Effects of Faults and Shunt Reactor Parameters On Parallel Resonance
Effects of Faults and Shunt Reactor Parameters On Parallel Resonance
Effects of Faults and Shunt Reactor Parameters On Parallel Resonance
IEEE Transactions on Power Apparatus and Systems, Vol. PAS-l00, No. 2, February 1981
EFFECTS OF FAULTS AND SHUNT REACTOR PARAMETERS ON PARALLEL RESONANCE
ABSTRACT
It is shown that more shunt reactor values may cause resonance on
paralleled EHV lines than has been reported previously [1, 2, 3, 4] when
faults on the opened circuit are analyzed. This is verified by approximate
electrostatic calculations and detailed system calculations for various
reactor and transformer configurations. The importance of evaluating
reactor zero-sequence currents and secondary-arc currents, as well as the
coupled voltages, is illustrated and a TNA simulation of parallel resonance
is also included.
heating. It is also shown that, for certain shunt-reactor values, the coupling from the adjacent circuit could prolong the secondary-arc current,
thus reducing the effectiveness of high-speed reclosure.
Starting in 1981, Tucson Gas & Electric plans to expand its 345-kV
system. Completion is scheduled for 1985. All configurations for theproposed four-year expansion were evaluated for possible resonance conditions, but only results for the 1981 system (Figure 1) are included in this
paper to illustrate basic study findings.
INTRODUCTION
This paper presents some of the results of a comprehensive evaluation of parallel resonance on the Tucson Gas & Electric Company System. In previous work, [3, 4] it has been shown that two values of shunt
compensation may result in resonance on an opened transposed line
when an adjacent circuit is energized whereas three values mayexist foran
opened untransposed line. These resonant conditions can occur for nofault conditions as well as for any fault on the energized circuit. In this
paper, it is shown that faults on the opened circuit can result in resonance
for other shunt-reactor values. When faults on both the opened and the
energized circuits are analyzed, a total of four shunt-reactor values can
cause resonance on a transposed line whereas 19 differentvalues can cause
resonance for an untransposed line. These values can be approximated
from the formulas derived in Appendices I, II, and III.
The effects of reactor design and transformer characteristics on
resonance are illustrated. It is shown that three-legged core-form reactors
exhibit different resonance characteristics from those of five-legged coreform, shell-form, or single-phase reactors. Shunt-compensated lines
exhibit different resonance characteristics when they are terminated with
a transformer that is equipped with a tertiary.- The tertiary often has the
effect of reducing the number of resonant shunt reactor values.
Coupled voltages on the opened shunt-compensated line are generally the parameters of concern. However, when the voltages are below
the reactor rating, the study showed that it is important to evaluate the
reactor zero-sequence current. When the reactor is a single three-phase
unit, a prolonged zero-sequence current can result in excessive reactor
McKINLEY
SAN JUAN
IF
CORONADO 500KV
Fig. 1. Planned 1981 Tucson Gas & Electric Co. 345 k Vsystem.
1200"
21 '2"
i 0
*-*-26'-o-o
.0
90'
1-30'
CIRCUIT NO. 2
954 KCMIL ACSR RAIL
80' PHASE CONDUCTORS
WITH 18" BUNDLE
7 NO. 8 ALUMOWELD
SHIELD WIRES
Vr
/I//I//[
CIRCUIT NO. 1
954 KCMIL ACSR CARDINAL
PHASE CONDUCTORS
WITH 18" BUNDLE
3/8" EHS STEEL
SHIELD WIRES
///////i/ ////////1/////
SAN JUAN
BCA
MLE29.96 ,I.
MILES*
B CA
ft
60 MVAR
30.27
GREENLEE
C BA
31.65
129.7210.09114.429.725
l
l ,t bI. *E 1 *l 33.58133.58134.73116.561
*< *l
6
C BA
b.
b.
b.
H20.16 MI
I-
35.50 Ml
-
5) 1981 IEEE
B.1
573
In Figure 2, the conductor arrangement for the 186-mile parallleled
lines for the 1981 system is shown. The corresponding transposition
scheme is illustrated in Figure 3. Additional lines and/or transformers
terminate at all the buses shown in the system diagrams. These were represented by their equivalent short-circuit impedances in the study.
