Chapter 6

Transients of Shunt
Capacitor Banks

In Chap. 2, capacitor switching transients in lumped circuits were 4. Possible secondary resonance when the capacitors are applied
studied. These transients are of importance, as large capacitor at multivoltage level (i.e., at 13.8-kV level as well as at 480-V
banks are finding applications and acceptability in industrial dis- level) in a distribution system
tribution and utility systems. Shunt power capacitors have practi-
cally replaced rotating synchronous condensers, and are used on 5. Restrikes and prestrikes in the switching devices
power transmission systems at voltage levels up to 500 kV, bank
sizes ranging from a few Mvar to 300 Mvar. The size and location 6. Autoclosing with precharge on the capacitors
is based on the load flow and stability studies of the transmission
network. These give rise to current and voltage transients, stress A transient, from its point of origin, will be propagated in either
the switching devices and insulation systems, and can be detri- direction in the distribution system and will be transferred through
mental to the sensitive loads. The switching transients of capacitor the transformer inductive/capacitive couplings to other voltage levels.
banks are at a frequency higher than that of the power frequency Transformer part-winding resonance can occur.1
(Chap. 2). Historically, capacitor-switching transients have caused The application of shunt capacitors can lead to the following
problems that have been studied in the existing literature. During additional side effects:
the period from the late 1970s to 1980s, switching of capacitor
banks in transmission systems caused high phase-to-phase voltages
■ Bring about severe harmonic distortion and resonance with
on transformers and magnification of transients at consumer-end load-generated harmonics
distribution capacitors. Problems with switchgear restrikes caused ■ Increase the transient inrush current of power transformers
even higher transients. Problems were common in industrial dis- in the system, create overvoltages, and prolong its decay rate1
tributions with capacitors and dc drive systems, and the advent
of pulse width modulation (PWM) inverters created a whole new ■ Stress the capacitors themselves due to switching transients
concern of capacitor switching. These concerns and also the phe-
nomena of secondary resonance, when a large utility capacitor ■ Increase the duty on switching devices, which may be
bank is switched, while a capacitor bank at the downstream low- stressed beyond the specified ratings in ANSI/IEEE standards2,3
voltage distribution system remains energized, are investigated in this
chapter; discussions are confined to shunt capacitor banks. The series ■ Discharge into an external fault, and produce damaging
capacitors and static var compensators (SVCs) used to enhance the overvoltages across current transformer (CT) secondary
power system stability limits are discussed in Chap. 15. terminals

Impact the sensitive loads, that is, drive systems and bring
6-1 ORIGIN OF SWITCHING TRANSIENTS ■
about a shutdown
The switching transients originate from:

1. Switching of a shunt capacitor bank, which may include 6-2 TRANSIENTS ON ENERGIZING A SINGLE
switching on to a fault CAPACITOR BANK
2. Back-to-back switching, that is, switching of a second Consider the transients on energizing a single shunt capacitor
capacitor bank on the same bus in the presence of an already bank. This assumes that there are no other capacitor banks in the
energized bank immediate vicinity that will impact the transient behavior, though
practically this will not be the case. On connecting to a power
3. Tripping or de-energizing a bank under normal operation source, a capacitor is a sudden short circuit because the voltage
and under fault conditions across the capacitor cannot change suddenly (Chap. 2). The voltage

123
124   Chapter Six

of the bus to which the capacitor is connected will dip severely.

This voltage dip and the transient step change is a function of the
source impedance behind the bus. The voltage will then recover
through a high-frequency oscillation. In the initial oscillation, the
transient voltage can approach 2 per unit of the bus voltage. The
initial step change and the subsequent oscillations are important.
As these are propagated in the distribution system, they can cou-
ple across transformers and can be magnified. Transformer failures
have been documented.4,5 Surge arresters and surge capacitors
can limit these transferred overvoltages and also reduce their
frequency.6 In a part-winding resonance, the predominant fre-
quency of the transient can coincide with a natural frequency of the
transformer. Secondary resonance, described further in this chapter,
is a potential problem, and the sensitive loads connected in the
distribution system may trip.
Example 6-1 In Chap. 2, switching transient of a series RLC cir-
cuit with sinusoidal excitation was discussed. A practical applica-
tion is illustrated in Fig. 6-1. To improve the load power factor, a
6-Mvar capacitor bank is switched on 13.8-kV bus by closing its cir-
cuit breaker CB1. The term “bank” by definition means an assembly Figure 6-1 Circuit diagram for the study of switching transients of
with all switching accessories, protective equipment, and controls a 6-Mvar capacitor bank.
required for a complete operating installation. The distribution can
be reduced to Thévenin impedance, as seen from 13.8-kV bus. For
switching transient studies, this is not permissible and complete
distribution system should be modeled in all its details. This is so
because much like lightning transients, the switching transients equivalent impedance, as seen from the 13.8-kV bus, is Z = (0.0115 +
will be transmitted and reflected from the impedance discontinui- j0.1056, per unit 100-MVA base). Thus, the transient analysis is
ties and will be transferred through the transformers windings too, reduced to excitation of a series RLC circuit. The reacrance in series
altering the time-domain profile of the calculated transient. With with capacitor bank shown in this figure is not considered.
this explicit qualification, this example illustrates the nature Figures 6-2 and 6-3 show EMTP simulation of current and volt-
of capacitor switching transients. With this simplification, the age transients in phase a. With the given parameters and looking

Figure 6-2 Switching current transient of 6-Mvar capacitor bank, Fig. 6-1.
 Transients of Shunt Capacitor Banks   125

Figure 6-3 Voltage transient, 13.8-kV bus 1, on switching of 6-Mvar capacitor bank, Fig. 6-1.

at Thévenin impedance of 13.8-kV bus, the following differential Taking inverse transform:
equation for the inrush current can be written as:
V 1 V 1
i= sin(1/T )t = LC sin t
5 . 08 × 10−4
di
+ 2 . 19 × 10−2 i +
∫ idt L 1/ T L LC
dt 83 . 6 × 10−6
C 1
= 11 . 26 × 103 sin ω t =V sin t (6-6)
  (6-1) L LC

If damping in a series LC circuit is neglected and the circuit is Thus, the peak inrush current can be written as:
excited through a step function V, the following equation can be
written: 2 ELL Ceq
imax,peak = (6-7)
3 Leq
di ∫ idt
L + =V (6-2)
dt C where imax,peak is the peak inrush current in amperes without damp-
ing, ELL is the line-to-line voltage in volts, Ceq is the equivalent
Taking Laplace transform: capacitance in farads, Leq is the equivalent inductance in henries.
(All inductances in the switching circuit, including that of cables
i(s) and buses, must be considered.) The frequency of the inrush
s2i(s) − sI(0) − I '(0) + =0 (6-3)
T2 switching current f is given by:
The initial current is zero, and from Eq. (6-2): 1
f= (6-8)
2π LeqCeq
di V
= = I '(0) (6-4)
dt L The voltage across the capacitor, which will also be the bus
Therefore: voltage, is given by:

V 1 i(s) V  1 
i(s) = (6-5) Vc (s) = = (6-9)
L ( s 2 + 1/ T 2 ) sC LC  s(s2 + 1/T 2 )
126   Chapter Six

