ARCING FAULT RESISTANCE (IT ISNT)

By Glenn Swift, Dave Fedirchuk, NxtPhase Corporation

[email protected], [email protected]

Tom Ernst, Minnesota Power

[email protected]

Presented to the

Georgia Tech Fault and Disturbance Analysis Conference Atlanta, GA

May 5-6 2003

ARCING FAULT RESISTANCE (IT ISNT)

Glenn Swift, Dave Fedirchuk, NxtPhase Corporation
[email protected], [email protected]

Tom Ernst, Minnesota Power

[email protected]

Abstract
Much has been written about the imperfection of distance relays when faced with arcing faults near the reach point on a transmission line when there is significant remote end infeed. Warrington, in 1931, showed that there is not really a problem, if the arc is modeled correctly. Here, the benefits of modern computers are used to confirm Warringtons conjecture, and show that it is not physically correct to simply postulate arbitrary arc resistance values and assume linearity of the arc resistance. An arcing fault is shown to have a relatively small apparent arc impedance compared to the line impedance, for a variety of conditions.

INTRODUCTION A transmission line arcing fault is often modeled as resistance, because the 60-hertz component of the current through the arc is essentially in phase with the 60-hertz component of the voltage across the arc. The first problem with this approach is that a particular arcs voltage/current ratio varies greatly with the current magnitude, and the second problem is that simply choosing a wide range of values for the arc resistance to cover all the bases is not a valid approach, as it turns out. Warrington recognized this many years ago(1), (2). This paper does not consider the high impedance arcing fault phenomenon, as when a broken conductor falls to the ground. That too is a highly nonlinear situation, but requires a different approach to detection since the current is usually well below load current. The situation addressed here is that of a lightning-induced or a pollution-induced 60-hertz arc across an insulator string. See Fig 1. A phase-to-phase arc induced by wind-blown conductors is not specifically considered, but the conclusions would be the same.

69 kV typically 2 ft. 500 kV typically 12 ft

AR C

Fig. 1. Sixty hertz arcing fault across an insulator string.

2

The reason for examining this condition is the need to set distance relays such that they do not over- or under-reach because of the arcing fault component of the sensed distance, or the error in this perception caused by remote end infeed. If this effect were significant, it would have to be allowed for either in a compromised setting for the distance relay zone, or by changing the distance-measuring algorithm if that were deemed reasonable and feasible. The effect, and three commonly used distance relay characteristics are shown in Fig. 2.
fault near reach s etting of relay X

apparent arc im pedance true arc im pedance

circular mho characteristic "tomato" mho characteristic quadrilateral characteristic

Fig. 2. Perception of the arc impedance in the distance relay plane.

THE NATURE OF A 60-HERTZ ARC Examples of voltage-current measurements taken on actual 60-hertz arcs, under actual operating conditions are rare. One excellent recording was made during staged-fault tests on a 500kV Canada-USA interconnection transmission line, in 1982(3). The single-phase arc length was 3.8 meters, i.e. about 12 feet, typical of an insulator string length for a 500kV line. Fig. 3 shows the arc voltage and current oscilloscope recordings. Zero time was the instant at which the fault was initiated, by the swinging of a very fine steel wire from ground to the conductor. The fault site was 48 km (30 miles) from the source end. Other details are available from the paper(3). Fig. 4 shows that it is reasonable to use phasor analysis to study the effects of arcs on the big picture. The sinusoidal fundamental components are quite accurate certainly for the first few cycles where first-zone impedance elements are expected to operate correctly.

ARC CURRENT - amps

5000

0 -5000 10 0 -10 0 1 2
TIME - c yc l e s

ARC VOLTAGE - kV

Fig. 3. Arc current and arc voltage from 500-kV line staged fault tests. It is tempting to calculate a resistance from the data of these curves. The rms voltage (in the first few cycles, and excluding the dc offset) is about 5400 volts, and the rms current is about 4400 amps. This gives an apparent resistance of 1.2 ohms. The trouble with actually using this figure is that it will vary wildly from this value at other fault locations. A key observation, though, is that it is low compared to the line impedance to the fault, and this will be shown to be almost always the case.
ARC CURRENT - amps

5000

-5000 10 0 -10 0 1 2
TIME - c yc l e s

ARC VOLTAGE - kV

Fig. 4. Sixty-hertz sinusoidal approximations (heavy lines) superimposed on the waveforms of Fig. 3 (light lines).
4

ARC MODEL Two commonly used models are the Warrington(4) model and the Blackburn(5) model. In both of these, it is assumed that phasor analysis holds, in other words that 60-hertz sinusoidal analysis is close enough. Figures 3 and 4 seem to corroborate this assumption. However, both models postulate that linear resistance is not a proper model. The Warrington model, the Blackburn model and the Linear Resistance model can all be expressed by a common equation: Common equation:

V = K LI N

where V is the ac rms arc voltage in volts, I is the ac rms arc current in amps, K is a constant, L is the arc length in feet, N is a constant. Warrington: Blackburn: Linear Resistance:

V = 8750 L I 0.4 V = 440 L I 0

(1) (2) (3)

