Transactions on the Built Environment vol 18, 1996 WIT Press, ISSN

Transcription

1 Time domain simulation of variable frequency drive locomotives for electromagnetic compatibility assessment purposes B.M. Steyn, J.D. van Wyk Infrastructure (Signals), Spoornet, South Africa Energy Laboratory, Rand Afrikaans University, South Africa Abstract Failure mode effects and criticality analysis performed on signalling systems forms a major input into the safety case for a particular signalling system. With the introduction of variable speed drive locomotives without fail-safe interference monitors the principle of FMECA had to be extended to the locomotive power electronics and associated control circuits. Therefore, time domain simulation models were developed and used during the FMECA process. After the simulation models are calibrated by experimental results, it can be used very cost effectively to determine the effect of failure modes and transient conditions. The models developed for a current fed inverter drive locomotive will be discussed together with measurement results used for calibration. Examples of simulation results obtained with the models, together with signal equipment susceptibility limits, are used to demonstrate the use of these simulation models in the FMECA. l Introduction In the past the specifications for new locomotives included requirements for electromagnetic compatibility in a very general sense. It required the locomotive to operate on lines equipped with electrical signalling without interference, and then a list of the installed equipment was included. In addition to this it also called for the impedance of the locomotive to be less than 4 ohm, to allow a certain level of ripple voltage on the 3kV d.c. overhead line, without interference to the 50 Hz AC track circuit. With the introduction of chopper controlled locomotives the chopper frequency was fixed to 298 Hz in the specification to avoid interference with the audio frequency track circuits in the range 350 to 450 Hz. To guard against a drift in chopper frequency it was required to be crystal controlled. During locomotive acceptance tests the Signal department was allowed a few tests to verify these requirements.

2 444 Computers in Railways With the acquisition of the most resent locomotives a functional specification for modern hi-tech locomotives was developed with the same, relatively broad, electromagnetic compatibility requirements. Because a functional specification does not specify the power electronic topology the configuration of delivered system is unknown at tender time [1]. The result was that locomotives with current and voltage fed variable frequency drives with induction motors were delivered recently within a short period. 2 EMC ASSESSMENT PROCEDURE As experience was gained with the AC drive locomotives, all over the world, it soon became apparent that electromagnetic compatibility was not easy to achieve [2,3]. Because different railways adopt different procedures and formal standards were lacking at the time, Spoornet had to develop their own procedures for the acceptance of new high-tech locomotives, which are outlined in Figure 1. Because signal equipment susceptibility was only known at 5OHz and the operating frequency of the j Locomotive EMC j, ~ i i! assessment procedure equipment, the first step was to determine the i y ' susceptibility at all frequencies up to about 10kHz. ^ _ J[, In South Africa many / installed track circuits are of j susceptibility equipment I j the single frequency type and therefore interference ' can cause both right side and wrong side failures. : Ensure The more modern track circuits are however immune j "tlzt' I to interference in the sense that they only possess % " right o side susceptibility i -/ in terms of traction ;! FaiiLT^modeeffe^T'l criticality analysis r^ interference. 1 -" t The second step is to prove, by measurement, that EMC the locomotive is compatible with the signalling acceptabl system under normal working conditions. If interference occur during these tests modifications ^ must be done on the locomotive or the signalling L_J^1_J system until compatibility is achieved. * _.. _.,.-, Figure 1 Simplified EMC Following this, failure modes and transient procedure in Spoornet conditions must be studied. Initially it was envisaged that measurement could be used to study the effect of transients and all the different failure modes, but it was soon realised that this was not possible either because of the cost, or dangerous conditions which might arise from introducing certain failures. It was therefore decided to develop locomotive simulation models which can be used to study the interference effect of a given failure. The rest of the paper will be devoted to a description of these time domain simulation models and their application in the failure modes effects and criticality analysis. 3 TIME DOMAIN LOCOMOTIVE SIMULATION MODELS Models for both a current fed and a voltage fed inverter were developed. Models for

