Fail Safe Operation of Audio Frequency Track Circuits for Railway Signalling C. Gautham Ram #1, A. Nithya #2, V. Jayashankar *3, P. R.

Transcription

1 Fail Safe Operation of Audio Frequency Track Circuits for Railway Signalling C. Gautham Ram #1, A. Nithya #2, V. Jayashankar *3, P. R. Goundan **4 # Summer Intern, IIT Madras, Chennai, India @nitt.edu @nitt.edu * Electrical Engineering Department, Indian Institute of Technology Madras, Chennai , India 3 jshankar@ee.iitm.ac.in ** Centre of Excellence in Wireless Technology, Indian Institute of Technology, Chennai , India 4 prgoundan@cewit.org.in Abstract An Audio Frequency Track Circuit (AFTC), a recent addition to railway signalling, confirms the absence of a train in a section of track. While the basic concept is established, several issues relating to proper operation of the AFTC need theoretical study. Installation practices like structure bonds and cross bonds, could lead to failure of AFTC operation, especially in the presence of breaks. An analysis is carried out, to study these influences on AFTC performance. Conjectured fault scenarios are identified which could lead to failure of the system, on the side of safety or otherwise. A three-wire model of the track with PSpice as a simulation tool is used for analysis and the reduction in margin due to structure and cross bonds is also established. Keywords Audio Frequency Track Circuit, S-bond, Mast, Structure Bond, Cross Bond, Fail-safe Operation. I. INTRODUCTION A track circuit detects the absence of a train within a section of railway track. Its objective is to achieve fail-safe operation with regard to variable track conditions and parameters. One of the methods of detection of train in a section of track is a DC track circuit. Here a DC voltage is sent from a transmitter across the two rails and received at a fixed location relative to the transmitter. In the absence of a train, a substantial voltage exists across the receiver. The presence of the train in the section is equivalent to a shunt across the rails and is sensed by a reduction in the receiver voltage. One of the obvious limitations of the DC track circuit is the need to isolate sections of tracks, thus leading to tracks with joints. Passenger comfort needs dictate joint-less tracks and this eventually leads to the concept of audio frequency track circuits. Conceptually, the difference between a DC track circuit and an AFTC is the use of AC excitation at a given frequency in a section, thereby making joint-less track operation feasible. The adjacent sections can operate at slightly different frequencies. Thus a small set of frequencies can be used to energise very large track lengths. The principle of AFTC operation and preliminary studies indicating its performance have been described in several works [1-4]. However some issues that need study with respect to fail-safe operation of the principle still exist, and need to be theoretically addressed. For instance, along the track, masts are connected to one of the rails at finite locations, for grounding purposes. Also, when multiple tracks occur in a yard, track-track connections exist across them. Both the mast connections and track-track connections function as traction power distribution requirements, and are referred to as structure bonds and cross bonds respectively. The influence of such bonds on AFTC performance has not yet been addressed and their influence needs to be evaluated. Finally, breaks can occur anywhere on the track and tend to impede regular flow of signal along the rail. This is bound to affect AFTC operation, and hence its effects are analysed to check for failure on the side of safety. II. IMPLEMENTATION OF AFTC Consider the railway track to be divided into sections of fixed lengths. An AFTC consists of suitable transmitterreceiver systems [1], on either side of each section, tuned to a particular frequency in the audio-frequency range. Consider three such sections of 750 m each, of a track spanning a length of around 10 km. The adjacent track circuit sections are separated from one another with the help of S-bonds. The transmitter consisting of a sinusoidal voltage source Tx and the sending end capacitor Ct, and the receiver consisting of capacitor Rx, are connected between the S-bond and the track [5]. S-bonds play a crucial role in the functioning of AFTC. The S-bond together with its tuning capacitors helps provide directionality. Here it is configured such that a voltage from Tx3 preferentially travels to the left, as shown in Fig. 1. Little or no voltage is transmitted to the right. The S-bonds thus ensure that each transmitter-receiver unit is unique to one particular section, and is unaffected by the presence or absence of trains in adjacent sections. Each transmitter Tx has a corresponding capacitor Ct and Rx in that section, which are tuned to the frequency of transmission F. The adjoining sections are operated at different frequencies, to prevent interference at the receiving end. In the analysis, these transmission frequencies are designed to be: F1 = 3855 Hz Ct1 = Rx1= 59.1 uf F3 = 4853 Hz Ct3 = Rx3 = 37.4 uf F5 = 5847 Hz Ct5 = Rx5 = 25.7 uf /09/$

