switches from one state to another. The track section was measured at two different points: data on the current were collected in the source cable and data on the voltage were collected between rails. When the resistance value was modify, the limits for the sweep were from 0 to 5 with the smallest step of 0.1 . To test the value of the resistance between rails required to switch the state from ON to OFF, it was performed a sweep from 0 to 2 . To check the change of status from OFF to ON, the sweep was from 5 to 4 . Figure 6 shows the 3D design of two consecutive track sections with the connections to the train on the left side.

20 Figure 6. 3D CST model of two consecutive DC track circuit sections

2.3.2 Lightning To simulate external interference, for instance by lightning, the resistance between rails should be removed, since the scenario may not need the presence of a train. As electromagnetic interference, lightning can be understood as a 30 kA current entering the railway infrastructure through the grounding impedance during 300 ns. This can be performed using a CST CABLE STUDIO® task called Transient, applied to an external port connected to the rail. To be imported, the lightning characteristics can be made transient in an ASCII (American Standard Code for Information Interchange) file, as shown in Figure 7. The 3D model of the scenario is like the previous one, but there are two modifications in the schema: R3 is removed and port 1 included.

Figure 7. Introduction of lightning as transient voltage

21 2.3.3 Short circuit between track circuit sections

An important point to highlight in the Swedish railway system is how to prevent one circuit from falsely powering another. This situation could occur in the event of insulation failure. When the electrical polarity from section to section (Figure 8) is reversed in a short circuit event, both consecutive relays will show an obstacle on the railroad and the system will activate an emergency brake call.

Figure 8. Layout of two consecutive track circuit sections To determine whether this method works, a set of tests can be performed. First, a second track circuit section must be added; at the same time, the poles of the battery and the connections of the relays must be changed (Figure 9).

Figure 9. 2D circuit CST model of two consecutive DC track circuit sections

22 To model the insulated joint, another variable resistance (R4) is added. In addition, the variable resistance R3 must be used in the first track section to simulate the presence of a train. They were performed the following tests: First test: configuring a huge value of R4, as a good working insulated joint, and setting a value of R3 low enough to detect a train. Second test: setting a huge value of R4 and configuring R3 high enough to not detect an obstacle on the first track section. Third test: decreasing the value of R4 to zero to simulate a short circuit between the I-rails of two consecutive sections due to a broken insulated joint and giving R3 the same range of values as in the first test. Fourth test: setting the same value of R4 as in the third test and the same value of R3 as in the second test. After all these tests are performed, a minimum value of resistance can be calculated to ensure the correct insulation between rails. 2.4 Validation of the Model A measurement campaign was carried out to fine-tune and validate the model, hoping to get the same switching point in both cases. When it was introduced a variable resistance R3 between the rails, the switching points were the next values in the relay transitions: • OFF to ON: in the step up from 4.8 to 4.9 , implies 1.5 V (Von-off), • ON to OFF: from 1.5 to 1.4 , implies 0.7 V (Voff-on). However, for the new voltage values for the relay characteristics for resistance R1 = 3 and R2 = 7.5 in the model developed in CST, they were used three degrees of freedom to suit the simulation to the real values. This modification was done to meet the requirements imposed by the results obtained in the measurements.

3. RESULTS

Figure 10 and Figure 11 compare the measurement values with the simulation in the virtual environment. In both transitions, from ON to OFF and from OFF to ON, the simulated values of the current are fairly close to the measured ones. It is important to note that the transitions between both states took place for the very same value of resistance between rails that was used to model the obstacle on track. This is a decisive validation of the simulation. The transitions in the simulation occur in the same situation as they do in the real case, and the simulated values follow the same trend as the real ones; thus, the scenario modelled can be considered reliable for simulations.

Figure 10. Comparison of simulated and measured current in the TC

23 There is a greater difference in the voltages than in the currents mainly because of resistances related to imperfections not considered in this study.

Figure 11. Comparison of simulated and measured voltage in the TC The study performed to simulate the lightning showed that this external force does not affect the track circuit system, contrary to what happens in other systems in the railway infrastructure [10]. The chokes placed in series with the relays avoid the transient lightning voltage that affects the proper operation of the train detection system. In Table 1 and Table 2, the orange fields represent when the system displays a danger signal, i.e., an obstacle is on the track. Even under the effect of the lightning, the relay still switches at the same value of resistance between rails. Table 1. Current values with the influence of the lighting

Four tests were performed to check the proper functioning of the insulated rail joint, two with proper insulation and two with a short circuit between track circuit sections. First test: configuring a well working insulated joint and the resistance between rails for a train detecting event; as expected, the result shows that only the first section is occupied. Second test: configuring a well working insulated joint and a non detecting train event; none of the sections detects anything. Third test: modelling a broken insulated joint with an obstacle in one of the sections; as the Swedish DC track circuit safety configuration inverts the polarisation of two successive track circuit sections, this test ends up with both sections detecting an obstacle on the railroad.