TRANSPOSED LINES WITH THE SHUNT REACTOR X0 =
XI
The San Juan-McKinley circuits (Figure 1) were analyzed to determine the coupled voltages for different fault conditions when one of the
circuits is opened. It is planned to have a shunt reactor on each circuit at
the San Juan station. Circuit No. 2 will have a reactor with Xo = X1 and
= 400. Figures 4 and 5 show the maximum coupled voltages on this reactor when Circuit No. 2 is open for various fault conditions as a function of
the shunt-compensation level. A detailed line and system model was used
for this analysis.
-----_
ENERGIZED CIRCUIT l l 1 __ _ _
_
300
Ill
--L-GFAULTON
_ __
ENERGIZED CIRCUIT t*t _
_ _
WITH ONE BREAKER F . X OPEN ON
I ------ --ENERGIZED CIRCUIT
300
4X
200 0
-NOFAULT'14
-_L-GFAULT ON
200 0 4~ ~
I
I'.>
I I I I I
> G-
30
20
-%
70
_/X/
80
90
Figure 4 shows the opened circuit voltages for no-fault, a single lineto-ground fault on the energized circuit with the breakers at San Juan and
McKinley closed, and a single line-to-ground fault on the energized circuit with the San Juan breaker opened and the McKinley breaker closed.
All three curves exhibit resonance at the same two levels of shunt compensation, 47.0 Mvar and 80.6 Mvar. As shown in Appendix III and Reference 4, these two resonance values can be approximately determined by
the formulas
and
L1=
30
40
50 60
MVAR (AT 345 KV)
70
80
90
Fig. 5. San Juan-McKinley Circuit No. 2 maximum open-circuit shuntreactor voltage versus total three-phase shunt compensation:
L-Gfault on opened circuit; L-L-Gfault on opened circuit; L-L
fault on opened circuit; L-L-Lfault on opened circuit.
Fig. 4. San Juan-McKinley Circuit No. 2 maximum open-circuit shuntreactor voltage versus total three-phase shunt compensation: no
fault; L- Gfault on energized circuit; L- Gfault on energized circuit
with one breaker open on energized circuit.
gz~~~~i
20
--
-A
50
60
40
MVAR (AT 345 KV)
10
_ _ \ __
<
1
-
10
1!
10
-co'(Ca'a'+ 2Cab')
(1)
(2)
w2 (Ca'a'- Ca'b')
where L 1 is the positive-sequence inductance of the shunt reactor.
b
2 (Ca'a+ CaIb)
(3)
Inputing the values for Ca'a' and Catb', the two resonant values are 5.49
henries and 3.95 henries or 57.5 Mvar and 79.9 Mvar. These approximate
values are, again, close to those calculated using the detailed model.
The coupled voltages for a double line-to-ground fault are also
shown in Figure 5. Resonance occurs at 69.4 Mvar with a peak voltage of
230 kV. The approximate formula for determining this resonant point is
Ll=
I2C
(4
(4)
4.60 henries or 68.7 Mvar for this condition. This is also a good comparison between the two calculations.
In Figure 5, the line-to-line and three-phase faults on the opened circuit are summarized. The resonant values for a line-to-line fault are 47.0
and 80.6 Mvar. These are the zero-sequence and positive-sequence resonant points, respectively, and are the same as those observed for no-fault
and energized-circuit fault conditions. Equations 1 and 2 can be used to
approximate the resonant points (Appendix III, Table A-I).
A three-phase fault exhibits only one resonant point and that occurs
Figure 5. This is the zero-sequence resonant
point which can be predicted from Equation 1.
574
In all, there are four different values of shunt compensation which
can result in resonance problems on a transposed line. Equations I
through 4 can be used to determine the approximate values, but a more
detailed system model is necessary to calculate the coupled voltages. The
difference between the two calculations is summarized in Table I for the
above example. The maximum difference observed was 1.3 percent.
TABLE I
San Juan-McKinley Circuit No. 2
Resonant Values of Shunt Compensation
(Mvar)
Estimation Method
46.4
57.5
68.7
79.9
+
L
300
FV
Difference Between
Two Calculations
Detailed Method
47.0
58.2
69.4
80.6
These values are close to those calculated above by the approximate method. Table II summarizes the differences between the two methods, the
maximum being 1.7 percent.
200__ _ _7
T11 11
I IIILII
>>
(percent)
1.3
1.2
1.0
0.9
NO FAULT
L-G
FAULT ON ENERGIZED CIRCUIT
--_
5,>~-_
-~
LI= -ijLII
LI
LI =
2+k
3k
L12
(5)
(6)
3k * L13
where L I1, L12, L13, and L14 are the four resonant values for k = 1.