2 Fig. 1 Ten km track model with track section operating at F3 (not to scale) A typical AFTC comprising of 2 S-bonds, 750 m section of track in between, with the transmitter on the right, and the receiver on the left is shown in Fig. 1. In the absence of a train, a finite voltage is obtained at the receiver Rx3. However, when a train enters the section, the axle shunts the rails, causing a fall in voltage at Rx3. It can be appreciated that, greater the no-train receiver voltage and lesser the withtrain receiver voltage, easier is the distinction between the presence and absence of a train. The smallest difference in the receiver voltage defines the margin available. This margin is calibrated in decibels and is an important value as the track circuit is generally implemented with passive elements alone and hence very fine tuning would not be possible. It is necessary that even if the system fails, it should do so on the side of safety only. This implies that in the absence of a train, the system can indicate otherwise. However, if in the presence of the train axle shunt, the system indicates its absence, the consequences can be disastrous, and the system is said to fail on the side of risk. II. MODELLING As excitation frequencies are low, a lumped parameter network of the track should be sufficient for analysis. Here we adopt a three-wire model as opposed to a simpler two-wire model [1], as the influence of structure bonds cannot be treated correctly in the two-wire model. Each 750 m section of track is treated as a pi-network of 10 elements. Fig. 2 shows the relevant model of an element. Here r xy represents the rail self-resistance, L xy represents the rail self-inductance, C x represents the rail-ground capacitance and R bx represents the ballast resistance. The train shunt resistance TSR has a nominal value of 0.5 ohms. The structure bond connected in each 75 m section is modelled as a series R-L circuit. The cross bond is employed between rails containing the structure bonds, of adjacent tracks, and is also modelled as a series R-L circuit. It again, occurs every 75 m. All rail parameters obtained from experimental studies [1,2] and the S-bond parameters, are given in the Appendix. III. SIMULATION PROCEDURE The AFTC is modelled with reference to Fig. 1. An excitation voltage of 10 V with a frequency F3 is applied at Tx3. The receiver Rx3 picks up all 3 working frequencies of the AFTC. The receiver voltages corresponding to each frequency were observed as maxima using Fourier analysis [7], but only the frequency F3, corresponding to the section under study, is analysed. The voltages are reported in decibels. The voltage at Rx3 is initially measured with no train in the section. The TSR is now placed at a distance of 75 m from Tx3 and the voltage at Rx3 is again noted. This exercise is repeated by changing TSR position at distances ranging from 225 m m at intervals of 150 m from Tx3. The resultant voltage profile is shown in Fig. 3. In order to study the influence of the structure bond, the R-L model representing the same is connected at every 75 m to the track, with the other end grounded. The simulation is repeated as described earlier. It is observed that there is a change in the voltage at the receiver as compared to the case without structure bonds (See Fig. 3). In the final stage, a similar track is placed parallel to the first at a distance of 3.5 m and R-L models of cross bonds are added between the tracks, every 75 m. The simulation is again run and the behaviour with and without train is noted. Fig. 3 shows this profile too. IV. STUDY OF AFTC OPERATION Fig. 2 Lumped parameter model of 75m of track A. Effect of Structure Bond The AFTC operational curve for a single track with no structure bonds is shown in Fig. 3. A plot of receiver voltage

3 versus train position is obtained as a U-shaped curve [1]. The no-train value at Rx3 was found to be mv and the highest with-train value was mv giving a margin of db, as recorded in Table I. With masts connected to return rails, the receiver profile shows a significant increase, and the graph changes to a decreasing curve, above the previous curve [9]. The no-train value was mv, while the maximum with-train value was mv. The result can be attributed to the fact that the structure bonds provide additional paths for current flow, thereby decreasing track impedance and increasing the current flow through the train shunt. The resultant increase in voltage across the train shunt appears at the receiver terminal. The difference between no-train and with-train circumstances goes down to db, form db thereby showing a reduction in AFTC performance. B. Multiple Tracks with Cross Bonds When multiple tracks are taken into account, the track return rails of different tracks are connected to one another, with strips of mild steel. Fig. 3 shows the corresponding AFTC performance. The no-train value rose to mv from mv. The with-train values, however opposed the trend, and fell from mv to mv. Decibel performance improved from db in the previous case, to db. The cross bonds function as additional diversions for signal flow, besides the structure bond. The receiver voltage magnitudes are lowered and a U-shaped curve is obtained. The performance is thus found to improve marginally. In effect the structural bonds reduce the available margin as ween in Table I. V. EFFECT OF BREAKS Fig. 3 Performance analysis of AFTC A. Single Track with No Structure Bonds The presence of track breakage could adversely influence AFTC performance. In order to study their behaviour in a step-by-step manner we initially consider the situation of a single track with structure bonds removed. In the presence of breakage, the with-train values remain nearly the same. But the no-train figure decreases from db to db. This result is shown in Table II, and we