High impedance fault tests on the Taipower primary distribution system

Electric Power Systems Research, 19 (1990) 105 - 114

105

High Impedance Fault Tests on the Taipower Primary Distribution System HUI-YUNG CHU, MING-TONG CHEN and CHING-LIEN HUANG

Department of Electrical Engineering, National Cheng Kung University, Tainan (Taiwan) SHI-LIN CHEN

Department of Electrical Engineering, National Tsing Hua University, Hsinchu (Taiwan) SHIH-SHONG YEN

Power Research Laboratory, Taiwan Power Company, Taipei (Taiwan) (Received J a n u a r y 15, 1990)

ABSTRACT

High impedance faults on the multigrounded three-phase four-wire primary distribution system cannot be detected effectively by existing overcurrent sensing devices. This paper summarizes some of the staged fault test results of different detection methods applied to the Taipower primary distribution system. The methods include the wiring modification of the ground overcurrent relay, the ground product relay, the ratio ground relay and the harmonic current approach. The test results are reported and discussed for the purpose of providing a practical evaluation of high impedance fault detection.

INTRODUCTION

The three-phase four-wire multigrounded feeder is widely applied to primary distribution systems [1]. In Taiwan, the three-phase four-wire l l . 4 k V or 22.8kV feeder is used. The feeder protection scheme normally uses overcurrent sensing devices including the phase overcurrent relay and the ground overcurrent relay. Because the feeder is multigrounded and an unbalanced load always exists, it is difficult to set a suitable sensitivity for the ground overcurrent relay. If a ground fault results from a broken conductor on high impedance surfaces (e.g. macadam, asphalt, etc.), the fault current is usually not large enough to initiate the ground overcurrent relay. The energized fallen conductor may be a potential hazard to life and public safety. The 0378-7796/90/$3.50

faults mentioned above are generally called high impedance faults (HIFs). Many studies of methods to detect HIFs have been presented, most of which have been implemented and tested. They can be divided into three categories according to their detection principles. The first category detects HIFs by calculating the ratio of two current components. It includes the ratio ground relay [2] and the proportional relaying scheme [3]. The former has been implemented in field tests [2] and the latter was used to prove the applicability of the ratio ground relay [3]. The second category is based on changes in the harmonic currents in the feeder. It includes the statistical algorithm for the unbalanced sequence current [4], the third-order harmonic current method [5] and the high frequency current method [6]. Although there is no exact theoretical basis on which to support the design of relaying logic, according to the test results of the past research [4- 6], the detection capability of such methods is good if arcing occurs during the period of the HIF. The last category detects HIFs with the aid of communication links [3, 7, 8]. It includes the high frequency impedance monitoring method, the subharmonic current generating method, the undervoltage relaying scheme and zero-sequence (or negative-sequence) component detection in the current or voltage signal. Because of the complexity of the distribution network and the extra cost of communication links with interrupting devices, no further research on methods based on communication links has been carried out up to now. If the distribution network is fully automated, ~ Elsevier Sequoia/Printed in The Netherlands

106

two-way communication may interconnect all points of the distribution system [9]. Therefore, detection methods based on communication links still have some potential. High impedance faults are also a problem in the Taipower primary distribution system. Many possible solutions had been studied. Each of them was tested on the Taipower three-phase four-wire 11.4 kV primary distribution system to improve the detection reliability. The first method is based on the concept of the ratio current detection method; it is implemented by modifications to the wiring of the ground overcurrent relay. The second method uses the current type of ground product relay by utilizing two different wirings. The principle of the relay is similar to that of the ratio current detection method. The third method uses the ratio ground relay. The last method is based on the analysis of harmonic components in the fault current; it is simulated in the laboratory. The test results are summarized and discussed for the purpose of providing a practical evaluation of HIF detection. GROUND OVERCURRENT RELAY WIRING MODIFICATION