24 Table 2. Voltage values with the influence of lightning

Fourth test: configuring a broken insulated joint and no train on the track; again the opposite supply between consecutive track circuit sections safety measures results in both sections detecting an obstacle. Finally, with our model, it was calculated the minimum resistance that ensures the correct insulation, 2 Ohms. That means that every insulation joint with a resistance over 2 Ohms will ensure the correct performance of the track circuits, in terms of insulation between track circuit sections. 4. CONCLUSIONS From our research it was reached the following conclusions. A reliable framework that enables the simulation of scenarios to test the operation of a track circuit was created; this can be exploited in future research. As a train detection system, the track circuit has a simple infrastructure which is robust against most of the interferences. The weakest point of the system is its dependence on the relay. If the relay fails, the fail-safe system will show an occupied track, causing delays in the flow of traffic. The system's efficiency can be increased by improving the track circuit, but this is difficult, due to its simplicity. However, the incorporation of another subsystem, for instance, the wheel axle counter, can complement the detection mechanism. The insulated rail joint is an important component of the infrastructure, as it is the second most likely system to fail. Good maintenance of the track and insulated rail joint is crucial to avoid situations which can lead to train delays with important economic repercussions. These simulations suggest a typical lightning strike directly on the track will not affect the track circuit. Although this research has focused on the track circuit, a deep understanding of the complete railway system is necessary to understand possible threats not just to the track circuit system but to railway operation in general and to be able to integrate subsystems in an infrastructure. Rolling stock is a key element in the interoperability of the railway’s subsystems, strongly related to the causes of abnormal behaviour in the signalling system. Most of the EMIs on track circuits are related to transient produced by the rolling stock or external sources, which should be studied in future research.

25 ACKNOWLEDGEMENT

The research has received funding from the European Community's Framework Programme FP7/2007-2013 under grant agreement No. "285259". The Consortium consists of CEIT, CAF I+D, CEDEX, IFSTTAR, York EMC Services, Trafikverket and Luleå Tekniska Universitet.

REFERENCES

[1] 0FELIA, Swedish railway failures database, Trafikverket. [2] American Railway Association. Signal Section the invention of the track circuit. The history of Dr. William Robinson's invention of the track circuit, the fundamental unit which made possible our present automatic block signalling and interlocking systems. New York, Signal Section, American Railway Association, 1922. [3] Banverket, Dokument: Signalteknisk Information. [4] Bombardier, Transportation Systems [online]. [5] [consulted: 18 of April 2013]. [6] CEIT Centro de Estudios e Investigaciones Técnicas de Gipuzkoa [online]. [consulted: 25 April 2013]. [7] Curt Fellman’s Banverket document BT 95088: “Signalteknisk information”, chapter 7.1. [8] EURAIL.COM B.V. Trains in Europe [online]. [consulted: 28 of April 2013]. [9] Høj NP, Kroger W. Risk analyses of transportation on road and railway from a European Perspective, Safety Sci. 2002; 40(1-4): 337-357. [10] Morant, A.; Wisten, Å.; Galar, D. et al. Railway EMI impact on train operation and environment, September 2012 [11] Nelson, T. “Electromagnetic interference in Distributed Outdoor Electrical Systems, with an Emphasis on Lightning Interaction with Electrified Railway Network”. Uppsala University thesis. [12] Niska, S. “Measurements and Analysis of Electromagnetic Interference in the Swedish Railway Systems” Doctoral thesis, Luleå University of Technology, 2008 [13] UIC 60-60E1, Steel Railway Rail Track standard. [14] University of Sydney News [online]. [consulted: 12 of May 2013].

26 Paper D

EL IMPACTO DE LA COMPLEJIDAD DE LA ELECTRÓNICA EN LA SEGURIDAD DEL SISTEMA FERROVIARIO

Emilio Rodríguez Ingeniero en Telecomunicación investigador en la Universidad de Luleå, Departamento de Ingeniería Civil, Ambiental y de Recursos Naturales, División de Operación, Mantenimiento y Acústica, Suecia.

Víctor Simón Ingeniero en Telecomunicación investigador en la Universidad de Luleå, Departamento de Ingeniería Civil, Ambiental y de Recursos Naturales, División de Operación, Mantenimiento y Acústica, Suecia.

Diego Galar Dr. Ingeniero Industrial catedrático en la Universidad de Luleå, Departamento de Ingeniería Civil, Ambiental y de Recursos Naturales, División de Operación, Mantenimiento y Acústica, Suecia.

Luis Berges Dr. Ingeniero Industrial profesor en la Universidad de Zaragoza, Departamento de Ingeniería de Diseño y Fabricación, España.

Jaime Tamarit Dr. en Ciencias Físicas director del Laboratorio de Interoperabilidad Ferroviaria en CEDEX, España.

RESUMEN:

La complejidad del sistema ferroviario aumenta cuando se utiliza más electrónica. Cuando se introducen nuevos trenes equipados con componentes electrónicos importantes en una infraestructura ferroviaria que no ha sido renovado en los últimos años, como la sueca, puede surgir uno de los problemas más relevantes de ferrocarril: el sistema de parada de emergencia activa los frenos del material rodante, causando largos retrasos con efectos en cascada. La causa es que el TCC (centro de control de trenes) puede detectar señales inesperadas debido a los campos electromagnéticos transitorios que puedan interferir en los circuitos de señalización y control.