3 Computers in Railways 445 the locomotive with a voltage fed inverter have been described elsewhere [4] and therefore the discussions in this paper will be made with reference to the models developed for the locomotive with a current fed inverter drive. In a study of railway EMC eventual integration of electromagnetic models for locomotives, traction transmission systems and signal systems, into a single simulation program is very advantages. Since the Electromagnetic Transient Program (EMTP) contains models for most of the systems mentioned, it was selected as the basis for the simulation. The program was initially developed with the aim to study electromagnetic transients in power transmission systems [5]. Because of this, lumped linear and nonlinear component models, as well as models with distributed parameters, are supported by the program. In addition to this, support programs enable the modelling of frequency dependant parameters. 3.1 Program overview Figure 2 shows the main components of the EMTP simulation program. The EMTP is a time domain simulation program which simulates electrical and control systems by solving the differential equations of the system model at discrete time intervals. The EMTP also contains a model for the simulation of electric machines which is based on the universal machine model. componen models Transmission line modeis EMTP core Set up difference equations Calculate initial conditions Solve difference equations j machine model Complex control systems may be MODELS included in the simulation. This can be done by either one of the two Figure 2 EMTP main components control system models. The first one is called "Models" and consists of a set of procedures and functions that can be organised in a structured program, similar to that of a modern high level language. The second option is a block orientated language known as TAGS (Transient Analysis of Control Systems). It consists of transfer function blocks expressed in terms of S- polynomial ratios and thus allows the Laplace description of a control system. The EMTP enables the use of data base modules whereby functional portions of a larger model can be grouped and tested on its own [6].This module can then be used in a larger model as a black box with certain inputs and outputs. 3.2 The locomotive model The block diagram of one half of the Class 14E current fed locomotive is shown in Figure 3, the second bogie being exactly the same as the one shown. Apart from representing the block diagram of the locomotive the blocks in Figure 3 also corresponds directly to the EMTP data base modules developed for the locomotive. On the block diagram, and thus the EMTP model, the substation is shown as an ideal split

4 446 Computers in Railways supply. The chopper is connected to the substation through the on board input filter inductor (Lf). In the simulation model this can also include the impedance of the overhead supply system. The two current source inverters, with their associated induction motors(ml, M2), are connected in series and is fed with the d.c. link inductor. The inputfiltercapacitor forms part of the chopper modules. Figure 3 Half of the Class 14E locomotive model In the figure the resistors labelled R^ are high value resistors included in the EMTP to avoid nodes floating with respect to earth when switches are in the open state. This is a common requirement of the EMTP. Resistors marked R^ are low value shunt resistors used to mathematically isolate components from each other or to measure current. The torque provided by the traction motors (Ml, M2) to the train (mechanical load) is represented in Figure 3 as current sources (T^ T^) 4 EMTP data base modules 4.1 Chopper The Class 14E chopper circuit is shown in Figure 4, and the part that was built into a module is marked as such. The chopper module consists of a conventional step down chopper, which is employed to control the current in the d.c. link inductor at the required level. Because there is a time step difference between the control system (TAGS) and the EMTP core program, the commutating diode is modelled as controlled switches. This is done in order to achieve correct commutation of the current from the GTO to the diode, and visa versa. When this module is called from a simulation

5 program, the values of the components inside the module can be passed to the module, enhancing the reusability even more. In the case of the Class 14E module the values of the input capacitor Q and the snubber components R^^ and C^ub must be supplied to the module. Computers in Railways Inverter Figure 4 Chopper circuit in the EMTP The inverter of the Class 14E locomotive consists of a conventional naturally pos commutated thyristor controlled circuit [7,8], and ^ GT5W -"- -- C5a is shown in Figure 5. The thyristors are modelled as _ RSb controlled ideal switches. C5b The conditions for Rib conduction is the same as u *'* Clb I Clb that of thyristors. They will continue to conduct even if c the gate drive is removed and will continue to do so until ^^ ashji^lca, the current goes below the holding current. They will Figure 5 Current source inverter circuit diagram also not conduct when they are reverse biased, even if gating signals are present. In this case however the diodes are not controlled, because no problem is experienced during commutation. 4.3 Chopper controller The block diagram of the chopper controller of the Class 14E locomotive as Figure 6 Chopper controller in the EMTP

