CHAPTER 16

Protection of Systems and Equipment against Transient Overvoltages

16.1 INTRODUCTION

Surges in power systems can very destructive. It is important, therefore, to protect power
equipment against them wherever possible, consistent with sound economist. In this chapter
we discuss how this is done.

First, we must strive to preclude lightning surges, generated outside the system, from gaining
entry. Second, we must minimize the effects of those that do center. We must accept
switching surges and generated by faults since are inevitable consequence of normal system
operation. However, we can reduce their effect by careful design; the reduction of switching
surges by pre-insertion resistors (Section 9.7.2) is an example of this procedure.

The protection of terminal equipment in substations is of particular

concern because, as we have seen in Chapter 15, damage there can be very
Costly in replacement and in outage time. A family of surge protection
Devices has been developed over the years to greatly assist in this task.

These devices and their application are discussed in Section 16.4. Special
requirements for HVDC arc examined in Section 16.6.

Yet another topic examined in this chapter is the protection of rotating

machines in power plants and industrial power Systems.

Finally, surges can intrude into control circuits causing destruction and/or
misoperation of relays, measuring equipment. monitors, etc. How such
problems are mitigated is the subject of Section 16.9.

16.2 PROTECTION OF TRANSMISSION LINES AGAINST LIGHTNING

It is common practice to protect overhead lines against lightning at least in
areas they are deemed to be vulnerable. The protection comprises
one or ground wires that are from tower to tower above the
phase conductors along the right-of-way. These wires are bonded to the
towers, so that they are at potential under normal circumstances.
The intent in the disposition of these is to have them intercept
lightning strokes that would otherwise have terminated on the phase conductors
themselves. Such protection may the extend length of a
transmission line, or it may be confined to the first few miles adjacent to the
substation or switching station. In the latter case the assumption is that
surges produced by any lightning strike to the unprotected section of the line
will have been reduced to manageable proportions by the time they reach
the ends of the line by flashover along the line and/or corona and other
attenuating effects experienced by the surge in its progress along the line.

One main concern in this section is with the location of the ground wires
with respect to the phase conductors. For them to be effective, they must be
more attractive than the phase conductors to the lightning.

As a leader stroke approaches the earth's surface, it is common experience

that it is attracted towards tall objects such as trees, tall buildings
and transmission tower structures. The concept of strike distance, S, has
been introduced [1]. What this means is that when the stroke tip reaches
within the distance S of a grounded object, the probability of the
terminating on that object is significantly greater than the probability of striking another object
a greater instance away. It has been observed that S
is a function of current; Love [2] has proposed the following relationship:

S=10 I 0.65

Where S is in meters and I in kiloamperes. Thus, if I =10 KA , S=44.7 m , and if ¿ 50 KA ,

S=127 m. Following Anderson’s approach [3], we construct the diagram shown in Fig. 16.1,
which shows the relative disposition of a ground wire G and a phase conductor ɸ as defined by
the shielding angle, α. Selecting some current I . arcs are drawn with G and ɸ as centers
and radios S corresponding to I .

The horizontal line at height βS, indicates

that strokes approaching within βS of the earth's surface arc disposed to
terminate on the earth rather than on G or ɸ. According to Anderson [3],
β=0.8 for EHV lines and 0.67 UHV lines. This line and the two arcs define
three regions. Strokes with magnitude I penetrating the arc OP will tend to
terminate on the ground wire; similar strokes penetrating the region QR Will
go to ground. Between is the arc PQ where strokes of current I will he
attracted to the phase conductor and thereby represent a shielding failure.

Viewed from above, there is an unprotected sward of width X which

parallels the transmission line.

For greater values of S than that showT1 in Fig. 16.1, corresponding to

higher values of I , the arc PQ Will be redutxd. At some value of I , which we
will designate I max , point P und Q will coincide. There should be no
shielding failures for lightning strokes with currents magnitudes above I max.
Conversely, as S is reduced for smaller currents than that chosen for Fig. 16.1
the PQ and its projection X Will increase. However, there is some
current we will designate I min, below which a stroke to the phase conductor
will be of little consequence because it will generate insufficient voltage to
cause a flashover of the line insulators. To a first approximation this voltage
corresponds to the critical flashover voltage of the line insulators. Thus, if
we neglect the impedance of the lightning channel,
2 V CFO
I min =
Z0

Where V CFO is the insulator critical flashover voltage and Z 0 is the surge impedance of the
line. In this equation we are arbitrarily neglecting the 60 Hz potential on the phase conductor.
The value of X corresponding to I min we will call X s.

For shielding failures to occur, the current must be in the range I min < I < I max, and the strokes
must terminate within a zone we have designated X s.

The probability of a current being within a particular range can be obtained from Fig. 14.9.That
is to say, probability I min < I < I max=Pmin −P max .
The number of strokes to earth/square km/year is defined by N=kT ,where T is the keraunic
level, defined in Section 14.3 and k is a constant in the range 1<k<0.19. Thus, the following
expression can be written for the shielding failure rate
kT x s
x x ( Pmin −Pmax )
10 2
N SF = /100 km/ year
100 km

xs
is the average value of X for the current range I min < I < I max, and is measured in meter. The
2
procedure just outlined must be performed for each phase conductor with respect to its most
protective ground wire in
order to obtain the overall shielding failure rate for the complete transmission
circuit.

Certain assumptions have been made in arriving at Eq. 16.2.3. For

example, it has been essentially assumed that lightning strokes coter the
environs of the line vertically. Other flash angles have been investigated by
Brown and Whitehead [4] lt is also assumed that the earth is flat in the
Vicinity of the line. Sloping terrain, adjacent trees, etc. can affect the
shielding. The height of the line may also vary along its right-of-way. It is
usual for a given span to assume the height shown in Fig. 16.1 is the height
at the tower, 2/3 the sag.

It is helpful to compute the shielding failure rate during the design stage
of a line, so that if the rate is unacceptable: design changes can be made to
correct it. This is done by reducing the shielding angle. lt is possible to
attain a zero shielding failure rate, at least by the criteria being used, if that
is what desired. Presumably the location of the phase conductors is fixed
by other considerations (clearances to ground, to the tower and two other
phases), the option, therefore, is to move the ground wire.
One would first compute S corresponding to X 1 , by combining Eqs. 16.2.1
and 16.2.2:
 2 V CFO 0.65
Smin =10( )
Z0
Point Q is now located, being Smin from the most exposed phase conductor
and β Smin above ground (refer to Fig. 16.1). The ground wire must now also
be Iocated a distance Smin from the point Q, which means on a circular arc
centered on Q and of radius Smin This assures that P and Q of Fig. 16,1
coincide, leaving no unshielded region. This is illustrated in Fig. 16.2.

TO attain complete shielding in this way may require installing two ground
instead of one, or increasing the height of the tower to elevate the
single ground wire (or both). Such design changes have an economic penalty
which must be weighed against the cost of outages which a finite shielding
failure rate implies.

16.3 LIGHTNING SHIELDING OF SUBSTATIONS

Ground wires are strung above substations for the same reason that they are
placed over transmission lines. They are frequently augmented by lightning
rods, extending above the substation fences and structural steelwork [5].