En este documento se propone un modelo en Matlab del sistema de alimentación eléctrica de la infraestructura ferroviaria sueca. En particular, se analizan los circuitos de vía, caso de estudio que se abordará en la continuación del proyecto. En el modelo de Matlab es posible integrar, en el punto que se desee, el subsistema que representa al tren como una carga activa y así poder analizar su influencia en el sistema global. De esta manera se estudia el comportamiento en baja frecuencia desde un punto de vista eléctrico, pero dichos resultados pueden también utilizarse como entrada para el diseño de un modelo electromagnético en frecuencias más altas. Tras el diseño del modelo se ha realizado una campaña de medidas en el norte de Suecia a partir de la cual se ha ajustado y comprobado el correcto funcionamiento del mismo.

Palabras Clave: Ferrocarril, alimentación eléctrica, señalización, circuito de vía, EMC, impacto de las EMI, operación, interoperabilidad y fiabilidad del ferrocarril

1.- INTRODUCCION

Todo sistema ferroviario tiene por objetivo maximizar ciertos indicadores entre los que se incluye la capacidad (es decir, el número de trenes por unidad de tiempo que se pueden introducir en el sistema), la fiabilidad, la seguridad, la comodidad, etc. minimizando los tiempos de posesión por parte de los operadores de las distintas secciones de circuitos de vía que forman el entramado ferroviario. Una vez que el tren abandona un tramo debería estar inmediatamente disponible para otro vehículo sin comprometer la seguridad del sistema. De hecho, la infraestructura es tremendamente sensible a fallos en el sistema debido a distintos tipos de peligrosidad que se pueden materializar si estos fallan: desde retrasos en los trenes debido al uso ineficiente de las vías que representarían el nivel más bajo de riesgo hasta una posible colisión entre trenes, como caso más adverso.

La organización del sistema ferroviario, privatizado y fuertemente regulado durante los últimos años, se ha estructurado en gestores de infraestructura y operadores que necesariamente necesitan una interacción fluida y eficiente de sus sistemas para prestar el servicio a pasajeros y mercancías. Sin embargo esta interacción se produce a través de cientos de elementos electrónicos y eléctricos cuyo funcionamiento simultáneo está amenazado por diversos factores entre ellos la EMI, lo que requiere incluir EMC como un objetivo más del sistema ferroviario.

En los últimos años se ha aumentado el uso de electrónica en el sistema ferroviario con el objetivo de mejorar la fiabilidad, disponibilidad y seguridad del sistema. Este incremento permite a los operadores del sistema, incluyendo tanto al maquinista como a los operadores del centro de control y la división de planificación y mantenimiento, obtener los datos y la información sobre el estado y la ubicación del tren, el estado de las vías del ferrocarril, la presencia de actividades de mantenimiento necesarias en la pista, y así sucesivamente. Como la operación segura del material rodante depende de varios factores, incluyendo el funcionamiento fiable de componentes y sistemas electrónicos, también aparecen posibles efectos adversos. Las EMI (interferencia electromagnética) pueden interferir entre sí ocasionando un resultado erróneo, por lo que si no se realizara correctamente la homologación de estos nuevos elementos el tren o el centro de control interpretará erróneamente una señal del sistema de señalización. La principal fuente de emisiones electromagnéticas es el material rodante, suponiendo una gran preocupación para los fabricantes de trenes y operadores de infraestructuras ferroviarias [1]. Los actuales estándares de compatibilidad electromagnética disponibles (EN50121-2 [2], EN50121-3 [3]) no cubren totalmente los problemas de interoperabilidad. Por otra parte, estas normas no se ocupan de las condiciones de los peores escenarios.

A pesar de que los fabricantes de trenes cumplen con la normativa vigente, los problemas surgen al integrarlos en las diferentes infraestructuras, lo que implica que el fabricante ha de validar adecuadamente su producto para el entorno correspondiente. Esta situación desemboca en problemas de seguridad, como paradas de emergencia, que a su vez se ven reflejados en retrasos en la línea con una reducción estimada del 10% de la disponibilidad de las líneas más concurridas. La duración de las pruebas de campo para resolver este tipo de problemas junto con el proceso de certificación puede variar entre 3 meses y 12 meses y cuesta entre 25 k€ a 1.5 M€ [1].

En este contexto, como socios del proyecto TREND (pruebas de compatibilidad electromagnética en material rodante para la interoperabilidad entre dominios) del 7FP (séptimo programa marco), tenemos el objetivo de hacer frente a esta situación mediante el diseño de un sistema de pruebas que permitan, tanto para transporte ferroviario de mercancías como de pasajeros, la homologación de la EMC (compatibilidad electromagnética), centrándose no sólo en las interferencias con los servicios de radiodifusión, sino también en los sistemas de señalización ferroviaria y en concreto en los circuitos de vía.

Fig. 1: Modelo completo con integración del tren.