Unbalanced load always exists in distribution systems. For this reason, the ground overcurrent relay (51N) carries a zero-sequence current (3•o) under normal conditions. To avoid maloperation, its current tap setting (CTS) is usually set at a higher level. Therefore, when a HIF occurs, the fault may not be detected by the 51N relay because of the low fault current. Figure 1 shows that 310 may flow back to the neutral point of the source transformer, INT, through the neutral line, INL, as well as the earth return, I G. In general, it is assumed that the magnitude of INL is proportional to 310, and most of the fault current, IF, may flOW back to INT through the earth return, IFG. In order to improve the sensitivity of the 51N relay, a current transformer (CT) was installed on the neutral line to reduce 31o (Fig. 1). For this wiring method, the operating current Ia of the 51N relay can be expressed as Ia

=

K1 x 310 - K2INL

¢-

~I

NEIGHBOURIN(3FEEDERS LOgO

CTI

PIIASE 31o ] RELAYS ~

I NT

[ \

INL I G

~-

' --~

%~INL*

'CT2

/I F

• &

"

t

-

~

IFL

ING

Fig. 1. Relaying method using the 51N relay with modified wiring. The CT and relay settings in this Figure are shown in the Table below.

CT1 CT2 CTS of 51N TDS of 51N

Singtron

Sunshine

500/5 50/5 0.2 2

100/5 100/5 0.5 2

tively. Therefore, if a HIF occurs, Ia may be larger than the value of the setting of the CTS of the 51N relay, and thus the 51N relay may detect the fault. The idea is very similar to the concept of the ratio current detection method which uses the ratio of the load unbalanced current and the neutral line current [3]. According to this idea, two staged faults were performed at Singtron Substation and Sunshine Substation in September 1976 and July 1977. Table 1 shows the record of the staged fault test at Singtron Substation. In the test, the values of K1 and/{72 were 100 and 10, respectively, and the CTS of the 51N relay which uses the conventional wiring was set at 80 A (referred to the primary current). After TABLE 1 Staged fault test record at Singtron Substation in September 1976 Fault cases

Relay wiring

I F (A)

CB trip

Wet asphalt

CTS-80 CTS-20 CTS-80 CTS-20

> 50 > 50 30 - 50 30- 50

4s 3s No No

Dry asphalt

(1)

where K1 and /(2 are the turns ratios of the phase CT and the neutral line CT respec-

CTS-80: 51N relay with conventional wiring and CTS = 80 A. CTS-20: 51N relay with modified wiring and CTS = 20 A.

107

the wiring modification, it was set at 20 A. The test results are summarized as follows. (1) If the conductor breaks on dry asphalt and the fault current is about 30-50A, the conventional 51N relay cannot detect the fault, and the 51N relay with the modified wiring modification cannot detect it either. This is because most of the fault current, IF, flOWS back to the source transformer via the neutral line, IFL. (2) Since the distribution system is multigrounded and many feeders are supplied by the same source transformer, IF may flow back to the source transformer via the neutral lines of the other feeders, I~L. In this test, the recorded value of /EL of the faulted feeder is about 4/5 of IF. Finally, staged faults were performed at Sunshine Substation. The values of K~ and K2 were set at the same value of 20. The neutral line of the protected feeder was separated from those of the other feeders. The test results show that the 51N relay still cannot obtain the expected effects, and IFL is approximately 1/3 of I F. The test results show t h a t the ratio of IFL to IF depends on the parameters of the feeder. The discussion is the same as in ref. 3. If the fault occurs at the front end of the protected feeder, IF may flow back to the source transformer via the neutral lines of the other feeders; thus, it is difficult to make a general rule for the setting of the 51N relay. If the 51N relay is modified to enhance its sensitivity, it may cause maloperation. Theoretically, the principle of ratio current detection is reliable and worth developing as a HIF detecting device. In these tests, the proportional parameters (K~ and K 2) could not be adequately set to match the parameters of the protected feeder, thus no satisfactory result was obtained. Therefore, if a protection scheme based on the ratio current principle is to be studied further, the proportional parameters must be suitably selected.

