DISTRIBUTION SYSTEM PROTECTION — 2

PROTECTION COORDINATION - PART 1

Overcurrent protection coordination is the process of selecting, arranging and setting the
protective equipment on the system so that it provides the maximum amount of overcurrent
protection while at the same time insures that a minimum number of customers are affected each
time a fault is cleared. It is important to recognize that proper protection coordination requires a
compromise between these two opposite goals.
An overcurrent protection scheme must have three attributes in order to operate effectively.
These are:
1. Speed
2. Sensitivity
3. Selectivity
Speed
Speed is the ability of the protective equipment to interrupt the fault current in a
minimum amount off time so as to insure the least amount of equipment damage. When a fault
occurs, all of the components carrying the short circuit current are subjected to thermal and
mechanical stresses which are proportional to both the square of the fault current and the
duration of the fault. Thus, in order to minimize the damage caused by fault currents we must
eliminate them as soon as possible.

Sensitivity
Sensitivity is the ability of a protective scheme to “see” all faults within its zone of protection.
For example, the phase overcurrent relays installed for feeder protection must be set with a high
enough pickup so that they do not misoperate for a normal overload or an unbalanced load.
Unfortunately, these relatively high settings may render the relays insensitive to many low level
faults. Consequently, they may not detect faults that occur at the far end of the feeder (where the
“available” fault current is low). So, in order to increase the “sensitivity” of the feeder protection
we could install a recloser at some distance along the feeder. Because the recloser normally
carries less load current than the feeder breaker we can set it to pick-up at a lower current as
well.
In Figure A-1 we compare the feeder relay pick-up (Relay P.U.) and the recloser pick-up
(Recloser P.U). Beyond a certain point, the amount of fault current, which will flow to the fault,
is below the relay pick-up setting. Consequently, the relay will not operate for faults beyond this
point: i.e. it is not “sensitive” enough to see the fault. However, the recloser will operate to clear
faults outside the relay’s “reach”.

By installing the recloser we make the entire feeder protective system more sensitive and better
able to identify and clear all faults on the feeder.

Selectivity
Selectivity is the ability of the protective scheme to determine which devices should operate
immediately and which should delay operation in order to allow other protection equipment to
operate. By making the protective scheme “selective”, only the faulted portion of the system is
isolated.

For example, a fault, which occurs in a distribution transformer should be cleared by the
transformer fuse, not the lateral fuse. In practice, the lateral fuse is selected to have a fuse
characteristic, which allows the transformer fuse to operate first. Thus, the lateral circuit remains
energized to supply all unfaulted distribution transformers along its route.
Figure A-2 illustrates the principle of selectivity. The 170 amp fault at the transformer will
cause the 20K transformer fuse to operate much faster than the 65K lateral fuse. If, for any
reason, the transformer fuse does not operate, the lateral fuse will operate eventually (after a
second or two). This disrupts the selectivity of the protective scheme but provides necessary
“back up” protection.

RELAY-FUSE COORDINATION
In the following example of protection coordination we will explain bow the operation of a
feeder breaker relay is coordinated with a lateral fuse.
Coordinating protective devices requires that we make a comparison of the time-current curves
for the various devices. These curves are usually drawn on the same set of axes, so we can
readily compare which device operates first for a certain value of fault current.
Fuse Time-Current Curves

In an earlier lesson * we described the operation of a fuse with only one time-current curve -the
fuse “total clearing” curve; that is, the total time it takes for the fuse to interrupt current. But the
fuse total clearing curve does not adequately describe the operation of the fuse for coordination
purposes. Actually, we must consider three different curves. These are:
1. The fuse “total clearing” curve
2. The fuse “minimum melt’ curve, and
3. The fuse “damage” curve

These three curves are plotted in Figure A-3

The fuse “total clearing” curve indicates the time it takes for the fuse to completely interrupt a
fault level current. We can think of this curve as being the sum of two periods of time: namely,
the amount of time for the fuse element to melt plus the amount of time it then takes to clear the
arc created after the fuse element melts.

The fuse “minimum melt” curve describes the amount of time necessary for the fuse element to
melt, for a given level of current.

The fuse “damage” curve describes how long it will take to cause damage to the fuse without
completely melting the fusible element. A fuse can be damaged in a variety of ways. Heat, oc-
casional overloads, or even a very short duration lightning surge may cause the fusible element
to partially melt, resulting in a derated fuse. This means that, for subsequent operation, the fuse
may blow at current levels considerably lower than expected. So, to provide a margin of safety
when coordinating fuses with other protective devices, it is assumed that, for any current level,
the fuse will be damaged in about 75% of the time it takes to melt the fusible element. Thus, the
“damage” curve is drawn by plotting 75% of the minimum melt time at each current level.
In the example of Figure A-3, for a fault current of 1000 amps, the fuse will be “damaged” in
about 0.075 second, it will melt in about 0.1 second, and it will clear in about 0.2 second.

Fault Selective Feeder Relaying

In many applications, the method of “Fault Selective” feeder relaying is chosen for feeder
protection. This type of relaying allows the feeder breaker to clear temporary faults on the
laterals before the fuse can operate or even be damaged. In the event of a permanent fault on a
lateral circuit, the fuse will operate and isolate only the faulty lateral. The feeder breaker will
successfully reclose to allow service to continue to all the unfaulted portions of feeder.

To see how all these operations are coordinated, refer to Figure A-4. Here, the lateral fuse total
clearing curve and damage curve are plotted, along with the overcurrent relay’s time delay and
instantaneous curves. Note that, when discussing the operating time of the feeder breaker we
must consider not only the relay operating time, but also the operating time of the breaker itself.
In this example, a breaker operating time of 5 cycles (about 80 milliseconds or 0.08 second) is
assumed, and is also plotted.

The relay’s instantaneous element characteristic is drawn (with the breaker operating time
included); its pickup current is set at 600 amps. The time delay element is of the extremely
inverse type and is set at a pickup of 480 amps, and a 1.5 time dial.