Resolving into partial fractions: current source converters, which produce characteristic harmonics
of the order of 5th, 7th, 11th, 13th, . . . (Chap. 15).
V 1 s  Harmonics generated by nonlinear loads may be described as
Vc (s) =  s − ( s 2 + 1/ T 2 ) (6-10) (1) having a Fourier series with fundamental frequency equal to the
LC(1/T 2 )   power system frequency, and a periodic steady state exists—this
Taking inverse transform: is the most common case; (2) a distorted waveform having sub-
multiples of power system frequency, and a periodic steady state
 1  exists—certain types of pulsed loads and cycloconverters produce
Vc = V 1 − cos t = V (1 − cos ω0t ) (6-11) these types of waveforms; (3) the waveform is aperiodic, but per-
 LC  haps almost periodic, and a trigonometric series expansion may
Thus, the maximum voltage occurs at: still exist, examples being arcing devices, arc furnaces, and sodium
vapor lighting; (4) the components in a Fourier series that are not
ω0t = π (6-12) a multiple of power frequency are called noninteger harmonics.
Harmonics generating loads in the power systems are always on an
These equations give a peak current of 4.58 kA and a frequency of increase due to the proliferation of power electronics. The effects
770 Hz, which corresponds well with the simulation results shown of harmonics on power system operation are manifold. The pres-
in Fig. 6-2 and 6-3. The high-frequency oscillation is clearly vis- ence of capacitors in a system does not generate harmonics in itself,
ible in these figures. It takes about six cycles before the oscillations but can accentuate these, create more distortion, and bring about
appreciably damp out. The damping is a function of the resistance resonant conditions.8
and losses in the system (rigorously some resistance of the capaci-
tors themselves should also be modeled). Figure 6-3 shows a peak
voltage excursion of approximately 22 kV, that is, two times the
6-3-1 Harmonic Resonance
bus-rated voltage of 13.8 kV rms. The inrush current, as read from Series and parallel resonant circuit concepts of Chap. 2 are applied
Fig. 6-2, is 4.6 kA, and its frequency can be approximately deter- to harmonic resonance that occurs when power capacitors are
mined around 700 Hz. applied in distribution systems which have nonlinear loads.
The damping due to the presence of the resistor has not been The shunt power capacitors act in parallel with the power sys-
accounted for, yet the calculated results are close to the simulation. tem. Thévenin impedance, as seen from the point of application of
These can be further examined. Calculate Q0: the capacitor, ignoring resistance, is the impedance of the parallel
combination:
X
Q0 =  10 (6-13)
R jω L(1/jω C)
(6-16)
( jω L + 1/jω C)
Thus, it does not have appreciable effect on damping. From Eq. (6-8)
Assuming that L and C remain invariant with frequency, resonance
1 will occur when the inductive and capacitive inductance of the
f0 = = 770 Hz (6-14)
2π LC denominator in Eq. (6-16) is equal and the denominator is zero.
This means that the impedance of the combination is infinite for a
Thus, inrush current frequency is 770 Hz. (6-15) lossless system. The impedance angle will change abruptly as the
resonant frequency is crossed. Thus, at the resonant frequency fn:
6-2-1 Prior Charge on the Capacitors
A prior trapped charge on the capacitors will prolong the transients. 1
j2π f n L = (6-17)
As per ANSI/IEEE specifications,7 the capacitors are provided with j2π f nC
an internal discharge device that will reduce the residual charge to
50 V or less within 1 min for capacitors of 600 V or less and 5 min Without the presence of power capacitors, the natural resonant
for capacitors over 600 V. frequency of the power system is fairly high, much above any load-
Thus, rapid switching in and switching out of the capacitors generated harmonic. As the frequency is increased, the capacitive
should be avoided to limit inrush current transients. Any switching reactance decreases and the inductive reactance increases. It may so
arrangement should block reconnecting the capacitor banks to the happen that at a load-generated harmonic, say 5th, (fn = 300 Hz),
power supply system without the required time delay. Up to twice Eq. (6-17) is satisfied. The power capacitor acts like one branch of
the normal inrush currents are possible when a circuit breaker is a parallel-tuned circuit while the rest of the system acts likes the
employed to reswitch capacitive loads. When the bank is inter- other parallel branch. Such a circuit, when excited at the resonant
rupted at or near current zero, the voltage trapped on the bank frequency (5th harmonic, in this case), will result in the magnifica-
may be near the peak value. Reclosing will produce high inrush tion of the harmonic current, which may even exceed the funda-
currents. When a capacitor bank is connected on the load side of mental frequency current. This will overload the capacitors and all
a feeder, the high inrush currents can be avoided by isolating the the system components with deleterious results. Resonance with
capacitor bank before reclosing takes place or reclosing must be one of the load-generated harmonics is a major concern in power
sufficiently time-delayed. systems, and this condition must be avoided in any application of
the power capacitors. Equation (6-17) can be simply written as:
6-3 APPLICATION OF POWER CAPACITORS
WITH NONLINEAR LOADS fn kVA sc
h= = (6-18)
Nonlinear loads like variable speed drives, wind power generation, f kvarc
switched mode power supplies, arc furnaces, and so on can give
rise to a variety of line harmonic spectrums. It is not the intention where h is the order of the harmonic, fn is the resonant frequency,
to go into the much-involved subject of harmonics, except to relate f is the supply system frequency, kVAsc is the short-circuit kVA at
it to switching transients and distortions that can be caused by the the point of application of the capacitor, and kvarc is the shunt
application of power capacitors with nonlinear loads. For the pur- capacitor rating. For a short-circuit level of 500 MVA, resonance at
pose of this analysis, we will confine the discussion to six-pulse 5th, 7th, 11th, and 13th harmonics will occur for a capacitor bank
 Transients of Shunt Capacitor Banks   127

size of 20, 10.20, 4.13, and 2.95 Mvar, respectively. The smaller the
size of the power capacitor, the higher is the resonant frequency.
Sometimes, this simple artifice is used to size capacitor banks
for a given distribution system to avoid resonance. A capacitor bank
size is selected so that the resonant frequency does not coincide
with any load-generated harmonics. However, the short-circuit
level in a power system is not a constant parameter. It will vary
with the switching conditions; for example, a generator or a tie-line
circuit may be out of service or part of the motor loads may have
been shut down, which will lower the short-circuit level. Thus, it
can be concluded that:

■ The resonant frequency will float around in the system,

depending on the switching conditions.
■ An expansion or reorganization of the system may bring
about a resonant condition where none exited before.

6-3-2 Frequency Scan

An elementary tool to accurately ascertain the resonant frequency
of the system in the presence of capacitors is to run a frequency
scan on a digital computer. The frequency is applied in incremen-
tal steps, say 2 Hz, for the range of harmonics to be studied, say
from the fundamental frequency to 2400 Hz. (This means 1200 Figure 6-4 Circuit diagram to illustrate harmonic resonance.
calculations for the entire range.) The procedure is equally valid,
whether there are one or more than one harmonic-producing loads
in the system, so long as the principle of superimposition is held
valid. Then, for the unit current injection, the calculated voltages
give the driving point and transfer impedances, both modulus and Table 6 - 1 Harmonic Spectrum of Six-Pulse Load
phase angle. The Ybus matrix contains only linear elements for each h 5 7 11 13 17 19 23 25 29 31
incremental frequency. Thus, an impedance plot can be made with
varying frequency, which gives resonant frequencies. System com- % 18 13 6.5 4.8 2.8 1.5 0.5 0.4 0.3 0.2
ponent models used for power frequency applications, for example,
transformers, generators, reactors, and motors, are modified for the
higher frequencies, which are not discussed in this book. bus through 5 MVA coupling transformer and divide into other
Example 6-2 For this example, the circuit of Fig. 6-1 is modi- system components connected to this bus, depending upon
fied and a 5-MVA transformer, which supplies a six-pulse drive their harmonic impedances. Harmonic load flow is akin to the
system (three-phase fully controlled bridge circuit, Chap. 15) fundamental frequency load flow in this respect. Apart from
load of 5 MVA, is added; see Fig. 6-4. The harmonic spectrum is overloading of the capacitors, the harmonic currents in the dis-
shown in Table 6-1, which is also a function of the source imped- tribution systems seriously derate the transformers, produce
ance (Chap. 15). A frequency scan of bus 2, both impedance additional losses, result in negative sequence overloading of the
modulus and angle, is shown in Figs. 6-5 and 6-6, respectively. A generators, give rise to transient torques and torsional oscilla-
resonance occurs close to the 11th harmonic, and Fig. 6-6 shows tions in rotating machinery, and impact the protective relaying,
that the angle abruptly changes at the resonant frequency. The cal- to name a few effects.
culated resonant frequency is 672 Hz. The calculation is made in an Generators have a limited I 22t rating, where I2 is the nega-
incremental step of 2 Hz. To capture the resonant frequency close tive sequence current (Chap. 10). Harmonics of the order of
to the actual, the incremental step should be small. Suppose an h = 6m + 1, where m is any integer, are forward going, that is,
incremental step of 10 Hz is selected, the actual resonant frequency harmonics of the order of 7th, 13th, 19th, . . ., and these rotate
may be at a variation of ± 5 Hz. at 1/h of the speed. Harmonics of the order h = 6m − 1 are reverse
Figure 6-7 shows the voltage spectrum of 13.8-kV bus 1 and going, that is, harmonics of the order of 5th, 11th, 17th, . . .
4.16-kV bus 2. This shows that the 11th harmonic voltage at In a synchronous machine, the frequency induced in the rotor
4.16-kV bus 2 is 17.2 percent, while at 13.8-kV bus 1, it is 11.8 is the net rotational difference between fundamental frequency
percent. Figure 6-8 shows distorted waveform of the voltage at and harmonic frequency. The 5th harmonic rotates reverse with
these two buses for one cycle. Figure 6-9 depicts the spectrum of respect to stator, and with respect to rotor, the induced fre-
current flow through the capacitor; the 11th harmonic current quency is that of 6th harmonic. Similarly, the forward-going 7th
is 130 percent of the fundamental current = 326 A at 13.8 kV harmonic produces a frequency of 6th harmonic in the rotor.
(the fundamental frequency current is not shown). Note that the The harmonic pairs 11th and 13th will produce a frequency of
injected 11th harmonic current, a current at 4.16-kV bus 1B, is 12th harmonic in the rotor. If the frequency of mechanical reso-
63 A based on the harmonic spectrum shown in Table 6-1. Thus, nance happens to be close to these harmonics during starting,
there is a magnification of approximately 17 times. This is the char- large mechanical forces can occur.
acteristic of a parallel-tuned circuit, that while the exciting current A similar phenomenon occurs in induction motors. Here, positive-
can be small, the capacitor forms a resonant tank circuit with the sequence harmonics of the order of h = 1, 4, 7, 10, 13, . . . produce
source impedance, resulting in much magnification of the injected a torque of (h − 1 + s)ω in the direction of rotation, and negative-
current. sequence harmonics, h = 2, 5, 8, 11, 14, . . ., produce a torque
A harmonic source can be considered as a harmonic generator. The of −(h + 1 − s)ω opposite to the direction of rotation. Again,
harmonic current injected at the source will flow to the 13.8-kV torque amplifications can occur.
128   Chapter Six

