Professional Documents
Culture Documents
Kershaw 1989
Kershaw 1989
1, January 1989 30 1
LEAKAGECURRENT
aa WM 032-5 A paper recommended and approved
~
by the IEEE Surge Protective Devices Committee of < 0.01 amp 100-500 amps 1-100 kA
Power Engineering Society for presentation at the
IEEEIPES 1988 Winter Meeting, New York, New York, CURRENT
January 31 -
February 5, 1988. Manuscricpt
Fig. 1.Comparison of the non-linear characteristics of metal-oxide
submitted September 1, 1987; made available for
printing November 13, 1987.
surge arrester and silicon-carbide surge arrester valve
elements.
0885-8977/89/0100-0301$01.00O1989 IEEE
302
Failure Mechanism Potential Metal-Oxide Arrester Application Problems
Figure 2 shows the voltage-current relationship of a 27-kVmetal- A Detroit Edison Company study (Appendix A) shows that pu
oxide distribution-class surge arrester at three voltage levels: nor- voltages as high as 168can occur on unfaulted phases ifthe neutral
mal operating voltage, duty-cycle-rated voltage, and a voltage conductor is inadequately grounded throughout the distribution
above rated voltage. Note that Current I, is out of phase with Volt- system and the wire size is too small. Although this would be un-
age El (operating voltage). At Voltage E,, the metal-oxide arrester usual, it can occur when converting an older ungrounded system
valve element is primarily capacitive and the watts loss-or power and emphasizes the importance of good grounding practices during
dissipation- is negligible.At Arrester Rated Voltage, E2,the resis- construction and maintenance.
tive component of corresponding Current I2 is apparent and the When earth resistivity or system conversion results in ineffec-
metal-oxide arrester valve element will heat as a result of the power tively grounded systems, special attention must be given to the
being generated. If the metal-oxide arrester operates for a long time TOV capability of a metal-oxide surge arrester. A higher duty-cycle
at Duty-Cycle-RatedVoltageE, or if the temperature rise is acceler- rating and MCOV may be required or, better yet, part of the system
ated by a n increase in environmental temperature or a discharge of may need to be rebuilt to bring it up to today's technology. If the
transient energy, I2 will increase toward I,, starting the runaway arrester duty-cycle rating and MCOV are i ncreased, the insulation
condition. Under ambient conditions, Current I, occurs when the coordination of the system must be rechecked to assure that the
arresterrated voltage is exceeded. At ElevatedVoltage E,, the resis- required protective margins are still met.
tive component of current is dominant and the watts loss is signifi-
cant. Applying a Metal-Oxide Arrester on a n Ungrounded System
The method used to apply a metal-oxide surge arrester on a n un-
grounded system is different from the method used to apply a sili-
con-carbide arrester on a n ungrounded system, particularly when
accidental grounds can exist for long periods of time. In these cases,
the metal-oxide arrester MCOVrating must be equal to-or greater
than-the maximum system phase-to-phase voltage. However if
fault-detection relaying is such that fault duration is precisely
limited a n MCOV rating approaching maximum line-to-ground
voltage may be used. This would result in a duty-cycle rating lower
than that recommended for a silicon-carbide arrester, but close
attention must be given to the metal-oxide arrester's TOV capabil-
ity so that it is not exceeded during fault conditions.
Ferroresonant Overvoltages
Detroit Edison Company developed a computer program for cal- For most four-wire multigrounded distribution systems, the wer-
culating the maximum expected overvoltages on the unfaulted voltage factor is 1.2pu or less and the voltage on the arrester during
phases of multigrounded neutral distribution systems during a single line-to-ground fault does not exceed 1.14pu above MCOV
sipgle phase-&ground faults. The program was based on a 1972
report on voltage ratings for the application of lightning arresters 13.2/.\/3x 105 x 1 2 = 1.14 pu above MCOV. (2)
on distribution systems published by the IEEE Working Group on 8.4
Surge protective Devices Committee.* The program output lists
the maximum overvoltages expected at various nodes along a distri- This 1.14 pu above MCOV can be sustained by all manufacturers’
bution circuit for earth return, neutral return, and multigrounded metal-oxide surge arresters for minutes, giving the circuit ample
neutral return. These calculations take into account the system time to open.
impedance, ground-fault resistance, and the earth resistivity. The Studies were also performed on lines consisting entirely of 636-
assumed earth resistivity was 100 ohm-meters. kcmil bare aluminum phase conductors. In each study, a different
Studies were performed using no. 6 copper, no. 2 copper and 110 type of neutral conductor inadequately grounded throughout the
ACSR neutral conductors. The overvoltages for neutral return and circuit was modeled. When a no. 6 copper ground wire was modeled,
multigrounded neutral return were highest for the no. 6 copper con- the maximum expected overvoltage was 1.68pu for a single-phase
ductor and lowest for the no. 2 copper Conductor. Because ofhigher fault occurring 10 circuit miles from the substation. This corre-
conductor impedance, the calculated overvoltages with a WO ACSR sponds to a L6 pu voltage above MCOV of the arrester in the exam-
neuty 7 conductor were higher than those with a no. 2 copper neu- ple above. If the system neutral was adequately grounded, the sys-
tral :a. ,refor neutral return and multigrounded neutral return by tem overvoltage would be lowered to 1.33pu.