find that the margin has come down from db to db, lowering AFTC performance. This 7.06 db fall in the no-train receiver voltage can also be used to detect the occurrence of breakage, in such a situation. TABLE I INFLUENCE OF STRUCTURE BONDS AND CROSS BONDS ON AFTC PERFORMANCE Analysis Condition Single track without structure bonds Single track with structure bonds Multiple tracks with structure bonds and cross bonds (No train) (Highest value with train) Margin in db Analysis Condition Single track without structure bonds Single track with structure bonds - cases 1, 3, 5, 7. Single track with structure bonds - cases 2, 4, 6, 8. Multiple tracks with structure bonds and cross bonds - cases 1, 3, 5, 7. Multiple tracks with structure and cross bonds - cases 2, 4, 6, 8 TABLE II INFLUENCE OF TRACK BREAKAGE ON AFTC (No train) Voltage in db (Highest value with train)

4 B. Single Track with Structure Bonds Eight different positions of breaks with respect to TSR are considered, in the top and bottom rails, as shown in Fig. 4. Table II summarises the results. With the TSR in the section, with no breakage, a value of 97 mv was registered at the receiver. In cases 1, 3, 5 and 7, the occurrence of breaks on the rail containing the structure bonds, gave substantially large values of receiver voltage, with the maximum being that of case 7, equal to mv. However, they were not as large as the corresponding no-train value of mv. The increase in receiver voltage with train is because the structure bonds provide alternate low impedance paths through the ground, in the presence of breaks [8]. Due to the structure bonds there is a distinct closed path from transmitter to receiver, and back. This enables a finite with-train voltage to be recorded in the receiver, even in the presence of breakage. So the increase in the with-train voltage from db to db is justified and AFTC performance goes down. In cases 2, 4, 6, and 8, the receiver voltages with and without train are diminished and become comparable. For instance when the receiver voltage in case 2 is mv, the corresponding without train value is surprisingly mv only. Hence the receiver gets a very small voltage, irrespective of the presence of train. Thus the system fails on the side of safety. Moreover fall in the no-train voltage from db to db also aids in breakage detection for these 4 cases. Fig. 4. Eight cases of break position with respect to TSR, with masts assumed to be connected to the top rail. C. Multiple Tracks with Structure and Cross Bonds The above 8 cases are discussed again, in the presence of a parallel track, with cross bonds. AFTC was assumed for one track alone, and analysis results are recorded in Table II. The results are same as from the preceding analysis. Cases 1, 3, 5 and 7 gave substantially large values of receiver voltage in the presence of train, as compared to the with-train value without breaks. On the other hand, cases 2, 4, 6 and 8 failed on the side of safety i.e. the receiver voltage values were highly diminished irrespective of the absence of train. Furthermore fall in the no-train voltage from db to 51.7 db aids in breakage detection. Reviewing cases 1, 3, 5 and 7 it is seen that besides the structure bonds, additional routes through the parallel track are provided with the help of cross bonds. The presence of additional paths in parallel, lowers the impedance to signal flow, hence with-train receiver potential is increased on addition of breaks. Having done analysis on normally conceivable faults, a situation is conjectured, which could drive AFTC operation towards failure on the side of risk. The corresponding analysis is as follows. Firstly, for a single track, retaining structure bonds only in the first and last stretch of 75m, and removing the rest, a break is introduced conforming to case 7. We find that the with-train receiver voltage rises from db to db while the no-train value decreases from db to db on addition of breaks. A similar study is done for multiple tracks wherein just two structure and cross bonds at the extremities of the 750 m section are retained. Results were obtained likewise viz. a with train value of db rose to db when the no train value decreased from db to db. Thresholds are set with respect to no breakage situation in the track. If the track breakage results in such a high value of voltage in the presence of a train that it exceeds the no-train value (with no breakage), the security will be compromised and the system would fail in the side of risk. The conjectured situation leads to a margin of only 4 db. Both these cases are however dangerous for AFTC operation. This is because further increase in the with-train value due to variations in ballast, electrification etc. may bring this value closer to the no-train value causing the AFTC to indicate absence of train in its actual presence. VI. CONCLUSIONS The performance of an AFTC was analyzed with a threewire track model. It was shown that a small set of frequencies could indicate the presence of a train in any section of track. Masts connected to traction return rails and connections across multiple tracks cause the AFTC performance margin to be reduced. This effect becomes more profound in the presence of breaks. Different conditions corresponding to track breakage were evaluated - some cases were shown to behave in the conventional mode while others resulted in fail-safe operation. A worst-case scenario was also conjectured which could lead to failure of the AFTC on the side of risk.