Figure 6-5 Impedance modulus, 4.16-kV bus 2, Fig. 6-4.

Figure 6-6 Impedance angle, 4.16-kV bus 2, Fig. 6-4

 Figure 6-7 Voltage spectrum, 13.8-kV bus 1 and 4.16-kV bus 2, Fig. 6-4.

Figure 6-8 Distorted voltage waveforms, 13.8-kV bus 1 and 4.16-kV bus 2, due to harmonic resonance, Fig. 6-4.
129
130   Chapter Six

Figure 6-9 Harmonic current spectrum through 6-Mvar capacitor bank, Fig. 6-4.

6-3-3 Harmonic Propagation and Mitigation The limits on rms voltage are given by:
Example 6-2 depicts a rather simple distribution system with respect 1/ 2
to harmonic amplification. When capacitors are located close to a h = hmax 
plant with significant harmonic-producing loads, harmonic reso- Vrms ≤ 1 . 1 =  ∑ Vh2  (6-21)
nance becomes a potential possibility, and a nearby distribution  h =1 
system, which does not have any nonlinear loads of its own, may The crest shall not exceed 1 . 2 × 2 times rated rms voltage,
be subjected to harmonic pollution. This means that the impact including harmonics but excluding transients.
can be propagated through interconnections and impact remotely.
There have been two approaches: h = hmax
V( rest ) ≤ 1 . 2 2 = ∑ Vh (6-22)
1. Consider capacitor placement from reactive power compen- h =1
sation point of view and then study harmonic effects.
Unbalances within a capacitor bank, due to capacitor element
2. Study fundamental frequency, voltage, and harmonic effects failures and/or individual fuse operations result in overvoltages on
simultaneously. other capacitor units. In a filter bank, failure of a capacitor unit
will cause detuning. The limitations of the capacitor bank loadings
For the power capacitors, the limitations due to harmonic loading become of importance in the design of capacitor filters. A capaci-
are well defined.7 Per unit kvar should not exceed 1.35: tor tested according to IEEE Std 1820 will withstand a combined
total of 300 applications of power frequency terminal-to-terminal
overvoltages without superimposed transients or harmonic con-
h = hmax
tent. The capacitor unit is also expected to withstand transient cur-
kvar( pu ) ≤ 1 . 35 = ∑ (Vh Ih ) (6-19) rents inherent in the operation of power systems, which include
h =1
infrequent high lightning currents and discharge currents due to
The rms current should not exceed 135% of nominal current nearby faults. Ref. 21 provides curves of voltage and transient cur-
based upon rated kvar and rated voltage, including fundamental and rents withstand for shunt capacitors.
harmonic currents. The impact of capacitor applications at transmission and sub-
transmission systems becomes fairly involved and is not the main
subject of discussion in this book. Figure 6-10 shows the frequency
1/ 2
h = hmax  scan of a 400-kV line with bundle conductors. IEEE Std. 519
I rms ≤ 1 . 35 =  ∑ Ih2  (6-20) defines the limits of harmonic indices, current, and voltage at the
 h =1  point of common coupling (PCC).9 This can be the point of metering
 Transients of Shunt Capacitor Banks   131

Figure 6-10 Impedance modulus of a 400-kV line.

or the point of connection of consumer apparatus with the utility impedance path to the harmonic of interest that is required to be
supply company. Within a plant, PCC is the point between the shunted through the filter to minimize its propagation through
nonlinear load and the other loads. The recommended current the system. Two or three ST filters can be used, each appropri-
distortion limits in IEEE Standard 519 are concerned about total ately tuned to a different harmonic that needs to be controlled and
demand distortion,9 which is defined as: bypassed. Referring to the equivalent circuit, the injected harmonic
current divides between the harmonic filter and the source imped-
h=hmax ance, as seen from the point of application of the capacitor bank.
∑ Ih2 The harmonic injected current flows partly into the capacitor and
TDD = h=2
(6-23) partly into the source.
IL
where IL is the load current demand for 15 or 30 min. The limits of Ih = I f + I s (6-24)
the TDD are specified at each harmonic, and also total permissible
TDD, as a function of Isc/IL, where Isc is the short-circuit current. Also, voltage across the combination of reactor and capacitor
The harmonics may be limited by (1) phase multiplication should be equal to the voltage across Zs.
(Chap. 15), (2) use of passive or active harmonic filters, and (3) modern
power electronics technology that limits the harmonics at the source. I f Z f = Is Zs (6-25)
The application of a single-tuned harmonic filter, also called a
band pass filter, is briefly discussed, with reference to Fig. 6-11. A Then:
band pass filter consists of merely a reactor in series with a capacitor,
and referring to Chap. 2, at its resonant frequency it offers a low
 Zs 
If =   I h = ρf I h
 Z f + Z s 
(6-26)
 Zf 
Is =   I h = ρs I h
 Z f + Z s 

In harmonic flow analysis and harmonic filter design, the com-

plex quantities rs and rf are of interest. To control the current injec-
tion into the utility system, these factors should be controlled, the
system impedance playing an important role. As the utility source
impedance may be considered fixed for a certain electrical distribu-
tion system, more current can be made to flow through the filter,
by increasing the size of the capacitor bank. The effectiveness of
harmonic filtering vis-à-vis source impedance is apparent from
Eq. (6-26). The lower the source impedance, the greater is the
harmonic current flowing into it, for the same filter size.
Generally, in the presence of load-generated harmonics, the
capacitors are applied as harmonic filters, and the series tuning
reactor serves the dual purpose of forming a low-impedance tuned
Figure 6-11 Equivalent harmonic injection circuit, distribution circuit as well as limiting the inrush current and its frequency on
system represented by single impedance, Zs. switching a capacitor bank.
132   Chapter Six

Example 6-3 The capacitor bank C1 shown in Fig. 6-4 is turned application of a band pass filter has not eliminated the resonance,
into a ST filter, the tuned frequency being 4.67 times the funda- but now it occurs at a frequency below the tuned frequency that
mental. The new frequency scan, impedance modulus, and angle can be placed at a point so that the load-generated harmonics, and
are now depicted in Figs. 6-12(a) and 6-12(b), respectively. The transformer inrush current harmonics are away from it. There will