apprc mately 2 percent and 1 percent, respectively. Because the In computing the maximum expected overvoltages for l/O-ACSR
overvdage calculations for earth return are independent of the neutral conductor, the overvoltage factor was 1.55 pu for fault 10
neutral wire, the maximum overvoltages were the same for allthree miles from the substation with insufficient grounding. If the sys-
wire sizes. tem neutral was adequately grounded, the system overvoltage was
Assuming a faillt approximately 1.5 miles from the substation, lowered to 1.29 pu. The overvoltage factor is reduced slightly more
the maximum calculated system overvoltage varied from 1.12pu for with a no. 2 copper neutral conductor than it is with a l/O-ACSR
multigrounded neutral return with a no. 2 copper neutral conduc- neutral condudor.
tor to 147 pu for neutral return with a no. 6 copper neutral
conductor.
If, for example, a 10-kV-operating-duty-cycle-rated metal-oxide
arrester with a n MCOV of 8.4 kV is installed on a 13.2-kVdistribu-
tion circuit regulating at 5 percent above normal, a 1.47pu overvol-
tage would stress the arrester at 140 pu above MCOV
13.2/.\/3 x 105 x 1.47 = 1.40 pu above MCOV. (1)
8.4
APPENDIXB
WYE-DELTA BANK
VA = VAN
NORMAL Y-A v~ = k'm PHASE B OPEN;
+
B UNBALANCED LOAD Vc = VCN NO LOAD
EFFECTIVE IB vA
v, 2 = v, ,v, = 0 = v,
PRESTER
vm, = VAp-VA
=7 . 2 m- F m
= +
7.2 (0.5 j0.866) - 6.235 (0 +jl)
= +
3.6 i6.235 - i6.235
\ I
Or
= 3.6O.l- t', 1 .,
VARR = vCN + vC U
= 7.2M&+ vE!
!l
= -
7.2 (0.5 j0.866) + 6.235 (0 + j l )
= 3.6 - j6.235 + j6.235
= 3.6U
assure that required protective margins are met. Did the.authors mean to say effective? Which is the disconnect with the largest single phase load that
that if the duty cycle rating and MCOV rating are increased, then the should be closed last? On the user side, the delta side, the single phase load
coordination should be rechecked? is between two phases. Do the authors mean to close the primary
Manuscript received February 12, 1988. disconnect last which has a line current substantially in phase with the single
phase load?
JOHN E. HARDER, Westinghouse E l e c t r i c Corporation, Manuscript received February 26, 1988.
Bloomington, Indiana: The authors have presented a
very u s e f u l paper describing several p o t e n t i a l S. S. Kershaw, G . L. Gaibrois, and K. B. Stump thank all of the
problem areas i n the application o f metal oxide discussers for their comments. Their contributions will make this paper a
arresters t o d i s t r i b u t i o n systems. This discussion much more useful tutorial on the application of metal-oxide surge arresters
focuses on one o f those concerns: the selection o f (MOSAs) on distribution systems.
the appropriate r a t i n g o f metal oxide arrester for It is appropriate to address Mr. Koepfinger’s discussion first. The IEEE
applications where the l i n e t o ground voltage on the Surge Protective Devices Committee (SPDC) was concemed that many
unfaulted phases may be higher than expected during a distribution system engineers think that metal-oxide surge arresters can
single l i n e t o ground f a u l t on the system. always be substituted for silicon-carbide (Sic) arresters based solely on
duty-cycle rating. SPDC members believe that it is important to call
As ably pointed out by the authors, the l i n e t o attention to conditions in which arrester application is independent of duty-
ground voltage i s affected by f a c t o r s such as l i n e cycle rating (e.g., conditions affecting the thermal stability of MOSAs). In
construction, wire size, effectiveness o f driven most cases, it is high-impedance, grounded distribution systems for which
grounds, etc. Consider a remote area where the selection criteria may be different. These conditions were addressed in
unbalances i n phase loading may r e s u l t i n a small the paper. For low-impedance (four-wire, multigrounded) systems, the
amount o f current through the driven grounds, application of MOSA’s and Sic arresters is usually the same. Fast
d e t e r i o r a t i n g t h e i r effectiveness. Many arresters i n interruption of faults and suppression of temporary overvoltages (TOVs)
that v i c i n i t y may be subjected t o excessive due to the saturation characteristics of the connected load are contributing
overvoltages during s i n g l e l i n e t o ground faults. In factors. As indicated in Appendix A of the paper, an exception can occur
addition, the t o t a l c l e a r i n g time f o r f a u l t s i n these when grounding is ineffective. The questions raised in the paper are
remote areas (with allowance f o r f a i l u r e o f the intended to help distribution system engineers to recognize potential
recloser closest t o the f a u l t t o operate properly?), problems, and, therefore, to be better able to make the proper decisions
may be r e l a t i v e l y long. One o f the problems, i t about MOSA applications.
