IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 3, JULY

Transcription

1 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 3, JULY A New Technique to Detect Faults in De-Energized Distribution Feeders Part I: Scheme and Asymmetrical Fault Detection Xun Long, Student Member, IEEE, Wilsun Xu, Fellow, IEEE, and Yun Wei Li, Member, IEEE Abstract Re-energizing an overhead distribution feeder safely is a major consideration for a utility s safe work practice. One way to improve the safety is to determine whether the feeder still experiences short circuits before it is energized. In this paper, a novel fault detection technique is proposed to detect if a de-energized distribution system still experiences short-circuit faults. The proposed method involves injecting a thyristor-generated-controllable signal into the de-energized feeder. The feeder voltage and current responses are analyzed to determine if a fault still exists. A thyristor gating control strategy and fault detection algorithm are also developed in this paper to detect all possible types of faults that can occur in a system. The effectiveness of the proposed method has been verified through theoretical analysis, computer simulations, and lab tests. Index Terms De-energized distribution line, fault classification, fault detection, power electronics, safe recloser. I. INTRODUCTION R E-ENERGIZING or reclosing to a de-energized overhead distribution feeder safely is a major consideration for a utility s safe work practice [1] [5]. After a feeder is de-energized for an extended period due to events, such as repair, maintenance, or storms, there is always the possibility that humans or animals may be in contact with feeder conductors unknowingly [6], [7]. A reclosing action in such a situation can easily lead to fatalities [8], [9]. Utility companies are therefore very interested in techniques that can determine whether a de-energized feeder is clear of short circuits so that operators can re-energize the feeder with confidence. Compared to detecting faults in an energized system, to detect faults in a de-energized system is more challenging, since it requires the generation and application of a voltage signal to the de-energized feeder. Furthermore, the signal must be high enough to mimic the normal medium-voltage stress (e.g., 25 kv) to be experienced by the feeder. One way to generate the signal is through charging or discharging a capacitor in the circuit. For example, in [10], a cable tester uses a capacitor and Manuscript received October 03, 2010; revised January 07, 2011; accepted February 08, Date of publication April 07, 2011; date of current version June 24, This work was supported by icore. Paper no. TPWRD The aurthors are with the Department of Electrical and Computer Engineering, University of Alberta, Edmonton, AB T6G 2V4, Canada ( xlong@ualberta.ca; wxu@ualberta.ca; yunwei.li@ece.ualberta.ca). Color versions of one or more of the figures in this paper are available online at Digital Object Identifier /TPWRD diodes to create a high voltage charging circuit. If there is a short circuit, the charging circuit will not be able to reach a very high voltage. In [11], a portable tester utilizes a capacitor to discharge a high-voltage pulse into the de-energized line and estimates the voltage and current response. These devices are portable but they require maintenance of power source such as batteries. Moreover, the energy of the injected pulse is usually fixed and limited, which means these devices cannot be adapted for a system with different voltage ratings or used for a high impedance fault. Another way to generate the detection signal is to borrow energy from the upstream power system and, therefore, an extra power supply is not required. The fault detector in [12] produces the detection signal by closing the recloser for one cycle. If the current is too high, indicating the existent of a downstream short circuit, a series impedance will be inserted to reduce the inrush current. Similar to autoreclosers, this method still injects significant inrush current into the downstream before the insertion of the series impedance. In [13], a short-duration voltage from an electronic switch is applied to test if the downstream circuit is shorted. However, the device may not have enough rating for working in MV system, and the strength of the pulse is not adjustable (different types of faults require different signal strength (e.g., a strong signal is needed to detect a high impedance fault). Furthermore, another drawback for the aforementioned methods is that they cannot identify different types of faults in a single device. Recently, a pulse-recloser technique has been developed for the purpose of reducing the inrush current caused by reclosing to a faulted feeder [14]. This technique uses specialized recloser and is intended for fuse-saving-oriented feeder reclosing operations. In theory, the technique can be applied to the