This overhead ground cape is solidly bonded to a “ground mat.” a mesh of

interconnected conductors placed beneath the crushed stone of the substation.
This mat is also linked to reinforcing rods of the concrete foundations
on which the substation components are placed. Rods are driven into
the ground and connected to this mesh to give as Iow a ground resistance as
economically viable. More is said about grounding in Section 16.8.

16.4 SURGE SUPPRESORS AND LIGHTNING ARRESTERS

16.4.1 Some Fundamentals

Inevitably the best shielding schemes will fail from time to time, and some
switching operations and faults may generate overvoltages which can be
potentially hazardous to the insulation of terminal equipment. To protect
the equipment against these contingencies, a family of devices has been
developed over the years, variously known as surge suppressors, surge
diverters, and lighting arresters. These devices are placed in parallel with
the object to be protected and in close proximity to it. They can be
permanently connected in such locations, or they can be switched into
operation, by the sparkover of a series gap. Their purpose is to assure that
under no circumstance will the voltage across them exceed What the
protected object can support.

The first important idea regarding such protective devices can perhaps
best be described as impedance division. A rod gap can be used as a crude
protective device by placing it in parallel with the insulation to be protected.

We consider this now as an extreme example to illustrated impedance

division. The thought behind its use as u protective device is as follows.

Under normal circumstances the open gap represents a very high impedance
dance, one might almost say an infinite impedance. When the gap flashes
over, it switches to a Iow impedance mode, for the voltage of the arc that forms is probably a
few hundred volts only, or a few thousand volts for a
long gap on a high-voltage system. The voltage is Iow because the arc
impedance is low compared with the impedance of the system, which is the
source of the surge. Since the gap is parallel with the equipment to be
protected, it maintains a low voltage across that equipment also. Thus we
note an effect whereby the surge voltage is divided between the system
impedance and the protector impedance. This can be placed on a more
rigorous basis by invoking Thevenin’s theorem.

One form of this theorem

states that when a switch is closed in a circuit, the current that flows can be
determined by dividing the voltage across the switch prior to closing, by the
impedance seen looking into the circuit at the open switch contacts. In Fig.
16.3, suppose that Z1 is the impedance of a system on which a surge is being
generated and that Z2 is a piece of equipment connected to the system. This
latter is parallel by a surge suppressing device of impedance Z p , which is
switched into the circuit when S is closed; S may be a gap of some kind that
the surge causes to spark over. According to Thevenin, the current through
S, when it is closed, is given by

Where V is the voltage across S prior to closing. Looking into the circuit at the switch we see
the impedance Z1 and Z2 in parallel, and they in turn are in series with Z p . The voltage V 1
across Z p can be obtained simply from Eq.
For the surge suppressor to do a satisfactory job of protecting the equipment
represented by the impedance Z2 , neither the voltage V, which it sees
initially at the time S closes, nor the voltage V , given by Eq. 16.4,2, should
exceed the voltage that the equipment is capable sustaining. One ,would
suppose that if V is satisfactory. V1 would also be satisfactory since V1<V
Actually, this is necessarily true; we have oversimplified the problem
an example Will show.
Suppose that Z1 represents the surge impedance of a transmission line
along which the surge approaches, and that Z2 is the surge impedance of a
transformer at its termination, let us also suppose that the Surge voltage
peak, Vpeok is well in excess of the electrical strength of the transformer
insulation. Without any protection this would be particularly serious, for
, as was shown in Section 9.3. When Z2>Z1, as would be the case here.
The surge would almost double on reaching the transformer terminals. But in
situation before us, we have a suppressor that sparks over at some safe
voltage V, well below Vp, when this occurs, the voltage will instantaneously
drop to the value V1, Of Eq. 16.4.2- Subsequently, however, as the surge
proceeds, the effective driving voltage will increase, and so therefore will
the voltage across transformer and suppressors.
The point we wish to make here is that the impedance of the protective
device and the impedance of the system as well as the value of the surge
itself, all play an important role in determining how effective the suppressor
shall be. This topic has been expanded by Sakshaug and Hednan (6).
The third important factor regarding protective devices relates to them
potential for storing and/or dissipating energy. When current is diverted
into a surge suppressor of whatever kind, and voltage is generated across it,
it is at the time absorbing energy. Tbc amount of energy involved depends
upon the magnitude and duration of the surge. The suppressors must be
capable of handling this energy by storing or dissipating it. without damage
to itself; it is not sufficient that it holds the voltage down to an acceptable level.
Analysis shows that, for a given surge, the maximum energy is dissipated in
a protective device when its impedance is equal to the impedance seen
looking into the system from the terminals of the protective device, In Fig.
16.3 this means when Zv. is equal to the parallel combination of Z1, and Z2.
For greater or smaller value values of Zp the energy is less; it is obviously
zero when Zp=0 v Zp=0.

16.4.2 Nonlinear Resistors

The rod gap was mentioned in the last section as form of overvoltage protection. Such have
the virtue of being very cheap, but they have
several disadvantages, the most important of which is that when they flash
over they throw a fault the circuit. lt is that this can be removed by
opening a circuit breaker, and that power can be restored by reclosing the
breaker after the rod gap has deionized, certainly this is preferable to
damaging a costly piece of equipment and perhaps sustaining a Lengthy
outage. Hut a device that can effect voltage limiting without creating a
fault is obviously more attractive. nonlinear resistor is such a device.
These resistors have the property that their resistance diminishes sharply
as the voltage across them increases. This characteristic is usually expressed
in the following way.

16.5 APLICATION OF SURGE ARRESTERS

16.5.1 Objective
The function of a surge arrester is to protect the insulation of other equipment, such as
transformers, reactors, etc., without putting itself at risk. The highest protective margin or
protection ratio is desirable, but we must recognize that as the margin increases, the energy
demands on the arrester also increase. Proper application demands a judicious compromise.
All contingencies must be considered: fast impulse surges, switching surges and temporary
overvoltages. The minimum recommended protective ratios are as follows:

Chopped wave withstand

≥1.20
Front −of −wave protective level
Full wave withstand (BIL) Switching surge withstand
≥ 1.20 ≥ 1.15
Impulse protective level Switching surge protective level

These ratios should be met, and if possible, significantly exceeded since the characteristics of
equipment deteriorate with time. Energy loadings much be investigated to be assured they are
not exceeded.

