Simulation of line fault locator on HVDC Light electrode line

Transcription

1 August 10, 2010 Simulation of line fault locator on HVDC Light electrode line Andreas Hermansson BACHELOR S THESIS Electrical Engineering, Electric Power Technology Department of Engineering Science

2 BACHELOR S THESIS Simulation of line fault locator on HVDC Light electrode line Summary In this bachelor thesis, cable fault locators are studied for use on the overhead electrode lines in the HVDC (High Voltage Direct Current) Light project Caprivi Link. The cable fault locators studied operates with the principle of travelling waves, where a pulse is sent in the tested conductor. The time difference is measured from the injection moment to the reflection is received. If the propagation speed of the pulse is known the to the fault can be calculated. This type of unit is typically referred to as a TDR (Time Domain Reflectometer). The study is performed as a computer simulation where a simplified model of a TDR unit is created and applied to an electrode line model by using PSCAD/EMTDC. Staged faults of open circuit and ground fault types are placed at three s on the electrode line model, different parameters of the TDR units such as pulse width and pulse along with its connection to the electrode line are then studied and evaluated. The results of the simulations show that it is possible to detect faults of both open circuit and ground fault types with a suitable TDR unit. Ground faults with high resistance occurring at long s can be hard to detect due to low reflection s from the injections. This problem can somewhat be resolved with a function that lets the user compare an old trace of a healthy line with the new trace. The study shows that most of the faults can be detected and a to the fault can be calculated within an accuracy of ± 250 m. The pulse width of the TDR needs to be at least 10 µs, preferable 20 µs to deliver high enough energy to the fault to create a detectable reflection. The pulse seams to be of less significance in this simulation, although higher pulse is likely to be more suitable in a real measurement due to the higher energy delivered to the fault. The Hipotronics TDR 1150 is a unit that fulfil these requirements and should therefore be able to work as a line fault locator on the electrode line. Date: August 10, 2010 Author: Andreas Hermansson Examiner: Lars Holmblad Advisor: Sören Nyberg, ABB AB Programme: Electrical Engineering, Electric Power Technology Main field of study: Electrical Engineering Education level: Bachelor Credits: 15 HE credits Keywords Line fault locator, TDR, Arc reflection, simulation, electrode line, HVDC Publisher: University West, Department of Engineering Science, S Trollhättan, SWEDEN Phone: Fax: Web: i

3 Preface I would like to thank Anna-Karin Skytt and Sören Nyberg at ABB AB who made this bachelor s thesis possible. Andreas Hermansson Ludvika 9 April 2010 ii

4 Contents Summary... i Preface... ii Abbreviations... iv 1 Introduction Background Project description Methodology Theory Travelling waves Description of the time domain reflectometer (TDR) Description of the impulse thumper Description of the arc reflection method Design Modelling of electrode line Modelling of line fault locator Time domain reflectometer Arc reflection Fault model Simulations Fault type Low resistance to ground High resistance to ground Open circuit width Connection Parallel/single line Electrode station configuration Arc reflection Results and analysis Fault type Low resistance ground fault High resistance ground fault Open circuit width Connection Parallel/single line Electrode station configuration Arc reflection Conclusions and future work References iii

5 Abbreviations HVDC High Voltage Direct Current IGBT Insulated Gate Bipolar Transistor LFL Line Fault Locator TDR Time Domain Reflectometer iv

6 1 Introduction 1.1 Background Caprivi Link is a HVDC Light project that will be delivered to Namibia during HVDC Light is a power transmission system developed by ABB that is based on voltages source converters using IGBT semiconductors and high frequency pulse width modulation to convert from AC to DC and back again. The Light version is available in the range from 50 to 1200 MW [1]. The project consists of converter stations, in Zambezi and Gerus, connected by two overhead lines approximately 970 km long. The link has a power rating of 300 MW operating at a DC voltage of 350 kv. This is the first HVDC Light project to be built with overhead lines. At a first stage one monopole is delivered but there is an option to expand to a bipole. This transmission link is very important for the power supply of the country as the networks in the region are very weak [2]. There are three different modes for the operation of the link during the monopole stage. These modes are Single line metallic return (the return current goes through a metallic return conductor), Single line ground return (the return current goes through ground electrodes) and Parallel lines ground return (the return current goes through ground electrodes). 970 km DC OH-line U dn = kv I dn = 857 A 50 km electrode lines to ground electrodes Outdoor AC filters Indoor AC filter IGBT valve Smoothing reactor DC capacitor Converter reactor Converter transformer Figure 1.1 Principal of operation in the ground return modes. The preferred operation of the link is in the Parallel lines ground return mode to minimize losses due to the resistivity in the pole lines. The return current will, in this case, go through ground electrodes located approximately 50 km from each converter station, see figure