problem of concern in this paper (i.e., detecting feeder faults after the feeder has been de-energized for an extended period). However, its application requires replacing the existing feeder breaker or recloser. This can be costly to utility companies if implemented on a large scale. In view of the limitations of existing techniques, a new detection technique is proposed in this paper. A thyristor-based device is used to inject a controllable signal into the de-energized downstream to stimulate the electrical response. A significant feature of this technique is that the signal strength is adjustable by changing the thyristor firing angle. Therefore, a low-voltage pulse can be created to satisfy the safety requirement, and a high-voltage pulse can be produced to break down an insulted /$ IEEE

2 1894 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 3, JULY 2011 gap of a high-impedance fault when necessary. The proposed idea has another important feature, which is the ability to detect different kinds of faults by using a single device (in parallel with a recloser). A fault could be asymmetrical or symmetrical and it may occur between phase to ground or phase to phase. The fault-type detection ability of the proposed technique can greatly facilitate the utility to make proper action. The actual embodiment of the technique is a low voltage power electronic device connected to MV feeders through common utility primary-to-secondary service transformers. The device is installed permanently at the recloser or breaker locations and can be operated locally or remotely. As a result, there is no need to replace the existing breaker or recloser. The rest of the paper is organized as following. The proposed fault detection scheme is presented in Section II. In this section, the characteristics of the detection signal in different types of faults are also investigated and the gating and detection algorithm for identifying different faults are developed. To verify the proposed technique, computer simulation results based on a 25 kv distribution system are provided in Section III. Moreover, a low-voltage single-phase experimental system prototype has been constructed, and the experimental results are presented in Section IV. Finally, an alternative structure of the detection device with the same functionality but with fewer thyristors is introduced in Section V. This paper is concluded in Section VI. Fig. 1. Single-line representation of the proposed fault detection method. Fig. 2. Waveforms of: (a) thyristor voltage and current and (b) measured voltage and current at the point X. II. POWER-ELECTRONICS-AIDED FAULT DETECTION As mentioned earlier, the challenge for fault detection in a de-energized system is how to generate and inject an appropriate detection signal. The signal is preferably to be adjustable in a large range of magnitude. For example, at the beginning, a low-strength signal is desired, which can be considered as an alarm so that human or animals contacting with the conductors can sense it and get away to avoid injury. Then, the detection signal strength should increase gradually to stimulate detectable electrical response. It should be able to reach a very high-voltage to break down the insulated gap if there is a high-impedance fault. The proposed technique is able to generate this kind of signal by adjusting the thyristor s firing angle. The single line representation of the proposed method is shown in Fig. 1. A thyristor is connected in parallel to a circuit breaker or recloser by a switch. The switch is off in the normal system operation. When maintenance or repair at downstream is completed and the de-energized feeder needs to be restored, the line operator can turn on the switch and control the thyristor to trigger at several degrees before the voltage crosses zero. The energized upstream line is thus momentarily connected to the de-energized side so that a detection pulse is created in the downstream. The thyristor automatically shut off when its current drops to zero [15], [16]. A step-down transformer is used to decrease the voltage of the distribution line to a low level for thyristor operation and then a step-up transformer is used to restore the signal back to the system voltage level. Point X in Fig. 1 is the location for measuring the stimulated voltage and current signals. The typical thyristor voltage, current and the measured waveforms at point X are shown in Fig. 2. Fig. 3. Three-phase thyristor bridge-based fault detection scheme. In a three-phase four-line system, a fault could be symmetrical or asymmetrical and it may occur between phases or between phase and ground. Therefore, the detection and classification of all kinds of faults is highly desired. To do this, a threephase thyristor bridge-based scheme is proposed. As shown in Fig. 3, a three-phase thyristor bridge circuit is connected in parallel to the circuit breaker. The upper thyristors are connected to one phase