distribution system is used in Taiwan, the CWC relay was chosen to detect HIFs and two staged faults were conducted to test it. The two wiring methods shown in Fig. 2 were used in those tests. One makes use of the difference between 3/0 and ~NL for the upper pole coil and INL for the lower pole coil, as shown in Fig. 2(a). In this wiring method, the current of the upper pole coil is equivalent to the ground current component IFG of IF. The other one u s e s INL of the protected feeder for the upper pole coil and the ground current Ia of the source transformer for the lower pole coil, as shown in Fig. 2(b). The staged fault for testing the CWC relay was conducted at Singtron Substation in April 1977 and the test results are listed in Table 2. The CTSs of the upper and lower pole coils were set at 0.5 A. For this setting, the product value of the CWC relay is 0.25. The time dial setting (TDS) was set at 1. It can be seen t h a t the CWC relay with the wiring method shown in Fig. 2(a) could not detect the fault which occurred on an asphalt road with a fault current of 40 A. When the fault current was 70 A, the CWC relay wired as in Fig. 2(b) could detect it. In order to simulate insulator flashover, the conductor was connected to the

<,

Lm.

RELAYS •

Io T2

/

!

CWC ~

INT INL

.7,INL* IG

I/ '.

CT3

/IF

i i

\

(a) |

<

11.4KV ~.~ ~IG~RING FEEDERS BUS I ~ CTI

i |

, iI PHASE

3Io

iI

TESTS ON THE CURRENT TYPE OF GROUND PRODUCT RELAY

Two types of ground product relay, the voltage type (CWP) and the current type (CWC), are available from Westinghouse Electric Corporation [1, 10]. Because the grounded

__~__ IFL

(5) Fig. 2. Wiring methods for testing the C W C relay. Recorded waveforms: I'L, /NL, /G, 310. (a) CTI = 200/5, CT2 = 5/10, CT3 = 20/5.

CT3 = 100/5;

(b)

C T I = 200/5,

CT2 =50/5,

108 TABLE 2 Staged fault test record at Singtron Substation in April 1977 Fault cases Asphalt-1 Asphalt-2 Insulator flashover

Prefault Postfault Prefault Postfault Prefault Postfault

I~L

1G

INL

3/O

IF

29 27 33 20 35 .

1.3 1.2 1.3 1.3 1.2

6.5 9 7 22 7.5 .

10.5 17 8 34 8

-40 -70 --

.

.

CS trip

CWC wiring Fig. 2(a)

No Fig. 2(b) 8.93 s Fig. 2(b)

.

0.17 s

Asphalt-l: length of grounding wire = 1.5 m. Asphalt-2: length of grounding wire = 2.0 m. An insulator flashover fault is simulated by connecting a fuse (5 A) between the conductor and the steel arm of the pole. TABLE 3 Staged fault test record at Sunshine Substation in July 1977 F a u l t cases

Feeder 17, 2.3 km, on asphalt Feeder 17, 2.3 km, on asphalt Feeder 151, 4.3 km, on asphalt Feeder 151, 4.3 km, on grass"

