EXCITATION SYSTEM PROTECTION
In order to protect a generator from damage due to a malfunction in the excitation system,
evaluate the potential failure modes and provide protection for these conditions. Since the
excitation system is normally operated under automatic voltage regulator (AVR) control,
the output of the excitation system changes in response to the varying load conditions
ranging from no load to short circuit. The excitation may be reduced to zero or increased
to maximum forcing voltage depending on the normal reaction of the AVR to maintain
voltage. To detect abnormal operating conditions in the excitation system is no easy mat-
ter. Abnormal response of the excitation system is usually zero excitation or forcing. How-
ever, it is possible for the excitation system to drive the generator into an area of operation
beyond its capability by operator adjustment or by load demand. This possibility adds to
the complexity of protecting the generator.
Figure 1: Normal Excitation System Operation
Possible failures to be addressed relating to the generator excitation system include:
• Loss of sensing to AVR
• External action directing the excitation system to an unsafe operating area
• Network conditions, such as over or under voltage, short circuits, etc.
• AVR failure
• Loss of excitation power
• Field ground
• Exciter diode failure
1
Figure 2: Loss of Excitation Control
Figure 3: Typical AVR Block Diagram
2
Figure 4: AVR Sensing Figure 5: AVR Sensing
Common failure modes include the loss of sensing to the AVR. This failure is commonly
caused by failure of a potential transformer fuse or a break in the connections between the
generator terminals and the AVR. Under this condition, the AVR measures a very low
voltage and attempts to raise it back to normal by increasing excitation to maximum forc-
ing voltage. For a generator operating alone, non-paralleled, the result will be increased
terminal voltage from 30 to 60% above rated. If the excitation power is taken from the
generator terminals, the increased terminal voltage results in higher than normal forcing
voltage, by the same 30 to 60%. Operation at this level of voltage is dangerous for the
generator windings and all electrical equipment connected to the generator stator wind-
ings. It is desirable to trip the excitation power or transfer to manual control or backup vts
as quickly as the fault can be detected. One may use a generator overvoltage relay (De-
vice 59) monitoring the stator voltage, a field overvoltage relay (Device 59F) monitoring
field voltage, a voltage balance relay (Device 60) monitoring the voltage between two sets
of potential transformers (vts), or (for three-phase sensing) a negative sequence voltage
relay (Device 47) monitoring the AVR sensing input.
Figure 6: Overvoltage Protection for Loss of Sensing
3
Figure 7: Voltage Balance Relay Loss of Sensing Protection
Figure 8: Possible Actions for Loss of Sensing
This loss of sensing voltage failure, if it occurs while the generator is operating in parallel
or with a network, will result in increased lagging VArs being supplied to the load bus.
Some small voltage increase may also be observed, but in general the other generators
operating on the load bus will maintain the bus voltage by absorbing the excess VArs
being generated by the unit with the loss of sensing. To detect this condition, one may use
any of the relays above except the generator overvoltage relay, but the most common way
to detect this failure mode in parallel operation is the field overcurrent relay with a time
delay. The generator can withstand full forcing current for up to 10 seconds, allowing time
for the normal generator response to heavy loads or fault currents. After this time has
passed, the field of the generator may overheat, damaging the winding insulation. Using a
current sensing relay set to 110 to 120% of full load field current with a 10 second time
delay will protect the generator and allow other relays time to operate first.
4
Figure 9: Generators in Parallel Operation
On larger generators, it is common to supply a low-set field overvoltage relay operating
only when the machine is being started. Using a trip setting 20% higher than the no-load
voltage at rated speed and a short time delay, this relay increases the protection during
no-load operation until the relay is blocked from operation by the closure of the generator
load breaker. The same idea may be applied to an overvoltage relay connected to the
stator through vts. At off-line operation, it may be possible to set a generator overvoltage
relay close to the generator nominal operating voltage and with a fairly short time delay, as
an alternative to the field overvoltage relay, but generally faster detection of loss of sensing
can be obtained on the field.