V = K LI1

Fig. 5 shows a comparison of these models. The Linear Resistance model is shown in parentheses because it should not be used. It can only be considered valid at a point where it agrees with either the Warrington or Blackburn model, not at other arbitrary points on its locus.
15 Warrington ARC VOLTAGE - kV 10 Blackburn 5 ( Linear Resistance )

0 0 1000 2000 3000 ARC CURRENT - am ps

Fig. 5. Comparison of several 60-Hz arc models. Incidentally, some recent work(6) in Germany re-examines Warringtons original 1931data(1) which was limited to currents of 1000 A, and also draws conclusions following laboratory work at FGH-Mannheim with currents up to 20 kA. A careful reading of the paper, though, suggests that Warringtons formula or Blackburns formula are not far off the mark.

For example, the paper suggests that Blackburns constant of 440 V/ft might be more accurate if changed to 366 V/ft or 412 V/ft or even 260 V/ft, but doesnt say which. One conclusion might be that we should be using only one significant figure: say 400 V/ft. It remains to more carefully specify the mathematics of the arc model, in phasor-usable form. Equations (1) and (2) specify the magnitude of the (assumed sinusoidal) voltage as a function of the magnitude of the (assumed sinusoidal) current. The phase must be considered, and the model simply specifies that the arc voltage is always in phase with the arc current no matter what that phase may be with respect to some reference. See Fig. 6. The mathematical detail to accomplish this, is not presented here. It is one of the factors that makes the circuit solution tricky.

i(t) v(t)

+ v(t) _
arc

i(t)

Fig. 6. Voltage-current relationship for Warrington or Blackburn model. It is interesting that, in the case of the Blackburn model, the circuit element is an ac constant voltage sink, analogous to an ideal generator, which is an ac constant voltage source. CIRCUIT ANALYSIS MODEL Circuit analysis is here complicated by the fact that the arc model is nonlinear, even if it is assumed to be a phasor relationship. The single-phase fault case (assumed here) is shown in Fig. 7.
Vsa=

0
C

sZL

mZL

(1-m)ZL line

rZL

Vra=

B A relay

ARC

Fig. 7. Three-phase circuit and corresponding sequence network, for a phase-to-ground fault through an arc. The symbols used in the figure are

 ZL m s

load angle line positive sequence impedance relative distance to fault sending end source impedance ratio for 100% of the line

receiving end source impedance ratio for 100% of the line Vsa phase A voltage, sending end Vra phase A voltage, receiving end Zratio zero-to-pos. sequence imp. ratio

Note that the effects of ground wires and tower footing resistance are already included by means of the zero-to-positive sequence impedance ratio, Zratio (which may be a complex number). See the Appendix for the sequence network used to solve the circuit. CALCULATIONS AND RESULTS Three examples are presented here. For each, the relay reach (m) is 80% of the line length; the remote end source impedance ratio (r) is 20%; and the Warrington arc model is used. The 500 kV line is that of reference (3); the 69 kV line is a hypothetical one representing the lower-voltage end of lines with distance relaying applied. The 500 kV line is 334 miles long, has a positive sequence impedance Z1 of 0.02+j0.498 /mi, a zero sequence impedance Z0 of 0.28+j1.674 /mi, an assumed load angle of 30o, and an insulator string length of 12 feet. The 69 kV line is 25 miles long, has a positive sequence impedance Z1 of 0.3+j0.8 /mi, a zero sequence impedance Z0 of 0.9+j2.4 /mi, an assumed load angle of 45o, and an insulator string length of 2 feet. Three significant parameters were varied: Arc length Source impedance Pre-fault power flow direction length of arc = length of insulator string the most probable case, length of arc = 3 x length of insulator string perhaps a wind-blown arc. 20% of the line length - the actual case for the 500 kV line, 200% of the line length, since this has a noticeable effect. toward far end of line, toward near end of line.

Table 1. Details of Studied Cases with Abbreviated Results of Calculations Results listed for reach point (m=0.8) only, and pre-fault load direction toward far end of line) (a) (b) Case Voltage Arc length Zsource Actual Relay sees Error Reactive reach kV feet % of line Z1 to fault Z to fault (b)-(a) error in % 1a 1b 2a 2b 3a 3b 500 500 69 69 69 69 12 36 2 6 2 6 20 % 20 % 20 % 20 % 200 % 200 % 6.9+j145.9 6.0+j16 9.4+j147.2 14.5+j149.8 6.3+j16.2 6.9+j16.6 6.8+j16.7 8.5+j18.2 2.5+j1.3 7.5+j3.9 0.3+j0.2 0.9+j0.6 0.8+j0.7 2.5+j2.2 0.9 % 2.7 % 1.2 % 3.8 % 4.4 % 13.8 %

Complete results are shown on R vs X diagrams in figures 8, 9, and 10. The fault was applied at every possible location along the entire length of the each line, that is, 0 < m < 1.0 . The tip-up in the perceived impedance is for a pre-fault load current toward the far end of the line, i.e. under-reach. A tip-down effect occurs for pre-fault current in the reverse direction. For clarity, it is not included in the figures except as noted by the dashed lines in Fig. 10.