Prefault Postfault Prefault Postfault Prefault Postfault Prefault Postfault

I~L

IG

INL

310

IF

CB

20 13 16 13 20 12 20 9

3 0 3 6 3 5 3 5

11 22 12 24 12 24 9 31

14 36 14 43 14 39 14 54

-60

No

72 -30 -90

trip

23 s No Yes

This fault is cleared by the existing relaying system. CWC setting: CTSs of I u and I 1= 0.5 A, TDS = 2. Length of grounding wire 1.5 m for all cases tested. s t e e l a r m of t h e p o l e b y a fuse. O w i n g to t h e large fault current, no fault signal was r e c o r d e d b e c a u s e t h e c i r c u i t - b r e a k e r w a s initiated instantaneously by the instantaneous t r i p u n i t ( I I T ) of t h e g r o u n d o v e r c u r r e n t r e l a y . T a b l e 3 shows the results of the staged fault t e s t a t S u n s h i n e S u b s t a t i o n i n J u l y 1977. I n t h i s test, t h e C W C r e l a y w a s w i r e d as i n Fig. 2(a); t h e C T S s o f t h e t w o c o i l s a n d t h e T D S w e r e set a t 0.5 A a n d 2 r e s p e c t i v e l y . W h e n t h e f a u l t c u r r e n t w a s less t h a n 60 A, t h e C W C r e l a y c o u l d n o t d e t e c t it. B u t , w h e n t h e f a u l t was on an asphalt road and the fault current w a s 72 A, t h e c i r c u i t - b r e a k e r t r i p p e d 23 s a f t e r t h e f a u l t . H o w e v e r , if t h e f a u l t o c c u r r e d o n g r a s s , t h e e x i s t i n g g r o u n d o v e r c u r r e n t r e l a y of the feeder could detect it b e c a u s e the fault c u r r e n t r e a c h e d 90 A. F o r t h e f a u l t o n a s p h a l t , t h e f a u l t w a v e f o r m s r e c o r d e d a t f e e d e r 151 o f t h e s u b s t a t i o n a r e s h o w n i n Fig. 3. The above staged fault tests provide the following conclusions. (1) W h e n a f a u l t o c c u r s , t h e c o m p o n e n t o f IF f l o w i n g b a c k v i a t h e n e u t r a l l i n e , IFL, is

a p p r o x i m a t e l y 2 / 3 - 4/5 of IF. I t d e p e n d s o n t h e n u m b e r of g r o u n d e d t e r m i n a l s a n d o n t h e g r o u n d r e s i s t a n c e of t h e feeder. (2) T h e f a u l t l o c a t i o n a l s o i n f l u e n c e s t h e a m p l i t u d e o f 3/0, a n d t h e 3/0 of t h e p r o t e c t e d f e e d e r does n o t i n c r e a s e p r o p o r t i o n a l l y w i t h IF. T a b l e 3 i n d i c a t e s t h a t t h e 310 o f f e e d e r 17 is 36 A w h e n t h e f a u l t c u r r e n t is 60 A, b u t t h e 3/0 of t h e s a m e f e e d e r is 3 9 A w h e n t h e f a u l t

" ~

q

T-~----J-- . . . . . . . . . . . . . .

. . . . . . . . .

310 ~

(

~

. . . . . . . . . . . . . . . . .

~ l

FAULT

CLEARED

:iLL y:---L;

L:

.

.

.

.

.

. _.,....

Fig. 3. Fault waveforms on asphalt at Sunshine Substation in July 1977.

109

current is 30 A. Therefore, the 51N relay may be limited to its detection capability, which is why its CTS is so difficult to set. (3) The CWC relay can detect the fault when the fault current is larger than 70 A. In the above tests, the principle of the CWC relay, using the two wiring methods, is similar to t h a t of the ratio current detection method. In practice, the wiring method which uses INL and 3/0 (Fig. 2(a)) is better than the other one because many feeders are supplied from the same source transformer and the I G is influenced by the load of the other feeders. In these tests, there was no maloperation of the CWC relay due to the front-end fault of other feeders, but the 51N relay with the modified wiring mentioned in the preceding section always caused maloperation. However, the detectable fault current of the CWC relay is still too high. Since the operating characteristics of the CWC relay are related to the phase angle of the currents between the two coils, and since the phase angle of the 3/0 frequently changes according to the unbalanced load of the feeder, it is difficult to set the sensitivity of a CWC relay.

RATIO GROUND RELAY TESTS

The ratio ground relay (CGR) was co-developed by Westinghouse Electric Corporation [11] and Pennsylvania Power and Light Company (PP&L) [2]. It was designed to improve

the sensitivity of the 51N relay for HIF detection. The detection principle of the relay is also based on the concept of ratio current detection. It contains two main elements: the operating unit which uses 13/0{2, and the restraint unit which uses the difference between the positive-sequence current squared and the negative-sequence current squared ({I1{2 - {12{2). The detection sensitivity of the CGR relay depends on the load conditions. The case of an open conductor was also considered. In tests on the PP&L distribution system, the reliability reached 74%-83% in the case of broken conductors [2, 3]. Owing to its high detection reliability, Taipower was interested in the CGR relay and performed three staged fault tests on it in 1982 and 1983. One of the tests was performed at feeder 10 of Singtron Substation in April 1982. In this test, the relay settings of the CTS and TDS were the same value, 1, as for the CGR relay, while the CTS was set at 0.8 and the TDS was changed according to different test conditions for the 51N relay. The test results are shown in Table 4. These results may be summarized as follows. (1) In the case of the fault on the asphalt road, the fault current was only 7 A and the fault resistance approximated 1000 fl. There was no change in 3/0; neither the 51N relay nor the CGR relay detected the fault. (2) When the conductor dropped into a ditch, the fault current was about 80A, and both the 51N relay and the CGR relay