16.5.2 Objective
The procedure will be illustrated with the example shown in Fig. 16.17. Since this is a 345 kV
circuit (362 kV, max. design), we would select an

Arrester with a MCOV of 362/ √ 3=209 kV . The breaker is to be closed onto the transmission
line without benefit of a pre- insertion resistor. In Section 9.7.1 we showed that the voltage
surge which travels down the line is given by

V ( t ) =( 1−ϵ−αt ) V (16.5 .1)

Where ∝=Z 0 /L and V is the voltage across the switch at the time of its closing. The voltage
starts at zero and climbs to V through the (1 - exp). The time constant is 1/∝ which is 37 μs,
which is short compared with the remote end it is refracted at the high impedance of the
transformer. Being conservative, we can treat this like an open circuit, so that the prospective
surge voltage doubles, approaching 2V after three or four time constants.
Consider first a normal energization. Assuming the worst condition—closing at the voltage
peak—the prospective crest at the transformer bus is (2 √ 2/ √ 3)362=591 kV . This
represents no threat to the transformer since is BIL is likely to be 900 kV with a corresponding
SIL of 0.83 X 900 = 747 kV. However, the arrester will be driven into conduction, so one must
check on the energy it absorbs. This can be done with EMTP if the appropriate arrester model
is available, or by a small INA study. We will stick to fundamentals and apply the teaching of
Fig. 16.5.
If we neglect corona and other attenuating agencies, the surge at the arrester bus will
attempt to rise to 591 kV and remain there until the reflected wave has gone back to the
source bus and returned a new reflection to the transformer, a total of 2.15 ms. The arrester
will, of course restrict the voltage at the transformer to a lower level determined by the
arrester characteristic and the load line. Figure 16.18 shows that this intersection occurs at Q 1,
which is 432 kV and 480 A. The power being dissipated is 423 × 480=203,040 kW and is
proportional to the rectangle OP1Q1R1. The energy spent during 2.15 ms ̶ 436.5 kJ. In terms of
the arrester it is 436.5/209 = 2.09 kJ/kV of MCOV, which is well within the arrester's capability
as indicated in Table 16.2.
Consider, now, the situation when the breaker recloses at a positive voltage peak when a
charge corresponding to negative voltage peak is trapped on the line. The voltage in Eq. 16.5.1
is now twice peak system voltage, and the potential at the arrester attempts to swing from ̶-V p
to 3V i.e. to 887 kV which exceeds the transformer SIL. More important, however, is the shift
to the right of the load line in Fig. 16.18 and the implications for energy absorption. The new
intersection is at Q2, i.e., 530 kV and 1020 A. It is gratifying to observe that the voltage is well
within

the SIL of the transformer, giving a protection ratio of 747/530= 1,41. However, the power
dissipation is proportional to the much bigger rectangle OP 2Q2R2, being 530 X 1020 = 540,600
kW. The energy spent during 2.15 ms = 1162 kJ. In terms of the arrester it is 1162/209 = 5,560
kJ/kV of MCOV. This figure is still within the arrester's capability according to Table 16.2, but it
is apparent that it could not take two such operations in quick successions. This points up the
benefits for the arrester of pre-inserting resistors on closing.

16.5.3 Protection Against Fast-Rising Surge

We need not be concerned with energy dissipation for fast-rising lightning surges because of
their short duration; the power may be high but not the energy. A review of Table 16.4 reveals
that the selected arrester with 209 kV MCOV (designated VN 217909) has a peak voltage of
665 kV for fast surges. This coordinates satisfactorily with the 90 kV BIL of the transformer to
give a protection ratio of 900/665 = 1,35.
There is an issue which must be discussed in this application and that is the need to place the
arrester as close as possible to the transformer it is protecting. Indeed, installing all protectors
as close as possible to the protected object is a cardinal rule of protection. Let us see why this
is so.
If a steep-fronted surge approaches a transformer that is protected by an arrester, the
arrester will spark over if it is a gapped arrester, and otherwise limit the voltage by going into
conduction if it has no gap. In either case it will let through some fraction of the surge which
will travel on and impinge on the transformer terminals. This is shown in Figs. 16.19ɑ and
16.19b, which show how a traditional gapped arrester would behave. At the same time a
reflected wave, also shown in the figures, will travel back down the line modifying the incident
surge. The transformer will present a comparatively high surge impedance to the ongoing
wave, the reflection adding substantially to the incident wave. The state of affairs will continue
until the reflected wave from the transformer terminals has returned to the arrester, been
itself reflected and arrived back at the transformer where I can effect a mitigating influence. If
the original surge has a very fast front, the transformer voltage could rise well above the
arrester champing voltage during

this time. If the arrester is right at the transformer terminals, this could not occur, hence the
need for a small separation distance.
Even a short length of lead to an arrester can be significant. Consider Fig. 16.20ɑ, which
shows a pole-mounted distribution arrester used to protect a cable pothead and equipment
connected to the cable. It would appear to be in close proximity to the pothead yet there is a
couple of feet of connections between the two. We note that these leads are directly in
parallel with the pothead and that in the event of an arrester operation, the discharge current
will flow through them, thereby adding their inductive volt drop to the discharge voltage of the
arrester. Using the figure from Section 13.8 of 0.4 μII / ft , and noting that a severe lightning
current might rise at 4 kA/μs, we see that 2 ×0,4 × 4=3,2 kV is added to the protective level
of the arrester, which is not insignificant in a distribution circuit.
The problem can be circumvented by making the connection first to the arrester and then to
the pothead, as indicated in Fig. 16.20b.

16.5.4 Protection Against Temporary Overvoltages

Temporary power frequency overvoltages (TOV) occur on power systems from time to time. A
single line-to-ground fault can create such voltages on the unfaulted phases, Ferranti rise due
to load rejection, and ferroresonance are other causes. TOVs have created problems in the
past for gapped SiC arresters if a surge caused sparkover while TOV was present. This was
because the arrester gaps were sometimes unable to interrupt the follow current in the
presence of the TOV. The devices were destroyed as a consequence.
ZnO arresters have some tolerance for TOVs, but it is limited. The power dissipation and
temperature increase rapidly with the increase in voltage. The 209 kV MCOV arrester cited in
the last two subsections can withstand a TOV of 304 kV for one second and maintain stability.
Arresters can withstand lower TOVs for longer periods and vice versa. The ambient
temperature and any dissipation immediately prior to the TOV will affect its tolerance. This is
made clear in Fig. 16.21.
The likelihood of TOVs should be studied where applying arresters. ANSI IEEE C62.2(12) can
be consulted with respect to magnitude. Duration should take cognizance of the operating
time of hack-up breakers.
16.5.5 Arrester/Equipment Insulation
The information of the last three subsections is conveniently summarized in the chart shown in
Fig. 16.22, which shows the protective margins for various applications. The illustrative
example is for a 230 kV SYSTEM. Equipment at this voltage normally has a BIL of 750 kV, but in
this diagram it has been reduced to 600 kV in recognition of aging.

16.5.6 Surge Arresters and Cable Circuits

The current which a surge suppressor must discharge is very dependent on the surge
impedance of the circuit to which it is connected. Imagine how the situation in Fig. 16.18
would change if the overhead line (Z 0 = 350 Ω) were replaced by a cable (Z 0 = 30 Ω). However,
there are mitigating factors, first, cables are usually relatively short; second, surges usually
enter cables from overhead lines, thus the initial surge in the cable is much lower than that on
the line. The voltage builds up subsequently, as we shall see later in this chapter, so the
potential energy dissipating demands on arresters near cables should be investigated.

16.5.7 Surge Arresters and Capacitor Banks

Capacitors are low impedance sources of energy which can convey much of their energy to
nearby arresters under adverse conditions. A normal energization close to a voltage crest is
shown in Fig. 16.23c. The MOV beings to conduct when the voltage approaches the arrester's
protective level, so that the overswing is modified significantly. The current is modest, so that
the

energy spent in the arrester, the integral of the product of current and voltage over the period
of conduction, is well within its capability.
The situation can be different, however, if a restrike occurs when the capacitor bank is being
disconnected (sec Section 5.3). This condition is illustrated in Fig. 16.25d. The restrike is
assumed to occur at the most inopportune time, when the source is at minus peak system
voltage and the charge trapped on the capacitor is at peak positive voltage. The capacitor will
attempt to swing to –p pu, rather than 2 pu, under these circumstances. The conduction
period will be longer and the current, and therefore the energy, will be much greater. Such
conditions should be modeled and checked to be certain that energy demands on the arrester
are not excessive.