7 1.2 Project description A fault on the pole line can be detected on-line by detecting the incoming travelling waves in each station, by comparing the arrival time of waves in the two stations the location can be determined. However, as the voltage on the electrode lines is approximately zero or very low, this method is not applicable. The aim of this study is to find equipment that can detect the location of the fault along the electrode line. It can be off-line equipment, i.e. does not need to locate the fault when the link is in operation. The fault locator will inject a pulse into the line and detect the time when the reflected pulse comes back. The simulations should answer a number of issues which affect the ability to locate a fault. How different types of faults, s to faults, pulse, pulse width affect the ability to locate a fault. The simulation should also elucidate the existence of any interference from parallel lines 1.3 Methodology Since the electrode line is still under construction and a similar overhead line is hard to get access to for field measurements, the study will be performed as a simulation using PSCAD/EMTDC. PSCAD/EMTDC is simulation software for design and verification of power systems. EMTDC is a simulation engine and PSCAD is a graphical user interface, see figure 1.2. PSCAD/EMTDC is suitable for simulating electromagnetic transients or instantaneous solutions. Simplified models of the line fault locators are constructed from manufacturer s datasheets and information from their sales representatives. The LFL models are applied to electrode line models with staged faults. Different parameters are varied one at a time to study the behaviour of the LFL. 2

8 Figure 1.2 The PSCAD simulation environment. 3

9 2 Theory 2.1 Travelling waves Any electrical charge applied to a conductor, such as a lightning stroke or the closing of a circuit breaker, causes a rise of the potential of the conductor. This produces travelling waves of voltage e and current i that propagates along the conductor at the speed of light. The voltage and current waves are related by a surge impedance Z. For a lossless transmission line, Z is dependent only to the inductance and capacitance according to equations (1). e Z = i Z = L C If a voltage and current wave hit a point of discontinuity in the impedance, while propagating along a conductor, voltage and current waves are reflected backward and transmitted forward from the point of discontinuity. I.e. consider two conductors in series with surge impedances Z 1 and Z 2. Let the original forward waves be denoted as e and i, the waves reflected backwards from the discontinuity point be denoted e and i and the waves transmitted forward be denoted e and i. Then the normal equations (2) describe the travelling wave and the boundary equations (3) the relationship at the point of discontinuity [3]. Normal equations e = Z 1 e' = Z 1 e'' = Z i i' 2 i'' (1) (2) Boundary equations e'' = e + e' i'' = i i' (3) 4

10 Then in terms of e and i we get Z e' = Z 2 Z 2 e'' = Z + Z Z i' = Z Z + Z Z + Z 2 Z1 i'' = Z + Z e i i e Now we can draw the following conclusions from equations (4) about the current and voltage waves at the point of discontinuity: If Z 1 =Z 2, the reflected current and voltage waves are zero and the transmitted current and voltage waves are equal to the original current and voltage waves. If Z 1 <Z 2, the reflected current and voltage have a lower with the same signs as the original waves unless Z 2 approaches infinity then the reflections have the same, see figure 2.1. The of the transmitted voltage is in the range: e < e <= 2e with the same sign and the of the transmitted current is in the range: 0 <= i < i with the same sign. (4) Figure 2.1 Voltage reflections at a point of discontinuity where Z1<Z2. If Z 1 >Z 2, the reflected current and voltage waves have a lower with the opposite signs unless Z 2 is zero then the reflected current and voltage waves have the same, see figure 2.2. The transmitted voltage wave is in the range: 0 <= e < e with the same sign and the of the transmitted current is in the range i < i <= 2i with the same sign. Figure 2.2 Voltage reflections at a point of discontinuity where Z 1 >Z 2 5