TABLE 4 Staged fault test record at Singtron Substation in April 1982 Fault cases

Relay

CTS/TDS

Trip

Dry asphalt

51N CGR 51N CGR 51N CGR 51N CGR 51N CGR 51N CGR 51N CGR

0.8 3

No

1/1 0.8 3 1/1 0.8 3 1/1 0.8 4 1/1 0.8 3 1/1

No No No Yes No • No Yes Yes No Yes Yes Yes Yes

Wet asphalt Vertical to ditch

Parallel to ditch

CT ratio = 200/5.

0.8 5

1/1 0.8 7

1/1

3Io (A)

I F (A)

0.6

7

0.6

7

2.4

80

14.0

350

110

FAULT

I NCEPT

I O N ".,,?

-

"

--

Fig. 4. Fault waveforms for the case of a fault in a ditch at Singtron Substation in April 1982.

detected the fault; however, the operation time of the CGR relay was longer than that of the 51N relay. Figure 4 shows the recorded waveforms at the substation for this fault case. However, because the value of the fault current of the HIF was generally less than 65 A, the CGR relay could not offer an exact result in this test. Similar results were obtained in the other two staged fault tests, but there is no detailed test record. Recently, Taipower ordered a modified CGR relay (CGRS relay), and have tested it on the Taipower primary distribution system [12].

at the substation included the normal phase current, Is, the fault phase current, I r, the neutral line current of the tested feeder, INL, the neutral line current of the source transformer, INT, the zero-sequence current, 3/0, and the fault phase voltage, Vr, while the signals at the fault location included the fault phase voltage, Vr, and the fault current, I~.. The fault surfaces included an asphalt road, a ditch and plowland. Figure 5 shows the arcing phenomena when the fault conductor fell on the wet asphalt road and on the plowland. To depict fault signals in the time domain, the fault waveforms were duplicated by a photo-recorder. Figure 6 indicates the recorded waveforms of the case of a fault on the wet asphalt road. Figure 6(a) shows the waveforms of the six signals recorded at the substation. It can be seen that there are apparent changes in the waveforms of IR, INT and INL. However, there is no big change in 3/o, Is and VR. Figure 6(b) depicts two waveforms recorded at the fault location. It also illustrates that IF changes irregularly when the arcing ground fault exists. The data are similar to those obtained in April 1982 and the published data [15, 16].

HARMONIC APPROACH TO HIGH IMPEDANCE FAULT DETECTION

From operational experience and the staged fault tests on the Taipower distribution system, it has been found that arcing often exists during the period of a HIF and results from an arcing ground fault. Since the arc impedance is nonlinear [13 - 16], the fault current possesses substantial harmonic components. There is a tendency to detect HIFs by analyzing the different harmonic components in the fault current [4- 6]. The detection principles of these methods are all based on the assumption that the arc occurs during the period of the HIF. In order to analyze the harmonic characteristics of the fault current during the period of the HIF, a staged fault was performed at Seichu Substation in December 1983. The fault signals at the fault location and the substation, respectively, were recorded by two tape recorders. Those signals which were recorded

(a)

(b) Fig. 5. Arcing phenomena when the conductor fell on (a) a wet asphalt road, (b) plowland.

iii

INL~

0'

~

Z

3I 0

-30 IR

P~U[T

CLEA~D " ~

<

vR ~ J V V W v % / V ~ % / V V V V

-60

(a)

! VR

I

i00 (a)

IF

1-

500

300 F R E Q U E N C Y (Hz)

CI-~ ~: iIHCn! I81PRTED

(b)

~,~ 'DIV

Z

Fig. 6. Fault waveforms on a wet asphalt road: (a) at the substation; (b) at the fault location.