A partir de dicho modelo se identifican los escenarios que reproducen las peores condiciones, tanto en estado estacionario como comportamientos transitorios, para la actualización los estándares. Estas condiciones más desfavorables se obtendrán a partir de un modelo en Matlab, Fig. (1), que integra las características de la infraestructura ferroviaria, la alimentación eléctrica y el material rodante. El análisis completo consta de una toma de datos, modelado, campaña de medidas para su verificación y análisis de seguridad e interoperabilidad entre los sistemas vecinos. Estos se clasifican en cuatro áreas de investigación: servicios de radiodifusión (que incluyen telefonía móvil, televisión y radio), GSM-R (sistema global para las comunicaciones móviles para ferrocarril), señalización y circuitos de vía. Este entorno físico completo permite un análisis preciso del modelo de acoplamiento de las EMI que afecta a todo el sistema de comunicaciones.

2. PROBLEMAS NO CUBIERTOS POR LOS ESTÁNDARES

Para identificar los posibles problemas que afectan al sistema de ferrocarril es necesario clasificar las amenazas que pueden causar las interferencias electromagnéticas y las potenciales consecuencias de las mismas.

Hay cuatro factores determinantes que causan interferencias: el material rodante, el sistema de alimentación y sistemas electrificados en general, los desperfectos en la infraestructura y los sistemas de comunicación y señalización. Estas amenazas tienen cinco áreas de impacto bien definidas: el propio material rodante, el sistema de señalización, los servicios de GSM-R, servicios de difusión y los circuitos de vía.

Existen dos tipos de mecanismos de acoplamiento distinguidos por sus características. El primero de ellos es la interferencia por radiación, afectando a la señal que cualquier receptor en uso es capaz de captar. El segundo mecanismo está relacionado con los propios aparatos transmisores y receptores, y como pueden ser dañados por las diferentes impedancias del sistema.

2.1 SERVICIOS DE RADIODIFUSION

En este ámbito se están llevando a cabo dos vías de estudio diferentes. En la primera se identifican aquellos problemas asociados con el uso de los transmisores y receptores en el entorno del ferrocarril y las mayores fuentes de interferencias.

Se han detectado tres fuentes principales de emisiones electromagnéticas: las producidas en los mismos trenes por distintas causas como pueden ser los mecanismos de tracción, sistemas de suministro de energía eléctrica o los motores de conmutación por corriente continua; la telefonía móvil pública, que con su creciente evolución demanda una gran parte del espectro electromagnético; y por último la difusión digital de radio y televisión.

Ciertos estándares especifican los límites de emisiones que el ferrocarril debería tener, pero están basados en medidas que cubren tanto los efectos producidos por las fuentes de energía continua como los transitorios, pero dichos estándares, EN 50121-2 [2] y EN 50121-3 [3], no especifican los distintos límites para uno u otro caso.

Para abordar el problema de la creciente demanda de espectro por parte de la telefonía móvil, se está adoptando el uso de las llamadas “picocélulas” que permiten un mayor aprovechamiento del espectro debido al a reutilización de frecuencias. Respecto la difusión digital de radio y televisión, la ya conocida técnica de OFDM (multiplexación ortogonal en el dominio de la frecuencia) ofrece un buen escudo ante posibles interferencias, pero en cuanto a emisiones hacia otras bandas de frecuencia cercanas no supone una gran mejora.

La segunda propuesta de estudio consiste en analizar dos problemas puntuales: las fuertes distorsiones sufridas en pantallas y monitores en las zonas metropolitanas cercanas, y la interferencia audible en ondas de difusión de onda corta y onda media. Tras un amplio estudio se ha concluido que la normativa respecto a la identificación de peligros relacionada con este tema era insuficiente, ocasionando un mal diseño de la compatibilidad electromagnética.

2.2 GSM-R

GSM-R pertenece al estándar ERTMS/ETCS (sistema europeo de gestión del tráfico ferroviario / sistema de control ferroviario europeo) asociado con Eurobaliza. Gracias a GSM-R la información de señalización llega a la unidad de señalización, emplazada a bordo del tren, con más fluidez facilitando una densidad alta de tráfico, con más velocidad y mayor seguridad.

Las principales interferencias que se sufren en este campo se deben, o bien a los transitorios eventuales que ocurren en las comunicaciones GSM-R; o por las interferencias constantes debidas a la presencia de las transmisiones públicas GSM (sistema global para las comunicaciones móviles) en canales adyacentes. Las primeras se producen por las radiaciones emitidas por la catenaria y el pantógrafo, captadas por la antena GSM-R.

Para analizar este problema, se procedió a detectar estos transitorios con una antena a bordo del tren, para posteriormente poder realizar una aproximación estadística al caso real para su estudio en el laboratorio.

2.3 SEÑALIZACION

Las corrientes armónicas que fluyen por el bucle creado por la fuente de alimentación – material rodante – fuente de alimentación pueden acoplarse al receptor de señalización y crear problemas de EMI.

Las principales fuentes de interferencias electromagnéticas se producen por los cambios de estado en el pantógrafo, operaciones de circuitos automáticos y las pérdidas de contacto entre el pantógrafo y la catenaria. Sin embargo, dichas interferencias electromagnéticas no se pueden identificar mediante los procedimientos desarrollados en los estándares EN 50121-2 [2] o EN 50121-3 [3].

