This action might not be possible to undo. Are you sure you want to continue?
Applications of a new EMTP line model for short overhead lines and cables
A.I. Ibrahim a,*, S. Henschel b, A.C. Lima b, H.W. Dommel b
Power Technologies Inc., 7677 East Berry Avenue, Greenwood Village, CO 80111, USA Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC, Canada V6T 1Z4 Received 3 May 2001; revised 20 September 2001; accepted 11 October 2001
Abstract This paper presents the applications of a new EMTP line model for the representation of short overhead transmission lines and cables. This line model overcomes the limitation of using a time step size not larger than the travel time. The interpolation errors inherent in this line model produce a ®ltering effect for higher frequencies. The new line is tested for two case studies. The ®rst one is a case of transmission line energization. Simulation results for this case show that the new line model gives results for a short line which are close to the constant parameter line model results, while results with a p-circuit show unrealistic high frequency oscillations. The new transmission line model is also tested for the case of a drive system, which involves power electronics devices. In this case, the new line model is used to represent a short cable and good results are obtained. The p-circuit and the constant parameter line models would give unrealistic voltage waveforms. q 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Transmission line modeling; Electromagnetic transients; EMTP simulation
1. Introduction For simulating short transmission lines and cables in electromagnetic transients programs of the EMTP type , nominal p-circuits are usually used when the travel time t is less than the time step size Dt. This p-circuit approximation is used because the constant parameter line models  as well as the frequency dependent line models [3,4] require that the time step should not be larger than the travel time. In this paper, a new modeling approach is presented which overcomes the time step size constraint for short lines. The constant parameter line model will be used to explain the approach, and to keep the derivation simple. The new line model was tested with two practical case studies. Simulation results show that the new line model may provide more accurate answers for transient studies as compared with the nominal p-circuit line model as well as the constant parameter line model.
2. Transmission line model for long lines Before introducing the new line model, let us review the lossless line model that is implemented in the EMTP, based on Bergeron's method . Fig. 1 shows the equivalent circuit from which the model's most important property is immediately clear: both terminals k and m are galvanically separated. The currents in the two terminals k and m are given by: " i k t i m t # " Gc 0 0 Gc #" vk t vm t # 1 " hk t hm t # 1
where the history vector is known from the currents and voltages of preceding time steps " hk t hm t # " 2 " Â 0 Gc Gc 0 #" vk t 2 t vm t 2 t # 2 " 0 1 1 0 2 #
ik t 2 t im t 2 t
* Corresponding author. Address: Power Technologies Inc., 8310 South Valley Highway, Suite 250, Englewood, CO 80112, USA. Tel.: 11-303741-7902; fax: 11-303-741-7909. E-mail address: firstname.lastname@example.org (A.I. Ibrahim).
and where Gc is the reciprocal of the surge impedance Zc: p G c C 0 =L 0 3
0142-0615/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved. PII: S 0142-061 5(01)00078-3
(2) with the interpolated values. and b t=Dt: By replacing the past values in Eq. Eq. These interpolated values can then be used to compute the history vector in Eq. Ibrahim et al. those conditions at the far end at time (t 2 t ). can be retrieved from `history' tables. With this rule. (1) and (2). (2) shows that the conditions at one end depend on what happened at the other end at travel time t earlier. The interpolation formula is as follows: x t 2 t ax t 1 bx t 2 Dt 4 where a Dt 2 t=Dt. The errors of the new line model are derived here for the open. It is therefore best to show them in the frequency domain. This paper concentrates on using the new line model for electromagnetic transient studies. The polyphase representation is discussed in more detail in Ref. Lossless line model using Bergeron's method. The equivalent circuit of this new line model for the single-phase case is shown in Fig. the EMTP automatically replaces them with Eqs. and the results are compared with those of the exact solution. The new line model Eqs. . Fig.I. whose