-30 H *a

To identify the characteristics of the fault signals in the frequency domain, a frequency spectrum analyzer was used. The frequency prefault and postfault spectra of INT are shown in Figs. 7(a) and 7(b) respectively. The operating frequency of the system is 60 Hz. From these Figures, it can be seen that the harmonic components increased after the fault, especially the second-order harmonic current. The results are similar for INL and I R. Figure 8 illustrates the frequency spectrum of I F at the fault location. It can be seen that IF contains substantial harmonic components. The values of the second-order harmonic cur-

<

-60

i00 (b)

FREQUENCY

INT: (a)

Fig. 7. F r e q u e n c y s p e c t r a of fault. rent

for different

shown

fault

in Table

tics are provided (1) IN T

During

and

cases

5. T h e by the

500

prefault; (b) post-

are

following

the fault,

I R change

300 (Hz)

analyzed

Table. the waveforms

abnormally

and

of INL, contain

TABLE 5 A n a l y s i s of r e s u l t s of the staged fault test data at Seichu S u b s t a t i o n in D e c e m b e r 1983 Fault cases

Arc condition

H.R. of IR (dB)

Dry a s p h a l t

Arc No arc No arc Arc Arc No arc Arc Arc No arc No arc Arc Arc Arc No arc

-28.6 . . -28.3 - 28.1 . - 40.7 - 40.3 . . - 27,4 - 28.1 - 30.4 .

Wet a s p h a l t

Ditch Tree Open c o n d u c t o r Plowland

Concrete

H.R. of INL (dB)

. .

.

. .

.

-21.3 . . -20.6 - 20.3 . - 34.4 - 35.6 . . - 20.0 - 21.1 - 26.8 .

H.R. of INT (dB)

and

characteris-

I F (A)

-26.6

15

-24.6 - 23.6

50 50

- 31.0 - 31.2

220 220

- 28.9 - 25.1 - 31.0

100 100 100

. .

.

. .

.

H.R. = h a r m o n i c ratio; it r e p r e s e n t s the ratio of the second-order h a r m o n i c c u r r e n t to the f u n d a m e n t a l c u r r e n t .

112

s

~

n

~

,zr2. +/i22

Z c3

-30

(a)

cos~am)

F-4

<

-60

100

300 FREQUENCY (Hz)

500

Fig. 8. Frequency spectrum o f I F a t t h e ~ u l t l o c a t i o n .

4

(b)

substantial harmonic components. Meanwhile, it is found that apparently the ratios of the second-order and third-order harmonic currents to the fundamental frequency current change. (2) The value of 3I0 hardly changes during the period of the faults, explaining why the 51N relay cannot detect HIFs. If an arcing ground fault occurs, from the test results of the staged fault, the harmonics in the current signal apparently increase. This phenomenon was used in the foregoing studies [4-6], and shows the change in the second-order and third-order harmonic currents explicitly. Since the third-order harmonic current may be interfered with by the normal harmonic source, for instance, the load of the SCR, it may be difficult to apply this method [5]. However, the even-order harmonic currents will not exist in the primary distribution system under normal conditions, therefore a fault detector based on the second-order harmonic current may be worth developing. In order to extract the second-order harmonic current from the fault signals, an algorithm was developed from the concept of Fourier analysis. Figure 9 shows the basic concept and the frequency response of the harmonic extraction algorithm. The developed algorithm was simulated by using the recorded data which were digitized by an analog-todigital converter and stored in the main memory of a microcomputer. Two fundamental cycles were chosen as the data window of the digitized data for the calculation. Figure 10 depicts simulated results of the case of a fault on plowland. Figures 10(a), 10(b) and 10(c) indicate the change in the second-order har-

6

%8

I0

12

/4

16

Fig. 9. The second-order harmonic extraction algorithm: (a) deduction principle; (b) frequency reponse plot.