11 2.2 Description of the time domain reflectometer (TDR) A time domain reflectometer, TDR, or pulse echo meter is a device which can measure s to changes in the impedance of a conductor. The technique is based upon the principle of travelling waves. A short rise time pulse is injected in the conductor and it is partly reflected by impedance discontinuities in the line. The trace is then displayed on an oscilloscope screen. By measuring the travel time t between the pulse is sent and a reflection is received it s possible to calculate the to the discontinuity that caused the reflection. To calculate the to the fault the wave propagation speed v p has to be known. The wave propagation speed for an overhead transmission line is close to the speed of light and can be calculated by equation (5) [3], where L is the inductance and C is the capacitance per unit length. Then the d to the fault is given by equation (6). The wave propagation speed can also be measured with a known length of the line by extraction from equation (6), this will give the exact propagation speed of the conductor. 1 v p = (5) LC d v p t = (6) 2 By inspecting the waveform of the reflection it s possible to determine what type of fault caused the reflection. From the conclusions in chapter 3.1 a short circuit is represented by a negative reflection and an open circuit represented of a positive reflection. For a high resistance shunt fault it s likely that the of the reflection is too small to be detectable. To resolve this problem some TDR units have a storage function so that an old trace can be compared with a new one, there deviations between the two traces can point out a fault [4]. 2.3 Description of the impulse thumper An impulse thumper is a high energy capacitive discharge unit that transmit a high energy pulse between the faulty conductor and ground. The unit basically consists of a power supply, capacitors and a high voltage switch. When the pulse reaches the fault location it creates an arc which makes an audible thump. The fault can then be found simply by listening after the thump created by the arc with a microphone or a magnetic loop antenna, however this method is not applicable on an overhead line [4]. 6

12 2.4 Description of the arc reflection method The arc reflection method combines the use of a high capacitive discharge unit and a TDR. The thumper and the TDR are connected via a filter that protects the TDR from the high voltage pulses created by the thumper. When the energy pulse reaches a high resistance fault it creates an arc that lowers the resistance of the fault. Simultaneously the TDR unit can measure the to the arc [4]. 7

13 3 Design 3.1 Modelling of electrode line The electrode line is connecting the converter station to the electrode station. The line consists of two parallel overhead lines and is approximately 50 km long. ch line is of the type Rail which has an aluminium conductor with a steel reinforcement, see table 3.1 for conductor data. The conductors are located 23 m above ground level and 1.76 m apart at the towers, no shield wires are used, see figure 3.1. ch conductor can be individually disconnected at both converter and electrode station making it possible to operate in ground return mode with only one conductor. This means that a faulty conductor can be measured without taking the electrode station out of service. C1 C Tower: Yiduell Conductors: Rail 0 Ground Resistivity: 50 [ohm*m] Relative Ground Permeability: 1.0 rth Return Formula: Analytical Approximation Figure 3.1 Cross section of the electrode line model. Table 3.1 Data for the two conductors in the electrode line. Conductor Aluminium area [mm 2 ] Equivalent copper area [mm 2 ] Current rating [A] DC resistance at 70 C/km [Ω] C1/C ,0722 The electrode line model consists of an ideal current source and two frequency dependent transmission line segments connected through ideal conductors. Table 3.2 shows the settings of the transmission line model used in PSCAD/EMTDC. An electrode resistance of 0.2 Ω has been used, which is the highest allowable resistance for the ground electrodes. Two transmission line segments are used so that a fault can be staged anywhere along the line by adjusting the individual length of the two segments. Figure 3.2 shows the electrode line model with one of the two conductors disconnected at both the converter station and the electrode station, a ground fault has also been placed between the two line segments. 8