Otro parámetro a tener en cuenta es la SNR (relación señal a ruido) disponible en el momento de la medida. La señal recibida en el tren procedente de la baliza / cable / punto de radiación debe de ser suficientemente distinguible por encima del ruido. Esta relación señal a ruido está condicionada por el ruido medioambiental y por el tiempo y la franja horaria de la captura de la señal. Este segundo factor condicionante es crucial en sistemas de señalización como el ERTMS, donde se usa una modulación FSK (modulación por desplazamiento en frecuencia) con dos frecuencias, dependiendo del tiempo y las características de la digitalización y la demodulación.

2.4 CIRCUITOS DE VÍA

Este sistema de detección de trenes, Fig. (2) y (3), aunque sencillo, es esencial para asegurar la seguridad del sistema ferroviario. Un fallo en la detección de un tren en un tramo de vía puede concluir en una colisión entre trenes con nefastas repercusiones [4]. Aunque esto ocurre pocas veces, algunos falsos positivos han causado lesiones y muertes, y en el accidente ferroviario Cowan, el 6 de mayo de 1990, 6 personas murieron y 99 resultaron heridas [5]. En definitiva, un buen desempeño del circuito de vía es esencial para el transporte seguro [6]. Es por eso que hay que prestar especial atención en el mantenimiento de este sistema para que no se vea afectada la interoperabilidad.

A parte de los estándares de compatibilidad electromagnética comentados anteriormente; EN 50388 [7], EN 50238 [8] y EN 50122 [9] aportan información acerca de cómo deberían analizarse los problemas de interferencia en los circuitos de vía. Aunque dichos documentos cubren ciertas condiciones básicas para desarrollar experimentos de análisis para condiciones estándar, adolecen en falta de información y procedimientos para tratar los peores casos.

Diversos puntos débiles deberían ser mejorados en estos estándares. Cuando no están totalmente definidas las condiciones de los diferentes escenarios, la generalización de los resultados de las medidas afecta a la incertidumbre de los resultados.

La meteorología puede tener un gran impacto en la variación de los resultados. No obstante, en una batería de pruebas deben ser grabadas tanto la batería de test como las condiciones meteorológicas. Sería beneficioso tener un conjunto de límites de emisión o una clasificación de la variación de los resultados en función de las condiciones meteorológicas. Parámetros como la humedad, la temperatura o la constante dieléctrica del suelo pueden sufrir grandes cambios en el mismo día en lugares con condiciones meteorológicas muy adversas.

Por otra parte, los escenarios más complicados están estrechamente relacionados con las repercusiones de los transitorios; tanto aquellos provocados por el paso del tren por la vía; como los causados por las fuerzas de la naturaleza, como los rayos. Las grandes longitudes de cable pueden actuar como antenas y captar los pulsos electromagnéticos de diversas fuentes. Para mitigar el impacto de los picos de voltaje de los transitorios, se hace necesario una correcta protección de los elementos y una puesta a tierra adecuada. A pesar de todas estas medidas preventivas, el daño provocado por los transitorios puede ocasionar interrupciones en el servicio, pérdidas de imagen en los monitores o producir falsas alarmas debido al exceso de calor en los rodamientos.

Fig. 2: Esquema eléctrico del circuito de vía.

Fig. 3: Foto del punto de la vía entre dos secciones de circuitos de vía y sus componentes.

3. MODELO DEL SISTEMA FERROVIARIO

La directiva 2008/57/EC de interoperabilidad en el ferrocarril en el marco de la Comunidad Europea [10] define el sistema ferroviario, como el sueco que se puede ver en la Fig. (4), como la siguiente lista de subsistemas: · Infraestructura · Control y señalización · Alimentación eléctrica · Material rodante · Operación y gestión del tráfico · Mantenimiento · Telemática

Nuestro primer objetivo es modelar la alimentación eléctrica de un único tramo de vía la de la infraestructura sueca, desde el convertidor hasta el final de la línea, como se muestra en la Fig. (5). [11]

Fig. 4: Esquema detallado de una sección de la alimentación eléctrica de la infraestructura sueca.

Fig. 5: Esquema detallado de una sección de la alimentación eléctrica de la infraestructura sueca. 3.1.- ALIMENTACIÓN ELÉCTRICA

Las especificaciones del sistema de electrificación constituyen un subconjunto de las especificaciones generales, por ejemplo, las especificaciones de voltaje son especificadas por normas internacionales como EN 50163 [12] y IEC 60850 [13].

En nuestra investigación se estudia la infraestructura sueca y española, por lo que en una primera instancia se ha diseñado un esquema común para ambas tal y como se muestra en la siguiente Fig. (6), que corresponde a una sección de vía desde el convertidor hasta el final de la línea. El objetivo de este modelo es la integración con el modelo del tren validado por el fabricante para obtener las señales resultantes de una simulación conjunta.

Fig. 6: Esquema general de la alimentación eléctrica.