Since Ca'a' = 1.527 x 10-6 farads and Ca'b' = .248 x I0-6farads, the
four resonant values for k = 1, using Equations 1 through 4, would be 46.3,
57.4, 68.5, and 79.6 Mvar. If K = 0.628, the resonant points calculated
from Equations 5 through 8 would be 29.1, 41.1, 57.2, and 79.6 Mvar. It
is apparent, then, that the resonant points for three-legged core-form
reactors will be significantly different than for those using other designs
which result in Xo = X1.
Figure 6 shows the four resonant points for San Juan-McKinley Circuit No. I when a reactor with Xo = 0.628X I is used. The four curves, which
were calculated using a detailed system model, illustrate that there are four
unique resonant points which occur at 29.6, 41.5, 57.8, and 80.4 Mvar.
10
30
20
-.-
50 60
40
MVAR (AT 345 KV)
70
80
90
Fig. 6. San Juan-McKinley Circuit No. I maximum open-circuit shuntreactor voltage versus total three-phase shunt compensation; no
fault; L- Gfault on energized circuit; L- Gfault on opened circuit;
TABLE II
San Juan-McKinley Circuit No. 1
Resonant Values of Shunt Compensation
(Mvar)
~~~Difference Between
(Mvar)
Estimation Method
29.1
41.1
57.2
79.6
Detailed Method
29.6
41.5
57.8
80.4
Calculation Methods
(percent)
1.7
1.0
1.0
1.0
Since both San Juan-McKinley circuits have very nearly the same
electrical characteristics, some comparison can be made between the calculations for the two circuits. The maximum voltages observed near the
first three resonant points are significantly lower in Figure 6 than they are
in Figures 4 and 5 because of the much higher zero-sequence resistance
that was modeled for the core-form reactor. Since the fourth resonant
point, 80.4 Mvar, is a positive-sequence phenomenon, the voltages calculated near that point are nearly the same in Figure 6as they are in Figures 4
and 5.
TRANSPOSED SHUNT-COMPENSATED LINES SWITCHED
WITH A TRANSFORMER
In the early stages of the development of a voltage class, transmission
lines are sometimes switched with transformers rather than by a separate
breaker. (This practice defers the buying of a circuit breaker, thus saving
the cost of the breaker carrying charges for a period of time.) The McKinley-Coronado line was investigated assuming this type of configuration.
This 345-kV circuit provides an interconnection between Tucson Gas &
Electric and the Salt River Project at Coronado where the line is to be connected to a 345/ 500-kV autotransformer with a tertiary. This line was analyzed to determine the values of shunt compensation which could lead to
resonance problems, assuming Xo = X1 for the reactors.
575
The formulas to estimate the resonant points for this case are derived in Appendix HI and shown in Table A-I, Appendix Ill. They are
,2 (Ca'a' + 2Ca'b') - L
LL
LI=
(10)
1
3LL 3LM
L =
w2Caa
(11)
a3LL3LM
W2 LI=1
(Ca'a'Cab')-
1(12)
(
LM
TRANSIENT PHENOMENA
In examining Figres 4and 5, it is evidentthatthe highestvoltages on
San Juan-McKinley Circuit No. 2 will occur for a fault on the energized
circuit when the shunt-compensation level is near 47.0 Mvar, the zerosequence or first resonant point. The transient voltages associated with
this event were simulated on the TNA assuminga46.5 Mvarshuntreactor
with Q = 220.
Faults would normally be cleared in approximately thtee to four
cycles by the breakers at San Juan and McKinley. The purpose of the TNA
simulation was to investigate the transient voltages on the opened circuit
for normal fault clearing and for delayed breaker opening. The results of
the three cases analyzed are described in Table 111. In each case, a fault was
initiated on Circuit No. 1. The breaker opening operations for each case
were:
A. San Juan breaker at three cycles and McKinley breaker at four cycles;
B. San Juan breaker at three cycles;
C. None.
When the fault was cleared normally, the maximum voltage observed on the opened circuit was 0.85 per unit (Case A). Thebreakerclearing times were much faster than the time constant of the circuit; therefore,
excessive coupled voltages did not develop.
