Overvoltage Protection of Light Railway Transportation Systems F. Delfino, R. Procopio, Student Member, IEEE, and M. Rossi, Student Member, IEEE Abstract In this paper the behavior of the power supply system of a typical Light Railway Transportation System (LRTS) subjected to overvoltages of atmospheric origin is analyzed, with the purpose to determine the optimum location of the surge arresters. A complete model of the whole power system has been developed and the transients have been simulated with the electromagnetic code PSCAD-EMTDC. Index Terms Electromagnetic transient propagation, overvoltage protection, electric traction. I. INTRODUCTION L IGHT railway is a generic term which covers a wide variety of railways and tramways. This modern electric mass-transit system implies economy of construction in one or more ways, such as light weight rails, wooden rails, light bridges, minimal earthworks, sharp curves and steep gradients, narrow track gauge and includes urban and suburban railways, also known as rapid transit, or trolley lines [-]. Reliable and effective protection against overvoltages (atmospheric and switching) on contact wires is most important in the smooth and constant running of the trains. To this end, the correct positioning of the surge arresters plays a fundamental role [-]. Several configurations for the protection system can be adopted, characterized by different locations and number of the surge arresters:. Surge arresters located at the DC busbars of each electrical substation;. Surge arresters located at the interface between contact line and feeder cables;. Surge arresters located both at the electrical substations and at the interface between contact line and feeder cables. The purpose of the present paper is to simulate the response of the whole system when subjected to an overvoltage wave, in order to determine the "best" configuration in terms both of protection effectiveness and cost economies. Thus, it is not aimed at modeling in an exhaustive way the field-to-line coupling originated by a lightning event [7], which would be so much complicated using a circuital code like PSCAD-EMTDC [8]. As a matter of fact, this would Federico Delfino, Renato Procopio and Mansueto Rossi are with the Department of Electrical Engineering, University of Genoa, via Opera Pia a (e-mail: federico.delfino@die.unige.it; rprocopio@epsl.die.unige.it; mansueto@die.unige.it). imply the implementation of the Agrawal method [9], which, even in its simplified version, would make the simulation of the system under test be so onerous from a computational point of view. In other words, the authors intention is to set up a simplified tool, which can give some useful information for the system design with the advantage of exhibiting good properties in terms of CPU costs. As will be examined more in details in the following sections, this kind of approach, although simplified, is able to identify among the three possibilities indicated above, the best arresters location. The paper is organized as follows: in section II a brief description of the LRTS is carried out and an equivalent circuital model is defined. In section III the results of the simulations are presented and thoroughly discussed and in section IV some conclusive remarks are drawn. II. SYSTEM MODELING A typical Light Railway Transportation System (LRTS) is powered at 7 V DC nominally with a maximum acceptable deviation of +% and % and distributed by an Overhead Contact System (OCS), supplied by parallel feeder cables. The OCS can include, for each track, a contact wire and two parallel feeder cables. The OCS of one track ( m long) is connected in parallel to the OCS of the other track at each end of the route, and, in principle, at every tram stop, substation and crossover. The running rails are used to carry the return current and constitute, with the return cables and the Electrical SubStations (ESSs) return negative busbars, the negative return system. Each ESS negative busbar is connected to the rail by means of insulated cables. Each LRTS ESS is fed from the Electricity Supply Board (ESB) MV network with supply voltage. kv AC and frequency Hz. The ESS absorbs electrical power to energize the OCS at 7 V DC and to supply auxiliary loads at / V AC. Each traction substation is connected to a MV generally grounded network. Usually, the electric system feeding the LRTS exhibits a modular structure. The borders of each subsystem are represented by the insulated overlaps, whose size is about cm. From the point of view of the overvoltage study, the analysis has been limited only to one subsystem, without any
loss of generality. This is confirmed by the fact that an overvoltage coming from one section cannot propagate along another one by the presence of the insulated overlap (being the air breakdown voltage kv/cm). The generic subsystem is shown in Fig.. Fig. A module of the LRTS network. The subsystem modeling considers both the AC and DC systems, which are briefly described in the following: ) The AC system is represented by a MV grounded network; ) Each traction substation is modeled by a windings transformer with two diode bridges connected in series forming a pulse rectifier configuration 7 V DC and 9 kw rated. ) The contact lines, the ESSs cables (length equal to m) that connect the OCS directly to the substations and the parallel feeder cables (length equal to m) represent the DC section. All the cables are characterized by a rated impulse voltage U NI =8 kv [] Surge arresters of Metal-Oxide type with a rated voltage of. kv DC and a lightning impulse residual voltage of.8 kv have been employed. The above-mentioned models have been implemented into the electromagnetic