6 448 Computers in Railways implemented in the TAGS portion of the EMTP is shown in Figure 6. Essentially the control circuit operates the chopper in such a way that the current in the d.c.-link inductor can be maintained at the required level. Depending on the magnitude of the difference between the actual link current (7,^) and the required current (ISET) a PWM control signal is generated. A logic circuit diagram is used to generated the gate drive signal for the controlled diode as discussed above in paragraph 4.1. When this module is called from the simulation program the chopper frequency (/^) and the relative phase shift (CHPHSE) must be passed. 4.4 Inverter controller ICOMPI The current source inverter is operated in the quasi square wave as well as the PWM mode to reduce torque pulsations at low frequencies [9]. Basically the PWM wave is generated by cutting notches out of the 120 conduction block and introducing pulses of equal width within the same 180 of the fundamental inverter frequency. The principle used to generate the PWM Figure 7 PWM generation phase currents for the inverter of the Class 14E model is shown in Figure 7 for one of the phases. As can be seen a specially derived ramp waveform (ICOMPI) is compared with the phase reference waveform I*, with the resulting PWM phase current labelled 1^^. Because the actual control circuit diagram is relatively complex and a description of it is considered outside the scope of this paper, it will not be printed here. 4.5 Mechanical load The electromagnetic torque (Te) produced by the motor is available to accelerate the masses of the rotor as well as any other mass coupled to the shaft (the train in this case). Figure 8 shows a mechanical equivalent, where the mass of the train is transformed to an equivalent inertia on the shaft [4]. The mechanical system can be implemented in the ATP as an equivalent electrical network as shown in Figure 8. ^Te Shaft Tl r \ L(uH) Mr Me Damping Figure 8 Mechanical load model and the electrical equivalent as used in EMTP

7 5 Measured and simulation results Computers in Railways 449 During locomotive compatibility tests, measurements are made on the signal equipment as well as on board the locomotive. A special test site is used for measurements and has been chosen to represent worst case conditions in terms of interference to signalling. The test section is configurable in terms of traction and can accommodate a great variety of signal and measurement equipment. Measurement on the locomotive under test is done from a test coach forming part of the test train. Many of the measured electrical signals are amplified and transmitted to the test coach with fibre optic links. The tractive effort is regulated to the required value during the tests with three to four diesel locomotives, applying dynamic braking. The motor terminal voltage and line current were measured during such a test and are presented, together with the results obtained from a simulation of the same conditions, in Figure 9. Simulated Motor Current, and Voltage I I j- T {L--1--JL-- IT h ~f i I ] ir >; 1000 i U Ifur 1 r 1 L JL!! I m "^ ::: =1 1 " pt jt~! t/*pv-t---r/ :rr:ja:;r J I 4"T It""!" I I - r ' ir"i fn ^ ^ ^ Figure 9 Measured and simulated motor current and voltage From this it can be seen that there is good correlation between the waveforms. Excursions of the simulated d.c. link current are smaller than the measured results and are due to a difference in d.c. link inductance of the actual and simulated systems. In practice the inductance of the d.c. link inductor varies from 25 mh at a low current to 7 mh at a current of 1300A. There is also a slight difference between the simulated and measured motor voltages. This can mainly be attributed to differences in motor parameters, parasitic components not modeled in the simulation model, as well as the limited bandwidth of the measurement equipment.

8 450 Computers in Railways 6 Application of the model 6.1 Failure condition The application of the simulation model can be demonstrated by the simulation of a transient condition. The case where the power connection to one of the bogies is disrupted during normal working conditions is investigated. A simulation was done for the case where the locomotive is operated with a d.c. link current of 800A, and the chopper frequency of 298Hz. Once steady state is reached the current to the second bogie is interrupted (Swl in Figure 3), and the simulation continues for a further period. A moving window FFT [10] is then performed on the input current of the locomotive to determine the frequency spectrum as a function of time. The results are shown in Figure 10. Figure 10 Time variation of the frequency components of a transient condition in the Class 14E From the figure a 600Hz component is observed which corresponds to twice the chopper frequency (choppers on the bogies are 180 out of phase). The other frequencies are relatively low as would be expected. At the instant of tripping, all frequencies increase because of the transient present in the line current. The amplitude of these frequency components decay with time as expected. A constant frequency component at ±300Hz (one chopper at 298Hz) emerges after the instant of tripping.