The detailed results for this line are given in Figure 7 which verifies
that the fourth, or positive-sequence resonant point, is the only one which
exisis for this shunt-compensated line. In Figure 7, it is apparent that
double peaking occurs at the resonant point because of the slight unbalance of the line since it is not completely transposed. If this line was not
transposed at all, the two peaks would occur at significantly different
values of shunt compensation. The peaks observed in Figure 7 occur at
114.4 and 115.4 Mvar and are very close to the 114.7 Mvar predicted by
0A/0B/0C
OA/j0B /0C
995.9 KV/DIV
199.2 KV/DIV
199.2 KV/DIV
200 MS/DIV
20 MS/DIV
20 MS/DIV
8.
Fig. Simulation of parallel resonance (Table III, Case B) following
an SLGFon San Juan-McKinley Circuit No. I and subsequent
clearing of the San Juan breaker.
0A/0B/0C
- NO FAULT
300__ L-G FAULT ON ENERGIZED CIRCUIT- _ _ _ _ _ _
._L-L-G FAULT ON OPENED CIRCUIT - _ _ _ _ _ _
200.
100---
-' 40_iL
Fig.
7.
50
60
70
80 90 100
MVAR (AT 345 KV)
110
120
130
McKinley-Coronado circuit maximum open-circuit shuntvoltage versus total three-phase shunt compensation:
no fault; L-Gfault on energized circuit; L-L-Gfault on opened circuit. NOTE: L-G fault on opened circuit is similar to no fault.
reactor
With both breakers closed (Case C), the voltages were lower than
with one breaker open (Case B), as observed in Figure 2, due to the lower
Xo/XI ratio of the system.
576
For the no-fault case in Figure 9, I reaches a maximum of approximately 2 percent at 47.0 Mvar. This parallels the very low voltages observed at this point in Figure 4. The 'o curves in Figure 9 do not exhibit the
fourth or positive-sequence resonance point as would be expected.
TABLE III
Initiation And Clearing of Single Line-To-Ground Fault
on San Juan-McKinley Circuit No. 1
with San Juan-McKinley Circuit No. 2 Deenergized
Per-Unit Voltage at San Juan on
Circuit No. 2
Maximum Maximum
Description
Case Steady Transient Transient Maximum
State After 3 After 10 Transient
Cycles
Cycles
0.29
0.85
t = 0; San Juan0.85
A 0.00
McKinley No. 1;
Phase A faulted at
San Juan.
t = 3 cycles; San Juan
breakers open.
t = 4 cycles; McKinley
breakers open.
2.69
0.29
7.07
t = 0; San JuanB 4.95
McKinley No. 1;
Phase A faulted at
San Juan.
t = 3 cycles; San Juan
breakers open.
0.29
1.20
2.97
t = 0; San JuanC 2.12
McKinley No. 1;
Phase A faulted at
San Juan.
SECONDARY-ARC CURRENTS
The secondary-arc current for a fault on the opened circuit could
jeopardize successful high-speed reclosure on that circuit. Single line-toground faults are often transitory in nature and, if sufficient dead time is
allowed, ihe arc current often dissipates. However, ifthe arccurrent is supplied by a nearby circuit during the dead time, it may not extinguish.
In Figure 5, the coupled voltages for a single line-to-ground fault on
San Juan-McKinley Circuit No. 2 are summarized. The corresponding
secondary-arc current is shown in Figure 10. While the voltage plot exhibits two resonant points, the secondary-arc current shows only one, with
the current being less than one ampere at the point of positive-sequence
resonance. At the 58.2 Mvar resonant point, the equivalent network is a
combination of the zero- and positive-sequence networks. Although the
zero-sequence driving voltage is small in this case, it isenoughto result ina
secondary-arc current of 23.4 amperes. A fault impedance of 100 ohms in
this case reduces the magnitude nearly 50 percent to 12.2 amperes at the
peak. Also, it can be said that the value of the fault impedance up to 100
ohms does not significantly change the secondary-arc current for shuntcompensation levels removed from the resonant value.
-cc
I1 1
I CIRCUIT
ITAI-L-G FAULTI ON OPENED
40----
5
2
a.
C'I)
ON
40-
30
LU
.
_
20----.
CU
-N
L-L-G FAULT
ON OPENED CIRCUIT
....L-G
FAULT4-CIRCUIT
1 0.
ON OPENED
10
20
30
40 50 60
MVAR (AT 345 KV)
..r
10----
0~
--
30
LO
70
80
90
Fig. 9. San Juan-McKinley Circuit No. 2 percent shunt-reactor zerosequence current versus total three-phase shunt compensation:
no fault; L-Gfault on energized circuit; L-Gfault on opened circuit; L-L- Gfault oh opened circuit.