Figure 10: Off-Line Loss of Sensing Protection
5
The most secure and high speed protection will be provided by the voltage balance relay,
device 60, but it requires two sets of potential transformers (vts). The relay compares
phase voltage between the two vts, looking for some unbalance in their outputs. If such an
unbalance exists, after a very short time one can take action to get around the problem by
transferring the sensing to the good vts, shutting down the AVR and transferring to manual
excitation control, or tripping down the generator. If only one set of vts is available, a de-
vice 47 relay is able to sense the loss of a phase, distinguishing this event from a phase-
neutral short circuit, and providing a fast tripping signal to transfer to manual control or to
trip the generator. If either of these methods is not acceptable, the less secure, slower
method of sensing is to monitor the field voltage or current using a dc overvoltage or dc
overcurrent relay and a time delay. Either of these relays must allow for sufficient time at
forcing excitation to clear faults, start large motors, or any other normal event requiring
extended operation at high field excitation levels.
Figure 11: Voltage Balance Relay for Loss of Sensing Protection
Figure 12: Negative Sequence Voltage Relay for Loss of Excitation Protection
6
It may be possible for operator action, remote control, auto synchronizer, or programmable
logic control (PLC) to adjust the excitation control system to operate outside of the safe
limits. Although not truly a protective relay, the excitation limiter is often provided to over-
ride the actions of such external influences and keep the generator in a safe operating
area. If the operating conditions or failure of one of these devices would cause the genera-
tor to operate in an overexcited area, the excitation limiter, sensing the level of field current,
is able to react by independently driving the excitation system operating point downward.
This form of control assumes the excitation system is operating correctly, but is being
commanded to overexcite the generator. This is the first line of “protection” for the genera-
tor for overexcitation, but it should always be backed up by a protective relay such as the
overexcitation relay that senses field current and trips the unit on overexcitation after a
short time delay or the dc overvoltage relay, which senses field voltage and trips after a
time delay.
Figure 13: Possible Outside Sources of Excitation Adjustment
Another source of over- or underexcitation is a change in the network condition. For
overexcitation, loss of the network or a substantial voltage reduction may cause the gen-
erator to attempt to pick up more load than it is capable of supporting. If the load is reac-
tive, the prime mover may be able to maintain operation without problem. This kind of
problem may come from voltage reduction programs, tap changer failure, or loss of a
transmission line. The results are the same at the generator. Loss or lowering of line volt-
age will cause the excitation system to increase VAr load to try to maintain the voltage. If a
VAr/power factor regulator is in service on the generator, it will adjust for the reduced
voltage until the lower voltage limit of operation is reached. Then the excitation system will
try to maintain voltage at the low voltage limit, increasing the lagging VAr load to support
the voltage. The excitation limiter again can save the generator by limiting the field current
to the safe operating value of the generator. This action should also be backed up by a
field voltage or current sensing protective relay with time delay.
7
Figure 14: Generator in Parallel with Network
Under short circuit conditions, the response of the excitation system depends on its de-
sign. If generator potential is used to provide excitation power, the impact of a fault on the
available excitation power depends on the distance from the generator and the type of
fault. Distance determines the available voltage for excitation at the terminals of the gen-
erator, and therefore, the available forcing voltage.
Figure 15: Fault Current Depends on Distance to Fault
8
Type of fault determines the available forcing voltage also based on which phases are
shorted and which phases are used for excitation power supply. For potential source
excitation, the fault will reduce the excitation forcing voltage a greater amount as the fault
moves closer to the generator location because the impedance of the lines between gen-
erator and short circuit increases with greater line length. If the excitation forcing voltage
falls too low, excitation voltage will collapse, and no short circuit current will be available to
operate overcurrent devices and selectively trip breakers. For potential source excitation
systems, it may be necessary to provide an undervoltage trip with time delay to make
certain the generator is tripped on a close-in fault.
Figure 16: Compound Exciter System with Fault
For compound source excitation or permanent magnet generator excitation systems, the
occurrence of a fault on the power system will result in the supply of current to the short
circuit, allowing the protective relays to trip breakers in a selective tripping scheme to
minimize the power outage and restore power to un-faulted loads, if possible. To back up
the protective relays providing this function, a field overcurrent or overvoltage relay may be
provided. This relay must incorporate a time delay sufficiently long to allow the protective
relays sensing the fault current to operate and clear the fault. Then, if the excitation re-
mains high beyond the worst case relay tripping time, the generator breaker and excitation
breaker can be tripped on overexcitation. The generator field must be capable of operating
with the forcing level of excitation for the duration of the time delay for the overexcitation
relay. An excitation limiter must also incorporate a time delay to prevent it from limiting the
field current too quickly during a fault. Usually, the limiter has an instantaneous limit, set to
allow a desired amount of fault current within the capability of the generator and its field.