Let us assume that a 1000 amp fault occurs on the lateral, as shown in the one line diagram of
Figure A-4. Both the time delay and instantaneous relay elements will sense the fault current and
begin to operate. You can see that the instantaneous unit will operate first, tripping the breaker in
about 0.15 second. Note that the instantaneous element of the relay operates quickly enough so
that the breaker trips before the fuse has time to be damaged.

Note also that there is a reclosing relay associated with the feeder breaker relay. The reclosing
relay, having monitored the breaker’s operation, now recloses the breaker, and also disables the
instantaneous element of the overcurrent relay.
If the fault was temporary the overcurrent relay may not see a fault after breaker reclosure, so the
breaker will remain closed. Service has been quickly restored; all customers experienced only a
momentary outage and the lateral fuse has not been damaged. The reclosing relay will eventually
time out and restore the instantaneous element to normal operation. Most faults on overhead
feeders are temporary in nature, so this operation is the usual case.

However, it the fault is permanent, the breaker will close back into the 1000 amp fault. Since the
reclosing relay has blocked operation of the instantaneous element, the breaker will not trip again
on the relay’s instantaneous element. But, the lateral fuse, having been selected to “nest”
between the instantaneous and time delay relay elements, will operate (according to its total
clearing curve) in about 0.8 second. This is before the time delay element has time to operate.
Therefore, selectivity is achieved: only the lateral circuit is interrupted.

If the fuse fails to operate, then the time delay element of the relay will clear the 1,000 fault in
about 2 seconds. The reclosinq relay will reclose the breaker a preset number of times. It would
then lock out until it is manually reset. Of course, if the relay operates as backup for the fuse, the
entire feeder will be out of service until the permanent fault can be located and repaired.

“Limits” of Coordination
In this example of fault-selective feeder relaying the fuse and relay are expected to operate as we
have described for fault currents above 600 amps and below 1780 amps (i.e. Note that 1780
amps is the current at which the relay instantaneous curve and the fuse damage curve intersect).
Outside these limits the desired coordination may not be achieved. For example a fault current
above 1780 amps will damage or blow the fuse before the instantaneous element can clear the
fault. Likewise, below 600 amps the instantaneous relay element will not even pickup, and the
fuse may operate. In either case, if the fault is temporary, operation of the fuse would cause a
needless interruption of service to customers fed by the lateral.

The 600 amps and 1780 amps are called the “limits of coordination”, since we can only expect
proper selectivity if the fault current is between these limits. Therefore, we would use the fuse
size and relay settings shown in Figure A-4 only if the maximum anticipated fault current on the
feeder is below 1780 amps.
Of course, when coordinating protective devices, it IS important for the system designer to
consider the maximum anticipated fault current when the system in reconfigured - (i.e. when
loads are switched from one feeder to another).

RECLOSER-FUSE COORDINATION
Figure A-5 shows the time-current curves for coordinating an automatic circuit recloser with a
lateral fuse. The coordination is similar to that of a breaker relay and a lateral fuse (see Figure
A-4). Both the fuse damage and total clearing curves are drawn as well as the recloser “fast”
(curve A) and “slow” (curve B) trip characteristics.

The recloser may be programmed for one or two “fast” (“instantaneous”) trips and three slow
(“time delay”) operations. Like the breaker relay instantaneous element, the recloser fast trip is
intended to protect the fuse from being damaged or operating during a temporary fault (i.e. the
recloser should initially trip first, then reclose, allowing service to continue to all customers.) The
fast trip is blocked after one (or possibly two) operations ; thus, if the fault is permanent, the fuse
will be allowed to operate, before the recloser can trip again on its slow curve.

Again, coordination is not always ideal; our “limits of coordination” are shown by points “a” and
“b”. For fault currents above the level shown by “b” the fuse will probably blow before the
recloser fast trip can operate. This would needlessly interrupt customers if the fault was
temporary. For fault current below the level shown by “a”, the recloser would trip out on its slow
curve and eventually lock out before the fuse can operate. This would, of course needlessly
interrupt all customers beyond the recloser location.

In practice, we may try to adjust the slow trip characteristic to be even slower so that, for low-
level faults, the fuse has time operate before the recloser “slow” trip can operate.
DISTRIBUTION SYSTEM PROTECTION — 2
RELAY - RECLOSER COORDINATION

Let us now consider how to coordinate a feeder breaker phase overcurrent relay with a recloser
out on the feeder.

As you know from an earlier lesson, a recloser is an automatic fault-interrupting device whose
operation is similar to that of a circuit breaker. However, the recloser cannot interrupt as much
fault current nor carry as much load current as a circuit breaker.

If an overcurrent relay in the substation is not sensitive enough (i.e. cannot be set low enough) to
protect the entire feeder, a recloser is normally installed. In most cases, reclosers are installed
toward the end of a long overhead feeder to provide fault protection for, perhaps, the last 20% of
the feeder.

“Available” Fault Current

During normal conditions, the amount of current flow on the feeder depends only on the
connected load. But, during a fault this is not the case. The amount of fault current, which flows
to the fault location, is independent of the connected load. Instead, fault current depends on other
factors, such as the impedance of the circuit to the fault location and the capacity of the power
“source”. Fault current also depends upon the impedance at the point of fault.

In the case of primary feeder faults the power “source” is the distribution substation transformer.
The higher the MVA capacity of the transformer, the greater the current, which will flow to any
fault location.