16.5.8 Arresters for Line Protection

Relatively recently, arresters have been introduced for the protection of transmission and
distribution lines as an alternative to overhead ground wires [13,14]. They are placed at
intervals on the poles or towers. How frequently they are installed, and whether they are
connected on all three phases or just the most exposed phase, is largely a matter of
economics. In one location a single arrester may be used on every third pole on one phase

only. On a critical section of a critical line, arresters may be placed on all those phases at every
tower.
The voltage across the line insulator at the point of installation is controlled by the MOV and
kept below the flashover voltage of the insulator, thus avoiding back flashover (see Section
14.3) which is the most frequent form of lightning outage. In the normal course of events, the
voltage across the line insulation is greatly influenced by the tower footing resistance. This is
not so in the application being discussed, a fact which is very important in regions where the
footing resistance is high.
The usual embodiment for a distribution line is shown in Fig. 16.24. The functions of support
ad protection are separated in two components. The structural member, whether post or
suspension type, is frequently a high strength fiberglass rod with polymeric rubber skirts. The
parallel MOV is also polymer-housed. This protection has been successfully applied at voltages
from 4.16 to 230 kV and appears to be technically and economically a valid alternative to
overhead ground wires.

16.5.9 Transfer of Surges Through a Transformer

It is common practice to put arresters on both sides of a transformer to protect the high and
low voltage windings. This creates a possible problem, but one that is readily solved with MOV
arresters. The possibility is that a switching operation will cause the energy of a high side
transient to be dissipated in the low voltage arrester. Even the sharing of the energy with the
high side arrester could overtax the arrester on the low voltage side. The characteristics of ZnO
arresters are such that increasing the MCOV of the low voltage arrester by 4% is sufficient to
avoid this contingency. At the same time the reduction of the protective margin on the low
voltage side will not jeopardize that winding.

16.6 SURGE SUPPRESSORS FOR DIRECT CURRENT CIRCUITS

There has been a considerable growth in HVDC transmission in the past twenty-five years.
Direct current has also found increasing use in industrial power systems. Both have been
largely attributable to advances in solid state devices, the kind of devices described in
Chapter7. As noted there, such components have a low tolerance for overvoltages, so
protection is a necessity. This has been a problem for traditional surge suppressors, for once
they have sparked over, a path is created for direct current which is difficult to interrupt. The
advent of the MOV has changed this situation. They can remain under voltage, drawing a
minimal current, in a d.c. circuit just as well as an a.c. circuit, but ready to carry a much greater
current and clamp the voltage severely in the appearance of an overvoltage.
The energy demands of surge suppressors in d.c. circuits can be much greater than that in
a.c. circuits because of the release of magnetic stored energy. This is particularly true of
switching transients. Alternating currents are interrupted at current zero when the inductive
energy is zero. As described in Chapter 7, a current zero must be created in a direct current
circuit, handing the entrained inductive energy is a major challenge [15-17].
Fortunately, the MOV has proved equal to the task. Depending on the application, it may
require several parallel colums of ZnO disks and these must be carefully matched to assure
equitable distribution of the surge current and associated energy. The high value of α
accentuates this need. All this notwithstanding, many surge arresters of this kind are in service
and doing an excellent job in protecting HVDC terminals. A guide for converter station
applications for metal oxide arresters has been prepared by Ekstron and Anderson [18]

16.7 SURGE CAPACITORS AND SURGE REACTORS

Very early in this book it was pointed out that one cannot suddenly change the voltage on a
capacitor, because to do so would require the momentary flow of the infinite current. If a
capacitor is therefore placed in parallel with a piece of equipment or a component, it will
provide some measure of protection against voltage surges, for the surge can only impress a
high voltage on the equipment or component as quickly as it can charge the capacitor. The
effectiveness of the capacitor in holding down the voltage depends upon the energy conveyed
by the surge and the capacitance of the capacitor. Such capacitors are particularly effective
against voltage ¨spikes¨ that is, short duration voltage surges. The energy of these surges is
such that it can be absorbed by the capacitor with little change of voltage. An example will
make this clear.
Suppose a lightning surge rising at 200 kV/μs enters a primary distribution circuit, which has
an insulation flashover voltage of 180 kV. The surge will reach 180 kV in 0.9 μs and then cause
a flashover, limiting the peak voltage to this value. The remainder of the surge will be diverted
to ground, while the sharp spike travels on down the line, much as is shown in Fig. 16.19. If the
line surge impedance is 400 Ω, this voltage spike will be associated with a similar current surge
with a peak of 450 A, which rises from zero at 500A/μs. The energy in this surge is
Suppose the instantaneous voltage on the line prior to the appearance of the surge was 20 kV,
and that a 0.25 μF surge capacitor on the line was charged to that value. The surge capacitor
will have stored

1
C V 2=1.25 ×10−7 ×(2 ×10 4)2
2
¿ 50 J

A simple calculation indicates that the extra 24.3 J can be accommodated in the capacitor with
an increase in its voltage of only about 5kV.
High-energy surges cannot be handled by capacitors. The required energy storage capability
makes their physical size and cost prohibitive, as a calculation will show. Suppose the energy in
a surge on a 250-V system amounts to 5000 J, and it is desired to hold the voltage to 500 V,
how big a capacitor would be needed?

1 1
energy= C V 2 = C ( 5002−2502 ) =5000 J
2 2
whence

C=53,000 μF

Capacitors fin their place in the protective scheme as absorbers of sharp spikes of surges, and
as waveform modifiers to soften the front of surges. The combination of a capacitor and some
other protective device, such as an arrester, can be very useful, since together they
complement each other. Arresters have a turn-up at short times in their volt/time curve, as
discussed earlier in the chapter. This is of much less consequence if a capacitor is present,
because this reduces the steepness of surge wavefronts, thereby eliminating the short time
part of the volt/time curve from consideration. To illustrate this point, suppose that an arrester
sparkover characteristic is reasonably flat after 1 μs. If a capacitor is put in parallel with the
arrester, such that even with the maximum surge conceivable, having a vertical wavefront, it is
impossible to charge the capacitor, through the surge impedance of the connected line or bus,
in excess of this sparkover value, in less than 1 μs, them this combination will keep all surges
down to the flat sparkover level. For example, suppose such a combination is used to protect a
large 13.8-kV motor, which is a common and recommended practice, and suppose that the bus
impedance is 100 Ω. It might be assessed that the highest voltage surge that could appear on
this bus would be 120 kV; this figure would be determined from flashover characteristics of
bus supports and so on. The arrester might typically have a sparkover of 39 kV. The surge
current would be

V
I= =1200 A
Zn

I 1200
the rate of rise of capacitor voltage= = V /s
C C
If the capacitor already is at peak voltage (13.8/ √ 3 × √ 2−11 kV ) when the surge arrives, the
voltage would not climb more than 28 kV in the first microsecond, if

1200
×10−6=28000∨C=0.043 μF
C

Typical installations of this kind employ 0.25 μF surge capacitors.