Los bloques principales de toda infraestructura son los convertidores, los transformadores (AT, Auto-Transformador, o BT, transformador amplificador, dependiendo de la tecnología) y las líneas de transmisión con sus acoplamientos correspondientes. Para su diseño es necesaria la siguiente información:

· Convertidor: Frecuencias y voltajes de entrada y de salida y la resistencia e inductancia serie. · Transformador: Potencia, voltaje y frecuencia de funcionamiento. En cuanto a las pérdidas, es necesario especificar las pérdidas de potencia en el test circuito abierto (sin alimentación), las pérdidas de potencia en el test de cortocircuito, la pérdida de voltaje en el test de cortocircuito y las pérdidas de intensidad en el test de circuito abierto. Y por último, también son relevantes las impedancias de las conexiones eléctricas e impedancias del mismo. · Las líneas de transmisión: Voltajes, distancias (configuración del poste), secciones y la conductividad del material para obtener el acoplamientos entre ellos. · Distancias: Entre postes, entre transformadores (d), entre convertidores y cuándo se realizan las conexiones de los postes a tierra.

La configuración de los postes es la siguiente se puede observar en la Fig. (7). [14]

Fig. 7: Configuración poste sueco siguiendo la norma de vía del "Golfo de Bothnia".

Teniendo en cuenta lo anterior, se obtiene el siguiente esquema de bloques que se diseñará en Matlab con la herramienta Simulink. Los bloques Z representan las impedancias de diferentes pares del circuito.

3.1.1. CONVERTIDOR

La función de esta parte del circuito de alimentación es transformar el suministro eléctrico doméstico (3 fases, 50 Hz, 70 - 220 kV) al suministro necesario para el sistema ferroviario sueco (1 fase, 16,7 Hz, 15 kV).

Una primera buena aproximación del convertidor se establece con un transformador lineal entre -15 kV y 15 kV, pero después de la campaña de medidas se ha sustituido por una fuente de alimentación real generada a partir de los 8 primeros armónicos más representativos de las señales medidas mostradas en la Fig. (8) [15].

Fig.8:Medidas en el sistema de suministro eléctrico de ferrocarril sueco antes del convertidor con el tren en la vía (izquierda) y sin el tren en la vía (derecha). El resultado de esta señal generada y la comparación con la fuente original se puede ver en la siguiente Fig. (9).

Fig. 9: Comparación entre la fuente original y la fuente simulada para intensidad y voltaje con el tren en la pista, respectivamente.

3.1.2. TRANSFORMADOR

Se encarga de asegurar el suministro de energía a lo largo de toda la línea de transmisión. En la infraestructura ferroviaria sueca la tecnología utilizada es el Auto-Transformador. Su circuito equivalente puede verse en la Fig. (10), la cual muestra la aplicación diseñada en Matlab para el cálculo de sus parámetros.

Fig. 10: Aplicación Matlab para el cálculo de los parámetros del Auto-Transformador.

Esta aplicación lleva a cabo los cálculos necesarios para hallar los parámetros equivalentes del Auto-Transformador. Las entradas requeridas por el programa son las siguientes características del Auto-Transformador: potencia, voltaje en el primario, frecuencia, pérdidas sin potencia, pérdidas en corto circuito y la pérdida de voltaje y corriente. Por otro lado, las salidas son los parámetros equivalentes a un transformador de dos bobinas con las especificaciones definidas en las siguientes Ecs. (1-4).

(1)

(2)

(3)

(4)

3.1.3. ACOPLAMIENTO ENTRE LÍNEAS DE TRANSMISIÓN

De igual manera, se ha programado otra aplicación en Matlab, Fig. (11), para el cálculo del acoplamiento entre las 10 líneas de transmisión de las que consta el poste. Las entradas para dicha función son las coordenadas que sitúan los cables en el poste, sus secciones y las conductividades de los materiales que los conforman.

Fig. 11: Aplicación Matlab para el cálculo del acoplamiento entre líneas de transmisión.

Para mayor exactitud, se modela un bloque para cada sección entre Auto-Transformadores, 10 km, dando como resultado el diseño que se muestra en la Fig. (12).

Fig. 12: Detalle del bloque de Matlab que modela el acoplamiento entre líneas de transmisión.

Para calcular estos parámetros se ha utilizado el método del espejo, considerando la imagen a una distancia h por debajo del nivel de los raíles de 1 metro.

Por último es necesario obtener la impedancia de cada línea de transmisión para construir la matriz de resistencias de todas las líneas de transmisión. Este proceso se lleva a cabo directamente con la parte real de la impedancia y la matriz de inductancias tal y como sigue.

La catenaria tiene una impedancia de 0.125 + j0.343 Ω/km [16]. La Resistencia es directamente 0.125 Ω/km y asumiendo una frecuencia de 16.7 Hz es posible obtener una inductancia de 3.3 mH/km por medio de la siguiente fórmula:

, (5)

donde Im(Z) representa la reactancia eléctrica de la inductancia y f la frecuencia 16.7 Hz. En el caso del raíl, la resistencia es R = 4.976 Ω/km y la inductancia mutua L = 0.965 mH/km. Se ha escogido una distancia de 10 km entre el Auto-Transformadores y una distancia de 450 km desde la fuente de alimentación al convertidor. Usando el método de línea multitransmisión, la inductancia mutua para cualquier par i, j de líneas de transmisión, tal como la Ec. (5), se puede calcular con la siguiente fórmula:

, (6)

-4 donde µ0 = 4 × π × 10 H / km, Dij la distancia desde la línea i a la imagen j y dij es la distancia desde la línea i a la línea j. La capacidad mutua se puede obtener de la Ec. (6) como:

, (7)

8 donde c = 3 × 10 m/s y Lij es la inductancia mutua.