code PSCAD-EMTDC [8]. This package allows to describe with high details all the system components. The response of the system has been tested in the three configurations described in Section I. As surge arresters do not protect the contact line, the overvoltages to be taken into account are those limited only by the intrinsic insulation of the contact line itself. Thus, the upper limit assumed for the overvoltages propagating from the contact line towards the cables corresponds to the impulse withstand voltage of the OCS insulators, namely kv. The voltage surge has been represented by a unidirectional double-exponential wave, rising to crest from zero in about µs and falling to half the amplitude in µs. This is the socalled standard lightning impulse or,/ µs wave. The analytical expression of this wave is the following: v.7 t. t () t.7( e e ) = [kv] () Before applying the voltage surge, the system has been led to its steady-state condition that corresponds to the DC operating voltage of 7 V. The initialization task lasts for. ms, so that all the overvoltage graphs should be considered from this time onward. III. NUMERICAL EXPERIMENTS The simulations have been carried out for the three cases applying the overvoltage both near the interface between the overhead contact line and the parallel feeder cable (point B in Fig. ) and near the interface between the line and the electrical substation cable (point A in Fig. ). Therefore, six different situations have been examined, according to the arresters location and to the overvoltage application point. We are interested in verifying the observance of the technical provisions of the voltage at the beginning of the cables and at the substations. In addition, the voltage across the diodes of the substation AC/DC converters has been evaluated. In the following, the results, relevant to all the six cases, are presented. A. Configuration : surge arresters located at the DC busbars of each electrical substation The location of the arresters is sketched in Fig.. Fig. LRTS configuration with arresters located at the DC busbars of each electrical substation. Let us first apply the voltage surge at point A. It is important to observe (see Figs. and ) that the voltages at the beginning of the ESSs cable and of the parallel feeder one are not so much mitigated, as could be expected by remembering the classical theory of the propagation of waves in a transmission line [9-]; this is because the time in which the wave travels along the part of line between the application point of the overvoltage and the interface with the cable is much shorter than the duration time of the overvoltage. Therefore, all the reflected waves are added and the resulting voltage is not so much smaller than the incoming voltage. The voltages at the nearest ESS and across the most stressed diode have been evaluated in three different situations, characterized by an earth resistance for the surge arresters equal to, and Ω, respectively. Obviously, the voltage drop on the earth resistance is added to the arrester threshold voltage (.8 kv), thus increasing the voltage at the point where the arrester is connected.
Fig. Voltage at the terminal of the ESS cable (contact line side) for the overvoltage applied at point A and arresters at the ESSs. Fig. Voltage at the nearest ESS for the overvoltage applied at point A and arresters (earth resistance Ω) at the ESSs. 8 8 - Fig. Voltage at the terminal of the parallel feeder cable (contact line side) for the overvoltage applied at point A and arresters at the ESSs. Fig. 7 Voltage at the nearest ESS for the overvoltage applied at point A and arresters (earth resistance Ω) at the ESSs., -, -, -, -, Fig. Voltage at the nearest ESS for the overvoltage applied at point A and arresters (earth resistance Ω) at the ESSs. Fig. 8 Voltage on the most stressed diode for the overvoltage applied at point A and arresters (earth resistance Ω) at the ESSs. As can be seen in Figs. -, the more the earth resistance grows, the higher are the voltage peaks at the substations and on the diodes.
, confl vd Main -, -, -, -, Fig. 9 Voltage on the most stressed diode for the overvoltage applied at point A and arresters (earth resistance Ω) at the ESSs. - Fig. Voltage at the terminal of the ESS cable (contact line side) for the overvoltage applied at point B and arresters at the ESSs., -, -, -, -, -, -, - Fig. Voltage on the most stressed diode for the overvoltage applied at point A and arresters (earth resistance Ω) at the ESSs. Applying the voltage surge at point B, one can again evaluate the voltages at the terminals of the cables (contact line side), at the ESSs and on the ESS diodes, which are plotted in Figs.,,, and, considering the arresters characterized by an earth resistance of Ω. In this case, thanks to the symmetry of the system, there is no difference between the two ESSs voltages. As can be observed, the most stressed cable is the parallel feeder one, since the overvoltage is applied at point B. As far as the voltage values at the cables terminals (contact line side) are concerned, they are higher than the rated impulse voltage U NI = 8 kv []. The voltages at the ESSs and on the diodes are limited by the surge arresters to values slightly lower than the corresponding ones of the case of voltage surge applied at point A. Anyway; in both cases the presence of the arresters prevents the ESS voltage from reaching values that could result dangerous for its devices. Fig. Voltage at the terminal of the parallel feeder cable (contact line side) for the overvoltage applied at point B and arresters at the ESSs. Fig. Voltage at the ESSs for the overvoltage applied at point B and arresters at the ESSs.