14 Table 3.2 Transmission line model settings in PSCAD/EMTDC. Frequency dependent (phase) model options Travel time interpolation Curve fitting starting frequency Curve fitting end frequency On 1 [Hz] [Hz] Total number of frequency increments 1000 Maximum order of fitting for Ysurge 20 Maximum order of fitting for prop. func. 20 Maximum fitting error for Ysurge 0.2 [%] Maximum fitting error for prop. Func. 0.2 [%] 01 [ohm] 1 1 Fault [ohm] Tlinezambezi Tlinezambezi Tlinezambe2 Tlinezambe2 01 [ohm] I1 01 [ohm] T TLinezambezi T TLinezambe2 Figure 3.2 The electrode line model in PSCAD consists of a grounded current source, two transmission line segments with a ground fault applied between them and a resistor connected to ground representing the electrode. 9

15 3.2 Modelling of line fault locator Time domain reflectometer The TDR models are based on the Hipotronics TDR 1150 and the Megger MTDR. Data for the two models are presented in table 3.3. Table 3.3 Data for the TDR units [5] [6]. Model TDR 1150 MTDR Max. pulse width [µs] Max. range [km] Sampling rate [MHz] Voltage input protection level The TDR model consists of a voltage source, a circuit breaker and a voltmeter, see figure 3.3. The circuit breaker is connected to a timing device so that the pulse width can be controlled. Initial simulations showed a DC voltage on the electrode line of approximately 10 V. A 10k Ω resistor is therefore connected parallel to the voltmeter to resolve this problem. The reading of the voltmeter is displayed on a graph that represents the oscilloscope of the TDR. The time interval of the graphs is chosen so that the whole length of the line is shown on the graphs. Two cursors are then placed at the graph, one at the beginning of the generated pulse and one at the beginning of the reflection. The time interval between the two cursors t is then used to calculate the using equation (6). 50 [ohm] BRK1 10 [kohm] BRK1 Timed Breaker Logic Open@t0 Figure 3.3 The TDR model, there is the measured voltage. 10

16 3.2.2 Arc reflection The Arc reflection model is based on the Hipotronics and the Megger PFL 40 series. Data for the two models are presented in table 3.4. Table 3.4 Data for the Arc reflection units [5] [6]. Model Voltage [kv] Stored energy [J] PFL 40 8/16/ The arc reflection model consists of a voltage source, a capacitor, two circuit breakers and a voltmeter. The voltage source handles the charging of the capacitor. The capacitor stores the energy released by the line fault locator. Equation 7 describes the relationship between the energy stored in a capacitor and the voltage and capacitance. Circuit breaker one separates the capacitor from the voltage source and circuit breaker two releases the pulse. The voltmeter represents the oscilloscope of the TDR unit, see figure 3.4. W 2 C V = (7) 2 1 [ohm] BRK1 BRK2 C Figure 3.4 Arc reflection model. 11

17 3.3 Fault model The fault model for ground faults consists of a fault module that can be switched on and off. The fault module is connected in series with a resistor to ground. Two values of the resistance are used: 1 Ω representing a short circuit to ground and a higher value of 231 Ω which is the highest likely resistance to trip the differential protection system of the electrode line, see table 3.5. The electrode line has a longitudinal differential protection system that is used to detect ground faults on the line. The protection system measures the current in both ends of the line with DC current transducers. The system is set to alarm when a difference of 17 A is detected [7]. This means that ground faults generating a current of 17A and above should be localised. The resistance of a fault that can be detected varies along the line due to the voltage drop in the line. Close to the electrode the voltage to ground is lower than the voltage at the converter station and a ground fault has to be of lower resistance to be detected by the differential protection system. Table 3.5 shows the voltage to ground (acquired by simulations in PSCAD/EMTDC) and the maximum resistance of a ground fault that leads to an alarm (calculated by ohm s law) at increasing s from the converter station. The calculations is based on a worst case scenario with one of the electrode line conductors disconnected and the other operating at it s maximum current rating of 920 A. Table 3.5 Voltage to ground and the corresponding fault resistance that leads to an alarm in the differential protection system at increasing s from the converter station. Distance [km] Voltage [kv] Resistance [Ω] 0 3, , , , ,