10
20
30
.40
bt
OU
/I
OU
JU
(1) Change reactor sizes, creating a non-interchangeability and a nonlinear compensation for varying outages.
(2) Relocate the reactors electrically to the station buses (as opposed to
being switched with line sections). This would have a rather severe
economic impact on existing installations and present certain operating difficulties since compensation will not be lost with its associated
line.
(3) Install neutral reactors or resistors with switching equipment to
change the resonant point or reduce the voltages and currents during
resonant conditions. The primary drawback of this option is its high
initial cost.
(4) Set operating policies directing that reactors be automatically disconnected from line sections that are removed from service. This option would require very minor additional expense since existing reactors and planned reactors are installed with circuit-switching devices
and supervisory control. This alleviates all but a few of the perceptible
resonant conditions and those which it does not absolutely cover
appear to be highly improbable.
CONCLUSIONS
Based on the results of the analyses performed in this study, the following conclusions concerning resonance on paralleled shunt-compensated EHV lines can be made:
(1) When all fault conditions on the energized and opened circuits are
considered, four distinct shunt-reactor values may cause resonance
on an open-circuited transposed line. For an untransposed line, 19
different reactor values may result in resonance.
(2) The design of the reactor can significantly effect the resonance characteristics due to the Xo/ X1 of the reactor and its associated losses.
(3) Transformer termination, when the transformer is equipped with a
tertiary, may act to eliminate the resonance problems associated with
the zero-sequence mode. For a transposed line, this would leave only
577
(4) TNA results illustrated the high transient voltages that could occur if
resonant conditions are not alleviated in a rather short time frame.
(5) Reactor zero-sequence currents and, consequently, reactor heating
could be significant even though the voltages may not exceed the
rating.
ACKNOWLEDGMENTS
This paper describes a portion of the engineering studies performed
on the planned Tucson Gas & Electric Company 345-kV system. Other
participants in this work included R. G. Rocamora of McGraw-Edison
Company and J. Weaver formerly of Tucson. Gas & Electric Company.
The authors wish to express their appreciation to them for their productive discussions and comments throughout the study.
REFERENCES
APPENDIX I
SOLUTION OF THE GENERAL PARALLEL RESONANCE
PROBLEM BY A PURE ELECTROSTATIC METHOD
This method of analysis is performed using only the transmission line
capacitance and the shunt inductive admittance matrices. This method
was first presented by M. H. Hesse and D. D. Wilson [1] in the initial analysis performed on the parallel-resonance problem. It has been shown by
LaForest, et al, [4] and confirmed in this paper, that this method provides
a very good approximation of values of shunt inductance which cause
resonance.
The previous analysis [4] assumed a diagonal matrix for the shuntinductance matrix and was performed for no-fault and energized-circuitfault conditions. The following analysis is performed with a full-shuntinductance matrix for no-fault, energized-circuit-fault, and opened-circuit-fault conditions.
[ell]
=~
CII-1
JwyluYI-I
I[C11_1] [e1]
(Al)
Where
[eab
[e 11]
e bj
(A2)
578
[C11-1I]
Ca'b'
rCa'a'
Ca b
For c'-G fault on the opened circuit (ec' = 0), this matrix is
"C
C.,'C']"
C b'b'
C b'c'
a'c
Ca
(A3)
C b'
Ca'a'-
LCatb -
is the partition of the capacitance Y bus matrix for the opened circuit;
[c11-1]
Cb'a'
LCI Cc'a'
b'b'
(A4)
b'c'
I
jwLB
LLA
I
juL'L
Cb'b'.