of the energized upstream and the bottom thyristors are connected to the neutral line. As mentioned earlier, to minimize the size and reduce the cost of power-electronics devices, transformers are used to let the thyristors operate at a lower voltage level. Even considering the added cost on transformers, the overall cost of this configuration will be lower. Once the detection signals are injected, voltage and current at point X are measured and analyzed in the signal detector to determine if there is a fault. The detection for different kinds of faults depends on the gating signals arrangement for the thyristor bridge. Basically, there are two control modes for phase-to-ground faults and phase-to-phase faults detection, respectively, as shown in Fig. 4. In the first mode, when an upper thyristor (T1) is turned on and all bottom thyristors are off, a detection pulse is injected to the corresponding phase of the de-energized line through the

3 LONG et al.: NEW TECHNIQUE TO DETECT FAULTS IN DE-ENERGIZED DISTRIBUTION FEEDERS PART I 1895 Fig. 5. Single-line equivalent system circuit with zero-sequence current injection. Fig. 4. Gating control logic in two fault detection modes.: (a) Phase-to-ground fault. (b) Phase-to-phase fault. TABLE I OVERALL CONTROL LOGIC AND DETECTED FAULT TYPES Fig. 6. Current waveforms under (a) no fault with balanced three-phase currents and (b) a single-line-to-ground fault with unbalanced three-phase currents. upper thyristor, creating a fault current if there is a phase-toground fault in that phase. In the other mode, T1 and T4 are fired simultaneously. If a fault between phase A and phase B exists, it provides a path for the fault current to flow from T1 and T4 to the neutral. With the aforementioned two modes for different types of fault detection, the overall fault detection logic is listed in Table I, which include three steps that can detect all types of faults. In the following two subsections, phase-to-ground faults and phase-to-phase faults are analyzed, respectively. A. Phase-to-Ground Faults Detection Statistics have shown that a single phase-to-ground fault is the most common fault in a distribution system, which accounts for 70% 80% of distribution line faults. On the contrary, a threephase-to-ground fault with balanced three-phase currents only account for 5% of faults [17]. For this reason, the analysis in this section focuses on a single phase-to-ground fault and other unbalanced phase-to-ground faults. In Step I of the fault detection algorithm, when T1, T3, and T5 (upper switches in a thyristor bridge) are turned on simultaneously and T2, T4, and T6 (lower switches in a thyristor bridge) are off, a pulse is injected into all three phases at the de-energized side. This signal, which is from the same phase of the energized side, can be considered as a zero-sequence voltage applied to the downstream. The equivalent circuit of this situation is illustrated as Fig. 5. If the firing angle of the thyristors T1, T3, and T5 is, the injected voltage distortion at point X can be expressed as where is the rated phase-to-neutral voltage of the distribution line, and are the zero impedance of (1) load, distribution line, system, and the step-down transformer, respectively. Since the system and transformers impedance is much lower than the load impedance, (1) can be simplified as Assuming the downstream impedance, and the relationship of the injected current can be obtained as Thus, the magnitude of the injected current can be further expressed as in (5), which suggests that the lower downstream impedance give higher values of current magnitude where. If there is no asymmetrical fault, the three-phase parameters should be balanced, and the impedance of each phase is determined by (6). Thus, the injected three-phase current pulses should be in phase and have the same magnitude as shown in Fig. 6(a) However, if a single phase-to-ground fault exists, the fault resistance will be parallel to the circuit as shown in Fig. 7. In this (2) (3) (4) (5) (6)

4 1896 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 3, JULY 2011 Fig. 7. Equivalent impedance under a single-phase-to-ground fault. case, the equivalent impedance in the faulted phase is smaller than the healthy phases as Fig. 8. Voltage and current pulses generated with different thyristor firing angles. (a) Voltage pulses. (b) Current pulses. (7) where is the fault resistance and the line reactance. As a result, the current pulse in the faulted phase is higher according to (5). The sample current waveforms of three phases with a phase A-to-ground fault are illustrated in Fig. 6(b). The severity of the current imbalance is highly related to the resistance of the fault. From the statistical data, about 85% 98% of the faults are low impedance fault [18]. Due to the increase of fault resistance, the imbalance of three phases becomes less severe. Thus, for high-impedance fault detection, the firing angle should be small enough to create a high voltage on the de-energized side to cause a sufficient fault current that differs from an unfaulted line. The criterion for detecting an unbalanced phase-to-ground fault is designed as follows. 