3.2.- INTEGRACIÓN CON EL MATERIAL RODANTE

El modelo global resultará de añadir al diseño explicado previamente el modelo validado del tren como una carga activa, Fig. (13). La distancia entre el convertidor y el final de la línea se puede variar dependiendo del número de bloques de acoplamiento que se incluyan en el modelo, los cuales definen la precisión del modelo dado que el tren se integrará entre cualquier pareja de éstos. El modelo completo fue validado tanto en ausencia del tren como con el tren en la vía.

Fig. 13: Esquema simplificado de material rodante conectado a la catenaria y a la vía.

El tren es una fuente de corriente armónica de dos puertos que conecta la catenaria con los raíles. Por lo tanto, la integración del modelo de tren introduce nuevos armónicos, estudiados dependiendo su situación en la vía y para los peores escenarios en términos de propagación de la señal. Esta respuesta dinámica que contiene armónicos y alto consumo de energía puede producir algunos efectos no deseados en las áreas circundantes que afectan a los sistemas electrónicos, potencialmente peligroso para la seguridad de los sistemas de señalización y otros componentes sensibles a EMI radiadas producidas por dichas señales.

4. CONCLUSIONES

Con la validación del modelo se cumple el objetivo del proyecto, poder realizar cualquier tipo de simulación previa o sustituyendo a una campaña de medidas. Por lo tanto con esta herramienta se pueden adelantar resultados o incluso evitar costosas y complicadas medidas en campo, como puede ser la emulación de un rayo en la vía. De esta manera se puede realizar una detección de fallos que a priori no son fáciles de conocer, especialmente cuando se integran varios sistemas produciendo resultados inesperados.

En este artículo se presenta una metodología que permite analizar las interacciones de los múltiples componentes del sistema ferroviario en pro de una adecuada búsqueda de fallos que no es observable con la mera operación debido a la fugacidad o falta de regularidad en la aparición de los modos de fallo esperados para una correcta diagnosis y prognosis. De hecho el método propuesto puede catalogarse dentro de los conocidos como NFF (fallo no definido) ya que permite descubrir y reproducir fallos cuya frecuencia o patrón de comportamiento no trazable.

Se presenta una aplicación de esta metodología sobre los circuitos de vía debido a su criticidad en el sistema ferroviario de cara a mantener un flujo de trenes con la máxima seguridad y fiabilidad. De hecho los circuitos de vía informan sobre la posesión de la vía y permiten una más fluida relación de operadores y gestores de infraestructura, liberando la vía y permitiendo otro vehículo tomar posesión de ella. Desafortunadamente estos sistemas, robustos en su funcionamiento no son inmunes a todas las interferencias electromagnéticas producidas por otros subsistemas como alimentación, comunicaciones del ferrocarril, etc. Sin embargo el número de escenarios posibles excede la capacidad de análisis en campo y reproducirlos es igualmente costoso. Por ello, el sistema propuesto permite analizar los efectos indeseados del mayor número de variantes que puedan suponer un problema para el sistema en general, más comúnmente conocidos como peores escenarios.

Gracias al trabajo realizado por TREND se ha podido desarrollar un modelo de integración que permite la simulación fiable de distintos escenarios, cuyos resultados son fácilmente analizables. Dicho modelo ha sido posteriormente validado a partir de los valores obtenidos en la correspondiente campaña de medidas, por lo que el modelo desarrollado cumple las expectativas de simulación necesarias para el estudio de las interferencias que puedan afectarle.

El circuito de vía es al mismo tiempo sencillo pero imprescindible debido a su robustez. Por evidentes razones de seguridad se ha de estudiar minuciosamente su integración en la infraestructura y prestar especial atención a su interoperabilidad y agentes externos que pueden interferir en su correcto funcionamiento. El modelo propuesto combina análisis sistémico recogiendo datos a través de campañas de medida con modelado a través de simulaciones reproduciendo casos no visibles durante las mediciones efectuadas en campo. Las campañas de medida in situ son necesarias pero no suficientes a la hora de analizar casos extremos debido al carácter fugaz de los mismos. Es por esto que un modelo de simulación por ordenador se hace imprescindible a la hora de analizar el mayor número de amenazas posibles a partir del mayor número de variantes que puedan suponer un problema para el sistema en general. Dicha conjunción de datos recolectados en vía (data driven) con las simulaciones por computador del comportamiento físico del sistema (physics driven) se conoce comúnmente como modelo híbrido y permite al sistema ferroviario incrementar su robustez considerando las más pesimistas condiciones imaginables a la par que no reproducibles (o reproducibles con un alto coste). Estos sistemas híbridos han permitido detectar algunas carencias en las normas vigentes donde algunos casos posibles aunque no probables no fueron considerados y por tanto la seguridad podría verse comprometida. Otra ventaja adicional de los modelos híbridos es su flexibilidad a la hora de introducir nuevos elementos y considerarlos en la globalidad del sistema complejo de cara a verificar su correcto funcionamiento sin tener que recurrir a costosas medidas y test cada vez que una modificación es introducida en el sistema. Esto es especialmente relevante en el sistema ferroviario debido a la rápida obsolescencia tecnológica de los componentes y por tanto la elevada frecuencia de actualización de los mismos en términos de software o hardware.