jwLB jwLA
jLB
JWLA
jjwLB
LjwLB
jwLB
jwLA
(A6)
Ca"bCaw
Ca'c
21
1
W2LB
C ab
Cb'b-
Ca''
Cb'c'-
21
Cc
21
(A7)
1-
tC b't
I
Ca'c'
Cal
Ca'c'
Cc'c'Ca'a'
LB
LO2
LA
_
(A9)
Ca`c`+ Cb'c
U LB
~~~~(A14)
2L A
For a'-c' fault on the opened circuit (ea'= ec'), this matrix is
22
Ca'
on
the
opened circuit
(2LA
W2LA w2LB
C ab'+ C b'c
(AIS)
C 1bb
(
t2LA
(eb'= ec'),
this matrix is
Cab'+Ca'c
W2
LB
Cb'b'+Cc'c'+2Cb'c'-
(A 16)
2
wLA
For a'-b'-c' fault on the opened circuit (ea'= eb' ec'), this matrix is
(A8)
For b'-G fault on the opened circuit (eb' = 0), this matrix is
, LALB
@
c'c'-1-B
Ca'b'+Ca'c- (02ULB
W2
ccCCTZ 21
(A 13)
Ca'c'+Cb'c'
For a'-G fault on the opened circuit (ea'= 0), this matrix is
2LB
t l-
For a'-b' fault on the opened circuit (e a'= e b'), this matrix is
For
(O2LB
W2LB
kCb b ~ 2LAC c
(A 12)
1]
Ca a+ Cb'b' + 2 Cabb'
j,
It has been shown [4] that the resonant points can be found by deterY1111]. This
mining the roots of the determinant (A) of [C11_11
matrix changes for different fault conditions. For no fault or any fault on
the energized circuit, this matrix is
LO2LA
(AI 1)
For b'-c'-G fault on the opened circuit (e b' = e c'= 0), this matrix is
is defined here as the shunt-inductive-admittance matrix. It has been defined as a full matrix so that various shunt reactor designs, neutral reactors, transformers, and combinations of these elements can be analyzed.
LA and LB must be determined for the appropriate configuration.
C a''a
21
[C ccc'
(A5)
[YII 11]
(A 10)
Cb'b'-
For a'-c'-G fault on the opened circuit (ea' = e c' = 0), this matrix is
ebi
For a'-b'-G fault on the opened circuit (ea'= e b' = 0). this matrix is
Leci
(02LB
cc?,,
Cc'b'
Feal
are
Ca'b'
is the partition of the capacitance Y bus matrix between the energized and
opened circuits:
[ei]
Ca'b'
2L
2L
2L
579
APPENDIX II
PARALLEL RESONANCE ON AN UNTRANSPOSED LINE
For a symmetrical conductor configuration, it is often adequate to
assume that C a'a' = C b'b' = Cc'c' and C ab' = C b'c'. Using these assumptions and solving for the roots of the determinants of the matrices defined
by Equations A7 through A 17 results in ten equations which define resonant conditions as given below:
"
2LB
(A23)
21-=
w2LA
(A24)
1C'bl
@ L
2+2(Cab'2
Ca'a'-
Ca'a'-Ca'c' (2
1
1
Ca'aa+ Ca'c'
2 9W2LA
2w2LB
VV4(C ac2-
(A 18)
2 =0
Ca'a'+
1I
C a'b'
2LA 2(2LB
Ca,.,
ab
,//4(Ca'b'-
//4(Cab
2)
o LB
,,,2LA 2w2LB
-1
1l
w2LB
__Cb
+
+/(C''+CC
a
2
)
L
LB
=0 (A25)
)2+/2 (Cab'b
Ca'c'-
22
LB
)2 =0 (A19)
Ca'a' Ca'b'
(A26)
w2LB
2/3 Ca
Ca'c'Ca'a' + 4/3 Ca'b'+ 42
1~
2
-2LB
=0
(A20)
Ca'a'+
1I
Ca'C'
2LA 202LB
\//i(Ca'ct
Caa'+Ca'b' w2LA
L)2
+ 2 (Catb'-
(A21)
w2LB
)2 = O
w2L=0LB
(A27)
Ca'a'+Ca'c'
W2LA
2
W2LB
(A22)
APPENDIX III
PARALLEL RESONANCE ON A TRANSPOSED LINE
For a transposed line, Ca'a' = Cb'b' = Cc'c' and Ca'b' = Cb'c' =
C a'c'. When these conditions are applied to the ten equations given in
Appendix III, four unique equations result for a tranposed line as summarized below:
For any fault on energized circuit, no fault, L-L fault on opened circuit, or
L-L-L fault on opened circuit:
For any fault on energized circuit, no fault, L-G fault on opened circuit, or
L-L fault on opened circuit:
Ca'a'+ 2Ca'b'-
LA
w2LB
(A28)
l
(, LA
2LB
2l
= 0
(A29)
Ca'a' -
Ca'a'a-Caab-
w2LA
2
(A30)
=0
I2LB
w2,LA w2LB
(A31)
Equations A28 through A31 have been derived for seven specific
reactor-transformer configurations. The four values of L 1, the positivesequence inductance of the shunt reactor, which may cause resonance are
listed for each configuration in Table A- 1 in the order of Equations A28
through A3 1.