18 4 Simulations Table 4.1 shows a list of the simulations. The simulations are done by changing one parameter at a time to be able to study changes. Table 4.1 Issues to investigate. No Item Issue Simulations 1 Fault type High resistance faults may be hard to detect. Is the of the reflection big enough to detect? 2 Distance to fault How does the from the measuring point to the fault affect the result? 3 width How do different pulse widths affect the result? 4 How do different pulse s affect the result? 5 Connection How does the connection of the TDR to the line affect the result? 6 Parallel line How does a parallel line in operation affect the result? Ground fault of 1 Ω Ground fault of 231 Ω Open circuit. 5 km 25 km 45 km 1, 10 and 20 µs 25 V 10 V Conductor-Ground mode. Conductor-Conductor mode. On Off 7 Electrode station configuration How does the connection of the line to the electrode affect the result? Connected Disconnected 8 Arc reflection Does a high voltage pulse increase the performance of the TDR? 30kV 2000J 4.1 Fault type As default settings, pulse of 25 V and a pulse width of 10 µs have been used. The TDR is connected between one conductor and ground as a default with the measured conductor disconnected at the electrode station. 13

19 4.1.1 Low resistance to ground Figure shows the voltage measured by the TDR from a low resistance fault of 1 Ω to ground at 5, 25 and 45 km Min 00 Max Time Figure 4.1 Low resistance ground fault at 5 km t = 33,0 µs Min -08 Max Time Figure 4.2 Low resistance ground fault at 25 km t = 165,0 µs Min Max Time Figure 4.3 Low resistance ground fault at 45 km t = 299,0 µs 14

20 4.1.2 High resistance to ground Figure shows the voltage measured by the TDR from a high resistance fault of 231 Ω to ground at 5, 25 and 45 km Min Max Time Figure 4.4 High resistance ground fault at 5 km t = 33,0 µs Min Max Time Figure 4.5 High resistance ground fault at 25 km t = 166,0 µs Min Max Time Figure 4.6 High resistance ground fault at 45 km t = 298,5 µs 15

21 4.1.3 Open circuit Figure shows the voltage measured by the TDR from an open circuit at 5, 25 and 45 km Min Max Time Figure 4.7 Open circuit at 5km t = 33,5µs Min Max Time Figure 4.8 Open circuit at 25km t = 165,0 µs Min Max Time Figure 4.9 Open circuit at 45km t=299,0 µs 16

22 4.2 width The pulse width is a way of controlling the amount of energy applied to a fault. A simulation of pulse width settings of 1 and 20 µs applied to a high resistance fault 25 km from the converter station is showed in figure 4.10 and The displayed values are the voltage measured by the TDR Min Max Time Figure 4.10 The reflection from a 1 µs pulse is too small to detect Min Max Time Figure 4.11 The reflection from a 20 µs pulse where t=165,5 µs. 17

23 4.3 The affects of higher or lower pulse is tested by applying a lower voltage of 10 V in addition to the previous 25 V simulations. The simulations are done with a high resistance fault, which is likely to be the hardest to detect, at 5, 25 and 45 km and a pulse width of 10 µs. Figure shows the result, where the displayed values are the voltage measured by the TDR Min Max Time Figure 4.12 High resistance ground fault at 5 km t=33,0 µs Min -59 Max Time Figure 4.13 High resistance ground fault at 25 km t=167,0 µs Min Max Time Figure 4.14 High resistance ground fault at 45 km t=298,5 µs 18

24 4.4 Connection By connecting the LFL to both conductors instead of one conductor and ground, another configuration is achieved. Simulations are done for a high resistance fault and an open line. Figure shows the high resistance simulations at 5, 25 and 45 km from the converter station. The displayed values are the voltage measured by the TDR Min Max Time Figure 4.15 High resistance ground fault at 5 km t=33,5 µs Min Max Time Figure 4.16 High resistance ground fault at 25 km t=165,5 µs Min -93 Max Time Figure 4.17 High resistance ground fault at 45 km t=299,0 µs 19