values are obtained from the previous time step. is equal to: " #" " # # hk t a 21 vk t 2 Dt bGc 1 2 a2 2 1 a hm t vm t 2 Dt " #" # a 21 ik t 2 Dt b 6 1 1 2 a2 2 1 a im t 2 Dt From Eq. t . otherwise. (4) produces errors that depend on frequency. they could be linearly interpolated to yield x(t 2 t ) provided that Dt $ t . . (5) and (6) can be transformed into the frequency domain as follows. we assume that a variable changes linearly as a straight line between two adjacent points.640 A. 1. The model equations were derived and implemented in MicroTran (UBC-version of the EMTP). we can observe that the conductance matrix also possesses off-diagonal elements. The variable x could be either a voltage or a current. if two solutions x(t) and x(t 2 Dt) were known. The new line model for short lines The concept of this new line model was suggested in Ref. 2. . (6). / Electrical Power and Energy Systems 24 (2002) 639±645 Fig. so that the history memory of this model is extremely short. 2. Therefore. The history vector of the new line model requires only values of the previous time step. Lossless transmission line model. we can obtain a new model for the lossless line: " # # " # " #" vk t ik t hk t 1 1 a2 22a Gc 1 1 2 a2 22a im t vm t hm t 1 1 a2 5 The history vector. which appear at the near end at instant t. Eqs. (7) and (8) by ejvt . Interpolation error analysis Interpolation with Eq. The integration method used in the EMTP is the trapezoidal rule of integration. (5) and (6) are only used if Dt . 4. we obtain the phasor equation: " # " #" # Ykk Ykm Vk Ik 9 Im Ymk Ymm Vm . so that the line equations can no longer be solved independently for both ends. As long as the step size Dt is less than the travel time t . (2). with I and V being phasors: " # " #" # Ik ejvt Vk ejvt 1 1 a2 2 2a Gc 1 2 a2 2 2a Im ejvt Vm ejvt 1 1 a2 " # Hk ejvt 1 7 Hm ejvt where " Hk ejvt Hm ejvt # bGc 1 2 a2 3 #2 V k e j v t 2 D t 4 5 21 a V m e j v t 2 D t 3 " #2 a 21 I k e j v t 2 D t b 4 5 1 1 2 a2 2 1 a I m e j v t 2 D t " a 21 8 After dividing Eqs. 3.and short-circuit responses. The use of this new line model for stability simulations with variable step size was described in Ref.
4. (13) and (14) represent the short. In a time domain simulation with Dt $ t . with that of the new line model. The time domain simulation for the currents Im and Ik at this point (Dt 1:3t and f 1000 Hz) are shown in Figs. 5 and 6. / Electrical Power and Energy Systems 24 (2002) 639±645 641 where Ykk Ymm 2Gc a2 1 2ab e2jvDt 1 b2 e22jvDt 1 1 10 a2 1 2ab e2jvDt 1 b2 e22jvDt 2 1 22Gc a 1 b e2jvDt a2 1 2ab e2jvDt 1 b2 e22jvDt 2 1 Ykm Ymk 11 The exact solution can be obtained from: 2 3 cos vt 21 " # " # 6 j sin vt j sin vt 7 Vk Ik 6 7 Gc 6 7 4 21 cos vt 5 Vm Im j sin vt j sin vt 12 If the receiving end is short-circuited. On the other hand. The current Im response in the time domain. The open-circuit ratio. Fig. for the same step sizes (dashed lines from top to bottom). The results obtained in the frequency domain can also be duplicated in the time domain at a given frequency using the EMTP. we can calculate the short-circuit response for the Fig. respectively. if the receiving end is open-circuited. The comparison between the time domain simulation and the frequency domain results shows that both answers are practically identical. which becomes more emphasized as the simulation step size is increased. Fig. 1. with that of the new line model. the receiving end current equals zero Im 0 and the opencircuit voltage ratio is given as follows: Vm Y 2 mk Vk Ymm 14 Eqs. 1.9t and 2. Ibrahim et al. 3 compares the magnitude of the short-circuit response for the exact model (solid line).and open-circuit responses of the transmission line model at any given frequency v . it can be seen that the short. the receiving end voltage equals zero Vm 0 and the short-circuit current ratio is given as follows: Im Y kk Ik Ymk 13 Fig. comparisons with frequency dependent line models should be done in future research. .I.3t .6t . for step sizes Dt equal 1. 3 and 4. 3. From these two ®gures.and open-circuit responses of the new line model show a strong ®ltering effect.2t (dashed lines from bottom to top). 4 compares the magnitude of the open-circuit response for the exact model (solid line). To prove this. The results with this ®ltering effect may come closer to reality. 5. From the results shown in Figs. The short-circuit ratio.A. this effect results in averaging of the transient voltage and current oscillations. The values were obtained for Z 100 V and t 50 ms: Fig.