9 If the 50Hz component is extracted from the graph of Figure 10 an amplitude versus time waveform can be constructed and shown in the graph of Figure 11, as a solid line. A smooth approximation of the wave decay is also drawn in on the graph to ease the estimation of the duration, and is shown as a dotted line. From this graph a duration versus amplitude curve can be constructed. The results of this are shown in Figure 12 together with the permissible limit for the 50Hz track circuit. As can be seen from the graph, the duration of the interference signal is more than that allowed for the 50HZ AC track circuit. It can therefore be concluded that tripping of a second bogie can present an interference problem. 6.2 FMECA 20 c15 <D 10 = 5 'm DC. Computers in Railways Time (S) Figure 11 Time variation of the 50Hz component in the locomotive current Permissible limit for 50Hz AC track <25 ~20-0) 15- = (0 10 * 5-0- The failure mode discussed must then be rated in the FMECA in order to determine the criticality of the failure. This is done in terms of 3 dimensions namely probability, severity and Time (S) detectability, and a ten point scale Figure 12 EMC assessment in terms of is used. The occurrence of this duration. failure is quite probable because tripping of the bogie under failure conditions is normal. The detectability of this failure is fairly good, since a 50% reduction in power will occur. The severity is extreme because the failure can cause a wrong side failure in the signalling system. Applying this, a criticality rating of approximately 400 is obtained. In signalling terms this is a very high rating and something must be done. 6.3 Possible solutions The criticality rating can be reduced in a number of ways. At this stage the possibility of delaying the operation of 50Hz AC track circuits on the line where the locomotive is to be used is considered. Another possibility is to trip the main circuit breaker on the locomotive when power to any one bogie is interrupted. This will remove the input filter components from the overhead line and prevent resonance as experienced in the example above.

10 452 Computers in Railways 7 Conclusion Many failures on the locomotive will not cause an interference problem on the signalling system, but some failures will. Measurement of the effect of these failures is in many cases not possible, and therefore calibrated simulation models becomes a cost effective alternative. It has been demonstrated that the use of simulation models for locomotives in the assessment of electromagnetic compatibility with the signalling system is possible and indeed very effective. 8 Acknowledgement The authors would like to thank the Signals section for permission to publish the paper and the Rand Afrikaans University for their assistance to develop the simulation models. It should be noted that the views expressed in this paper are the opinion of the authors and do not necessary reflect the policies of the Signals section of Spoornet. 9 References 1 Ad Sparrius. Anthropocentric Project Management, Ad Sparrius Systems Engineering and Management, South Africa, pp Ford, R: Can Traction and signalling coexist, Proceedings of The Institution of Railway Signal Engineers, 1994/5, pp Ford, R.: Eurostar testing delayed by signalling interference glitch, Modern Railways, December 93, pp Steyn, B.M., J.D. van Wyk. An Experimentally calibrated locomotive simulation model for the determination of harmonic content in the traction current, Proceedings of the IEEE /CT/fS" KT, Bologna, September 1994, pp Leuven EMTP Centre, Alternative Transient Program rule book, Updated September 1991, Printed July 1987, Belgium. 6 Dommel H.W., et. al. Electromagnetic Transient Program Reference Manual (EMTP Theory Handbook), Bonneville Power Administration, Portland, USA, Ausust Carroll, D.P., S.S. ABD-El-Hamid et al, A simplified analitical model for a current fed force commutated converter, IEEE Transactions on Industry Applications, Vol IA-16, No. 4, July 1980, pp Subrahmanyam, V. Steady state analysis of an induction motor fed from a current-source inverter using complex-state (Park's) vector, IEE Proceedings, Vol. 126, No.5, June 1979, pp Creighton, G.K.: Current-source inverter-fed induction motor torque pulsations, IEE Proceedings, Vol. 127, Part B, No. 4, July 1980, pp Steyn, B.M. Electromagnetic compatibility of power electronic locomotives and railway signalling systems, Chapter 8, D.Eng. diseretation, Rand Afrikaans University, November 1995.