1) Measure all three phase currents in Step I. 2) Calculate the magnitude of the injected current in each phase. Are they identical? If yes, there is no asymmetrical fault. If no, an asymmetrical fault exists. i) Among the three current magnitudes, if one of them is larger and the others are the same (e.g., ), it indicates a single-phase-to-ground fault and the faulted phase is identified. ii) If two of them are the same and larger than the third one (e.g., ), it indicates a double-phase to ground fault and then the two faulted phases are detected. The method using the unbalanced currents is not applicable for three-phase symmetrical faults detection. Thus, another idea using harmonic impedance is proposed for this purpose, and the detail of three-phase symmetrical fault detection will be discussed in a companion paper. Note that the firing angle of the thyristors determines the pulse energy. Reducing the firing angle can increase the pulses magnitude and extend the conduction period (see Fig. 8). Therefore, an initial alarming pulse with a small voltage is generated by controlling the firing angle close to 180. This alarming signal can avoid potential hazards to the humans or other objects in contact with the power lines. After that, the firing angle decreases gradually to achieve high voltage for high-impedance Fig. 9. Signal injection for phase-to-phase fault detection. fault detection. If the device is fully turned on, the maximum amount of energy is essentially applied. B. Phase-to-Phase Faults Detection. Unlike the previous scheme for phase-to-ground fault detection, the detection signal for phase-to-phase fault is created by simultaneously turning on one upper thyristor and one or two bottom thyristors from different phase legs. In this case, one phase connects to the energized side and the others connected to the neutral line. If a phase-to-phase fault exists, a current pulse will show up on both phases in reverse directions. Otherwise, there is no path for current flowing in the de-energized phases. For example, as shown in Fig. 9(a), T1, T4, and T6 are turned on and others are off. If there is no fault between phase A and phase B, the current in phase A is determined by (4) and no current flow in phase B and phase C. If a fault exists between phase A and phase B, the current will show up on both phases in reverse direction. The equivalent impedance of phase A is where is the fault resistance, are the line inductance of phase A and B, respectively. Since the load impedance is much larger than the line impedance, most current from phase A flows back to neutral through the distribution line instead of the load. The current of phase B can therefore be determined as: (8) (9)

5 LONG et al.: NEW TECHNIQUE TO DETECT FAULTS IN DE-ENERGIZED DISTRIBUTION FEEDERS PART I 1897 Fig. 10. Voltage and current pulses when a phase A-to-phase B fault exists. (a) Voltage pulses. (b) Current pulses. The voltage and current waveforms of phase A and phase B with a phase A-to-phase B (low impedance) fault are shown in Fig. 10. Based on the analysis of signal characteristics in a phase-tophase fault, the present reverse signal could be used as an indicator. The criterion is therefore designed as follows: 1) In Step II, the magnitude of phase B and phase C currents are measured. If, a fault exists between phase A and phase B; If, a fault exists between phase A and phase C; Otherwise, there is no fault between phase A and any other phases. 2) In Step III, the magnitude of phase C current is measured. If, a fault exists between phase B and phase C; Otherwise, there is no fault between phase B and phase C. 3) All possibilities of phase-to-phase faults can be checked after Steps II and III. Finally, combining both the phase-to-ground faults and phase-to-phase faults detection schemes, the overall fault detection algorithm is summarized in Fig. 11. Different kinds of asymmetrical faults can be identified after conducting the three steps. If a large or small injection current is needed, the firing angle can be adjusted accordingly. Note that the neutral ground connection is important for the proposed scheme. A loose ground connection will not affect the line-to-ground fault detection, but it will affect the line-to-line fault detection since there is no current return path in this case. To check the ground connection, a simple procedure can be performed by firing T1 and T4 simultaneously at a very small angle before the voltage zero crossing. The present of a current pulse indicates good ground connection. III. COMPUTER SIMULATION RESULTS Computer simulations using PSCAD are performed to verify the above analysis. The rated voltage of the distribution line is 25 kv. As shown in Fig. 12, a three-phase