, -, -, -, -, Fig. Voltage on the most stressed diode for the overvoltage applied at point B and arresters at the ESSs. B. Configuration : surge arresters located at the interface between contact line and feeder cables The location of the arresters is depicted in Fig.. Fig. Voltage at the terminal of the ESS cable (contact line side) for the overvoltage applied at point A and arresters (earth resistance Ω) at the interface OCS/cables. - Fig. 7 Voltage at the terminal of the parallel feeder cable (contact line side) for the overvoltage applied at point A and arresters (earth resistance Ω) at the interface OCS/cables. Fig. LRTS configuration with arresters located at the interface between contact line and feeder cables. Let us simulate the application of the voltage surge at point A. The results of this test are reported in Figs., 7, 8 and 9 and are relevant to Ω surge arresters earth resistance. These results should be compared with the corresponding ones of case. As can be seen from Fig., the voltage at the ESS cable terminal (contact line side), although lower than that of Fig., is higher than 8 kv, in spite of the presence of the surge arresters. This is due to the fact that the discharge current through the arrester causes a voltage drop, which is unavoidably added to the arrester residual voltage, thus originating the voltage behavior of Fig.. In Fig. 7 the voltage at the terminal of the parallel feeder cables (contact line side) is plotted. This figure reveals that the voltage is slightly higher than the residual voltage. 8 Fig. 8 Voltage at the nearest ESS for the overvoltage applied at point A and arresters (earth resistance Ω) at the interface OCS/cables.
Moreover, in this case, the voltage at the nearest ESS is no more limited by the surge arrester (which is not present at its busbar). Fig. 8 highlights that the crest value reached by the voltage is much higher than the one obtained in case (see Fig. ). Such value could result too heavy for the ESS devices. This fact obviously reflects on the diode inverse voltage that is shown in Fig. 9. - - - - Fig. Voltage at the terminal of the parallel feeder cable (contact line side) for the overvoltage applied at point B and arresters (earth resistance Ω) at the interface OCS/cables. 8 - Fig. 9 Voltage on the most stressed diode for the overvoltage applied at point A and arresters (earth resistance Ω) at the interface OCS/cables. If we apply the overvoltage at point B, we get the following results (Figs. - ). Figs. and show the voltage at the ESS and parallel feeder cable terminals (contact line side), respectively. As can be expected, the higher value of voltage is on the parallel feeder cable. In Figs. and the voltages at the ESSs and on the diodes are plotted. These voltages result higher than the corresponding ones obtained in case. g Voltage ( [kv] g) Fig. Voltage at the ESSs for the overvoltage applied at point B and arresters (earth resistance Ω) at the interface OCS/cables., -, -, -, -, -, - -, Fig. Voltage at the terminal of the ESS cable (contact line side) for the overvoltage applied at point B and arresters (earth resistance Ω) at the interface OCS/cables. Fig. Voltage on the most stressed diode for the overvoltage applied at point B and arresters (earth resistance Ω) at the interface OCS/cables.
As a general comment, one can conclude that this configuration does not seem to be effective, since the presence of the arresters at the interface between contact line and cables could not be sufficient to limit the cable voltages. Moreover the ESS devices could be damaged since they are not protected by surge arresters. C. Configuration : surge arresters located both at the interface between contact line and feeder cables and at the DC busbars of each electrical substation The location of the arresters is depicted in Fig.. In all these cases the earth resistance of the OCS arresters is Ω, (since it is highly impractical to obtain a lower value) while the earth resistance of the ESSs ones is assumed to be Ω. - Fig. - Voltage at the terminal of the parallel feeder cable (contact line side) for the overvoltage applied at point A and arresters located both at the interface OCS/cables (earth resistance Ω) and at the ESSs (earth resistance Ω). Fig. LRTS configuration with arresters at the interface OCS-cables and at the ESSs. Figs. and show the voltage at the terminal both of the ESS cable and of the parallel feeder cable (contact line side), respectively as a consequence of the application of the voltage surge at point A. The benefits deriving from the connection of the surge arresters to the ESS busbars are highlighted in Fig. 7 that shows the voltage at the nearest substation. In order to investigate the dependence of the voltage peaks on the earth resistance, the analysis has been repeated with OCS arresters earth resistance of Ω. In Table I the values of the peaks in both cases are compared, highlighting that there is a not negligible difference of such peaks in the two cases. Fig. 7 - Voltage at the nearest ESS for the overvoltage applied at point A and arresters located both at the interface OCS/cables (earth resistance Ω) and at the ESSs (earth resistance Ω). TABLE I VOLTAGE PEAKS ON THE DIFFERENT DEVICES WITH OCS ARRESTERS