seems, i s t h a t the industry has very l i t t l e Mr. Smith’s comments about the use of MOSAs on spacer-cable circuits
assessment concerning the numbers o f i n s t a l l a t i o n s are noteworthy. A study conducted by an SPDC working group in 1972
l i k e t h i s which may f o r whatever reason be concluded that higher arrester voltage ratings are required on spacer-cable
vulnerable, and the magnitude o f overvoltages and systems than normally used on open-wire circuits. (1) Although the working
times involved. group recommendations have, for the most part, been ignored, the authors
recommend that they be reviewed for MOSA applications.
Once the magnitude o f the overvoltage and i t s We agree with Mr. Smith’s comments regarding phase-to-ground
duration are established, there remains the question voltages exceeding phase-to-phase voltages on unfaulted phases during line-
o f selection o f the most appropriate arrester to-ground faults. The possibility that these higher voltages may exist must
rating. The time a t overvoltage c a p a b i l i t y o f metal be recognized before sound applicationschoices can be made; however, due
oxide arresters varies among the manufacturers. to their short duration, they should not exceed the thermal capability of the
Thus, f o r a given time a t overvoltage requirement a arresters.
10 kV arrester or manufacturer A may be satisfactory, Ungrounded wye-delta systems should, indeed, be ungrounded wye-delta
while a 12 kV a r r e s t e r o f manufacturer B may be banks. The 2.7-pu overvoltage cited is the worst case that could occur
required. I t seems cumbersome f o r users t o have t o during a phase-to-phase fault on the secondary with one primary open. We
specify one a r r e s t e r r a t i n g f o r most o f the agree that a permanent ground on the primary wye neutral is not a
d i s t r i b u t i o n system, and another r a t i n g f o r c e r t a i n recommended practice for the reasons given; this leads to the corrective
remote areas ... and the r a t i n g depending upon the
manufacturer o f the arrester. Users, i t seems, need
measures recommended in Items 4 and 5 (temporary grounding or
impedance grounding).
t o be able t o specify an a r r e s t e r r a t i n g which can be Regarding the last paragraph of Mr. Smith’s discussion, the correction
used system wide. has been made; decreased has been changed to increased.
Mr. Jackson’s discussion also pertains to the section on ungrounded wye-
I t would seem h e l p f u l t o have the industry delta banks. The basis for the load unbalance ratio of four times is a
standardizing bodies develop application guidelines calculated number that assumes the primary phase-to-ground voltage is
for arresters nn d i s t r i b u t i o n systems which would limited to 1 pu with one primary phase open. The four-times-unbalanced-
e s s e n t i a l l y eliminate the f a i l u r e o f arresters on load factor was supported specifically by Mr. Smith and, in general, by Mr.
unfaulted phases due t o excessive voltages, from Dalton in their discussions. The primary phase feeding the largest
which more sophisticated users could take exception secondary load-connected across the transformer winding associated with
i f they preferred a lower rated arrester. the feeding primary winding-should be closed last.
All of the discussions reemphasize the concerns about applying arresters
I t seems t o t h i s discusser t h a t the industry needs t o on all types of overhead and underground distribution systems. The authors
(1) quantify the extent o f the problem and (2) agree with Mr. Harder that the extent of the problem in these areas is
e s t a b l i s h application guidelines which accommodate unknown, but system conditions that could cause arrester failure must be
the expected overvoltages. What i s the author’s addressed. A metal-oxide surge arrester application guide, now being
opinion concerning how t o best deal with t h i s facet developed by the IEEE Surge Protective Devices Committee, will discuss
o f the problem? some of the distribution system conditions that may jeopardize arrester
operation. However, because potential problem conditions are functions of
Manuscript received February 17, 1988. specific configurations and parameters, it may be impossible to establish
guidelines that will cover all conditions.
David W.Jackson (R. W. Beck and Associates, Waltham, MA): In their The authors appreciate the supporting comments from Messrs. Dalton
recommendations for dealing with possibly destructive overvoltages which and Simpson as well as those from the other discussers.
might occur on ungrounded wye-delta systems with unbalanced loads the
authors have suggested limiting any unbalanced load on one phase to no Reference
more than four times the load on either of the other phases. They have also 1. Working Group of the IEEE Surge Protective Devices Committee,
suggested closing the disconnect on the large single phase load last. Several “Voltage Rating Investigation for the Application of Lightning
questions arise. Arresters on Distribution Systems,” IEEE Transactions on Power
What is the basis for the load unbalance ratio of four times? Has this Apparatus and Systems, Vol. PAS-91, pp. 1067-1074, 1972.
limitation been explored analytically or experimentally to prove that it is Manuscript received April 13, 1988.