thyristor bridge-based signal generator is connected to the upstream feeder (phase A in this case) and neutral through a single phase transformer. The three bridge legs are connected to the de-energized downstream through step-up transformers. As the current pulses through the transformers are not balanced, three Fig. 11. Fig. 12. Details of the three-step fault detection algorithm. Configuration of the computer simulation system. separate step-up transformers are used instead of a three-phase transformer to reduce the interference between phases. Other parameters are listed in the following:. Utility: L-L RMS 25 kv, Yg connected.. Transformers: step-down transformer, single phase, kv/0.48 kv, 5 MVA; step-up transformer, single phase, 0.48 kv/14.43 kv, 5 MVA.. Signal generator: three-phase thyristor bridge. Firing angle is Feeders: both Line 1 and Line 2 are 5 km, the positive sequence ; the zero sequence.. Load: 4 MVA, lagging,, Fed with a 5 MVA, 25 kv/0.6 kv, Yg/Yg transformer %. Faults: resistance, the default. According to the gate signals control logic (see Table I), three detection steps are carried out to detect different kinds of faults. These thyristors fire once in every four fundamental cycles. It is noted that shorter intervals for faster detection are also possible as the signal attenuates to zero within one fundamental cycle. The interference between different steps is negligible as shown in the results. Considering the three steps as one operation period, the firing angle reduces from 170 after each period until the potential faults are identified. Most faults (expect for some high impedance faults) can be detected within 1 s. When there is no fault, the measured currents are shown in Fig. 13. In Step I, as a zero-sequence voltage is applied to three

6 1898 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 3, JULY 2011 Fig. 13. Three-phase currents in three steps without fault. Fig. 15. Current waveforms of phase C with different fault resistances. Fig. 14. Three-phase currents in three steps under a phase C-to-ground fault. Fig. 16. Current difference between phases under the phase C-to-ground fault. phases, the currents are identical. In Step II, a signal is injected from phase A, and both phase B, C are connected to neutral. Since there is no fault between A-B or A-C, the signal has no path to flow into phase B and C so that the currents of phase B and C are zero; Step III is similar to Step II, with the only difference being that the signal is injected from phase B rather than from phase A. A. Phase-to-Ground Fault If a single-phase-to-ground fault exists in phase C and ground, the current in Step I will change. As shown in Fig. 14, the current of phase C increases significantly due to the fault. As discussed earlier, under the same firing angle, the magnitude of the current depends on the fault resistance in the existing system. When the firing angle is 150, the current waveforms with different fault resistance are shown in Fig. 15. It is apparent that the magnitude of the fault current decrease with the increase of fault resistance. Therefore, if is small, the thyristor firing angle should be large enough to decrease the current to a safe value. On the other hand, If is large, the current of faulted phase is comparable to the currents of unfaulted phase, which increase the difficulty of detection. Thus, the firing angle should be reduced to obtain an obvious difference between faulted and healthy phases. According to the criterion of identifying an unbalanced phase-to-ground fault, the current difference between two phases is calculated. The single phase C-to-ground fault does not affect phase A and phase B; thus, the currents of phase A and B are identical when a zero-sequence voltage pulse is applied to the downstream in Step I. As shown in Fig. 16, is almost zero regardless of firing angles and fault resistance. Since the current of phase C is larger, and Fig. 17. Three-phase currents in three steps under a phase A-to-phase B fault. have almost the same values. As discussed, reducing the firing angle can make the difference of current between phases more significant. B. Phase-to-Phase Fault If there is a phase-to-phase fault between phase A and B, the three-phase current waveforms in three steps are shown in Fig. 17. In Step I, all three-phase currents are identical, but in Step II, a reverse signal shows up in phase B when a pulse is injected from phase A. It indicates a fault existing between phase A and phase B. The waveforms in Step III verify the existent of a fault since a reverse signal shows up in phase A when a pulse is injected from phase B. The current of phase B in Step II not only depends on the injected current from phase A, but it is also affected by other three parameters of power system as shown in (9): the fault location (the length from the fault detection point, which determines the line impedance of interest), the load, and fault impedance.