En última instancia los resultados obtenidos en este proyecto de investigación son utilizados para la actualización de los actuales estándares, los cuales no cubren totalmente los problemas de interoperabilidad. Para ello se identifican los peores escenarios, tanto en estado estacionario como comportamientos transitorios, especialmente en el ámbito electromagnético, y se detallan en los estándares las condiciones que los definen con total exactitud.

5. AGRADECIMENTOS La investigación ha recibido financiación del 7FP de la Unión Europea FP7/2007-2013 bajo acuerdo de subvención número "285.259". El consorcio está formado por CEIT, CAF I+D, CEDEX, IFSTTAR, York EMC Services, Trafikverket y Luleå Tekniska Universitet.

6. BIBLIOGRAFÍA [1] ERA EMC Report 2010. [2] BSI/CENELEC, Railway applications – Electromagnetic compatibility Part.2: Emission of the whole railway system to the outside world, EN 50121-2: 2006. [3] BSI/CENELEC, Railway applications – Electromagnetic compatibility Part 3-1: Rolling stock, EN 50121-3-1: 2006. [4] Høj NP, Kroger W. Risk analyses of transportation on road and railway from a European Perspective, Safety Sci. 2002; 40(1-4): 337-357 [5] Rodríguez, E., Simón, V., Galar D., Niska S., “Dependability issues of Track Circuits - A hybrid approach”. Conference paper accepted in The Second International Conference on Railway Technology: Research, Development and Maintenance, Ajaccio, Corsica, France, 2014. [6] The University of Sydney News. Fellows of Senate-Senate. [8 May 1990] [consulted: 12 of May 2013] [7] BSI/CENELEC, Railway Applications - Power supply and rolling stock - Technical criteria for the coordination between power supply (substation) and rolling stock to achieve interoperability, EN 50388: 2012. [8] BSI/CENELEC, Railway applications – Compatibility between rolling stock and train detection systems, EN 50238: 2003. [9] BSI/CENELEC, Railway applications - Fixed installations - Electrical safety, earthing and the return circuit - Part 1: Protective provisions against electric shock, EN 50122: 1997. [10] Directiva 2008/57/EC del Parlamento Europeo de Interoperabilidad en el Sistema Ferroviario de la Publicación de la Comisión de la Comunidad Europea del 17 de Junio de 2008. [11] Rodríguez, E., Karki N.R., Galar D., Valderas D., Niska S., “Fault detection of Railway EMC problems using MATLAB models”. Conference Paper, CM 2013 and MFPT 2013, Cracovia, Polonia, 2013. [12] BSI/CENELEC, Railway Applications - Supply voltages of traction systems, EN 50163: 2004. [13] BSI/CENELEC, Railway applications – Supply voltages of traction systems, IEC 60850: 2007. [14] Banverket-Projektering 400 850, Número de dibujo BB10-725 065-01, Acoplamiento Eléctrico: 2006-06-30.7. [15] Rodríguez, E., Galar D., Karki N.R., Niska S., “Simulation of electrical power supply system in railway infrastructure. Integration with rolling stock in Sweden”. Conference paper, International Conference on Power & Energy Systems: Advances in Power Systems, Kathmandu, Nepal, 2013. [16] Banverket-Dokument, Impedanser för KTL 132 kV, 30kV och 15 kV ML.

7. LISTA DE ACRÓNIMOS

7FM: Seventh Framework Programme, 7PM, Séptimo Programa Marco AC: Alternating Current, corriente alterna AT: Auto-Transformer, Auto-Transformador BT: Booster Transformer, transformador amplificador DC: Direct Current, corriente continua EMC: Electromagnetic Compatibility, compatibilidad electromagnética EMI: Electromagnetic Interference, interferencia electromagnética ERTMS/ECTS: European Rail Traffic Management System, sistema europeo de gestión del tráfico ferroviario / European Train Control System, sistema de Control Ferroviario Europeo FSK – Frequency Shift Keying, modulación por desplazamiento de frecuencia GSM-R: Global System for Mobile communications in Railway, sistema global para las comunicaciones móviles para ferrocarril Hz: Hercio I-raíl: Isolerade Rail, raíl aislado kA: Kiloamperio kV: Kilovoltio kVA: Kilovoltio-amperio mH: Milihenrios NFF: Non Fault Found, fallo no definido ns: nanosegundo OFDM: Orthogonal Frequency Division Multiplexing, multiplexación por división de frecuencias ortogonales SNR: Signal to noise ratio, relación señal a ruido S-raíl: Svetsade rail, raíl soldado TCC: Train Control Central, centro de control de trenes TREND: Test of Rolling stock Electromagnetic compatibility for cross-domain interoperability, pruebas de compatibilidad electromagnética en material rodante para la interoperabilidad entre dominios Watt: Vatio Z: Impedancia