Figure 6-12 (a) Impedance modulus with 6-Mvar capacitor bank turned into a single tuned band-pass filter, tuning frequency 4.67th of fundamental,
Fig. 6-4 (b) Impedance angle.
 Transients of Shunt Capacitor Banks   133

Figure 6-13 Voltage waveform, 13.8-kV bus 1, with 6-Mvar capacitor bank turned into a single-tuned band-pass filter, tuning frequency 4.67th of
fundamental, Fig. 6-4.

be some swings in the tuned and resonant frequency with varied Example 6-4 The circuit of Fig. 6-1 is modified with the addi-
system switching conditions, and in a practical filter design, these tion of another capacitor bank C2, similar to bank C1 of 6 Mvar, as
must be considered with filter loading in each case.8 Figure 6-13 shown in Fig. 6-14. C1 is already energized and its circuit breaker
shows 13.8-kV bus voltage with the applied filter; this can be com- CB1 is closed. Bank C2 is switched on the same bus by closing its
pared with Fig. 6-8. circuit breaker CB2. The equivalent switching circuit is shown in
Fig. 6-15. To calculate the switching transients, it is necessary to
6-4 BACK-TO-BACK SWITCHING calculate the inductance between the banks accurately.
The back-to-back switching involves energizing a capacitor bank
on the same bus when another energized bank is present. The
■ Inductance of cables C1 and C2 between the banks = 2-3/C
inrush transient in this case mainly consists of interchange of cur- 500 KCMIL, 11 m, Lc1 + Lc2 = 4.6 mH
rents between the two banks, and the current supplied from the
supply system can be ignored. This will not be true if the supply
system impedance is comparable to the impedance between the
banks being switched back to back.
The switching inrush current and frequency can be calculated
from:

2 ELL C1C2
imax, peak = (6-27)
3 (C1 + C2 )L m

And the frequency of the inrush current is:

1
f= (6-28)
C1C2
2π L m
(C1 + C2 )

where C1 and C2 are the sizes of the capacitors in farad and Lm is

the inductance between them in henries. Apart from stresses on the
circuit breaker, the high-frequency transient can stress the other Figure 6-14 A circuit diagram for the study of back-to-back
equipment too. switching of capacitor banks.
134   Chapter Six

Figure 6-15 Calculation of inductance for back-to-back switching of capacitor banks.

■ Inductance of bus duct between the circuit breakers, networks. These can cope with high recovery voltages without
13.8 kV, 1200 A, 2 m = 1.4 mH restrikes. Installing closing resistors is one way to reduce switching
transients; see Sec. 6-13-1. The vacuum switchgear, which is com-
■ Inductance of each bank itself = 5 mH monly used at medium voltage, cannot be provided with resistor
switching, though two breakers with a separately mounted resistor
■ Total inductance = 16 mH can be used. The interruption of capacitor currents and effects on
recovery voltages are covered in Chap. 8.
The source inductance of our example, as calculated earlier, is A bank is considered “isolated” if the rate of change of the tran-
588 mH. Thus, the inrush current can be calculated by ignoring sient inrush current does not exceed the maximum rate of change
the source inductance. Using these equations, the inrush current is of symmetrical interrupting capability of the circuit breaker at the
435 kA and its frequency is 6144 Hz, which are very high. applied voltage.
Again, for a 15-kV indoor oil-less breaker, rated short-circuit
6-5 SWITCHING DEVICES FOR CAPACITOR BANKS current of 40 kA, K factor is 1.0. Then maximum rate of change is
given by:
The derating of switching devices is required for capacitor switching
duties, whether these are contactors, circuit breakers, or load break
switches. The current requirement for rating the switching device  di 
should consider the effect of overvoltages (generally 10 percent),   = 2ω( 40 . 0 × 103 ) × 10−6
 dt max (6-29)
capacitor tolerances (105 to 115 percent), and harmonic compo-
nents. For example, review the application of an indoor oil-less = 21 . 32 A/m s
metal-clad 15-kV circuit breaker. The rated current of a 6-Mvar
bank at 13.8 kV is 251 A. IEEE Standard C37.06-1987 specifies that
a “general-purpose” 1200-A circuit breaker has a capacitor switch- An inrush current of 4.58 kA (peak) and a frequency of 770 Hz
ing current rating of 250 A only.2 A definite-purpose circuit breaker was calculated when switching 6-Mvar bank in Example 6-1. This
for capacitor switching, rated 1200 A, and 40-kA rms short-circuit gives a rate of change of:
rating, has a rated capacitance switching current of 630 A for iso-
lated capacitor bank or for back-to-back switching. Furthermore, 2π (770)( 4580) × 10−6 = 22 . 16 A/m s (6-30)
the inrush current and its frequency should not exceed 15 kA and
2000 Hz, respectively. This means that for the proper application
of even a definite-purpose breaker, the inrush current and its fre- Therefore, switching of even a single 6-Mvar capacitor bank in
quency should be reduced for back-to-back switching applications. Fig. 6-1 cannot be considered isolated bank switching. It requires
For high-voltage outdoor circuit breakers, similar derating applies. a breaker of higher interrupting rating or, alternatively, the size of
For example, a general-purpose 362-kV breaker, rated for 2000 or the capacitor bank should be reduced. Another option will be to
3000 A and short-circuit rating of 65-kA rms symmetrical has an provide an inrush current limiting reactor.
overhead line or isolated capacitor bank switching current of 250 A, The year 2000 revision of ANSI/IEEE Standard C37.06-1987
and a definite-purpose circuit breaker must limit the inrush cur- has made K factor of oil-less indoor circuit breakers = 1. (This is
rent to 25 kA, frequency 4250 Hz, on back-to-back switching. Gas an attempt to harmonize ANSI standards with IEC standards.) This
circuit breakers, SF6 or air-blast type, have a much better capability reference may be seen for new rating tables of oil-less indoor circuit
of capacitor switching and are now used exclusively for high-voltage breakers.2
 Transients of Shunt Capacitor Banks   135

6-6 INRUSH CURRENT LIMITING REACTORS Example 6-5 The limitation of switching inrush current and its
frequency for isolated bank switching is demonstrated in this exam-
When applying an inrush current limiting reactor, the following
ple by providing an inrush current-limiting reactor. A series reactor
considerations apply:
of 3.8 mH, with an X/R = 50 is provided in the switching circuit of
6-Mvar capacitor bank in Fig. 6-1. Then from Eqs. (6-7) and (6-8),
1. An inrush current-limiting series reactor will tune the
the inrush current is reduced to 1633 A, and its frequency to 264 Hz.
capacitor to a certain frequency. When load-generated harmon-
An EMTP simulation is shown in Fig. 6-16, which shows that in
ics are present, caution is required that the series reactor does
the first positive 1/ 2 cycle, the simulated value is close to the cal-
not tune the capacitor close to a load-generated harmonic,
culated value. The current increases in the negative 1/ 2 cycle, and
unless the intention is to turn the capacitors into single-tuned
this can be attributed to the exchange of energy between two energy
filters, which requires detailed considerations. This is to avoid
storage elements, that is, the inductance and the capacitance. Also,
overloading of the capacitors, which will act as filters and
it is noted that the decay of the current is prolonged due to added
create a low-impedance path to a load-generated harmonic.
reactor. The transients are much reduced.
2. A series reactor does reduce the capacitive reactance, yet, To reduce the inrush current on the back-to-back switching to
the net leading kvar output from the combination increases be within the circuit breaker capability, its magnitude should be
rather than decreases. This is so because the voltage drop no more than 15 kA and frequency should be equal to or less than
across the reactor is additive to the capacitor voltage and the 2000 Hz.
terminal voltage of the capacitor rises. As the reactive output Equation (6-7) can be used to calculate the minimum inductance
of a capacitor changes in proportion to the square of the voltage, between the banks to limit the transients. Both, inrush current limi-
there is a net increase in the leading reactive kvar from the com- tation as well its frequency limitation should be considered. This
bination. This brings another consideration that the capacitor- calculation shows that minimum inductance should be 14.32 mH to
rated voltage may have to be increased. Voltage at the junction reduce the frequency to 2000 Hz, and the inrush current will fall
of the capacitor and reactor will be: to 5.089 kA.
The ohmic losses in reactors can be substantial. A reactor of
n2 4.3 mH, as calculated earlier and a Q factor of 40, has resistance of
  Vc = (6-31)
n −1
2 0.04 Ω. The full-load current of 6-Mvar bank is 251 A. This gives
where n = fn /f, fn is the frequency in presence of the reactor. Also, an energy loss of 7.56 kWh, which, on a yearly basis, translates into
the reactive power of the capacitor bank will be: 66 MWh of energy loss. For indoor reactors, the heat dissipation
and ventilation for large reactors becomes a major consideration,
n2 and proper ventilation and heat dissipation measures should be
  S f = × reactive power without reactor (6-32)
n −1
2 considered.