This approach is particularly useful where semiconductor devices, such as silicon rectifier
installations, are concerned, because the semiconductor components have very little turnup in
their volt/time curve and therefore present a coordination problem. For reasons mentioned in
Chapter 7, it is sometimes necessary to put a resistor in series with the capacitor in some
applications in this area, to form and RC ¨snubbed circuit¨. Any such resistor reduces the
effectiveness of the capacitor in the context we have just discussed, since the surge current
diverted into the capacitor will at once produce the IR drop of the series resistor. The situation
is further aggravated if the resistor also possesses inductance. The purity of components for
these applications, be they resistors or capacitors, cannot be overstressed.
Capacitors designed specifically for surge duty have minimal inductance and resistance. The
so-called extended foil construction is used, whit multiple connections to the foils. Electrolytic
capacitors are not well suited for this task, but it is possible to capitalize on their cheapness
and high capacitance-to-volume ratio by placing a small, high-quality surge capacitor in parallel
with a bank of electrolytics. A useful scheme which allows the use of polarized electrolytic
capacitors on a three-phase a.c. circuit, and yet economizes in their use, is shown in Fig. 16.25.
If a surge appears between any pair of phases, regardless of its polarity, it will be
communicated to the electrolytics with a high resistance to bleed off any excess charge
acquired by them, which would otherwise remain trapped by the diodes.
Cable has a quite significant capacitance and can be used to advantage for surge protection.
It was shown in Section 9.3 that when a surge reaches a

transition point in a circuit where there is a change in surge impedance, the surge undergoes a
change by the initiation of reflected and refracted waves from the transition point. More
specifically, the refracted surge, which penetrates beyond the transition point, is related to the
incident wave by Eq. 9.3.8:

2 Z2
refracted wave=V 1= ( ) V
Z1 + Z 2 1

where V 1 is the incident surge or wave, and Z1 and Z2 are the surge impedances on either side
of transition point. It is apparent that if line 1 is an overhead transmission line ( Z1 typically 400
Ω) and line 2 is a cable ( Z2 typically 50 Ω), only a fraction of the incident surge will penetrate
the cable and travel to the terminal equipment. This presentation, however, is incomplete, for
the surge that enters the cable will be reflected at the terminal equipment, return to the
junction, and be reflected again. In fact it will travel up and down the cable many times, until it
is attenuated by losses. During this time the voltage will build up at the terminal equipment.
This will be demonstrated by an example.
Suppose that Z1 =400 Ω and Z2 =50Ω , and further suppose that the cable is 100 ft long and
terminates at a transformer whose surge impedance is so high compared with that of the
cable, or line, that it can be considered an open circuit. For the purpose of this example we
assume that the surge has a vertical front and a flat top, that is a square wave of amplitude V 1.
The wave that initially penetrates the cable will have an amplitude
2× 50
V 3= V =0.222 V 1
50+ 400 1
It will typically travel at 400ft/μs on the cable and will therefore reach the transformer
terminals in 0.25 μs. The reflected wave from this termination will be equal to the incident
wave and will add to it, thus the arrival of the wave at the transformer will be signaled by the
voltage there taking a jump of 0.444 V1.
Then 0.25 μs later the reflected wave will have returned to the junction point, where new
reflected and refracted waves will be initiated. The reflected wave will be

Z 1−Z 2
( )
V −0.173 V 1
Z1 + Z 2 3

(This is derived from Eq. 9.3.7; Z1 and Z2 are interchanged because the wave is traveling on
the cable towards the line.) This wave is also reflected from the transformer with sign
unchanged and returns to the transition point after a round trip of 0.5 μs to initiate another
pair of waves. Sequential waves will be

Z 1−Z 2 2 Z 1−Z 2 3
( Z1 + Z 2) (
V 3,
Z 1 + Z2 )V 3 , etc ,

Thus the voltage at the transformer terminal will rise in a staircase fashion, the steps being
each 0.50 μs long, but becoming progressively smaller. Letting

Z 1−Z 2
=ɑ
Z 1 + Z2

the transformer terminal voltage will be given by

2 V 3 (1+ɑ + ɑ 2 + ɑ3 +…+ ɑn )

It will be found that this series converges on a value V 1/V3, that is, the terminal voltage is
asymptotic to 2V1. This is shown in Fig. 16.26ɑ. This is the value it would have attained
instantaneously, had the cable not been there, since the termination is effectively an open
circuit. The value of the cable, therefore, is not in reducing the terminal voltage but in
modifying its rate of rise as would a capacitor. Indeed, much the same result would be
achieved with a capacitor having the same capacitance as the cable. In that case the voltage at
the transformer would rise smoothly as a (1-exponential) curve, as shown in Fig. 9.13, rather
than with the steps of Fig. 16.26ɑ.
 An additional advantage will accrue for cable or capacitor if the surge has a finite front and
tail, since the voltage will not have reached twice the peak

value of incident wave before the surge declines. This is illustrated for the cable in fig, 16.26b.
Reactors are sometimes used to protect against voltage surges, particularly within pieces of
conversion equipment, in a manner that complements the use of surge capacitors. The
capacitor parallels the equipment to be protected; the reactor is placed in series with the
equipment. When a surge reaches such a series combination, the initial tendency is for the
surge voltage to appear principally across the reactor, since this offers the highest impedance
to the rapidly changing surge current. Put another way, the surge energy is initially absorbed in
the magnetic field of current in the reactor. Like capacitors they can take care of low-energy
surges, or act temporarily to protect a device, until such time as a gap can flashover or an
arrester operate.
When voltage is impressed across a reactor the following equation applies:

−d Φ
V=
dt

where Φ is the magnetic flux linkage of the reactor. Rearranging and integrating,

∫ V dt=Φ 2−Φ 1(16.7 .1)

The quantity on the left has the dimensions of volt seconds—how many volts the reactor can
support, for how many seconds. To support the voltage the flux linkage must be changing and
Eq. 16.7.1 indicates the change in flux linkage required for any desired number of volt seconds.
This is particularly useful in determining the core cross section for reactors with a magnetic
core. For example, suppose we wish to support a voltage of 20 kV for 5 μs with a core material
that saturates at 1.4 W/m2 (or 14000 G). Let the cross section of the core be A m 2 and let there
be n turns on the reactor. Assuming all the flux links with all of the turns and that the core is
initially demagnetized, the change in flux linkage to the point of saturation will be

Φ 2−Φ1=Φ 2=1.4 × An
the volt seconds called for ¿ 2 ×104 ×5 × 106−0.1, whence

0.1
An= =0.071 m 2 turns
1.4

Thus a reactor with 20 turns would require a core cross section of 3.55 X 10 -3 m2 or 35.5 cm2.
There are obviously combinations of A and n that would satisfy these particular
requirements. The choice in any practical circumstances will be principally an economic one
determined by the cost of the core versus the cost of the coil. The design is constrained by two
other relationships,

∫ H dl=I
and the B/H curve of the core material.
We have been describing here a saturable reactor, one which presents a high inductance
until is saturates, but thereafter presents the inductance of a comparable air cored device.
They can absorb voltage, and therefore delay its appearance elsewhere, until saturation is
reached.
The discussion so far has been idealized in one or two important ways that we should be
aware of. It was pointed out in Section 8.5 that the penetration of flux into a conducting solid
is opposed by eddy currents. The net flux is always the consequence of the applied ampere
turns and those produced by these eddy currents. To minimize the effects of the eddy currents
the core must be finely laminated, especially where a rapid change of flux is called for as with
steep-fronted surges. If this is not done, it will not be possible to take advantage of the
presence of the core. In extreme cases, powder cores or ferrites must be used.
Another important fact is that the core and winding structure of the reactor or choke
represents a complicated system of capacitances, which effectively shunt the inductance. In
the initial stages when the surge first reaches the equipment, these, in combination with
similar capacitances in the apparatus being protected, are the determining parameters in the
voltage distribution. Conditions are quite analogous to those discussed in Section 11.1.4. This
points up the advantages to be gained by a combination of series reactor and shunt capacitor
to form a surge filter in front of a piece of vulnerable equipment. Such combinations are used
very effectively, especially in control circuits.