200

true relay reach

100
Bolted f ault Insulator arc

error =2. 5 +j1. 3

error =7.5 +j3. 9

3 x insulator arc length

0
R

100

Fig. 8. Case 1: 500 kV line with no arc, 12-foot insulator arc, and 36-foot arc.

20

true relay reach

error =0.3 +j0.2

error = 0.8 +j0.7

10
Bolted f ault Insulator arc 3 x insulator arc length

0
R

10

Fig. 9. 69 kV line.

20
error = 2.5 + j2.2 X

true reach point

error = 0.8 + j0.7 load toward sending end

10

Bolted f ault Insulator arc 3 x insulator arc length

0
R

10

load toward sending end

Fig. 10. 69 kV line with high source impedance.

9

CONCLUSIONS For the representative cases studied, the error in reactive reach is less than five per cent, except under very unusual cases such as extra-long fault arc length and extra-high source impedance and for a relatively low voltage line. It is probable that errors in measurement or errors in other settings (e.g. Z1/Z0 ratio) will be greater than the error due to apparent arc impedance. Therefore, it is concluded that this factor need not be taken into account in setting distance relays. Also, the additional resistance-direction reach to be covered by a distance relay if intended to allow for apparent arc resistance - need not be as large as may have been assumed in the past.

REFERENCES 1. Warrington, A.R.van C.; Reactance Relays Negligibly Affected by Arc Impedance, Electrical World, Sept. 19, 1931, pp. 502-505. 2. Protective Relays, their Theory and Practice, Volume 1, Warrington, A.R.van C., Chapman and Hall Ltd., London, 1968. See page 200. 3. Kappenman, J.A., Sweezy, G.A., Koschik, V., Mustaphi, K.K.; Staged Fault Tests with Single Phase Reclosing on the Winnipeg-Twin Cities 500kV Interconnection, IEEE Transactions on Power Apparatus and Systems, Vol. PAS-101, No. 3, March 1982, pp. 662-673. 4. Protective Relays, their Theory and Practice, Volume 1, Warrington, A.R.van C., Chapman and Hall Ltd., London, 1968. See page 198. 5. Protective Relaying, Blackburn, J.L., Marcel Dekker, Inc., New York, 1987. See page 436. 6. Terzija, V.V., Koglion, H.J.; New Approach to Arc Resistance Calculation, Proceedings of the IEEE Power Engineering Society Winter Meeting, Jan. 28-Feb. 1, 2001, pp781-787.

Appendix: Sequence Network

The circuit of Fig. 7 can be analyzed directly, using for example EMTP, or EMTDC, but the nonlinear arcing fault models of interest in this paper may not be available. Therefore, MathCad was used, for which it was decided that a sequence component network would be the most efficient way. Only the single phase to ground fault was considered since it is by far the most common. In MathCad, it is quite convenient to use the Given, Find routine to solve nonlinear complex-number equations. The symmetrical component equivalent circuit used is shown below, as Fig. 8. The solution is found by solving three equations in three unknowns (VF, IF, and I1).
10

I1
Vsa= V 0

relay sZL Zsend

f ault (1-m)ZL Zrecv rZL Vra=

V

mZL

Zsend ZT ZratioZsend

Zrecv

ZratioZrecv + VF = f (3 IN) = f (IF) -

arc

IN

Fig. 11. Sequence network. The symbols used in the figure are ZL m s r load angle line positive sequence impedance relative distance to fault sending end source imp. ratio receiving end source imp. ratio Zsend Zrecv Vsa Vra Zratio total imp. to fault, sending. end total imp. to fault, receiving end phase A voltage, sending end phase A voltage, receiving end zero-to-pos. sequence imp. ratio

The details of the calculation are not presented here.

AUTHORS Glenn Swift: Glenn graduated from the University of Alberta in 1953, and received the PhD degree from the Illinois Institute of Technology in 1968. He has worked for several firms including Metropolitan-Vickers in the UK, Canadian Westinghouse, Ontario Hydro and Federal Pioneer Ltd. After thirty-six years as a professor of Electrical Engineering, he helped establish a division of NxtPhase Corporation, a manufacturer of relays, recorders and optical current and voltage transformers. Dave Fedirchuk: Dave graduated from the University of Manitoba in 1972. He has been a Professional Engineer for twenty-five years, with Manitoba Hydro, engaged in relay settings determination and disturbance analysis work. For the last 6 years he has been working for NxtPhase Corporation as Product Manager for their line of relays and disturbance recorders. Tom Ernst: Tom graduated from the University of Minnesota in 1978 with a BSEE. He worked for Northern States Power Co. (Xcel energy) in Minneapolis, Minnesota for 18 years where he held various positions in substation relay engineering, and plant electrical and I&C engineering. In 1993, he left NSP to work as a utility consultant designing substation projects for Municipal, Cooperative and Investor Owned utilities. Tom moved to Duluth, Minnesota in 1997 to take his current position as Supervising Engineer Relay Engineering for Minnesota Power. 11