Figure 6-16 Reduction in current transient of 6-Mvar capacitor bank, Fig. 6-1, with a series inductor of 3.8 mH introduced in the capacitor bank.
136   Chapter Six

Figure 6-17 Surge suppressors across CT secondary windings to

limit the voltages on discharge of capacitor currents on a fault.

6-7 DISCHARGE CURRENTS THROUGH

PARALLEL BANKS
In Fig. 6-4, assume a fault occurs at F. The charged capacitors Figure 6-18 A circuit diagram for the study of secondary resonance.
in the system will discharge into the fault. If the capacitors are
ungrounded (the usual case for the industrial distributions), the
maximum discharge current will occur for a three-phase fault. If
the capacitors are grounded, the maximum discharge current will where fc is coupled frequency, fm is the main switching frequency,
occur for a three-phase to ground or single-phase to ground fault. Ls and Cs are inductance and capacitance in the secondary cir-
The total discharge current will be the sum of the discharge cur- cuit, and Lm and Cm are the inductance and capacitance in the
rents from individual banks given by Eq. (6-7). Again the capability main circuit. The lower the ratio fc/fm, the greater is the mag-
of the breaker to withstand this discharge current should not be nitude of the coupled transient. High-transient voltages of the
exceeded. The effect on bushing CTs and linear couplers should order of five times the rated voltage are possible, as shown in
be calculated. Fig. 6-19.10
These transient voltages can result in failures and fuse opera-
I transient f tions in low-voltage power factor correcting capacitors and nui-
Vcts = × VAohms × t (6-33)
CT ratio f sance trips of power electronic-based devices, such as adjustable
speed drives.
where Vcts is secondary voltage developed across BCT, VA ohms is CT
burden in ohms, ft is transient frequency, and f is supply system
frequency.
The voltage should be limited to safe levels. Sometimes, the pro-
vision of surge suppressors across CT terminals becomes necessary
(Fig. 6-17).
For example, for a total discharge current of 20 kA, frequency
of 2000 Hz, relay burden of 0.5 Ω and CT ratio of 1000/5, the
secondary voltage across CT from Eq. (6-33) is 1.67 kV, which will
impact all the relays and instruments connected across the second-
ary of the CT.

6-8 SECONDARY RESONANCE

Figure 6-1 is now redrawn as shown in Fig. 6-18, and a 2-MVA
transformer with secondary load and a 200-kvar capacitor bank
are added. Such a circuit gives rise to the possibility of second-
ary resonance. The switching of a large high-voltage capacitor can
cause a much higher per unit transient voltage at the location of a
smaller low-voltage capacitor bank. This occurs in the secondary
circuits, which have their resonant frequencies close to the natural
frequency of the switched capacitor bank. The initial surge can trig-
ger oscillations in the secondary circuits, which are much greater
than in the switched circuit:

fc L mCm
= (6-34)
fm L sCs Figure 6-19 Overvoltages due to secondary resonance.
 Transients of Shunt Capacitor Banks   137

The studies show that: Example 6-7 This example is an EMTP simulation of switching
transients of a transformer. The schematic single-line diagram is
1. Transients will predominate when the size of the switched shown in Fig. 6-22. The EMTP transformer models are described
capacitor bank is much larger than the capacitance of the low- in Chap. 14. The purpose of this example is to demonstrate that the
voltage capacitor and cable, which is generally the case. transformer inrush transients escalate in amplitude and their decay
is prolonged when a transformer and capacitor bank are energized
2. Highest transients will occur when the energizing frequency together.
is close to the series-resonant frequency formed by the step- Figure 6-23 shows the switching inrush current transient on
down transformer and low-voltage capacitance. energizing the transformer when the 5-Mvar capacitor bank shown
on the secondary is disconnected.
3. Problems predominate when there is little damping, which is
Figure 6-24 depicts the transient escalation in magnitude and time
very common in industrial distributions as the resistance is low.
when the transformer and 5-Mvar capacitor bank on the secondary
of the transformer are simultaneously energized. The transient pro-
Example 6-6 The 200-kvar capacitor connected to the 2-MVA, file shows some periodic escalations due to some higher-frequency
13.8 to 0.48-kV transformer in Fig. 6-18 remains in service, while resonance. This may not be a very practical situation. A load depen-
6-Mvar capacitor at 13.8-kV bus is switched. The simulation in dent or some other form of switching control of the capacitor bank
Fig. 6-20 shows that a peak voltage of 1220 V line-to-neutral is is adopted so that leading reactive power is not supplied into the
developed across the secondary of 480-V transformer. This over- system and capacitor is switched off on loss of voltage.
voltage is 3.1 times the rated system voltage. The calculations show Switching of a large capacitor bank when the system is unloaded
that fc / fm ≈ 3.8. Figure 6-21 shows the transient current that flows gives rise to steady-state overvoltages, which can be calculated from
through the 200-kvar capacitor; its peak value is 2200 A, approxi- the following expression:
mately 6.5 times the full-load current of 200-kvar capacitor.
When capacitors are applied at multivoltage level in a distribu- kvarc × % Z
tion system, apart from the overvoltages due to the resonance illus- %∆V = (6-35)
kVAT
trated earlier, these can also prolong the decay of the transients.
It is, therefore, best to apply capacitors at one voltage level where %Z is the transformer percentage impedance and kVAT is
only. If these are applied at multivoltage levels, a rigorous switch- the transformer kVA rating. The operating voltages may change
ing transient analysis to predetermine the resonance points and by 10 percent for short periods, while adjustments are being
eliminate them is required. This requires lots of rigorous modeling made to the new operating conditions with the switched capaci-
not generally undertaken when placing capacitors in the distribu- tors. This becomes important for the application of surge arresters
tion systems. (Chap. 20).

Figure 6-20 Voltage transient on 480-V bus 2, phase a, on switching of 6-Mvar capacitor bank, 200-kvar capacitor bank already energized, Fig. 6-18.
 Figure 6-21 Current transient through 200-kvar capacitor bank.

F i g u r e 6 - 2 2 A simple circuit for switching transient of a trans-

former, with and without 5-Mvar capacitor bank.

Figure 6-23 A 20-MVA transformer switching current transient without secondary 5-Mvar capacitor bank.
138
 Transients of Shunt Capacitor Banks   139

Figure 6-24 Current transients 20-MVA transformer and 5-Mvar bank switched together.