16.8 SURGE PROTECTION OF ROTATING MACHINES

16.8.1 Protection of Generators

A recent survey [19] suggest that North American utility companies are somewhat ambivalent
on the subject of generator protection. Some use surge arresters, some use surge capacitors,
some use both and some use neither, choosing to rely on the arresters on the high voltage side
of the generator step up transformer (GSUT). There are a number of possible reasons for this.
First, generators are somewhat shielded from the transients of power systems by the
intervening presence of the GSUT. Also, the isolated phase nature of the
generator/transformer bus in most installations, makes bus faults unlikely. This bus is another
reason for less than enthusiastic embracing of surge protective devices; it makes them difficult
to locate and service. Moreover, there has been a belief in some quarters, at least in the past,
that the bus system and the generator are more reliable than the protecting devices. The
failure of an arrester or surge capacitor in a generator bus system creates a major perturbation
on the operation of the plant.
 At the same time it must be recognized that surges can transfer through the GSUT by
magnetic and electric coupling (see Section 11.1.5), and reach the generator. The increasing
use of generator breakers between the generator and the GSUT is introducing another
potential source of transients. The introduction of ZnO arresters has made the coordination
easier and improved reliability. As a consequence of this somewhat fluid situation and the
failure of a generator, EPRI instigated a study on generator protection [20]; the survey
mentioned above was a part of it.
The EPRI investigation [20.21] examined a large number of contingencies that could
conceivably stress generator insulation. It was found that several could generate substantial
overvoltages, but that their probability of occurrence was low. The voltage division that the
series connection of the generator and GSUT effects for both electric and magnetically induced
surges, should be considered. It will be most onerous in those installations where several
generators are feeding a single transformer, but only one machine is connected. Surge
capacitors on the generator bus can help as far as electrostatically coupled surges are
concerned. These should be placed on the generator side of the breaker if a breaker is present
so as to additionally soften transients caused by the breaker itself. There seems no other good
reason for using surge capacitors. Sloping of surges is less important since many generators
have single turn windings.
Arresters on the generator bus can clip impulses and switching surges alike. Attention should
be paid to the matter discussed in Section 16.5.9. Also, when there is any danger of generator
self-excitation of ferroresonance causing abnormally high bus voltage, the capability of the
arrester to handle the energy involver should be checked.

16.8.2 Protection of Motors

Motors are exposed to transients, both switching transients and sometimes lightning
transients. As discussed in Section 11.3, they have multiple turns per phase and are therefore
vulnerable to interturn stress when impacted by fast rising surges [22]. For these reasons it is
desirable to protect them, or at least the bigger machines and the more critical ones- The cost
of the protection is small compared with the cost of repairing or replacing a damaged motor. A
combination of surge capacitor and surge arrester is ideal, the first being used to slope the
front of fast surges, the second to limit surge magnitude. Harder [23] gives some guidance on
the selecting of arresters. The importance of closely connecting the protective device with the
motor being protected is stressed by McLaren and Abdel-Rahman [24].

16.9 TRANSIENT VOLTAGES AND GROUNDING PRACTICES

In this section we consider how good grounding practices contribute to more reliable and safer
utility and industrial power systems. We are concerned with minimizing the damage to
equipment and with false operations of equipment, under transient conditions, and also with
the avoidance of shock hazard to personnel. Space precludes a very detailed discussion of
grounding practices, although references presented [25-27] indicate where such data can be
found. Instead the focus will be on the underlying principles.
Under steady-state, quiescent operating conditions in a power system, the integrity of
equipment is preserved by adequate insulation, and safety of personnel is secured by kepping
sufficient separation between people and high-voltage conductors. Preferably a grounded
barrier in the form of a metal enclosure, wire fence, or the like, is placed between the circuit
and any personnel. ¨Grounded¨ barrier is unambiguous; it refers to a physical equipotential
surface, at the same potential as local ground and any other grounded objects around.
Under abnormal conditions, as when a fault occurs, this situation may no longer prevail;
¨ground¨ becomes a relative thing. To take a simple illustration, consider Fig. 16.27. Suppose a
short circuit occurs across the conductors from the transformer, perhaps by a foreign body
falling across them. This could be simulated by closing the switch in Fig. 16.27. We know that
whatever voltage existed across the switch before it was closed will be annihilated by the
closing. But since the flux linkage with a circuit cannot change instantaneously, this voltage will
reappear distributed around the circuit. Because of its capacitance, the transformer will tend
to hold its voltage momentarily, so that considerable differences of potential will suddenly
appear across the ends of short lengths of bus, as between point P and local ground. This
becomes particularly disturbing when one of these conductors is part of the ¨ground¨
structure, as is often the case.
Another example is shown in Fig. 16.28. This relates to events occurring when a structure
such as a utility pole, becomes part of a circuit by a flashover from the line or perhaps by being
itself struck by lightning. Some consideration has already been given to this in Section 14.3. It is
apparent
that a high surge current flowing down the structure will develop a considerable resistive volt
drop, which an individual in the position shown will pick up. In addition, and this may be more
important the loop formed by the person and that part of the structure and ground that be
spans, is mutually coupled with the circuit in which the surge current is flowing. Thus, as this
current is building up, a considerable voltage, M(dl/dt), can be induced in the loop. Indeed, the
man may be involved without actually touching the structure, by simply standing beside it, for
sufficient voltage could be induced in the incomplete loop to cause a flashover from the man
to the structure. Again, the surge current will presumably flow into the ground and spread
radially away from the pole. If high ground resistance should create a high radial voltage
gradient, a person would be shocked by simply spanning a stride on such a gradient with his
two feet. The unfortunate propensity of cattle for standing under trees during thunderstorms
has had some sad consequences. But it has been observed on some occasions that those
standing radially have been killed, while those standing circumferentially, and therefore having
their feet approximately on an equipotential, have survived, when the tree was struck by
lightning.
The preceding examples illustrate that under abnormal conditions, when transient currents,
sometimes of high magnitude, take unusual paths, high transient voltages can appear between
points that are normally at or close to the same potential. This can endanger personnel and, by
causing further flashovers, can damage equipment, thereby adding considerably to the
problem of the initial disturbance. Some bouncing around of potentials is unavoidable, even
during normal switching operations on a system. But the adverse consequences of this and of
abnormal voltage disturbances can be greatly reduced by correct design.
The object of this design is to reduce transient potential gradients as far as possible. It is of
little consequence if even a safety guardrail around an equipment flies up to a high potential
with respect to some remote ground, provided all other adjacent structures and circuits follow
suit. This is achieved in a substation, vault, of where have you, by having a number of sturdy
cooper strips run around the equipment and exposed metal structural members of the room,
to which these items are firmly bonded at frequent intervals. The periphery of the building or
room should be circled by such a ground conductor, which would attach to structural steel
work, water pipes, and the like. This would be connected within the room to cubicles, machine
frames, transformer tanks, bus ways, and so on. This copper harness should itself be attached
to the ground mat of the building. The ground mat may comprise the steel reinforcing in the
concrete of the floor, or a mesh of carefully bonded copper cables specifically put down for the
purpose. Such an arrangement provides a low impedance path for ground fault currents and
therefore minimizes the potential difference that can exist between any pair of points that a
man could reach. Analysis of grounding schemes of this kind are described in references [25]
and [26].
In an industrial plant, ground fault currents are usually confined to such networks and to
cable sheaths. On utility systems true ground is often involved. As yet we have said nothing
about connections from the ground mat to true ground, although such connections are clearly
necessary. In a switchyard, the mesh just described will be buried in the ground. At its
periphery it is frequently connected to metal rods driven into the ground, typically 5-10 ft long.
The effectiveness of these rods is critically dependent on the resistivity of the soil. This varies
widely with location and weather conditions, from a few thousand to hundreds of thousands
of ohm centimeters. Some typical values are given in Table 14.4. The presence of moisture
greatly improves the conductivity, especially if certain minerals are also present. Dry sandy
soils offer a high resistance, which is acute situations must be modified by adding salt. This
eventually leaches away and must be replaced. It is also found that ground resistivity increases
sharply when the ground freezes. The poorer and soil conditions, the closer the intervals at
which ground rods must be placed. When a fault or surge current is dissipated to ground
through a rod, the voltage gradients is highest adjacent to the rod, where the current density is
highest. The closer together the rods are placed, the less will the local potential fluctuate. The
overall outcome of any ground mat design should be as low a resistance path for ground fault
currents as is economically possible. Suggested maximum values are 1 Ω for a large station and
5 Ω for a small substation, where the fault current is likely to be lower. But it will be clear that
these values can represent a significant part of the ground fault impedance, and therefore
impress a significant fraction of the system voltage between station ground and true ground.
In a survey of substation ground resistances [27] of a large number of U.S. utilities, it was
found that only 80% of those reporting did in fact have ground resistance less than 5 Ω. This is
indicated in Fig. 16.29, taken from the survey report.
Sometimes when we solve a problem, we do so only to have it reappear somewhere else, or
in some other form. This is true is designing grounding systems. It is possible, for example, by
the expedients described above, to have a site or building made secure from high potential
gradients under the most adverse conditions, at the expense of introducing very high gradients
in adjacent areas. The potential picture is of a flat-topped plateau on which the substation sits,
but beyond its perimeter the gradient is very steep. Under