7 LONG et al.: NEW TECHNIQUE TO DETECT FAULTS IN DE-ENERGIZED DISTRIBUTION FEEDERS PART I 1899 Fig. 18. Sensitivity study of a phase-to-phase fault. (The bases are: Line 1 length 5 km, Load 4 MVA, Rf = 200 ). Fig. 20. Equivalent circuit of the experimental testing system. Fig. 19. Three-phase currents in three steps under multiple faults. Fig. 21. Measured voltage waveforms in the lab experiment. Fig. 18 shows the ratio of with the different values of the three parameters in Step II. Apparently, the fault resistance has a significant impact on the current division. If the phase-tophase fault has high impedance, the current of phase B through the fault will be largely reduced. Another significant impact is from the load. A large load will lead more current flow through the load on phase B rather than from the grounded conductor of phase B where the measurement point X is located as shown in Fig. 9(b). The impact of the Line1 length is very limited as shown in Fig. 18, which confirms the analysis in (9). Moreover, the situation of multiple faults is also tested, which includes: 1) a phase C-to-ground fault; 2) a phase A-to-phase B fault; and 3) a phase A-to-phase C fault. The test results are shown in Fig. 19. In Step I, phase C current is much larger than that of phase A and phase B, which indicates a phase C-toground fault. In Step II, a pulse is injected from phase A and reverses pulses show up on both phases B and C, meaning that the A-B fault and A-C fault exist. Step III verifies that a fault between A-B exists. If necessary, a pulse can also be injected from phase C to verify if a fault between A-C or B-C exists. Therefore, all three different faults can be successfully identified with the proposed scheme. IV. EXPERIMENT VERIFICATIONS An experiment based on a 120-V single phase system has been carried out in the laboratory, and the prototype consists of: 1) a thyristor-based signal generator; 2) a NI-DAQ-based data-acquisition system; and 3) a lumped model-based equivalent circuit. The firing angle of the thyristor can be adjusted from 170 to 150 and the firing interval can be either 2 or 4 fundamental cycles. In the DAQ device, six channels including three voltage measurement channels and three current measurement channels have been utilized (note that regular current/voltage probes or transducers are sufficient for the proposed device). Due to the limitation of lab equipment, all of the transmission lines and transformers have been replaced by equivalent inductors in the single-phase system. The loads have been replaced by an equivalent R-L model as well. All parameters have been scaled down based on the computer simulation model:. power source: 120 V;. transformers: 0.03 p.u., 0.76 mh;. signal generator: a thyristor. Firing angle is adjustable from 170 to 150 ;. feeders: 0.4 mh;. load: 10 mh;. faults: resistance. The equivalent circuit of the lab test prototype is shown in Fig. 20. The fault resistance and the load size are adjustable. Fig. 21 shows that the typical voltage waveforms in the test. represent the 120 V, 60 Hz single-phase voltage source and is the voltage across the thyristor. When the thyristor is fired at a certain degree before the source voltage crosses zero, becomes zero instantaneously. At this moment, the de-energized part of the circuit is connected to voltage source. is the voltage pulse created by the thyristor. When the thyristor current becomes zero, the thyristor turns off naturally and the passive circuit returns to its de-energized state. The corresponding currents, including the load current, the fault current, and the total current, are shown in Fig. 22. The currents are highly related to the characteristics

8 1900 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 26, NO. 3, JULY 2011 Fig. 25. Alternative cascaded structure. TABLE II CONTROL LOGIC OF THE CASCADED STRUCTURE Fig. 22. Measured current waveforms in the lab experiment. Fig. 23. Measured current waveforms with different firing angles. difficult to detect. However, in the proposed methods, asymmetrical high-impedance faults can be identified by unbalanced fault currents. Fig. 24. Measured current waveforms with different fault resistances. of the passive circuit, such as the load impedance, the length of the line, and the fault impedance. It can be seen that with a small impedance fault, the fault current is close to the total current and the load current is relatively small as the load impedance is much