The impedance to the fault location consists primarily of the impedance of the line conductors.
Thus, the impedance of the circuit to the fault location varies in proportion to conductor length.
The farther the fault is out on the feeder, the greater the impedance to the fault and consequently
the lower the fault current. We refer to the maximum amount of fault current, which would flow
to any particular location as the “available” fault current at that location.
A graph of “available” fan it current versus the distance along the feeder is illustrated in Figure
B-1. Here, we have assumed a particular MVA “source” at the substation. We have also assumed
that the impedance at the fault location is low. At the source, the available fault current is
maximum - 3500 amps. However, the further we move from the substation the less fault current
is “available” at the fault location (since the line Impedance is increasing). Thus, a fault-
interrupting device with a low interrupting capacity might be capable of clearing a fault if
installed at location “B”, but would be inadequate if installed at location “A.”, where the
“available” fault current is much higher.

Assuming the pickup settings shown in Figure B-1, you can see that the C/B (i.e. “circuit
breaker”) relay will not operate for any fault beyond point “D”. Beyond this point, the available
fault current is less than 600 amps, and hence, neither the instantaneous nor the time delay relay
element will “see” the fault. The C/B relay is thus not “sensitive” enough to detect faults beyond
point “D”.

In this case a recloser could be installed to protect for faults beyond location “D”. However we
would not install the recloser precisely at location “D”, since there could conceivably be a fault
location, which would be just upstream from the recloser, and have a current too low to be
picked up by the relay. So a margin of safety is provided by locating the recloser within the
“reach” of the relay - i.e. perhaps at location “C”. This ensures that all feeder faults will be
“seen” by either the relay, the recloser, or both.

Should a fault occur between the relay and the recloser, only the relay will pickup on either its
instantaneous or time delay curve and trip the breaker. If a fault occurs beyond location “D”,
only the recloser is sensitive enough to pick up and clear the fault.

So what happens if a fault occurs in the “overlap” area - that is, between locations “C” and “D”?
Both the recloser and the relay time delay element will pick up. However, the time-current
curves are coordinated so that the recloser will trip and clear the fault before the relay has a
chance to operate.
TIME-CURRENT CURVES

Figure B-2 shows the breaker relay’s time delay curve as well as the recloser’s “fast” and “slow”
trip curves. Assume a permanent fault occurs between locations C & D of Figure B-l. According
to the available fault current curve, this fault may have a magnitude of, say, 800 amps. An 800
amp fault will be seen” by the relay time delay element, as well as the recloser (see Figure B-2).
Note that the relay instantaneous element has too high a setting to sense the fault. The recloser
will trip once on its fast trip curve, then reclose into the fault. The recloser may again trip on its
fast trip curve, and reclose. At this point, the fast tripping will be blocked, and the recloser will
trip on its “slow” curve a total oil perhaps 3 times before finally locking out.

Although the relay time delay element picked up on each reclosure, its time-current characteristic
is chosen to be slow enough to allow the recloser to trip first. After recloser lockout, only those
customers beyond the recloser location are without service.

“NOTCHING”
A problem which sometimes occurs when trying to coordinate a recloser and a relay is
“notching”. Notching is the process of the relay disc advancing while the fault is on the system
and then failing to fully reset during the tripping and reclosure of a downstream recloser.

Referring again to Figure B-2, our 800 amp permanent fault in the “overlap” area of the relay
and recloser zones will cause both the recloser and the relay time delay element to pick up. You
can see that the relay disk will rotate a small amount before the “fast trip” operates the recloser.
However, on subsequent “slow trips” of the recloser, the relay disk has time to rotate even more.

If the relay cannot reset fully between operations of the recloser, it is possible for the relay to
“notch” its way toward the closing position - i.e. this effectively reduces the relay’s operating
time. Thus, the relay may misoperate before the recloser locks out.
Notching is usually not a problem as long as the relay’ s time delay element is set well enough
above the recloser’s slow trip curve. This minimizes the distance traveled by the relay disk
during successive reclosures.

RECLOSER - FUSE COORDINATION

Any lateral fuse that is installed beyond the recloser (see Figure B-1) would have a time-current
curve which is “nested” between the recloser fast and slow trip curves, as illustrated in Figure
B-3. This is similar to the nesting of the fuse curves when coordinating a breaker relay with a
lateral fuse (see Figure A-4). Thus, the recloser “fast trip” will protect the fuse in the event of a
temporary lateral fault (for fault currents below the “limit” indicated). In the event of a
permanent lateral fault, the fuse should operate before the recloser trips on its slow trip curve or
locks out.

TRANSFORMER BREAKER RELAY - FEEDER BREAKER RELAY COORDINATION

Figure B-4 illustrates a substation arrangement in which a transformer is connected to the low
voltage bus through a circuit breaker. Four feeders leave the substation, as shown. The feeder
breakers are equipped with time delay and instantaneous overcurrent relays (denoted by #51 and
50, respectively). The transformer secondary breaker has only a time delay element installed.

Assuming that the feeders are each rated to continuously carry 400 amps, the continuous rating
of the transformer secondary breaker must be 1600 amps.

Typical relay settings might be:

1. Feeder breaker time delay – 600 amp pickup

(1.5 x continuous rating)
2 Feeder breaker instantaneous - 1600 amp pickup
(4 x continuous rating)
3 Secondary breaker time delay – 2400 amp pickup
(1.5 x continuous rating)
U

Suppose a fault occurs on one of the feeders, close to the substation, as shown in FigureB-4. In
this case the fault current could be quite high – say 5000 amps. The time current curves for the
relays are drawn in Figure B-5. Note that the breaker operating time (80 ms) has been added to
the relay operating time, so that the time scale represents total “relay plus breaker” operating
time.
For the 5000 permanent fault, what will be the sequence of operation?
A 5000 amp fault is large enough so that both feeder breaker and secondary breaker relays will
pick up and start to operate. The feeder breaker relay instantaneous element will operate first and
clear the fault in about 0.11 seconds. The feeder breaker will then reclose into the fault (the
reclosing relay controlling the feeder breaker will block any further instantaneous trips). The
feeder relay time delay element will then cause the breaker to trip in less than 1 second. After
two or three more time delay trips and closings, the breaker relay will “lock out”.