these conditions, personnel within the station will be quite safe, but those close by outside
would be imperiled. The solution here is to extend the grounding system beyond the
perimeter of the station, tapering it to give an acceptable gradient.
Another way in which the benefits of a good grounding system can be vitiated is by bringing
into the area, or taking from it, insulated ground. This can be done inadvertently; for example,
a water pipe coming into a building, so that when a disturbance causes building ground
potential to change suddenly, a considerable voltage can appear between the water pipe and
adjacent objects. This problem can be remedied by ensuring that the pipe is connected to the
building ground mat, but it must then be recognized that the pipe has become part of that
ground mat system.
 In like manner, dangerous transient voltages can be taken out of an area by insulated
conductors connected to ground in the area, which then pass to adjacent areas. The hazard is
then in the adjacent areas. The most frequent offenders here are control and communication
circuits. Sometimes it is not possible to connect such circuits directly to local grounds, in which
case capacitors can be used to tie down the ends of the circuits against transient disturbances.
However, this too may be unacceptable because of circulating currents that are introduced. It
may then be necessary to install a fairly elaborate shielding scheme. These problems are
discussed in Chapters 8 and 18.

16.10 PROTECTION OF CONTROL CIRCUITS

Following the old adage that prevention is better than cure, we consider first some principles
for keeping transients out of control circuits. As mentioned, this is of vital importance for the
correct operation of such circuits, aside from the need for their protection. Kotheimer [28,29]
has addressed himself to this problem, particularly as it affects relay equipment in high voltage
switchyards. These references are drawn on liberally in the following discussion.
The first point to recognize is that two general classes of electrical conductor systems coexist
in every switching station:

1. The HV power circuits, including all of the HV buses and apparatus, primary circuits of
instrument transformers, and all devices operating at high potential such as line traps,
arresters, and gaps. Also included are the station ground grid, all apparatus grounds,
and all bushing capacitances and stray capacitances to ground.
2. Control circuits, including all potential and current transformer secondary circuits, all
battery, d.c. control, a.c. auxiliary power, protective relaying, supervisory, alarm, and
communication circuits.

During HV switching operations, high-magnitude transient currents circulate between the

buses and the ground system through the various bushing capacitances and stray capacitances
in the station. These currents are in close proximity to the various control circuits at the base
of most HV apparatus and therefore favorably disposed to induce transients therein. The
object of good design is to reduce the coupling between these circuits by layout and shielding.
Coupling can be electrostatic, electromagnetic, or conductive. The first two of these are
discussed at some length in Sections 8.2 and 8.3. Conductive coupling exists when two circuits
share a common conductor for part of their length, In the circuits being described, this is most
likely to exist through sharing of common ground connections. The problem is best illustrated
by specific examples.
One of the most prevalent sources of pickup is the switching of a section of EHV bus by an air
break disconnect. Figure 16.30ɑ shows a typical installation, with the area of interest circled.
The physical arrangement of the apparatus can be understood from Fig. 16.30b. Consider the
case when all the disconnect switches except C and D are open, the breakers are likewise
open, and the transformer is being excited from the 161 kV side. Disconnect switch C is
opened and a restriking are occurs between its arms, making and breaking the charging
current flowing into the series combination of the circuit breaker voltage dividing capacitance
and the CT bushing capacitance. Reignitions and restrikes are to be expected until the
disconnect reaches a considerable gap, which takes time to achieve since disconnects are slow-
moving devices.
We saw in Section 5.4 that it is readily possible to have twice peak system voltage across the
switch contacts when it re-ignites. It is possible to compute the inductance of the local
discharge loop from its dimension and, knowing the capacitance involved, calculate the
magnitude and frequency of the resulting current. An equivalent circuit which neglects all
losses is shown in Fig. 16.30c. Here LT, the inductance of the loop between the CTs, may be
typically 120 μH. If the capacitance of each CT is 800 pF and the breaker is 125 pF,

C A CB
CN= =108 pF
C A +C B

The effective capacitance involved in the circuit oscillation is

CN CE
C T= =95 pF
C N +C E

The surge impedance of the circuit will be

1/ 2 1 /2
LT 125 ×10−6
( ) (
CT
=
95 ×10
−12 ) =1140Ω
With an instantaneous voltage of 700 kV across the switch when reignition occurs, the current
would be 620 A. The frequency is