580
TABLE A-1
LI
Configuration
Configuration
I_=_L___
(02(Ca'+ 2Ca b )
w2(Ca'a' + Ca'b')
092C a'a8
12
LL
.1
w2(Ca'a'- Ca'b)
L 14
2(l + k) 1
1
2k
21 +5k) L
6k
R
R+3
R+ I
R+3
R+2L
R+3
12
13
I
k
41
2+ k
3k
42
I + 2k
3k 3k
L43
L44
2(1+ 2k) L
3k
2L14
One reactor bank on line
with Xo = XI, but with a
neutral reactor (LN), assuming R L/ LN
LM
L12
3kLl3
w2aa-3LL 3LM
k . L
2+k
-
Ca,
2
1
2
u2(Ca'a'+Ca'b') 3LL3LM
withX0=k.XI
=L
2(Ca'a' + 2CatbI)I
2(1 + k)
3k
+ 2k)
L41
L42
3k
2(1 + 5k)
L43
6k
2L
44
L43
42
581
Discussion
J. J. La Forest & S. A. Miske, Jr. (General Electric Company, Schenectady, NY): The authors have presented an interesting paper introducing additional facets of the parallel-line resonance phenomena.
We would like to comment ori a number of items in the paper that need
further clarification.
The authors illustrate the effect of faults on resonant voltages.
Presumably, the authors have assumed zero resistance faults. In the case
of the faults on the opened circuit, our analysis indicates that fault
resistance significantly reduces resonant overvoltages. Is it to be assumed that the opened circuit fault contingencies postulated by the
authors would involve direct and prolonged metallic connection:
The faults on the energized line considered by the authors increase
the zero sequence voltage available to excite the opened circuit. Such a
condition is a temporary one, and the resultant overvoltage would
likely be limited by the reactor saturation and/or by the opened circuits' arresters. It would appear that energy dissipation required of
these arresters would be within practical limits provided the fault is
removed in a reasonable number of cycles. This would be especially
true if the arresters were of zinc oxide design. Do the authors agree?
The authors calculated voltages (Figure 6) for Circuit No. 1 with
the three-legged, core-form reactor are-quite interesting. Our calculations confirm that the zero-sequence resistance of the reactor greatly
reduced zero-sequence resonant voltages. It would appear then that this
reactor type is of advantage for Circuit No. 1.
The authors discussion on shunt reactor zero-sequence currents
appears to be misleading and self-contradictory. In Figure 9 of the
paper, the zero-sequence currents for the Circuit No. 2 reactor are
shown. For the single line-to-ground fault on that circuit, a current of
42.5% is indicated for a 58.2 MVAR reactor. The authors point out
that such zero-sequence currents, if prolonged, are of concern as they
may subject three-phase reactors to excessive heating. However, the reactor in question was modeled by the authors assuming Xo = Xl
and X/R = 400. By implication then Ro = Rl, thus there would be no
heating to the reactor that is unique to zero-sequence current flow. In
addition, the 42.5% current is obviously less than rated current for
the reactor. Therefore, it would appear the reactor could easily withstand this condition indefinitely. Is this conclusion correct?
In the case of the three-legged, core-form reactor on Circuit No. 1,
Ro is greater than R1, indicating that zero-sequence current flow will
cause more reactor. heating than would the same amount of positive
sequence current. Unfortunately, the authors did not display the zerosequence currents for this reactor type. Therefore, we have taken the
L-G FAULT ON
OPENED CIRCUIT
____L-L-G FAULT
--
ON
OPENED CIRCUIT
La
50
40
_-
30
z
us.I 20
10
10
20
30
40
50
A-
60
__
70
80
90
kV)
Fig. A. San Juan McKinley Circuit 1; Percent Shunt Reactor Zero Sequence Current Versus Total Three Phase Shunt Compensation.