During successive closings of the feeder breaker into the permanent fault, the secondary breaker
relay will continue to “see” the fault. However, its operating time is too slow to trip the
secondary breaker before the feeder breaker locks out.
35

This illustrates what is meant when we speak of sensitivity and selectivity. Both feeder and
secondary relays are “sensitive” enough to see the fault condition and begin operation. Together
they provide “selectivity”, causing only the feeder breaker to trip to isolate the fault.

Let us suppose that for some reason the feeder breaker does not open as intended (i.e.
malfunction of either the breaker or the relay). The secondary breaker will continue its operating
sequence if the fault is not cleared, tripping the secondary breaker in 9 seconds and locking open.
The secondary breaker has thus provided “back up” protection for the feeder breaker. Although
selectivity is lost (all four feeders will be interrupted) the fault is eventually cleared (after 9
seconds)
PROTECTION COORDINATION - PART 3
Protection Coordination on Secondary Systems

As discussed in earlier lessons, most protective equipment is installed on the primary distribution
system. Secondary protection is not normally required, since secondary faults usually burn clear
and do not affect as many customers as primary faults.

However, in the case of self protected distribution transformers and secondary spot networks (or
secondary network systems), protective devices are installed on the secondary side and these
must be coordinated with other equipment.

Self-Protected Distribution Transformers

Figure C-1 shows a self-protected distribution transformer, equipped with its own built-in
lightning arrester (for overvoltage protection), primary expulsion fuse, and secondary circuit
breaker.

The time-current curves for coordinating this equipment with the other protective devices, (such
as the primary lateral fuse and the customers service entrance breaker) are drawn as a function of
primary amperes.

The transformer’s internal fuse is sized to operate for faults within the transformer. As you can
see from the time-current curves, a transformer fault should cause the internal fuse to operate
before the primary lateral fuse can operate or be damaged. Thus, only the secondary circuit is
interrupted not the entire primary lateral (i.e. “selectivity” is achieved).
In the event of a heavy overload or a fault on the secondary circuit, the transformer’s internal
secondary circuit breaker should trip (it is thermally activated) well before the internal
transformer fuse can be damaged or operate (this internal fuse is sized to carry several times
normal load current continuously).
By the same token, an overload within the customers residence should cause the customers
service entrance breaker to trip (or the customer’s fuses to operate) before the transformer
secondary breaker.

Secondary Spot Network

Figure C-2 illustrates a secondary spot network, in which three network transformers feed a
common secondary bus. The major components of the network protector are shown
schematically; i.e. the automatic air circuit breaker, controlled by a “reverse power relay”, and a
“backup” fuse. Also shown are cable “limiters”, which protect the system in the event of a fault
on a service cable.

An “ideal” coordination scheme for the secondary spot network is illustrated in Figure C-3,
where we show the insulation damage curve for a 4/0 service cable, as well as the time-current
curves for the 4/0 cable limiter, the backup fuse, and the over current relay in the substation.

The reverse power relay is a directional-type relay, which operates and trips the breaker when it
senses any current flowing toward the network transformer and the primary system (i.e. it
operates on “reverse” current). This would be required in the event of a primary feeder (or
internal transformer) fault, to prevent the fault from being “backfed” through the other feeders
and the common bus. Since the reverse power relay operates by sensing direction, it has no time-
current curve to coordinate with the other protective devices.
However, the “backup” fuse does have a protective characteristic, as shown in Figure C-3. This
backup fuse is installed for the sole purpose of backing up the reverse power relay and its
breaker. Thus, it operates only for primary faults (i.e. on a feeder or in the network transformer).
It is not intended to operate for faults on the secondary bus or service cables.
OVERVOLTAGE PROTECTION
Introduction
As mentioned at the beginning of this topic, the distribution system must be protected from two
different conditions;

1. overcurrents, and
2. over voltages

Lightning and Its Effects

On the primary distribution system the chief cause of overvoltage is lightning. A lightning stroke,
which strikes near or comes into contact with a primary conductor, will cause an almost
instantaneous rise in voltage. Overvoltages due to lightning can be quite severe, often exceeding
one million volts.

A lightning stroke is actually a surge of current, as illustrated by the typical wave shapes shown
in Figure D-l. Stroke current magnitudes may range from as low as 200 amps to as high as
100,000 amps (100 kA). The wave shape at the top of Figure D-l represents the current surge
used by the industry (i.e. equipment manufacturers) to simulate a lightning stroke for equipment
testing purposes. It rises to its maximum current value in only 8 micro seconds (8 millionths of a
second), decreases to half value in 20 Micro Second, and to zero in 100Micro second. Compare
this with a 60 Hz wave, which reaches its peak in about 4,000 microseconds (4 milliseconds).
You can see that a lightning surge (which is over in about 100 Micro Second) represents an
extremely short time frame when compared to the normal 60 Hz current or voltage. This
representative wave used to simulate a lightning surge is referred to as an “8 x 20 Micro Second
wave”.
As you might expect, all lightning strokes do not conform to this representative wave shape.
Some strokes have relatively high magnitudes and short durations (see Figure D-l). These strokes
usually produce shattering or explosive effects with little burning. Other strokes may last for
thousands of microseconds with relatively low current magnitudes. This type of stroke,
commonly called “hot lightning”, produces considerable burning, melting and fires. In many
cases a lightning stroke is a combination of both these types.
The high magnitude current of the stroke itself is normally of such short duration that it will not
cause the overcurrent protective devices (relays, fuses, reclosers) to operate. However, the
extremely high voltages generated by the stroke current may cause insulation failure of system
equipment.

When a lightning stroke contacts a distribution line the impedance that it sees (known as the
surge impedance), is on the order of 300 ohms. When the lightning stroke strikes the line, half of
its current travels in one direction and half in the opposite direction, as illustrated in Figure D-2.
If the lightning stroke is, say, 60 kiloamps it will generate a voltage equal to 30 kA x 300 ohms,
or 9,000,000 volts (9 megavolts).