25 Figure shows the open circuit simulation of the conductor conductor mode at 5, 25 and 45 km. The displayed values are the voltage measured by the TDR Min Max Time Figure 4.18 Open circuit at 5 km t=33,5 µs Min Max Time Figure 4.19 Open circuit at 25 km t=165,0 µs Min Max Time Figure 4.20 Open circuit at 45 km t=298,5 µs 20

26 4.5 Parallel/single line Parallel lines give a risk of induced voltages that might affect the ability to make accurate measurements. Therefore a simulation is made with a change in the current in one conductor. Figure 4.21 shows the induced voltage at the converter station when di/dt changes from 2 A /ms to 0 A /ms in the parallel conductor. The displayed value is the voltage measured by the TDR Time Figure 4.21 The induced voltage at the converter station from a parallel line. A simulation with a single line as a reference and for future projects was also made with a high resistance ground fault at 5, 25 and 45 km. The results are shown in figure , where the displayed values are the voltage measured by the TDR Min Max Time Figure 4.22 High resistance ground fault at 5 km t=32,0 µs 21

27 Min Max Time Figure 4.23 High resistance ground fault at 25 km t=169,0 µs Min -30 Max Time Figure 4.24 High resistance ground fault at 45 km t=306,5 µs 4.6 Electrode station configuration In case of problems with the remote controlled disconnectors at the electrode station a simulation is made with the disconnectors closed. The simulations are made with a high resistance fault at 5, 25 and 45 km, see figure The displayed values are the voltage measured by the TDR. 22

28 Min Max Time Figure 4.25 High resistance ground fault at 5 km t=33,0 µs Min Max Time Figure 4.26 High resistance ground fault at 25 km t=166,0 µs Min -75 Max Time Figure 4.27 High resistance ground fault at 45 km t=299,0 µs 23

29 4.7 Arc reflection Simulations with a high voltage pulse representing the Arc reflection method are done to a high resistance ground fault at 5, 25 and 45 km, see figure The displayed values are the voltage measured by the TDR. y 4k 3k 3k 2k 2k 1k 1k k k -1k -1k -2k -2k -3k -3k -4k 069k k k Min k Max k Figure 4.28 High resistance ground fault at 5 km t=33,5 µs y 4k 3k 3k 2k 2k 1k 1k k k -1k -1k -2k -2k -3k -3k -4k 069k k k Min k Max k Figure 4.29 High resistance ground fault at 25 km t=166,0 µs 24

30 y 4k 3k 3k 2k 2k 1k 1k k k -1k -1k -2k -2k -3k -3k -4k 069k 372k 302k Min 069k Max k Figure 4.30 High resistance ground fault at 45 km t=297,5 µs 25

31 5 Results and analysis To be able detect a line fault with a LFL the reflection from an injected pulse has to be well defined with significant and a preferable low rise time. High of the reflection makes it easier to distinguish from smaller reflections caused by discontinuities in the line. A low rise time makes it easier to point out a more precise arrival time for the reflection and there by give a more accurate to the fault. The results from the simulations are presented in tables with the travel time for the pulse and the of the reflection. Further is the to the fault calculated by equation (6) and an error is calculated by subtracting the actual from the calculated. The propagation speed that has been used to calculate the s is 302,35 m/µs which been obtained as a mean value through simulations at 45 km. The odd result of a higher propagation speed than the speed of light probably comes from an inefficiently long solution time step setting of 0,5 µs. Simulations show that by shortening the solution time step a more accurate propagation speed can be obtained. However as a shorter solution time step increases the simulation time the longer time step has been chosen. 5.1 Fault type Low resistance ground fault The results of the low resistance simulations (figure ) are shown in table 5.1. From the results can we see that the reflections are of reversed polarity as indicated in section 2.1. The s of the reflections are high at a close but attenuate as the to the faults increases. The rise time of the fault reflections is low at a close but increases with increased. The error seams to increase with increasing and the maximum error is 201 m which is approximately 0,4% of the total length of the line. Over all the low resistance fault of 1 Ω is highly detectable in the simulation. Table 5.1 Results from the low resistance simulation. Fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , ,