monic current, /2, and the relative change in the second-order harmonic current with respect to the fundamental frequency current, A I 2 / A I 1, of INL, Isw, and IR. From these Figures, it can be seen that the changes in INL and I s are highly likely to be increases. In this simulation, the changes in the curve are obvious when the fault current is about 30 A. The results concur roughly with those of the high frequency current method [6] (20-50 A) and the third-order harmonic current method [5] (15 A). The detection method based on the harmonic current assumes an arcing ground fault. Although the arc always occurs during switching, for example, of a capacitor bank, and even-order harmonic currents may be generated, the arc usually diminishes in a few cycles. The switching of a capacitor bank has been recorded many times in order to study the transient phenomena. These phenomena, limited to time, will disappear after a certain number of cycles, as shown in Fig. 11. Moreover, the inrush current during the closing of a pole transformer may generate substantial harmonic currents; they will influence the detection capability of this approach. Fortunately, the transient inrush current may diminish in a few cycles, so the interference caused by transients may be overcome in the use of harmonic methods to detect HIFs. However, some loads in the systems, such as arcing welders and arcing furnaces, may generate even-order harmonic currents at the

113

~-~

~.:-~

FAULTINCEPTION

TIME ~CYCLE)

(a)

TIME (C YCL E)

INL

]hit

(b)

i

"X~-gAULf INCEPTION

i

i

i

SO

TIME(C YCLE~

(c)

I~

1~

Fig. 10. Simulation results of the second-order harmonic approach for the fault signals (a) /NL, (b) INT and (c) I R on plowland: ©, I2/I1; A, I2; x, I,.

beginning of melting, and this problem must be taken into consideration in further studies.

CONCLUSIONS

Z

U

m

0

!

100 TIME (ms)

w

200

Fig. 11. Transient phenomenon o f 6 0 0 k V A R capacitor bank switching.

Many staged fault tests for high impedance fault detection have been performed on the Taipower primary distribution system. The conclusions drawn from the test results are as follows. (1) The detection method based on the wiring modification of the ground overcurrent relay is similar in principle to the ratio current detection method. In this study, it was

114 not f o u n d to detect h i g h i m p e d a n c e faults effectively b e c a u s e the p r i m a r y d i s t r i b u t i o n system is m u l t i g r o u n d e d and the fault c u r r e n t m a y flow b a c k from the e a r t h r e t u r n or the n e u t r a l lines. F o r f u r t h e r study, the proportional f a c t o r s m u s t be set c a r e f u l l y a c c o r d i n g to the p a r a m e t e r s of the p r o t e c t e d feeder. (2) The c u r r e n t type of g r o u n d p r o d u c t relay m a y i m p r o v e the d e t e c t i o n of h i g h imped a n c e faults, the principle of its design also being similar to t h a t of the r a t i o c u r r e n t det e c t i o n method. Its o p e r a t i n g c h a r a c t e r i s t i c is related to the p h a s e angle of the c u r r e n t s flowing b e t w e e n the u p p e r and lower pole coils. But the i m p e d a n c e in the feeder load c h a n g e s f r e q u e n t l y and the p h a s e a n g l e of the zero-sequence c u r r e n t c h a n g e s a c c o r d i n g l y , so the d e t e c t i o n c a p a b i l i t y of the r e l a y is influenced by the load conditions. (3) A c c o r d i n g to the f o r e g o i n g studies, the r a t i o g r o u n d r e l a y a c h i e v e d good results. However, in the staged fault tests performed on the T a i p o w e r system, no s a t i s f a c t o r y results were obtained. T a i p o w e r ordered a modified r a t i o g r o u n d r e l a y (CGRS relay) for f u r t h e r testing. (4) U s i n g the h a r m o n i c d e t e c t i o n principle to detect high i m p e d a n c e faults, several methods h a v e been developed. The d e t e c t i o n m e t h o d simulated in this paper, based on the c h a n g e in the second-order h a r m o n i c current, is not disturbed by the n o r m a l h a r m o n i c source. The s i m u l a t i o n results show t h a t the m i n i m u m fault c u r r e n t d e t e c t a b l e is a b o u t 30 A. Therefore, this m a y be a possible app r o a c h to d i s t i n g u i s h high i m p e d a n c e faults.