85 ms. Schematic of a lossy line model. The main and auxiliary contacts of the circuit breaker close at times speci®ed in Ref. the transmission line is divided into two sections.642 A. we found that the constant parameter line model provides results which come Fig. A single-phase to ground fault is assumed to occur in phase A at 10 ms. Case studies In order to demonstrate the usefulness of the proposed line model.15 and 15. a time domain simulation of transmission line energization and adjustable speed drive (ASD) cases are used. as shown in Fig. the p-circuit and the new line model. The results and analyses are presented in this section. 6. 8. while the shorter section is to be represented by different line models.  for an unfaulted line energization case. was implemented in MATLAB environment and it is presented in Ref. The current Ik response in the time domain. 6. From the ®eld test results for the case of the unfaulted line energization. given point using Eq. The lossy line model was implemented in MicroTran (UBC-version of the EMTP). 8. Fig. between 7. / Electrical Power and Energy Systems 24 (2002) 639±645 two lossless line sections. 6. The lossy line model To represent the losses approximately. 7. . 7. (13). The results obtained are practically equal to the short-circuit response that is measured in the frequency domain as I m =Ik < 0:98: The interpolation error analysis for the new line model. Transmission line energization Fig. 5.I. . as compared with other line models and the exact solution. The data come from the Jaguara±Taquaril tests conducted by a Brazilian utility company [8±10]. To show the performance of the new line model as compared with other EMTP line models. This approach is also used for the polyphase case . lumped resistances are inserted at both ends and in the middle of A transmission line energization case is shown in Fig. The connected subroutine for this implementation is presented in Ref. including the constant parameter line model. Ibrahim et al. . Transmission line energization.1. The longer section is represented by a constant parameter line model.
The basic components of this adjustable drive system are shown in Fig.2. From Fig. This is because the nominal p-circuit model is not accurate enough to represent the high frequency transients involved in the switching action. It has a step-up transformer 480:1350 V. so as to produce as much as possible a ripple free voltage (or current). we can see that the new line model is more accurate than the p-circuit for the representation of short lines. Also. ASDs generate harmonics.31 ms. / Electrical Power and Energy Systems 24 (2002) 639±645 643 Fig. Fig. Harmonic instability can arise when anti-resonances in the supply system coincide with a multiple or sub-multiple of the resonance in the drive system. 11. ASDs are used because they allow a more ¯exible motor operation. 9. The simulation step size is chosen to be 4 ms. Fig. The inverter has usually switches which are capable of conducting and interrupting the current at predetermined times. The input voltage (or current) in the inverter will contain characteristic harmonics of the diode or thyristor bridge. above and below rated speed. we can observe that the response of the nominal p-circuit line model does not match the results of the constant parameter line model. which may coincide with a multiple (or sub-multiple) of the supply frequency during some speci®c operating frequencies. 10 shows the response produced by the new line model for a step size slightly larger than the travel time of 4. 10. Since the converter is operating at a very low switching frequency. as the constant parameter line model requires that the simulation step size should not be larger than the travel time of the shortest line. in ASD which are often found in oil exploitation and mining . Ibrahim et al. which are analyzed in this study. the modeling with ideal switches is suf®ciently accurate in Fig. 11. From Fig. close to the ®eld test results . together with the characteristic harmonics generated by the inverter.I. This response is simulated for the same step size as for the new line model. Experience has shown that. The motor is connected through a 2. The constant parameter model response. 9. The drive system studied is a square wave inverter with 480 V output. From these results.5 km long submarine cable. all semiconductors were treated as ideal. The response of the nominal p-circuit line model is shown in Fig. The response of the constant parameter line model is shown in Fig. The new line model response. for example. 6. The harmonics on the inverter side are re¯ected through the DC link to the converter input. The DC link has the function of ®ltering the recti®er output. we can see that the response of the new line model agrees very well with that of the constant parameter line model. . The performance of the new line model for the case of the faulted line energization is compared with the response of the constant parameter line model. 10.A. it fails to duplicate the maximum peak overvoltage which are shown for both the new line model and the constant parameter line model. Power electronics case study Power electronics devices are becoming very common in industrial power systems. Besides small resistances to account for conduction losses. The travel time of the short line is 4. with a 135 HP motor rated 950 V. 12. The aforementioned phenomena stress the importance of representing both converter stages (recti®er and inverter). low harmonics are generated. for studies of the complete drive system. This leads to inter-harmonics. The p-circuit line model response.31 ms. 11.