higher than the fault impedance. As shown in Fig. 23, the current pulse magnitude is highly correlated with the firing angle. The current increases as the firing angle decreases. For safety considerations, the firing angle is initiated at close to 180, and then decreases gradually until all potential faults can be detected. The impact of fault resistance is illustrated in Fig. 24. The current decreases with the increase of fault resistance. A bolted fault can result in a large current, which is easily detected by a signal with low strength. The high-impedance fault is more V. ALTERNATIVE CASCADED STRUCTURE An alternative signal generation device structure is also proposed in this paper as shown in Fig. 25, which has four thyristors cascaded together. Similarly, the device is connected to an energized upstream phase line and the neutral through a transformer. In the de-energized side, the phase conductors are connected to the joints between two thyristors. The cascaded structure only uses four thyristors rather than six thyristors in the previous structure. To detect different types of faults in this single device, the control logic of gating signals is designed as in Table. II. Similar to the bridge-based topology, the process for fault detection includes three steps. In Step I, a zero-sequence signal is implied to all three phases for phase-ground fault detection; In Step II, a signal is injected from phase A and returns to the neutral from phase B or C if there is any fault between phase A and other phases; There is a little difference from the bridge-based scheme in Step III of the four-thyristor device, where a signal is injected

9 LONG et al.: NEW TECHNIQUE TO DETECT FAULTS IN DE-ENERGIZED DISTRIBUTION FEEDERS PART I 1901 not only from phase B but also phase A since the first thyristor must be turned on in order to connect phase B to the energized side. So faults between A-C or B-C will be detected in Step III. Despite the different control logic of gating signals of the cascaded thyristor topology, the criteria for fault detection are the same as those in the bridge-based topology, as the signal responses under different types of faults are still the same. VI. CONCLUSION In this paper, a thyristor-based fault detection scheme is proposed to detect the existence of a fault in the de-energized downstream system, which can ensure a smooth and safe reclosing without causing a hazard to the downstream devices and personnel. The proposed technique functions by generating a controllable detection signal through the thyristor- based device connected in parallel with a recloser, and using the stimulated voltage and current signals to detect whether a fault still exists in the downstream. The device can be run manually or automatically whenever reclosing is needed. For auto operation, communication is not an issue since the recloser is linked with communication means if it can be reclosed remotely. To identify different types of faults, the characteristics of injected signals under phase-to-ground faults and phase-to-phase faults are studied and the criteria for identifying a faulted line from healthy lines are developed accordingly. The three-step fault detection algorithm is then developed, where all phase-to-ground faults (including asymmetrical and symmetrical faults) will be detected in Step I, and the phase-to-phase faults are detected in Steps II and III. The whole three-step testing procedure will take between one to a few seconds, which is sufficiently fast for the fault detection in a de-energized system. Both computer simulation and lab experiment results are obtained to verify the effectiveness of the proposed technique. An alternative structure with fewer thyristors is also proposed to reduce the size and cost of the power-electronics device. The proposed technique is used for reclosing after the feeder has been de-energized for an extended period; however, it also can be improved for the fuse-saving oriented reclosing operations. This paper mainly focuses on the introduction of the proposed power electronics-aided fault detection technique, such as the faulty line electrical signal characteristics, gating signal generation algorithm, and fault detection steps. More practical implementation issues, including the detection of three-phase symmetrical faults and distinguishing a fault from stalled motors or downstream shunt capacitors, are presented separately in a companion