582
We encountered a similar case of line compensated by a reactor
bank having phase reactors and neutral reactor; the [YII-ll matrix
for the reactor bank is given by:
-2p
2y 2 +y y n
p
p
[Y
II-II I
Y2
3Yp
2+y
2
p
+ y
n
p n
-2
-Yp
-yp
where
yp
1
juLp
1
jwLn
-Y2p2
Y2p
(B6)
+a2
wL + -
wL
WL
wL
Ln
wLp
wL+-L
p
2o( =
a
a3
2w(C,al
Catc)
+a
2w(C
Ca,)
+ -UL
2 /2
2wl(Calat + calbf)
2wC,,l
aIaIwCa,b
wL,)
WL,
w3 )2b2 +
(B29b)
(B30a)
, )5
a
(B30b)
a1 5
a w(C,a, + 8
b
2
a3=
(B31)
WLn)
Ca,b,)
+w
2
+
(_ alc
C__
,b_)
C2
3 Ca'b + C lbl C
w(4C8ac, + 5 Ca,,b) + w2 (Cav
~~~~~~~~2
4 C2
+
Ca
_L
n1
2w(C , +
~~~~~aa
w)
2'(C
alb
ab
a c
Ln
1(
a4
a5
2w(C
(B2 I b)
L) 2
+ CaYcI+
a
a
2u(C,,
a aaa
(2
CIa
(B22a)
(B22b)
(B23a)
(B23b)
a5
alai
(B29a)
a3
a}2
(
(C
a'a'
C,
C,
2 C
C2
a4
+13+a
(B2 I-a)
(B28)
2w(Ca,aI + Ca ,
(B27)
a1
where
a2
Caib)
2w C ,
Cbl)
2w(C,a,
1 wLn
(B24)
u(C
WL
+ 2C a
wL + -22 Ln = 2w
p
Calai
a,,)
Cbt +3
2w(C, +
WL n
=1+
wL
w(Cala
a1
ab2i%)
pn
(Ca,a,
(B20)
-wL
2
wL +
p
(B 18)
3C,
3wL
wL+-3wL
Lp
+ a2
((Ct I +
3
=2)
aa),
+-uL
Cat)
a=
3 (C
wL + 3WL=
p
n
2y p +yy
pn
(For the sake of easy reference we have given numbers to our equations which correspond to those used by the authors in their Appendix.)
We have analysed equation (Al) with [YiI-II of (B6). The roots
of the resulting determinant (A) are given below. It may be noted that
equations (B 18) to (B3 1) correspond to (Al 8) to (A3 1) in their significance and both these methods are equivalent.
L
3
uL + -L
2 n
p
2
ca)
(C,, +
32 Ln
uL + -
Cab)
2
' )+
wL
w(Ca
a'a
-abi
negative.
a2
(B25)
(B26)
583
In Tables I and II the authors have shown the close agreement between the electrostatic (estimation) method and the detailed method as
regards the locations of the roots. Would the authors comment on the
"magnitude of the voltages" at resonance as calculated by these two
methods?
584
for the resonance conditions are actually simpler than given in Equations B28 through B3 1. In fact, we have found all the simple expressions given in Table A-i to be invaluable in determining potential
resonance problems prior to performing detailed system voltage and
current calculations. It should be pointed out that the magnitudes of
voltages and currents near a resonance condition are very much dependent on reactor and system losses. Consequently, Equation Al
should not be used to determine the coupled voltages on the open
circuit.
Mr Huntzinger's comments on hung-pole conditions are well
taken. We have since investigated this phenomenon and concur that the
coupled voltage values under these conditions are considerably higher
than under the open-circuit fault conditions which were presented
in this paper. As mentioned previously, an example of a hung-pole
condition is given in Figure 11. The main point to be made is that this
condition can occur on single-circuit, three-phase lines without coupling
from another circuit. In fact, the field tests given in Reference I included measurements of voltages on open phases when only one or two
and the third equation for each configuration given in Table A-i. For
two phases open, the applicable equations are A8, A9, A10, A21, A22,
A24, A26, A29, A31, and the second and fourth equations in each
configuration in Table A-l.
The switching equipment mentioned in conjunction with a neutral
device - a bypass switch around the reactor or resistor - is not absolutely necessary. While this solution is fairly inexpensive, it is not as
economical as the one chosen. It should be noted that at least one other
utility has chosen the neutral-reactor solution.
Mr Lester's comments on the initialization of a resonant condition from the closing of a grounding switch are well taken.
The recovery voltage across the arc path which results from the
extinction of the arc current generally takes the form of the voltage
wave shown in Figure 11; i.e., a slowly building 60-Hz waveform.
The envelope effect of the waveform results from the shunt-reactor
inductive reactance being nearly equal to the line capacitance. This
is similar to the nature of the voltage waveform which results upon
the deenergization of a shunt-compensated transmission line as discussed in Reference 1.
REFERENCE
EHV projects.
Mr. Colapret is a member of IEEE and a
registered engineer in the State of Arizona.
Top Oscillogram:
1.54 pu/Div.
50 ms/Div.
Bottom Oscillogram:
.62 pu/Div.
50 ms/Div.