Voltages of this high magnitude may cause the immediate failure of the insulation on system
equipment connected to the line. In order to minimize insulation damage, lightning (or “surge”)
arresters are installed to protect the insulation of transformers, breakers, station buses, and other
critical equipment. A surge arrester is a device that will conduct the high surge current
harmlessly to ground and prevent destructive overvoltages from impinging on equipment
insulation.

Lightning is also the most common cause of faults on the overhead distribution system. The
tremendously high voltages generated by the stroke current will often cause the porcelain
insulators supporting the conductors to flashover (these porcelain insulators are usually not
protected by arresters, since this type of insulation is “self-restoring”; that is, it will not be
damaged by a flashover). This flashover of the insulators creates an arc, which establishes a
conducting path to earth. Note that, by this time the lightning current surge will have dissipated,
and it is now 60 Hz fault current, which is being conducted to ground as a result of the insulator
flashover. In effect the lightning overvoltage has established a phase to ground fault on the
system. It is now the task of the system overcurrent protection (relays, fuses, reclosers) to
interrupt the 60 Hz fault current.

It is important to distinguish between the very short duration surge current due to the lightning
stroke and the 60 Hz fault current which passes afterwards (i.e. after an insulator flashover).
Lightning arresters
On the primary distribution system, lightning arresters are installed to protect the insulation of
various equipment, such as

1. Substation transformers
2. Distribution transformers
3. Capacitor banks
4. Voltage regulators
5. Circuit breakers
6. Underground cables and switchgear

Although porcelain insulators (on the overhead feeders) do not require lightning protection, in
some applications arresters may be installed, as sketched in Figure D-3. For example, on systems
subjected to frequent thunderstorms, it may be desirable to prevent insulation failure and
consequent frequent operation of the overcurrent protection.

Arresters are connected line-to-ground; with one arrester per phase on three-phase equipment.
For station circuit breakers, power transformers, outdoor voltage regulators and reclosers,
arresters are usually connected on both the incoming and outgoing sides of the equipment to
protect for surges entering the equipment from either direction. Arresters should be located as
near as possible to the bushings of the equipment they are installed to protect.

Two types of arresters are in common use. These are the silicon carbide (“gapped”) type and the
zinc oxide (“gapless”) type.

Silicon Carbide Arresters

The silicon carbide arrester consists of a group of “gaps” in series with a “non-linear resistor”.
The non-linear resistor (known as the “valve element”) is composed of silicon carbide. The gap
assembly and silicon carbide are contained in a porcelain housing, as shown in Figure D-4.
The silicon carbide has an operating characteristic as shown in Figure D-5, in which voltage
across the arrester has been plotted against current through the arrester. Since resistance is simply
voltage divided by current we can think of the resistance of the silicon carbide element as the
slope of the operating curve. You can see that this slope will decrease as the voltage across the
arrester increases. Thus, at relatively low voltages (i.e. normal system voltage) the silicon carbide
exhibits a high resistance. Conversely, at the very high voltages typical of lightning surges, the
resistance of the silicon carbide is quite low.

Since the silicon carbide material has a voltage-current characteristic, which is not a straight line
(resulting in a variable resistance), we refer to this material as a “non-linear” resistance. The
silicon carbide is also referred to as a “valve element” because it allows current to pass when the
resistance is low and inhibits current from passing when the resistance is high.

Silicon carbide arresters must contain gaps to isolate the non-linear resistance from the system
during normal operating conditions. This is because, at normal system voltage, the resistance is
not high enough to completely prevent current from “leaking” through the arrester. Without an
open gap, as shown in Figure D-5, this leakage current would flow continuously, causing thermal
damage to the arrester. However, during a lightning surge, the gap will sparkover, and
consequently introduce the non-linear valve element into the circuit. The low resistance of the
valve element (at the high surge voltage) will allow the surge current to dissipate safely to
ground. However, you should recognize that once the surge current is discharged, there is still a
60 Hz fault on the system. These gaps are designed to open against this 60 Hz current before the
overcurrent protective devices can operate.

The protective characteristic for a distribution-type silicon carbide arrester is illustrated

in Figure D-6. Upon the incidence of a lightning surge, the arrester voltage rises very quickly, but
when the dielectric strength of the gap is exceeded (within about 1 microsecond) the gap will
flashover and begin conducting current. The voltage which causes the gap to break down is
referred to as the “sparkover voltage” (in our example, sparkover occurs at 39 kV). When
sparkover voltage is reached, the arrester voltage will drop momentarily due to the very low
resistance exhibited by the silicon carbide valve element. Over the next several hundred
microseconds, the arrester will “discharge” the surge current. The arrester voltage during dis-
charge is simply the surge current times the resistance of the silicon carbide. The low value of
resistance will effectively limit discharge voltage. In the example shown in Figure D-6, the
discharge voltage is limited to a maximum value of 36 kilovolts.

Of course, the discharge voltage depends on the magnitude of surge current; the higher the surge
current the higher the maximum discharge voltage. At the manufacturer’s facility, selected
arresters are tested with simulated impulse waves at various current magnitudes. For example, a
typical discharge test (shown in Figure D-7) uses a 2.5 kiloampere simulated stroke current,
which causes the arrester voltage during discharge to rise to a value of 28.2 KV.

After discharging the lightning current surge, the gap is still conducting 60 Hz current, referred
to as the “power follow” current. Since the arrester voltage is now decreasing toward its normal
system voltage, the resistance of the silicon carbide will increase significantly. This high value of
resistance will limit the follow current and allow the arrester gap to effectively interrupt the flow
of current. This is known as the arrester “reseal”. The maximum voltage which allows the
“follow current” to be limited enough so that the gap can interrupt is referred to as the reseal
voltage. In the example of Figure D-6, the arrester reseals at 9 kV.