ACKNOWLEDGEMENTS The a u t h o r s are g r a t e f u l to the N a t i o n a l Science Council, Taiwan, for financial s u p p o r t of project No. NSC74-0404-E006-06. T h e y also a c k n o w l e d g e T a i w a n P o w e r C o m p a n y for supp l y i n g the empirical d a t a and p e r f o r m i n g the s t a g e d f a u l t tests. T h e y w o u l d especially like to t h a n k Mr. C. C. Liang, D i r e c t o r of the P o w e r D i s p a t c h i n g Division, T a i w a n P o w e r C o m p a n y , for his v a l u a b l e c o m m e n t s on this work.

REFERENCES 1 J. L. Blackburn and J. V. Kresser, System grounding and protective relaying, Applied Protective Relaying, Westinghouse Electric Corp., Relay Division, Coral Springs, FL, 1979, Chap. 11, pp. 11.1-11.12. 2 R. E. Lee and M. T. Bishop, Performance testing for the ratio ground relay on a four-wire distribution feeder, IEEE Trans., PAS-102 (1983) 2943-2949. 3 J. Carr, Detection of high impedance faults on primary distribution systems, IEEE Trans., PAS-IO0 (1981) 2008 - 2016. 4 Power Technologies, Inc., Detection of high impedance faults, EPRI Research Project 1285-1, Final Rep. EL2413, Electr. Power Res. Inst., Palo Alto, CA, June 1982. 5 Hughes Aircraft Company, High impedance fault detection using third harmonic currents, EPRI Research Project 1285-2, Final Rep. EL-2430, Electr. Power Res. Inst., Palo Alto, CA, June 1982. 6 B. M. Aucoin and B. D. Russell, Distribution high impedance fault detection utilizing high frequency current components, IEEE Trans., PAS-101 (1982) 1596 - 1606. 7 H. L. Graham, A. J. Carlson and T. A. Granberg, Broken-conductor and high-impedance fault detection by high frequency impedance monitoring, IEEE PES Winter Meeting, New York, 1980, Paper No. A80 064-6. 8 R. E. Lee and L. A. Kilar, Summary and status report on research to detect and de-energize high impedance faults on three-phase four-wire distribution circuits, IEEE PES Summer Meeting, Vancouver, BC, 1979,

Paper No. A79 516-6. 9 B. D. Russell, New developments in systems for transmission and distribution substation control and protection, Electr. Power Syst. Res., 5 (1982) 21 - 34. 10 CWC, CWP ground product relay, in Training Inst. of Taiwan Power Co. (ed.), Protective Relays Handbook, Taiwan Power Co., Taiwan, 1979, pp. 61-76. 11 W. A. Elmore, Application Consideration of CGR broken conductor detection relay, RPL81-1, Westinghouse Electric Corp., Relay Instrument Div., Coral Springs, FL, May 1981. 12 M. T. Chen, H. Y. Chu, C. L. Huang and F. R. Wu, Performance evaluation of high impedance fault detection algorithms based on staged fault tests, Electr. Power Syst. Res., 18 (1990) 75-82. 13 H. Y. Chu, H. L. Jou and C. L. Huang, Detection methods of high impedance fault in primary distribution system, J. Electr. Eng. (Taiwan), 28 (3) (1985) 7 - 23. 14 T. H. Lee, Dynamic behavior of electric arcs, Physics and Engineering of High Power Switching Devices,

MIT Press, Cambridge, MA, 1975, Chap. 6, pp. 233- 262. 15 IEEE PES Committee Report, Sine-Wave Distortion in Power Systems and the Impact on Protective Relaying,

IEEE, New York, Pubn. No. 84TH0115-6PWR, 1984. 16 R. H. Kaufmann and J. C. Page, Arcing fault protection for low-voltage power distribution systems--nature of the problem, Trans. AIEE, Part III, 79 (1960) 160 - 167.