1
f 0= =1.34 MHz
2 π ( LT C T )1/ 2

These high-frequency currents flow not only in the buswork, but also in the CTs and their
ground connections, and in the ground grid. This gives rise to capacitive and magnetic coupling
to secondary and control wiring due to their close proximity to the ground conductors.
The obtain some idea of the magnitude of the induced voltages, consider Fig. 16.31ɑ, which
shows the high-frequency equivalent circuit of a high voltage CT. The capacitor C 1 represents
the bushing capacitance (790 pF), C2 represents the capacitance of the primary to the internal
shield (about

50 pF), and C1 is the distributed capacitance of the secondary winding to the shield, core and
case (about 500 pF). The magnetic flux of the high frequency current in the ground conductor
and ground grid links the loop formed by these conductors and the control circuit leads. The
loop is completed by the capacitance C3. By estimating some dimensions of this loop, it is
possible to make a rough calculation of the voltage induced, Equation 8.3.10 is helpful in this
regard; the flux linking with the loop is given by

μ 0 I d−r
Φ= ln W /m(8.3 .10)
2π r

(this assumes that the ground conductor is part of a long line), r is the radius of the ground
conductor (1 cm, say), and d is the width of the loop of the control circuit (1 m, say). The
induced voltage will be
d Φ μ0 d−r
V= =
dt 2 π
ln( r )( dIdt )V / m
which has a maximum value

4 π ×10−7 d−r
V max =
2π (ln
r )
ω I max V /m(16.10 .1)

Substituting the dimensions and setting ω=2 π × 1.34 ×106 and I max=620 A .

V max =2 ×10−7 ( ln 100 ) 2 π ×1.34 × 106 × 620=4800 V

If the ground conductor is 2m long, the peak induced voltage would be almost 10 kV.
This voltage appears on both secondary conductors with respect to ground, that ism as a
common mode voltage. The control cable may be taken several hundred feet to the control
house, forming a short transmission line with the ground amt. The induced voltage will be
impressed on this ¨line¨ and travel down it to the control house.
A similar situation to that described can exist in coupling capacitor potential devices as
indicated in Fig. 16.31b. The physical layout is clearer in Fig. 16.32ɑ. Two schemes for
drastically reducing the pickup are shown in Figs. 16.32b and 16.32c. In the first example the
loop is eliminated; transient currents are conducted by a tubular ground conductor, while
control leads pass down the field-free regions inside. In the second approach the loop is short
circuited by using a shielding secondary cable with the shield grounded at the base of the
potential device and at the receiving end. The shield must be a good conductor.
Power systems abound with relay, control, and monitoring circuits of many kinds. To avoid
damage to these circuits or misoperation from voltage surges, it is essential that they be
protected. The need for protection has become significantly more acute in recent years by the
introduction of printed circuits. Such circuits operate at a low voltage and energy level yet are
more sensitive to overvoltage. Most frequently the surge voltage incursions are induced from
neighboring power circuits, in the manner just described. The first step, therefore, is to
minimize the coupling with these circuits by effective shielding through the various methods
mentioned above and described in Chapter 8. Even when this has been done, there sometimes
remains a residue of voltage disturbances, which can still create serious problems.
The approach to protection for control circuits is basically the same as that for the power
circuits themselves: we must consider coordination, impedance division, and energy. Except
for those circuits that are exposed to lightning, the voltage and energy are much smaller than
the voltage we associate with power system surges. But, as pointed out previously, the ability
of the circuit components to cope with voltage and energy is also drastically reduced.
MOVs are made in minute sizes for applications in this area. Their high value of α allows
them to clip voltage transients and absorb their energy. Surge capacitors can be used within
their limitations, especially in conjunction with other devices. A surge capacitor, in this context,
may be no bigger than a dime, since the surge energy level in many circuits is low. The small
size of the components has the advantage of minimizing stray inductance. The capacitor, for
example, can be tightly connected across the terminals of a relay card.

16.11 SURGE PROTECTION SCHEME FOR AN INDUSTRIAL DRIVE

SYSTEM

To pull together the many aspects of circuit protection that have been discussed in this
chapter, and example will now be considered of an integrated protection scheme in which
most of these aspects have to be considered. Protective devices described earlier are shown in
a typical environment. The example selected is an industrial drive that might typically be found
in a steel mill. It is shown schematically in Fig. 16.33 and in more detail in Fig. 16.34.
The motive power derives from a direct current motor whose speed is controlled by
regulating the voltage of the bus to which it is connected. This regulation is achieved by a
bridge rectifier. The particular drive shown is made reversible by having two sets of oppositely
poled rectifiers, which allow the polarity of the bus to be reversed. It also permits dynamic
breaking by regeneration to assure rapid reversal of the motor. The source of power is an a.c.
line which delivers energy to the rectifier through a stepdown transformer. Note that where
one thyristor is shown it may represent several in series/parallel combination, depending upon
the size and voltage of the motor
Starting at the line, it is seen that arresters are located line-to-ground at the high voltage
terminals of the transformer. These would be either station or intermediate type, with a
voltage rating appropriate to the rating of the transformer (15 kV class in this case). They
would limit voltages from lightning or switching surges to their sparkover voltage. In most
installations the transformers are located outside the building close to the wall. This outside
area would usually have additional lightning protection in the form of overhead structural
steelwork and/or ground wires, with lightning rods on the building above and strategically
placed around the switchyard if that is extensive.
It was pointed out in Section 11.1.5 that short but potentially very high voltage transients can
be electrostatically coupled through a transformer from one winding to another. These are
effectively suppressed by putting capacitors on the low voltage side of the transformer
connected from each phase to ground. These are shown in Fig. 16.33 as C 1. Where the
transformation ratio is very high this might be supplemented by a grounded metal shield
between the windings. Longer duration surges coupled electromagnetically through the
transformer, as might appear from an induced lightning surge which just failed to spark over
the arrester, are absorbed by the
large capacitor C2, which was described in Section 16.7. As an added safeguard, a surge
suppressor (SS) may be placed in parallel.
The remainder of the protective equipment is applied primarily to protect the thyristors from
surges they themselves generate in the course of their normal operation. These have been
discussed in Section 7.6 and will there be reviewed here only briefly. We note in Fig. 16.34
inductors marked L, in series with the controlled rectifiers. These assure equitable division of
current between the parallel strings of thyristors, if such are used, but they perform an
important function during commutation in the manner described in Section 16.7. When a
string such as 6 is turned on, the voltage formerly across it is not suddenly transferred to 3.
Instead it is absorbed momentarily by the Ls and then transferred in an orderly manner to C 4
with R4 acting as a damper. At the time 6 is gated, 4 will already be conducting (see Fig. 7.16b),
thus a momentary short is applied between phases R and Y. The protection capacitors C 1 will
therefore discharge into each other, and once again the inductors, L, will modify the current.
This transient will be damped by the presence of R 3 and C3. At the end of commutation the
transients evoked by the turnoff of the outgoing phase will be similarly modified.
 An installation of this kind would be grounded according to practices described in Section
16.8. Outside the building the transformer tank, arrester, and circuit breaker framework would
be connected by short, heavy copper braids to the ground mat. Inside the building a harness of
copper strip would join the rectifier and control cubicles, the motor frame, and any exposed
metallic structure of the building. This harness also would be connected at seveal points to the
ground mat.