Arresters are in fact rated by their reseal voltage, as shown in the table of Figure D-8. For
example, an arrester rated 15 kV may spark over at a maximum voltage of 63 kV, discharge at a
maximum voltage of 54 kV (with a 10 kiloamp current surge) but it will reseal at 15 kV, the
arrester rating.

The arrester should reseal and eliminate the power follow current in a fraction of a millisecond.
You should recognize that this is long before the overcurrent protective devices can sense the
fault and operate. As apparent from Figure D-8, silicon carbide arresters are classified their
rating, their maximum sparkover level, and their maximum discharge voltage as a function of the
surge current.
Zinc Oxide Arresters
Zinc oxide arresters are quite different from the silicon carbide type, since they do not require
gaps to begin conducting surge current and to then interrupt power follow current. Instead, the
zinc oxide arrester consists simply of its non-linear resistance, which is always connected to the
system.
When an overvoltage occurs, the zinc oxide arrester “eases into” conduction and, when the
voltage subsequently drops, this type of arrester “ease out” of conduct ion. This operation is
possible due to the extremely non-linear characteristic of zinc oxide.

Figure D-9 shows a comparison of the operating curves for zinc oxide and silicon carbide. Notice
that, when normal line-to-neutral voltage is impressed across the arrester, the zinc oxide unit
conducts but a small fraction of an ampere; that is - leakage current. However, the silicon carbide
unit (if it were not equipped with gaps) would conduct several hundred amps at normal system
voltage.

Likewise, after a current surge has been dissipated and the arrester voltage decreases toward its
normal value, the follow current will be limited to only a fraction of an ampere by the zinc
oxide’s very high resistance at system voltage. The silicon carbide arrester requires gaps to
interrupt the large follow current.

Unlike a silicon carbide arrester, the zinc oxide arrester has no “sparkover voltage” at which the
arrester begins conducting current. This is because the zinc oxide arrester always conducting
some small amount of current, even during normal conditions.

One example of zinc oxide arrester characteristics is given in Figure D-10. They are classified by
their “maximum continuous voltage capability”. This is the maximum continuous 60 Hz voltage
that the arrester can withstand without suffering a thermal failure. The zinc oxide arrester’s reseal
voltage represents the voltage at which only an insignificant leakage current will pass through
the arrester. Like the silicon carbide arrester, zinc oxide arresters are rated by this reseal voltage.
Keep in mind that the reseal voltage and the maximum continuous voltage capability are not the
same value.
Arrester “Classes”
Arresters are grouped into three general “classes” - Station, Intermediate, and Distribution.
As a general guide to arrester class usage versus equipment size, the following appears to prevail
as typical practice:

Station Class Component protection of 7.5 MVA and above and large
or essential rotating machines

Intermediate Class Component protection of 1-20 MVA substations and

rotating machines

Distribution Class Protection of distribution class apparatus, small rotating

machines, and dry-type distribution transformers.

Voltage ratings for the three classes are given in the table below: (extracted from ANSI Standard
C62.2).

You will notice that the voltage ratings “Overlap” from class to class; however, the protective
characteristics are quite different.

For example, a 9 kV distribution class arrester is not the same as a 9 kV intermediate class
arrester, as shown by their respective sparkover and discharge voltages, given below:

Silicon Carbide Arresters:

Sparkover kV Discharge kV at 10 kA
9 kV (Int ) 31 23.5
9 kV ( Dist ) 39 36

The intermediate class arrester will sparkover at a lower voltage and the discharge voltage will
be limited to a lower value, providing increased equipment protection.
INSULATION COORDINATION
This segment examines the process of applying lightning arresters to protect critical equipment
on the distribution system. Overvoltage protection on underground residential distribution
systems is also discussed, as well as “other” (i.e. non-lightning) sources of overvoltages.

Insulation Characteristics and BIL

In order to apply an arrester to protect a piece of equipment, we must first know something about
the nature of the equipment’s insulation. The insulation characteristic describes “how much” and
“how fast” voltage can be applied to the equipment without breaking down or damaging the
insulation.

For example, the insulation withstand curve for a typical 7200-volt single-phase distribution
transformer is shown in Figure E-1. Note that the insulation can withstand a higher voltage (110
kV) if it is applied within a microsecond or two, and a lower voltage (95 kV) if the voltage is
applied over a period of several microseconds.

The insulation withstand characteristic is determined by factory tests which use a variety of
standard wave shapes designed to certify the strength of the equipment’s insulation.

One such test is the “basic impulse insulation level”, known as “BIL”. The BIL standard wave
shape is drawn in Figure E-2, and is intended to simulate a lightning impulse. The wave peaks in
1.2 microseconds and decays to one half of peak voltage in 50 microseconds. The 7200 volt
transformer (to be applied on a 12.47 kV four wire system) would be tested with a wave of the
shape shown in Figure E-2, and a magnitude (“E”) of 95 kV.

Another insulation test is the “chopped wave” test. This test simulates the effect of a “fast front”
surge. This may be due to a lightning stroke which does not fit the standard wave and causes a
rise in voltage that is quicker than that simulated by the BIL voltage test.
The chopped wave test is conducted using a voltage that is approximately 15-20% higher than
the BIL test voltage (i.e. 110 kV if the BIL is 95 kV - see Figure E-l). The voltage is built up at a
certain predetermined rate and, after cresting at approximately 1:2 microseconds, is “chopped” to
zero.
Manufacturers build equipment to meet certain industry standards. For instance, the table of
Figure E-3 shows standard insulation withstand levels for oil-type transformers in different
voltage “classes”. The single-phase 7200-volt transformer we discussed earlier (i.e. for
application on a 12.47 kv four-wire system) is in the 15 kV class, and has a BIL (full wave)
withstand of 95 kV and a chopped wave withstand of 110 kV. As illustrated in the table, the
transformer can also withstand a low frequency (60 Hz) overvoltage of 34 kV rms for 1 minute.
Note that each voltage class has different standard withstand levels.