5 VII. APPENDIX This section provides the S-bond and track parameter values used for the simulations. Table III covers the series impedance parameters of rails, cross bonds and structure bonds. The admittance parameter values are expressed in terms of their rail-rail equivalents. TABLE III SERIES MODELLING PARAMETERS Parameter Rail Cross Bonds Structure Bonds Resistance 9.5 /km 50 m 5 Inductance 1.25 mh/km uh 4.8 uh Length 75 m 3.5 m 2.7 m Admittance parameters: Rail-Rail Capacitance = 0.5 uf/km [1,4]. Ballast/ Rail-Rail Conductance = 62.5 us/km [4]. S-bond parameters: Sinusoidal Excitation = 10 V Rail-bond coupling coefficient k = 0.5 Total Inductance of S-bond = 34.2 uh Total Resistance of S-bond = 60 m ACKNOWLEDGMENT The authors wish to acknowledge M/S HBL Power Systems for making available experimental track parameters and models for simulation study. REFERENCES [1] R. J. Hill and P. C. Coles, A user-friendly simulator for modelling audio frequency track circuit operation, in Proc. of the 1993 IEEE/ASME Joint Railroad Conference, April 6-8, 1993, pp [2] R. J. Hill, D. C. Carpenter, B. Mellitt, J. Allan and J. C. Brown, Calculation and measurement of rail impedances applicable to remote short-circuit fault currents, in IEE Proceedings-B Electric Power Applications, November 1993, vol. 140, no. 6, pp [3] R. J. Hill, S. Brillante and P. J. Leonard, Railway track transmission line parameters from finite element field modelling: series impedance, in IEE Proceedings - Electric Power Applications, Nov. 1999, vol. 146, no. 6, pp [4] A. Mariscotti and P. Pozzobon, Determination of the electrical parameters of railway traction lines: calculation, measurement, and reference data, in IEEE Transactions on Power Delivery, Oct. 2004, vol. 19, no. 4, pp [5] G. D'Addio, P. Ferrari, A. Marisconi, P. Pozzobon, Integrated modelling of audio frequency track circuits, in International Conference and Exhibition on Electromagnetic Compatibility, Dec , 1999, pp [6] R. J. Hill, S. L. Yu, N. J. Dunn, Rail transit chopper traction interference modeling using the SPICE circuit simulation package, in IEEE Transactions on Vehicular Technology, Nov. 1989, vol. 38, no. 4, pp [7] M. C. Falvo, E. Fedeli and R. Lamedica, A measurement campaign on audio frequency track circuits of Italian high speed railway systems, in International Symposium on Power Electronics, Electrical Drives, Automation and Motion, SPEEDAM, May 23-26, 2006, pp [8] N. Nedelchev, Influence of earth connection on the operation of railway track circuits, in IEE Proceedings - Electric Power Applications, May 1997, vol. 144, no. 3, pp [9] R. J. Hill, D. C. Carpenter, T. Tasar, Railway track admittance, earthleakage effects and track circuit operation, in Proc. Of Railroad Conference, April 25-27, 1989, pp