32 5.1.2 High resistance ground fault The results of the high resistance simulations (figure ) are shown in table 5.2. From the results can we see that the reflections are of reversed polarity as for the low resistance fault. The s of the reflections are lower than for the reflections from the low resistance faults as expected from equations (4), due to the higher resistance. The rise time of the fault reflections is low at a close and increases with increased as for the low resistance faults. The maximum error is 126 m which is about 0,25% of the total length of the line. The low of the reflections at long s would make them hard to detect in a real world measurement. This is due to reflections from small in homogeneities in the electrode line that would cause a background noise and disguise the fault reflection. A function that can somewhat resolve this problem is the possibility to store a trace from healthy measurements. The new trace can then be compared with an old trace displaying any deviations between them, figure 5.1 illustrates this function. Table 5.2 Results from the high resistance simulation. Fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , , However considering table 3.5 in section 3.3, high resistance faults at long s would not be detected by the differential protection system and is therefore not required to be detected by the LFL [8] Eb Min Max Time Figure 5.1 A high resistance fault (blue) plotted over a stored trace of a healthy line (green) makes it easier to pinpoint difficult faults. 27

33 Table 5.2 Results from the high resistance simulation. Fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , , Open circuit The results of the open circuit simulations (figure ) are shown in table 5.3. From the results can we see that the reflections are of the same polarity as the original pulse, indicating that the fault is of a high resistance series character. The s of the reflections are higher at a close but attenuate as the to the faults increases as for the ground faults. The rise time of the fault reflections seams to be lower than for the ground faults although the error shows no sign of improvement with a maximum error of 201 m. The simulations show that an open circuit fault ought to be easier to detect than ground faults due to the high s of the reflections. Table 5.3 Results from open circuit simulation. Fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , , width The results of the pulse width simulations (figure ) that were applied to a high resistance ground fault are shown in table 5.4. The s of the reflections are higher with a longer pulse width, this is probably from the increased amount of energy delivered to the fault. The shorter pulse width of 1 µs is hardly detectable and a pulse width of at least 10 µs preferable 20 µs seams suitable for a longer overhead line. A possible application for a short pulse width is if the fault is close to the measuring point and longer pulse widths would disguise it. The rise time of the fault reflections seams to be unaffected by the longer pulse width. Further shows the error no difference between the two detectable traces. 28

34 Table 5.4 Results from the pulse width simulation. Fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , The results of the pulse simulations (figure ) that were applied to a high resistance ground fault are shown in table 5.5. The s of the 10 V pulse reflections are roughly in the same area as for the 25 V pulse s calculated as fractions of the original pulses. The rise time of the fault reflections seams also to be unaffected by differences in the pulse, this is somewhat expected due to the linearity of the fault model. Further shows the error no bigger difference between the different pulse s. The simulations show no difference between the pulse s and no conclusions can be made stating that higher or lower pulse is better. A possible reason for choosing lower pulse would be measurements on low voltage control cables reducing the risk of interference. However as the LFL is to be used on an overhead line this is not an issue of interest and higher pulse should be preferred due to the higher energy delivered to the fault. Table 5.5 Results from the pulse simulation. Fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , , Connection The results of the simulations with the TDR connected between the two conductors (figure ) are shown in table 5.6 for the high resistance fault and in table 5.7 for the open circuit. The reflections of the high resistance faults are lower at short s although they are more constant throughout the length of the line. The rise time of the fault reflections are also quite consistent with a well defined reflection even at the longer. The error shows no significant improvement to previous simulations. 29

35 Table 5.6 Results from the high resistance simulation. Fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , , The reflection s of the open circuit faults are high and the attenuation is low even at long s. The rise time of the fault reflections are quite consistent with a well defined reflection even at the longer. The error shows no significant difference in improvement. Table 5.7 Results from the open circuit simulation. Fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , , Parallel/single line The simulation with a parallel line in operation (figure 4.21) shows that a moderate change in the current will induces a voltage in the conductor connected to the TDR. The induced voltage is over 200 V which will probably make accurate measurements difficult and could cause possible damage to the TDR unit. A simulation was made with a single electrode line as reference and for future projects (figure ). The simulation was applied to high resistance ground fault and the result is shown in table 5.8. The simulation show that a single line has slightly less attenuation of the reflected wave compared to the case with two conductors. It should therefore be easier to detect faults on a single line, although the simulation also shows a significant decrease in the propagation speed at increasing line length. This results in a big error if no compensation in the propagation speed is made. 30