Selection of Arrester Rating

At first one might think that, to protect our 7.2 kV single-phase transformer we would simply go
to the arrester charts (see Figure D-8) and choose a 7.5 kV arrester.

However, if we did this the 7.5 kV arrester would probably not reseal after a sparkover. Failure to
reseal could result in burn-out of the arrester.

Let us examine why a 7.5 kV arrester is too low a rating to apply to a 7.2 kV distribution
transformer. Recall that the arrester rating is the maximum voltage that will allow the arrester to
reseal. But during a line-to-ground fault on the 12.47 kV system, chances are that the system
voltage (on the unfaulted phases) will not fall as low as 75 kV. This is because the fault will
cause the neutral of the system to “shift” away from ground, so that voltage across an arrester
(connected to an unfaulted phase) will increase.

This phenomenon of “neutral shift” is illustrated in the vector diagrams of Figure E-4. Here, we
show the two extremes: “zero shift” on a “perfectly grounded” system and maximum, or “total”
shift on an ungrounded system.
If a three-phase four-wire system were grounded “perfectly” (i.e. zero impedance in the neutral
return circuit), then a single line to ground fault on “A” phase will cause the AF voltage to
collapse to ground and a large fault current to flow. But, since there is no impedance in the
“return path”, ground voltage is equal to neutral voltage. Thus, there is no rise in voltage on the
unfaulted phases. The arrester voltage during such a fault is simply the line to neutral system
voltage (7200 volts in this example).

Quite by contrast, if the system is ungrounded, a single line to ground fault will result in zero
current flow (i.e. “infinite” impedance in the ground path). Thus, AF voltage is equal to ground
voltage, causing the arresters on the unfaulted phases (connected line-to-ground) to experience
the full line-to-line voltage. For example, if the system were 13.2 kV ungrounded wye, the
arrester voltage would rise from 7.62 kV (line-to-neutral on a 13.2 kV system) to the full 13.2 kV
due to the neutral shift (an increase of 73%).

So - in the “perfectly grounded” case the arresters must reseal against “line-to-neutral” voltage,
while in the ungrounded case they must reseal against “line-to-line” voltage.

In practice, for grounded systems, the “neutral shift effect” will be between these two extremes,
depending upon the effectiveness” of the system grounds. Recognizing that no system is
“perfectly” grounded, we use a neutral shift “factor” of 1.25 for a four-wire “multi-grounded”
system. This is the factor by which we would multiply line-to-neutral voltage to determine how
the arrester would be rated.

Therefore we conclude that, for our 7.2 kV transformer (installed on a 12.47 “multi-grounded”
system), the arrester voltage will not drop back to 7.2 kV after sparkover and discharge. Rather,
it will drop back to about 7.2 kV x 1.25 = 9.0 kV. Thus, we must use arresters rated at least 9 kV
to assure that they will reseal.

Likewise, for a 7.62 kV single-phase transformer (installed line-to-ground on a 13.2 kV

ungrounded wye system), the arrester voltage will not drop any lower than 13.2 kV following
arrester sparkover. Thus, we must choose an arrester rated at least 13.2 kV (i.e. a 15 kV rated
arrester would be chosen) to protect a 7.62 kV transformer.
Insulation Coordination
Once an arrester has been selected we can see how well it will protect a piece of equipment by
comparing the arrester characteristic with the insulation withstand characteristic of the
equipment. This procedure is called “insulation coordination”.

Using the same example of the 7.2 kV distribution transformer protected by a 9 kV arrester, we
simply plot the transformer’s insulation withstand curve (see Figure E-l) on the same graph as
the 9 kV arrester curve (see Figure D-6).

The combined plot illustrates the principle of insulation coordination and is given in Figure E-5.

The graph reveals that the maximum system voltage (limited by the arrester) will not approach
the insulation withstand levels of the transformer.

Industry guidelines recommend a minimum 20 percent “safety margin” as calculated by the

following formula:

% margin = Insulation withstand —Arrester protective level * 100

Arrestor protective level

Using this expression we can calculate two safety margins.

Chopped Wave Margin

% margin = 110kv — 39kV x 100% = 182%

39kv
(Using chopped wave withstand and arrester “sprakover”)

BIL Margin

% margin = 95kV — 36kv x 100% = 164%

36kv
(Using BIL withstand and arrester “discharge” voltage )
In this example, we see that the margins are well in excess of the 20% guideline. Of course, these
“margins” of safety may be affected by other factors - such as transformer aging and
overloading, as well as the lead length between the arrester terminals and the transformer
terminals.

URD Protection
The overvoltage protection of underground residential distribution (URD) systems is somewhat
more complicated than that of overhead systems due to two factors. First, the surge arresters are
usually located on riser poles, relatively remote from the equipment they are installed to protect.
Secondly, URD systems are usually looped, with a normally open tie point. This open point
introduces traveling wave reflections and a “voltage doubling” effect. This is usually not a
problem on overhead systems due to the use of arresters at each equipment location.

Figure E-6 illustrates the use of arresters located on the riser poles; these riser poles serve as the
interface between the overhead primary circuit and the URD system. If a lightning stroke hits the
overhead conductor, the arrester will spark over at, say, 39 kV. This voltage will “travel” into the
URD system via the cable circuits.

When such a “traveling wave” encounters a change in impedance, a portion of it will be reflected
back toward its source. The magnitude of the reflected voltage depends on the impedance
change. An open point is seen as an infinite change in impedance, so the traveling wave will be
reflected back in its entirety. This will produce a wave twice as large as the original wave; in our
case, 78 kV.