6-9 PHASE-TO-PHASE OVERVOLTAGES 7.5-Mvar capacitor bank: The bank is ungrounded and
switched at an instant when the voltage in phase a peaks.
The switching of capacitors can produce phase-to-phase overvoltages There is no inrush current-limiting reactor.
at a remote location. Consider a capacitor bank switched on a bus
connected through a transmission line to a transformer. The surge Loads at 13.8-kV bus and low-voltage transformer secondary:
will double at the transformer and the surges produced on energizing Passive load models consisting of appropriate RL elements and
the capacitors are of opposite polarity on two phases; the transformer their connections.
at the remote end will experience phase-to-phase surges. The surge
arresters, normally connected phase-to-ground, will limit the phase- 2.5-MVA transformer: It is the same type of model as for main
to-ground components, depending on their protective level, but the 40-MVA transformers. This transformer is high-resistance
phase-to-phase overvoltages will be twice this value.11 Restrikes in grounded; grounding resistance = 70 Ω.
switching devices controlling ungrounded wye-connected banks can
The results of EMTP simulations are shown in Figs. 6-26
also generate high phase-to-phase voltages (Chap. 20).
through 6-30.
Example 6-8 An electrical power system is shown in Fig. 6-25 Figure 6-26 shows the 13.8-kV three-phase voltages at bus V3
for the simulation of capacitor bank switching transients. The fol- on capacitor bank switching. A sudden voltage dip and subsequent
lowing models are used: recovery profile can be seen from this figure. The peak excursion is
approximately 18 kV from phase to ground.
Source impedance at 230 kV: This is simply a passive coupled Figure 6-27 illustrates phase a voltage on the primary of 2.5-MVA
RL branch, modeled with positive- and zero-sequence source low-voltage transformer at V4 for about 4 ms from the instant of
impedance. switching. The thick band is due to multiple reflections in the
230-kV, 80-km line: A CP model is used, as discussed in Chap. 4. 400-ft cable. The voltage escalates and has high-frequency super-
imposed oscillations due to multiple reflections at the cable ends.
40-MVA, 230- to 13.8-kV transformer: Transformer bushings, To capture these high-frequency oscillations, a time step of 1 ms is
windings, and interwinding capacities are modeled; see used in EMTP simulations.
Chap. 14, for transient models of transformers. The wye Figure 6-28 illustrated the secondary voltage V5 of 2.5-MVA
windings of the transformer are grounded through a resis- transformer, simulation for 30 ms only. The voltage has high-
tance of 20 Ω to limit the ground fault current to 400 A. frequency oscillations and rises to 2900 V peak from line to
Surge arresters: The surge arresters considered are gapless zinc- ground. Some resonant phenomena and amplification is occur-
oxide surge arresters, rated voltages as shown in Fig. 6-21. ring inside the transformer.
Their models are discussed in Chap. 20. Figure 6-29 shows the 7.5-Mvar capacitor three-phase inrush
currents which have high-frequency components, and finally,
400 ft of 3/C 500 KCMIL cable between the 13.8-kV bus and Fig. 6-30 depicts the 230-kV line current.
2.5-MVA 13.8- to 0.48-kV transformer: This is a model with This is continued further in Chap. 20 to illustrate the control of
characteristic impedance, length, and attenuation (Chap. 4). high voltage on the secondary of 2.5-MVA transformer.
140   Chapter Six

Figure 6-25 A power system configuration for the study of switching transients.

6-10 CAPACITOR SWITCHING IMPACT ON 6-11 SWITCHING OF CAPACITORS WITH MOTORS

DRIVE SYSTEMS Capacitors are, generally, applied for the power factor correction
An investigation of the failure of a drive system due to frequent of induction motors. The loaded induction motors operate at a
cycling of a capacitor bank in the utility system, which raised the power factor of 80 to 90 percent approximately. The power factor
dc link voltage of a drive system to trip, is discussed in Ref. 12. This decreases for low-speed motors due to leakage reactance of stator
has been a major concern in the industry. overhang windings.
Generally, the induction motors do not operate at full load,
Example 6-9 This example is an EMTP simulation of esca- which further lowers the operating power factor. Even though the
lation of the dc link voltage of a six-pulse converter feeding
power factor of the motor varies significantly with load, its reactive
a dc motor load, Fig. 6-31, when a 6-Mvar capacitor bank is
power requirement does not change much. Thus, with the appli-
switched. The simulation in Fig. 6-32 shows that the voltage rises
cation of power factor improvement capacitor, the motor power
to approximately 2 pu. This can be controlled by all the means
factor from no-load to full load will not vary much. Figure 6-33
that reduce the capacitor inrush transients; see Sec. 6-13.
 Figure 6-26 Voltage transients, 13.8-kV bus V3.

Figure 6-27 Voltage transient primary of transformer T2, V4.

141
 Figure 6-28 Voltage transient on secondary of T2, V5.

Figure 6-29 Current transients through 7.5-Mvar capacitor bank.

142
 Transients of Shunt Capacitor Banks   143

Figure 6-30 Line current through 230-kV line.

shows three methods of location of the capacitors. (a) The capaci-

tors are connected directly to the motor terminals; (b) the capacitor
and motor are switched as a unit, and (c) the capacitor is switched
independently of the motor contactor through a separate switching
breaker interlocked with the motor starting breaker.
The switching of the capacitor and motor as a unit (second
method) can result in problems due to:

■ Presence of harmonic currents

■ Overvoltages due to self-excitation
■ Excessive transient torques and inrush currents due to out-of-
phase closing

Overvoltages due to self-excitation are important for the proper

application of capacitors with induction motors.
The magnetizing current of the induction motors varies with the
motor design. Premium high-efficiency motors operate less satu-
rated than the previous U- or T-frame designs.13 The motor and
capacitor combination in parallel will circulate a current between
motor and capacitor corresponding to their terminal voltage. In this
manner, the network is said to self-excite.
This is shown in Fig. 6-34a and b. The same size of capacitor
applied to a standard and high-efficiency motor has different results
because of the motor magnetizing characteristics. In case of high-
efficiency design motor, it raises the terminal voltage to 680 V.
A capacitor size can be selected for the power factor improve-
F i g u r e 6 - 3 1 A circuit diagram for the study of escalation of dc link ment based on the following equation:
voltage in a drive system on switching of 6-Mvar capacitor bank.
kvarc ≤ 3I 0 sin φ0 (6-36)
144   Chapter Six

Figure 6-32 DC link voltage and current transients, Fig. 6-31.

Figure 6-33 Various methods of connection of capacitors with motors.

where kvarc is the maximum capacitor that can be applied, I0 is the ■ The motor is meant for repeated switching, jogging, inching,
motor no-load current, and φ0 is the no-load current power factor or plugging
angle. This means that the capacitor size does not exceed the motor
no-load reactive kvar. However, this no-load excitation motor data ■ A reversing or multispeed motor is used
is not readily available and the size of capacitor application for a ■ A high inertia load may drive the motor, that is, a broken
specific motor design should be based on the manufacturer’s rec- belt of a conveyor
ommendations. The capacitors should not be directly connected to
the motor terminals when:
6-12 INTERRUPTIONS OF CAPACITANCE CURRENTS
■ Solid-state starters are used A breaker may be used for line dropping and interrupt charging
currents of cables open at the far end or shunt capacitor currents.
■ Open transition starting methods are applied; see Chap. 11 These duties impose voltage stresses on the breaker. A single-phase
 Transients of Shunt Capacitor Banks   145

Figure 6-34 Self-excitation of motors with capacitors—typical motor saturation characteristics, (a) standard design, (b) high-efficiency design.

circuit for capacitive current interruption is in Fig. 6-35a. The dis- at its crest is trapped on the capacitors. Therefore, after half a
tributed line capacitance is represented by a lumped capacitance cycle following the restrike, the voltage across the breaker poles
C2, or C2 may be a power capacitor. C1 is the source capacitance. is the difference between the supply side and the disconnected
The current and voltage waveforms of capacitance current inter- side which is at peak voltage of the equalizing process and a sec-
ruption in a single pole of a circuit breaker under following three ond restrike may occur. Multiple restrikes can occur, pumping the
conditions are shown in Figs. 6-35b, c, and d.14 capacitor voltage to three, five, and seven . . . times the system
voltage at each restrike. The multiple restrikes can terminate in two
■ Without restrike ways: (1) these may cease as the breaker parting contacts increase
the dielectric strength; and (2) these may continue for a number of
■ With restrike cycles, until these are damped out.
A distinction should be made between reignitions in less than
■ With restrike and current chopping 5 ms of current zero and reignitions at 60-Hz power frequency.
Reignitions in less than 5 ms have a small voltage across the circuit
After interruption of the capacitive current, the voltage across breaker gap and do not lead to overvoltages. Most breakers exhibit
the capacitance C2 remains at the peak value of the power fre- current chopping; see Chap. 8. It is usually caused by the instability
quency voltage: of arc at low magnitude of currents due to high arc voltages. The
chopping currents of modern breakers have been reduced, but it
2 un is clear that the capacitor charge will not be at peak if chopping
u2 = (6-37) occurs. This will lower the recovery voltage peak across the breaker
3 gap. See Chap. 8 for recovery voltage profiles.
In Fig. 6-35d, the current is chopped before it reaches its natural
The voltage at the supply side oscillates at a frequency given by zero. The voltage on the disconnected side does not remain at the
supply side C1 and L1, about the driving voltage un. The difference of service frequency peak voltage, but at a lower momentary value ua
these two voltages appears at the breaker pole. This can be more than of the driving voltage. Depending on the supply source impedance,
double the rated voltage, with no prior charge on the capacitors. If the chopping current ia creates an oscillation with an overvoltage,
the gap across poles of a circuit breaker has not recovered enough and the voltage us across the breaker contacts can rise to:
dielectric strength, restrike may occur. As the arc bridges the part-
ing contacts, the capacitor being disconnected is again reconnected
to the supply system. This results in a frequency higher than that of L1 2
u s = ua + u0 + λ ua2 + i (6-38)
the natural frequency of the source side system being superimposed C1 a
on the 60-Hz system voltage. The current may be interrupted at a
zero crossing in reignition process. Thus, the high-frequency voltage where l is the damping coefficient <1.
F i g u r e 6 - 3 5 (a) A single phase circuit for capacitive current interruption. (b), (c), and (d ) Interruption with no restrike, with restrike, and with current
chopping, respectively.