36 Table 5.8 Results from the single line simulation. Fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , , Electrode station configuration The results of the simulations with the disconnectors closed at the electrode station (figure ) are shown in table 5.9. The reflection s show the same result as for the high resistance faults with the disconnectors open. Further shows the simulations the same attenuation and rise time as the high resistance simulation. At long s the fault reflection can be disguised by the reflection from the electrode and measurements should therefore preferable be made with the disconnectors open. Table 5.9 Results from the electrode station simulation. Actual fault [km] width [µs] Reflection Travel time [µs] Calculated Error , , , Arc reflection The results from the Arc reflection simulations (figure ) applied to a high resistance fault are shown in table The simulations show no difference from the low voltage pulse simulations on high resistance faults. This is probably because of the linear fault model and a better nonlinear model should be developed if this method is to be simulated further. 31

37 Table 5.10 Results from the Arc reflection simulation. Fault [km] [kv] width [µs] Reflection [kv] Travel time [µs] Calculated Error , , ,

38 6 Conclusions and future work The results of the simulations show that it is possible to detect faults of both open circuit and ground fault types with a suitable TDR unit. Ground faults with high resistance occurring at long s can be hard to detect due to the low reflection s. This problem can somewhat be resolved with a function that lets the user compare an old trace of a healthy line with the new trace. The study shows that most of the faults can be detected and the to the fault can be calculated within an accuracy of ± 250 m, although further studies with field measurements and staged line faults should determine a more precise accuracy of the final LFL. The pulse width of the TDR needs to be at least 10 µs, preferable 20 µs to deliver high enough energy to the fault to create a detectable reflection. The pulse seams to be of less significance in this simulation, although higher pulse is likely to be more suitable in a real measurement due to the higher energy delivered to the fault. The Hipotronics TDR 1150 is a unit that fulfil these requirements and should therefore be able to work as a LFL on the electrode line. Further shows the simulation with the TDR connected between the two conductors that this is a possible way of decreasing the rise time of the reflections and thereby increasing the accuracy, however further field studies can clarify this. The simulations show a risk of high voltage transients if measurements are done with a parallel line in operation. This is due to induction from current changes in the parallel conductor and could cause possible damage to a TDR unit. In the case with a single conductor the simulation shows less attenuation of the reflections and a fault should therefore be easier to detect. Although a significant decrease of the wave propagation speed at increasing to the fault was detected. This results in a substantial error in the calculated if no compensations are made to the propagation speed. Simulations with the electrode line connected to the electrode shows that high resistance ground faults close to the electrode station can be disguised by the reflection of the electrode. Measurements should therefore preferable be made with the disconnectors open at the electrode station. The simulation of the Arc reflection method, which combines the use of a TDR and a high energy impulse generator, shows no improvement to the stand-alone TDR unit. This can be explained with the linear fault model and a better fault model has to be developed to study this method further. 33

39 References 1. ABB AB (2010) sy introduction for laypersons [Electronic] ABB AB Available: < [ ] 2. ABB AB (2010) Caprivi Link Interconnector [Electronic] ABB AB Available: < [ ] 3. Hileman, Andrew R (1999) Insulation Coordination for Power Systems Marcel Dekker 4. Gill, Paul (2008) Electrical Power Equipment Maintenance and Testing CRC Press 5. Hipotronics (2010) Products - Cable Fault Locating Equipment [Electronic] Hipotronics Available: < [ ] 6. Megger (2010) PFL40A-1500 [Electronic] Megger Available: < > [ ] 7. ABB AB 1JNL HVDC Protection System Unpublished manuscript ABB AB 8. ABB AB 06MR0005 Rev.00 Caprivi Link Interconnector Converter Stations Project Volume 3 Sec Unpublished manuscript ABB AB 34