The insulation coordination of underground equipment must take this doubling phenomenon into
account. Using the (distribution - class) arrester and transformer characteristics of Figure E-5, we
can calculate protective margins of:

Chopped Wave:

110 kV — 78 kV = 41% and

 78 kV
B IL
95 kV 72 kV = 32%
72 kV
(i.e. 72 kV reflects the doubling of the discharge voltage)

Note that the safety margins are substantially reduced over those calculated for the overhead
system, but they are still above the 20% guidelines.

In many cases, the arresters installed at the riser pole will be of the intermediate class. Since the
arrester sparkover and discharge voltages are lower for intermediate class arresters (as compared
to distribution class arresters), the protective margins will increase.

If the insulation withstand capability of the installed equipment is greater than the doubled
voltage, with the adequate margin, there will generally be few problems. Otherwise, an arrester
must be installed at the open tie point.

This can be accomplished by installing arresters, which resemble elbow connectors. On “dead-
front” pad-mounted transformers, these can be connected to the transformer open tie point to
provide surge protection. Alternatively, some transformer manufacturers are now providing built-
in surge arresters under oil. In either case, these arresters are of the zinc oxide type.

Arrester Lead Length

An important consideration in the application of surge arresters (on both underground and
overhead systems) is the length of the leads connecting the arrester terminals to the protected
equipment’s terminals, and also the lead length from the arrester to ground. Inadequate
overvoltage protection can result if these leads are too long.

The voltage across the protected equipment is the sum of the arrester voltage and lead voltage.
For particularly long lead lengths, the lead voltage may be high enough to cause damage to the
equipment. For this reason, arresters are located as near as possible to the equipment they are
installed to protect.
“Other” (i.e. “System—Generated”) Sources of Overvoltage
Up to this point, we have discussed lightning as the primary source of distribution system
overvoltages. We have also examined the application of arresters to limit lightning overvoltages
to levels, which will not damage the insulation of the system equipment.

However, there are a variety of other conditions, which may cause overvoltages. These are
generated not by the weather - but rather by phenomena on the distribution system itself. It is im-
portant for the system designer to carefully consider these “other” sources of overvoltage since
they may cause the undesired operation of arresters (or damage to the arresters).

Although this section was not included in the videotape presentation, its discussion is felt to be
warranted at this time. “Other” sources of overvoltage include:

1 capacitor bank switching

2. a phenomenon known as “ferroresonance”
3. the “chopping” of current, and
4. accidental contacts with high voltage systems

Capacitance Switching
With switching capacitors, a particularly troublesome condition can occur when de-energizing
the bank (see Figure E-7). After the breaker opens, the interruption of the arc will take place
when the sinusoidal current passes through zero, that is when the capacitor voltage is at a
negative maximum. Remember voltage lags current by 90 degree in a capacitive circuit. So,
while the capacitor voltage remains charged at a negative maximum, the voltage on the source
side of the breaker will continue its cycle. You can see that, when the system voltage is at a
positive maximum, the voltage difference across the breaker will be twice its normal value; that
is, voltage across breaker system voltage - capacitor voltage = V - (-V) = 2V.

As we mentioned in a previous lesson, this buildup of voltage may exceed the strength of the
breaker’s insulation, resulting in a restrike, and further magnification of voltage. Voltages
approaching 3 times normal are possible during capacitor switching operations.
This overvoltage may be enough to sparkover the arresters and, if the voltage continues to
escalate, the arresters may not reseal, resulting in damage. Many utilities have avoided this
problem by installing resistors in the breakers, as described in an earlier lesson.

Ferroresonance

Another potential source of excessive voltage occurs when the inductance and capacitance of a
circuit interact in series. This condition is referred to as “ferro” resonance, since the inductance
involved belongs to equipment with an iron core, for example, a pad-mounted transformer.

An example of a condition, which could give rise to ferroresonance, is illustrated in Figure E-8.
The capacitance is provided by the cable connecting the ungrounded windings of the transformer
to a riser pole. In this particular example two of the fused cutouts on the riser pole have blown,
due to a fault. It might appear at first that with only one fuse connected, the transformer is not
energized. In a way, this is true. But, there is still a path for current flow through two of the trans
-former windings and the cable capacitance, as indicated by the arrows. Note that the winding
inductance and cable capacitance are effectively connected in series; the voltage across the
transformer inductance (VL) leads the current by 90 degrees while the voltage across the cable
capacitance (Vc) lags by 90 degree (See the vector diagram) . These voltages add to give the
applied system voltage, V. You can see that both the winding voltage and cable voltage can far
exceed the system voltage. This could cause the surge arresters on the open phases to sparkover.
If the condition is sustained, repeated operation can destroy the arresters.

In these underground applications, many utilities have avoided ferroresonance problems by

installing wye-grounded transformers. The grounded primary neutral then effectively “shorts
out” the series connection with the cable capacitance, thereby preventing the creation of the
resonant circuit.
Current Chopping
A third source of overvoltage which may occur in breakers and current limiting fuses is a
phenomenon called “current chopping”. As already discussed, a current limiting fuse, when
called upon to clear fault current, forces a current interruption much sooner than the normal
current zero (see Student Workbook DSO 7.1, Figure D-5). This so-called “current chop” can
produce an extremely high arc voltage; this may be sufficient to sparkover a surge arrester
located on the source side of the fuse. The discharge of energy through the arrester is much
slower than a 1ightning discharge and may cause excessive heating and consequent damage to
the arrester valve element.

Accidental Contact With HV Systems

Another possible cause of system overvoltages is accidental contact with higher voltage systems.
Sometimes, overhead primary circuits are installed underneath subtransmission circuits on the
same pole. If a broken high-voltage conductor falls on the primary circuit, the increase in voltage
may cause the arresters to fail along the entire length of feeder.

In the application of surge arresters, system designers will typically consider all possible sources
of system overvoltages, including lightning, switching surges, ferroresonance, and so on. In
applying arresters, the system designer must assure that equipment is adequately protected from
lighting surges, and, at the same time that the other system-generated overvoltages do not
damage or destroy the arresters.