146
 Transients of Shunt Capacitor Banks   147

Figure 6-36 Sequence of creating trapped charges in an ungrounded three-phase capacitor bank. (a) Phase a clears first. (b) Phases b and c clear in series.

6-12-1 Disconnecting A Three-Phase Bank ■ Using point-of-wave switching (synchronous breakers).

Disconnecting a three-phase capacitor circuit is more complex. The ■ Implementing the application of surge arresters.
instant of current interruption and trapped charge level depends
on the circuit configuration. In an ungrounded three-phase wye- ■ Dividing the capacitor bank into smaller size banks. The
connected bank, commonly applied at medium- and high-voltage smaller the size of the capacitor bank being switched, the
levels, let phase a current be first interrupted. This will occur when lesser is the transient.
the voltage of phase a is at its peak. Figure 6-36a shows that phase a
is interrupted first. The charge trapped in phase a is 1 per unit and ■ Avoiding application of capacitors at multivoltage levels
that trapped in phases b and c is 0.5 per unit. to eliminate possibilities of secondary resonance. Metal-oxide
The interruption of phase a changes the circuit configuration and varisitors (MOVs) can be applied at lower voltage buses.
connects the capacitors in phases b and c in series. These capacitors
are charged with equal and opposite polarities. The current in phases b ■ Converting the capacitor banks to capacitor filters; this is
and c will interrupt simultaneously as soon as phase-to-phase current a must when harmonic-generating loads are present. This is
becomes zero. This will occur at 90° after the current interruption in one effective way to mitigate transients, eliminate harmonic
phase a at the crest of the phase-to-phase voltage so that an additional resonance, and control harmonic distortion. Active filters can
charge of 3 / 2 is stored in the capacitors, as shown in Fig. 6-36b. be used, depending on the size of the capacitive compensation
These charges will add to those already trapped on the capacitors in and harmonic mitigation required for a particular system.
Fig. 6-36a, and thus voltages across the capacitor terminals are:
■ Providing current-limiting reactors and chokes, which is a
Ea = 1.0 per unit must for back-to-back switching and to have acceptable capac-
Eb = 0.37 per unit itor switching duties on circuit breakers and switching devices.
Ec = 1.37 per unit ■ Considering steady-state voltage rise due to the application
of capacitors. The transformer taps may have to be adjusted.
Further escalation of voltages occurs if the phases b and c are
not interrupted after 90° of current interruption in phase a. The
voltage across the interrupting breaker pole will be the sum of sup-
6-13-1 Resistance Switching
ply system voltage and the charge trapped, maximum occurring in In ac current interruption technology, the use of switching resis-
phase c, equal to 2.37 per unit. The capacitor bank is considered tors in high-voltage breakers is well implemented to reduce the
ungrounded and the stray capacitance to ground is ignored in the overvoltages and frequency of transient recovery voltage (TRV);
above analysis. It is hardly possible to take into account all modes see Chap. 8. In medium voltage cubical-type or metal-clad circuit
of three-phase interruptions with restrikes.14 breakers, as the switching resistors are not integral to the breakers,
Advancements in the contact materials in vacuum interruption two breakers can be used, as shown in Fig. 6-37, to preinsert the
technology has reduced the restrikes and prestrikes of initial years, resistor for a short duration.
and so also the chopped currents.15 However, it cannot be said Figure 6-38 shows a basic circuit of resistance switching. A resis-
that the restrikes are totally eliminated. The application of surge tor r is provided in parallel with the breaker pole and R, L, and C are
arresters provides a safeguard against these overvoltages and con- the system parameters on the source side of the break. Consider the
sequent insulation failures.16 current loops in this figure. The following equations can be written:
di 1
6-13 control of switching transients u n = iR + L + ∫ ic dt (6-39)
dt C
The capacitor-switching transients can be controlled by:
1
■ Using series inrush current limiting reactors. C
∫ ic dt = ir r (6-40)

■ Using resistance switching. i = ir + ic (6-41)

148   Chapter Six

The same result could have been arrived at directly from Chap. 2.
Again reverting to the switching of 6-Mvar capacitor bank, and
substituting the values, the transient disappears for a resistance
of 1.32 Ω.
The simulation is shown in Figure 6-39a and b for current and
voltage, respectively. A 1.32-Ω resistor is inserted for four cycles only.
This clearly shows the effect of resistance switching on damping out
the switching inrush transients. The current and voltage waveforms
are at fundamental frequency; only minor excursions are visible.

6-13-2 Point-of-Wave Switching or

Synchronous Operation
In all the examples, the transients are calculated with the switch
closed at the peak of the voltage wave. A breaker can be designed to
open or close with reference to the system voltage sensing and zero
crossing.17,18 An example of switching a transmission line with syn-
chronous closing is discussed in Chap. 7. The switching device must
have enough dielectric strength to withstand system voltage until its
contacts touch on a closing operation. The consistency of closing
Figure 6-37 Two metal-clad indoor breakers organized for within ± 0.5 ms is possible. Grounded capacitor banks are closed
resistance switching. with three successive phase-to-ground voltages reaching zero, for
example, 60° separations. Ungrounded banks are controlled by
closing the first two phases at a phase-to-phase voltage of zero, and
then delaying the third phase 90°, when phase-to-ground voltage
is zero. The results obtained approximate resistance switching. See
Chap. 7 for an EMTP simulation of synchronous closing.

6-13-3 Surge Arresters

Chapter 20 is devoted to surge arresters. Here it can be said that
gap-type surge arresters in series with nonlinear resistors can be
exposed to high stresses when protecting capacitor banks. If a
transient triggers the arrester, the capacitors will discharge totally
and the energy is dissipated in the arresters. Conversely, gapless
metal oxide arresters will not discharge the capacitors below the
system-rated voltage, as there is smooth transition from conducting
to insulating condition.
The possibility of overvoltages due to lightning, switching surges,
and temporary overvoltages requires a detailed evaluation before
the application of surge arresters close to a capacitor bank.19
Figure 6-38 Diagram to explain resistance switching. On an incoming surge, the capacitors will absorb the charge, not
much dependent on the rate of rise of the incoming voltage. Shunt
capacitor banks have low surge impedance and to an extent may be
This gives: self-protecting in case there are other surge arresters properly rated
for lightning surge duty. Depending on the substation configura-
d 2ir  R 1  dir  1 R  tions, additional surge arresters may be necessary.
+ + + + i =0 (6-42)
dt 2  L rC dt  LC rLC r
The overvoltage protection should be considered at the follow-
ing locations:
The frequency of the transient is given by:
■ On the switching circuit breakers to control the TRV when
2 the shunt capacitors are switched out. See Chap. 8.
1 1 1 R 1 
fn = − − (6-43)
2π LC 4  L rC ■ At the primary of transformer to limit phase-to-phase
voltage due to capacitor switching. See Chap. 14. Also,
on the secondary of the transformer windings close to the
In power systems, R is << L. If a parallel resistor across the switched capacitors, an example of which has been discussed
1
contacts of value r < L /C is provided, the frequency reduces in this chapter.
2
to zero. The value of r at which frequency reduces to zero is called ■ At the end of transformer-terminated lines to limit the
the critical damping resistor. The critical resistance can be evalu- phase-to-phase overvoltages due to capacitor switching.
ated in terms of the system short-circuit current, Isc:
■ On neutrals of ungrounded capacitor banks.
1 un
r= (6-44) ■ On coupled low-voltage systems; Example 6-6 given in this
2 I scω C
chapter.
 Transients of Shunt Capacitor Banks   149