paper. [4] J. F. Witte, S. R. Mendis, M. T. Bishop, and J. A. Kischefsky, Computer-aided recloser applications for distribution systems, IEEE Comput. Appl. Power, vol. 5, no. 3, pp , Jul [5] A. T. Johns, R. K. Aggarwal, and Y. H. Song, Improved techniques for modeling fault arcs on faulted EHV transmission systems, in Proc. Inst. Elect. Eng., Gen., Transm.. Distrib., 1994, pp [6] V. L. Buchholz, M. Nagpal, and J. B. Neilson, High impedance fault detection device tester, IEEE Trans. Power Del., vol. 11, no. 1, pp , Jan [7] S. P. Ahn, C. H. Kim, R. K. Aggarwal, and A. T. Aggarwal, An alternative approach to adaptive single pole auto-reclosing in high voltage transmission systems based on variable dead time control, IEEE Trans. Power Del., vol. 16, no. 4, pp , Oct [8] M. M. Eissa and O. P. Malik, A novel approach for auto-reclosing EHV/UHV transmission lines, IEEE Trans. Power Del., vol. 15, no. 3, pp , Jul [9] C. G. Wester, High impedance fault detection on distribution systems, in Proc. Rural Elect. Power Conf., St. Louis, MO, 1998, pp. C [10] A. T. Trihus, Cable tester, U.S. Patent , Mar. 3, [11] D. A. Rhein and L. R. Beard, Fault detection system including a capacitor for generating a pulse and a processor for determine admittance versus frequency of a reflected pulse, U.S. Patent , July 22, [12] D. D. Wilson and Hedman, Fault isolator for electrical utility distribution systems, U.S. Patent , Jan. 25, [13] R. Maier and W. Roethlingshoefer, Method and arrangement for detecting short-circuit in circuit branches of electrical power system networks, US Patent , Sept. 6, [14] S&C, S&C IntelliRupter PulseCloser: Outdoor distribution 15.5 kv and 27 kv, in S&C Descriptive Bull Mar. 22, [15] W. Xu, G. Zhang, C. Li, W. Wang, G. Wang, and J. Kliber, A power line signaling based technique for anti-islanding protection of distributed generators Part I: Scheme and analysis, IEEE Trans. Power Del., vol. 22, no. 3, pp , Jul [16] W. Wang, K. Zhu, P. Zhang, and W. Xu, Identification of the faulted distribution line using thyristor-controlled grounding, IEEE Trans. Power Del., vol. 24, no. 1, pp , Jan [17] M. T. Aung, J. V. Milanovic, and C. P. Gupta, Propagation of asymmetrical sags and the influence of boundary crossing lines on voltage sag prediction, IEEE Trans. Power Del., vol. 19, no. 4, pp , Oct [18] D. C. Yu and S. H. Khan, An adaptive high and low impedance fault detection method, IEEE Trans. Power Del., vol. 9, no. 4, pp , Oct Xun Long (S 08) received the B.E. and M.Sc. degrees in electrical engineering from Tsinghua University, Beijing, China, in 2004 and 2007, respectively, and is currently pursuing the Ph.D. degree in electrical and computer engineering from the University of Alberta, Edmonton, AB, Canada. His main research interests include power-line signaling, distributed generation, and fault detection. Wilsun Xu (F 05) received the Ph.D. degree from the University of British Columbia, Vancouver, BC, Canada, in From 1989 to 1996, he was an Electrical Engineer with BC Hydro, Vancouver, and Surrey, BC, respectively. Currently, he is with the Department of Electrical and Computer Engineering, University of Alberta, where he has been since His research interests are power quality and distributed generation. REFERENCES [1] IEEE Guide for Automatic Reclosing of Line Circuit Breakers for AC Distribution and Transmission Lines, IEEE Std C , Apr [2] G. D. Rockefeller, C. L. Rockefeller, J. L. Linders, K. L. Hicks, and D. T. Rizy, Adaptive transmission relaying concepts for improved performance, IEEE Trans. Power Del., vol. 3, no. 4, pp , Oct [3] M. M. Eissa and O. P. Malik, A new digital directional transverse differential current protection technique, IEEE Trans. Power Del., vol. 11, no. 3, pp , Jul Yun Wei Li (S 04 M 05) received the B.Sc. degree in engineering from Tianjin University, Tianjin, China, in 2002, and the Ph.D. degree from Nanyang Technological University, Singapore, in In 2005, he was a Visiting Scholar with the Institute of Energy Technology, Aalborg University, Denmark. From 2006 to 2007, he was a Postdoctoral Research Fellow in the Department of Electrical and Computer Engineering, Ryerson University, Toronto, ON, Canada. After working with Rockwell Automation Canada in 2007, he joined the Department of Electrical and Computer Engineering, University of Alberta, Edmonton, AB, Canada, as an Assistant Professor. His research interests include distributed generation, microgrid, power converters, and electric motor drives.