EARTH RESISTANCE OF AND Ω AND VOLTAGE SURGE APPLIED AT POINT A. Earth resistance (Ω) Parallel feeder voltage peak (kv).. ESS cable voltage peak (kv).8 Nearest ESS voltage peak (kv).8 Fig. - Voltage at the terminal of the ESS cable (contact line side) for the overvoltage applied at point A and arresters located both at the interface OCS/cables (earth resistance Ω) and at the ESSs (earth resistance Ω). Applying the overvoltage at point B, one can again plot the voltage at the contact line terminal of the ESS and of the parallel feeder cables, as well as the ESSs voltage, as shown in the following three figures. Also in this case the test has been repeated considering an OCS arrester earth resistance of Ω. The obtained results, here not reported for the sake of brevity, confirm what already
noticed about the strong dependence of the voltage peaks on the earth resistance of the OCS arresters. It can be highlighted that the effect of locating the arresters also along the OCS (i.e. at each interface OCS/cables) is strongly dependent on the value of their earth resistance (as such resistance grows up, the voltage peaks increase). - Fig. 8 - Voltage at the terminal of the ESS cable (contact line side) for the overvoltage applied at point B and arresters located both at the interface OCS/cables (earth resistance Ω) and at the ESSs (earth resistance Ω). - Fig. 9 - Voltage at the terminal of the parallel feeder cable (contact line side) for the overvoltage applied at point B and arresters located both at the interface OCS/cables (earth resistance Ω) and at the ESSs (earth resistance Ω). Fig. - Voltage at the ESSs for the overvoltage applied at point B and arresters located both at the interface OCS/cables (earth resistance Ω) and at the ESSs (earth resistance Ω). Thus, since the currents involved in a lightning event are very high, this configuration, which is more expensive than the first and can create problems of maintenance, could not be as effective as expected. In conclusion, the first solution seems to be the best. IV. CONCLUSIONS In this paper, the response of the power supply system of a typical LRTS to an overvoltage wave has been investigated. A detailed model of the system has been derived using the electromagnetic code PSCAD-EMTDC, in order to determine the optimum location and number of the surge arresters. The comparative study has pointed out that the solution where the arresters are only at the ESSs is the best one, since it represents a good compromise between a sufficiently safe configuration and a cheap one. Of course, the study has been carried out developing a tool whose aim is not to describe in detail the field-to-line coupling due to a lightning event, but to give useful guidelines to a designer with low computational costs. V. REFERENCES [] Light Rail Transit Association, UK Development Group, Fact Sheet N., July. [] R. Buckley, Tramways and Light Railways of Switzerland and Austria, National Atlases from LRTA, London,. [] EN - (March ): Railway applications Insulation coordination. Part I: Basic requirements. Clearances and creepage distances for all electrical and electronic equipment. [] EN - (March ): Railway applications Insulation coordination. Part II: Overvoltages and related protection. [] EN - (March 999): Railway applications Fixed installations DC switchgear. Part IV: Outdoors DC in-line switch-disconnectors, disconnectors and DC and earthing switches. [] EN (November 99): Railway applications. Supply voltages of traction system. [7] F. Delfino, P. Girdinio, R. Procopio, M. Rossi, Technique for computing the response of a line of finite length excited by HF electromagnetic fields, IEE Proceedings on Science, Measurements and Technology, vol. 9, n., September, pp. 89-9. [8] PSCAD-EMTDC version., The Professional's Tool for Electromagnetic Transients Simulation, Manitoba HVDC Research Centre Inc.,. [9] C. R. Paul, Analysis of multiconductor transmission lines, John Wiley & Sons, New York, 99. [] F. Tesche, M. Ianoz, T. Karlsson, EMC analysis methods and computational models, John Wiley & Sons, New York, 997. [] E. F. Vance, Coupling to shielded cables, Wiley Interscience, NY, 978. Federico Delfino was born in Savona, Italy, on February 8 th, 97. He graduated cum laude in Electrical Engineering from the University of Genoa, in 997 and received the PhD degree from the same University in. He is currently a Researcher at the Department of Electrical Engineering of the University of Genoa. His research interests are mainly focused on electromagnetic field theory, numerical techniques applied to EMC and TL theory. Renato Procopio was born in Savona, Italy, on March th, 97. He graduated cum laude in Electrical Engineering from the University of Genoa, in 999. He is currently working on his PhD on lightning return stroke current modeling and TL theory as well as on Power Quality improvement in Distribution networks. Mansueto Rossi was born in Savona, Italy, on April th, 97. He graduated cum laude in Electrical Engineering from the University of Genoa, in 999. He is currently working towards the PhD degree at the Department of Electrical Engineering of the University of Genoa. His research interests are focused on numerical modeling of electromagnetic fields and EMC.