Figure 6-39 (a) Capacitor switching current with resistor switching, Fig. 6-1. (b) 13.8-kV bus voltage with resistance switching.
150   Chapter Six

6-14 SHUNT CAPACITOR BANK ARRANGEMENTS Table 6-2 M

 inimum Number of Units in Parallel
Shunt capacitor banks are formed from standard unit capacitor sizes per Series Group to Limit Voltage on the
available in certain kvar and voltage ratings. The unit sizes are gener- Remaining Units to 110% with One
ally limited to 100, 200, 300, and 400 kvar, and their voltage ratings Unit Out
are: 21.6, 19.92, 14.4, 13.8, 13.28, 12.47, 9.96, 9.54, 8.32, 7.96,
7.62, 7.2, 6.64, and so on, all in kV. To form a capacitor bank at a Number Grounded Double-Wye,
certain system voltage, a number of units are required to be con- of Series
Wye or Ungrounded Equal
nected in series and parallel combination, as shown in Fig. 6-40. Groups Delta Wye Sections
A capacitor bank for 500 kV can be formed with the following
alternative capacitor units: 1 — 4 2
2 6 8 7
■ 14 series strings of 21.6-kV capacitor units are required.
3 8 9 8
■ 20 series strings of 14.4-kV capacitor units are required.
4 9 10 9
■ 30 series strings of 9.54-kV capacitor units are required. 5 9 10 10
■ 38 series strings of 7.62-kV capacitor units are required. 6 10 10 10
7 10 10 10
There are imitations on the formation of series strings and the
number of capacitor units in a parallel string. Figure 6-40 shows 8 10 11 10
that each capacitor unit is protected by a fuse. A minimum number
9 10 11 11
of capacitors in parallel should be placed per series group to limit
the overvoltages on the remaining units to 110 percent, in case one 10 10 11 11
capacitor goes out, say due to operation of its fuse. Table 6-2 shows
11 10 11 11
the minimum number of units in parallel per series group to limit
voltage on remaining units to 110 percent with one unit out. 12 and above 11 11 11
Table 6-3 provides analytical expressions for the calculations of
fault current and voltage on each remaining group in series with
one group faulted. Table 6 - 3 6 0-Hz Fault Current and Voltages
Individual capacitor fusing is selected to protect the rupture/ with One Unit Shorted
current rating withstand capability of the capacitor can, and the
Voltage on Each Remaining
fuse operates to prevent a rupture of the capacitor can. The maximum
Group in Series with Faulted
Bank Configuration Fault Current Group
S * VLG
Wye, grounded I
S −1 φ S −1
3S 3VLG
Single, ungrounded I
3S − 2 φ 3S − 2
6S 6VLG
Double, ungrounded I
6S − 5 φ 6S − 5
S = Number of series groups.
I φ = Nominal phase current, A.
*
For S = 1, the current is the system line-to-ground fault current.
VLG = Line-to-ground voltage V. (Maximum value is used where appropriate).

number of capacitors that can be connected in parallel in a series

group is dictated by the energy liberated and fed into a fault
when a fuse operates. Figure 6-41 depicts that for a fault in one
of the parallel units, the stored energy in the parallel capaci-
tor units is fed into the fault and the protecting fuse must be
able to handle this energy, depending on its characteristics. This
energy release may be approximately calculated from the follow-
ing equations:

E = 2 . 64 J/kvarc rated voltage (6-45)

E = 2 . 64(1 . 10)2 J/kvarc 110 % voltage (6-46)

E = 2 . 64(1 . 20)2 J/kvarc 120 % voltage (6-47)

F i g u r e 6 - 4 0 Typical arrangement of the formation of a high- Normally with expulsion-type fuses, commonly used for out-
voltage capacitor bank. door rack-mounted capacitor banks, 3100 kvar of capacitors in
 Transients of Shunt Capacitor Banks   151

Figure 6-41 Limitation of number of capacitor units in parallel with expulsion-type fuses.

total can be connected in a parallel group, that is, a maximum of rated for line-to-line voltage. A wye-ungrounded group can be
ten 300 kvar units. When current-limiting fuses are used, this limit formed with one series group per phase when the required oper-
can be increased. ating voltage corresponds to standard capacitor unit rating. For
example, for 13.8-kV service voltage, one series group is ade-
quate with capacitors rated for 7.96 kV ( 3 × 7 . 96 = 13 . 8 kV )..
6-14-1 Connections and Grounding of Grounded wye neutrals and multiple series groups are required
Capacitor Banks for higher voltages above 35 kV. As a rule of thumb, when the
The capacitor banks can be connected in a variety of three-phase system is effectively grounded, the capacitor bank neutral is
connections, depending on their size, optimum utilization of the grounded. The grounded capacitor banks have the following
capacitor units, fusing, and protection. The following are the com- characteristics:
mon connection methods:
■ These provide a low-impedance path to ground for the
■ Multiple series groups, grounded wye connection harmonic currents. The resulting harmonic currents may
cause interference with the communication systems if
■ Ungrounded wye connections
the power lines are paralleled with the communication
■ Grounded double-wye neutrals circuits.
■ Delta connections ■ An open phase produces zero sequence currents which
may cause ground fault relay operation.
These connections are shown in Fig. 6-42. Delta connection
is common for the low-voltage systems, with one series group ■ Third-harmonic resonance could be a problem.

Figure 6-42 Grounding arrangements of three-phase capacitor banks.

152   Chapter Six

■ As the neutral is grounded, the recovery voltages on current 4. In Problem 3, calculate the inductance to drop the inrush
interruption are reduced. current and frequency of inrush current to 15 kA and 2000 Hz,
respectively.
■ A low-impedance path is provided to lightning surge currents
and gives some protection from the surge voltages. Sometimes, 5. Find the value of R for resistance switching in Problem 1
the bank can be operated without surge arresters. with bus section circuit breaker open, and then repeat with the
bus section circuit breaker closed.
On the other hand, the ungrounded banks have the following
characteristics: 6. Ignoring the nonlinear load, what is the Thévenin impedance
of the system, as seen from 13.8-kV bus A? Use complex R + jX
■ Banks do not provide any surge voltage protection and calculation to combine the component impedances.
provide no path to ground for third-harmonic currents.
7. Consider that the nonlinear load shown within dotted lines
■ The entire bank, including the neutral, must be insulated for has a harmonic current spectrum, as shown in Table 6-1. Find the
line-to-line voltage. current in the capacitor bank at each of the harmonics 5th, 7th,
and 11th.
When a bank is too large to meet the requirements of 3100 kvar
per group as a maximum for the application of expulsion-type fuses 8. Calculate the current in the Thévenin impedance at each of
or large enough to meet the minimum series group requirements, the harmonics 5th, 7th, and 11th. From this calculation, find the
the bank can be split into two wye sections. The two neutrals are harmonic current injected into 138-kV utility system at each of
usually ungrounded. the harmonics.

9. Why can the power factor of an induction motor not be

Problems corrected to unity? Explain self-excitation and what gives rise
1. A distribution system is shown in Fig. 6-P1. Calculate to it. How it can be avoided?
the voltage and current transients on sudden switching 10. Form a 58-Mvar, 34.5-kV, neutral grounded, wye-connected
of 10-Mvar capacitor bank, when the bus section circuit capacitor bank. Select a number of units in a parallel in each
breaker is closed. Ignore the nonlinear load shown in dotted group and total number of series units in parallel, using the
lines. guidelines given in this chapter.
2. Repeat Problem 1 with bus section circuit breaker open. 11. In Problem 10, calculate the fault current when one unit is
3. In Fig. 6-P1, the 10-Mvar capacitor bank is divided into shorted. Calculate the voltage in each remaining group in series
two banks of 5 Mvar each. If the total inductance between the with faulted group.
banks is 15 mH, calculate the inrush current and its frequency 12. Figure 6-13 does not show the voltage waveform of
on energizing a 5.0-Mvar bank when the other 5.0-Mvar bank 4.16-kV bus. Will it be the same as in Fig. 6-8? Will it be
is in service. much improved because of filter?

Figure 6-P1 A power system configuration for problems.