13. t ). 12. Fig. / Electrical Power and Energy Systems 24 (2002) 639±645 Fig. 16 shows the simulation results for the constant parameter line model for a step size of 18 ms (Dt . This response is simulated with the same step size of the new line model of 20 ms (Dt . 13 shows the voltage waveform of the ®eld test recording conducted at the given drive system. The ®ltering effect for high frequencies seems to be similar to the damping effect caused by skin effect in the actual cable. . t ). Fig. 15. The ®eld test waveform is asymmetric due to grounding problems while measuring the voltage. Fig. Fig. The ®eld test recording for the drive system. These measurements were done on an oil platform where ¯uctuations can occur in the ground voltage . 14 shows the simulation results with the new line model used for the representation of the submarine cable. because this version uses the critical damping adjustment (CDA) scheme to eliminate numerical oscillations. we can observe that the responses of these two models are unrealistic.I. 13 with the responses of the p-circuit and the constant parameter line models. 14. 14. Simulation results for the constant parameter model. Fig.644 A. Other EMTP versions may require more complicated modeling to dampen numerical oscillations. Fig. Fig. 16. while the simulation step size is chosen to be 20 ms (Dt . The power electronics case. Ibrahim et al. at the sending end of the submarine cable between phases A and B.31 ms. t ). From Fig. Fig. Simulation results for the p-circuit line model. 15 shows the simulation results for the p-circuit line model. Simulation results for the new line model. When we compare the ®eld test recording of Fig. we can observe that the proposed model gives very reasonable answers. The travel time for the short cable is 18. MicroTran.
Simulation and Control. 1992. In a comparison with actual ®eld test measurements. 2nd ed.I. New York: Wiley. Lima A. Henschel S. Vancouver. BC. Full frequency-dependent phase-domain transmission line model.A. / Electrical Power and Energy Systems 24 (2002) 639±645 645 7. It overcomes the limitation of using a time step size not larger than the travel time. International Power System Transients Conference. Dommel H. Int J Elect Power Energy Syst 1999. Canada: Microtran Power System Analysis Corporation. Acknowledgements The authors gratefully acknowledge the ®nancial assistance of the Natural Sciences and Engineering Research Council of Canada. University of British Columbia. Dommel HW. Modelling the electrical drive system for oil exploitation. Dommel HW. Castellanos F. June 1997. Electromagnetic Transients Program (EMTP). Computer simulation of ®eld tests on the 345 kV Jaguara±Taquaril line (in Portuguese). An intelligent support system for the analysis of power system transients. and of BC Hydro & Power Authority. BC. presented at Á õII Seminario Nacional de Producao e Transmissao de Energia Eletrica in Belo Horizonte. Digital computer solution of electromagnetic transients             in single. Case studies for electromagnetic transients.12(3):1331±9. Dommel H. Canada: Microtran Power System Analysis Corporation. The new line model with Dt $ t will be extended to the more accurate frequency dependent line model that has been used extensively for cases with Dt . Ibrahim et al. Pedroso A. IEEE Trans Power Syst 1997. The error analysis for the short. available from Dommel HW). IEEE Trans Indust Electron 2000. Canada. Brazil. Ibrahim A.PAS-101:147±57. Bergeron L. September 2000.21:191±8. Transmission line model for variable step size simulation algorithms. 1993. Vancouver.and multiphase networks. PhD Thesis. p. Marti J.and open-circuit responses shows that the new line model has a ®ltering effect for higher frequencies. Conclusions This paper describes the application of a new EMTP transmission line model for the representation of short overhead lines and cables. through funding provided for the NSCERC-BC. Dommel HW. t . Hydro Industrial Chair in Advanced Techniques for Electric Power Systems Analysis. Ibrahim AI. 1961. Water hammer in hydraulics and wave surges in electricity. whereas both the p-circuit and the constant parameter line models produced unrealistic high frequency oscillations. IEEE Trans Power Appar Syst 1974. Department of Electrical and Computer Engineering. Numerical modelling of frequency dependent transmission line parameters in an electromagnetic transients program. EPRI Report EL-6788s. IEEE Trans Power Appar Syst 1982. Stephan R. EMTP Field Test Comparisons.47(3):549±56. Marti J. 234±9. EMTP theory book. Stephan R. 1990. 1973 (Portuguese paper or English translation by Cameron DI. Modelling adjustable speed drives with long feeders. . Lima A. Paper BH/GSP/12.PAS-93:1401±9. References  Dommel HW. Meyer W. Dommel HW. Cunha CAF. The new line model was tested with two practical case studies. it could be shown that the new line model is a suitable for an accurate representation of cables. Accurate modelling of frequency-dependent transmission lines in electromagnetic transients simulations. BC.PAS-88(4):388±99. This model is an extension of the constant parameter line model of the EMTP. IEEE Trans Power Appar Syst 1969. Vancouver.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.