Then a time delayed limit is set to reduce field current to a safe longer-term level, with the
time delay set to allow the overcurrent tripping to occur before the longer-term limit is
enabled.
9
Figure 17: Overexcitation Relay for Fault Backup Protection
If the AVR should fail, it will ordinarily fail with full forcing voltage or no voltage to the field. If
the generator is operating as a single unit (Islanded), with or without load, the generator
voltage will rapidly react to such a condition by dropping toward zero or rising rapidly to
30% or more above nominal voltage. Because both of these conditions also occur during
normal AVR operation, an over/undervoltage relay device must incorporate sufficient time
delay to prevent “nuisance” tripping during events such as startup of the generator, motor
starting, load loss, or faults. Because this results in some slow protection, it is common to
use more sensitive protection with shorter time delays when the generator breaker is open
and the generator is under no load. It is also possible to detect the AVR failure using dc
voltage or current relays, with time delays, and it is common to use more sensitive protec-
tion during no-load operation with such relays also. Relays used to provide off-line genera-
tor protection can be the same relays used for on-line loss of sensing, external influences,
network problems, or short circuit backup relaying, but it will not be possible to determine
from the relay tripping, the cause of the trip. By using separate relays specialized for de-
tecting certain faults, and separate relays for off-line and on-line operation, diagnosis of the
cause can be determined much more quickly, speeding up resolution of the problem and
restoration of the generator to service.
10
Figure 18: Off-Line Overexcitation Protection
If the AVR should fail with the generator on line, paralleled to a network or to other genera-
tors, the failure of the AVR will have little effect on generator terminal voltage.
Figure 19: Overexcitation Limits of Safe Generator Operation
11
Instead, the output VArs will increase in the lagging direction (supplying VArs to the bus)
for overexcitation, or output VArs will move from lagging through unity power factor over to
the leading power factor side (absorbing VArs from the bus) to sustain voltage during loss
of excitation. It is possible to apply a directional power relay connected in quadrature
(voltage leading current by 90 degrees) to sense excessive VAr flow at the generator
output terminals and provide a time delayed trip signal. It is more common to apply a dc
overcurrent relay with time delay, backing up the excitation limiter which can have no effect
in controlling a failed AVR. Set the overcurrent relay to 110% of full load excitation current,
with a time delay sufficient to allow normal short circuit clearing time.
Figure 20: Reverse VAr Relay for Underexcitation Protection
If the AVR fails in the underexcited mode while paralleled to other generation, the other
generators will supply VArs to the underexcited machine to maintain its terminal voltage. A
power factor meter will indicate leading power factor operation, or a VAr meter will indicate
leading VArs. Power flow from the generator to the load may continue relatively undis-
turbed. A common misconception leads many to depend on the reverse power relay for
protection from loss of excitation, but this is not true unless the characteristic of the reverse
power relay allows it to sense the leading VAr load. It is much better practice to maintain
the separation of fuel/real power and excitation/reactive power and provide separate relay-
ing for reverse power and reverse VArs.
12
Figure 21: Underexcitation Limits of Generator Operation
The principle danger to the generator from underexcitation is the reduction of synchroniz-
ing torque, required to maintain the rotor at synchronous speed and prevent slipping of
poles, also called out of step operation. If pole slip occurs, caused by rotor angle exceed-
ing a 90 degree load angle, the rotor accelerates toward the next rotating magnetic stator
position, slips past it to the next, and so on. Each time the rotor slips past a stator mag-
netic pole, large currents are induced in the rotor iron or the amortisseur windings. For
generators equipped with the amortisseur windings, the generator will operate as an
induction generator, increasing in speed as far as it can be driven by the available prime
mover torque. Under this condition, stator overcurrent protection may trip the generator, as
the stator current in this condition increases up to 200% of rated. If the generator has no
amortisseur windings (some two-pole rotors), high current will be induced in the rotor,
heating the rotor beyond its design limits. In addition, voltage is induced in the field, poten-
tially causing high voltage which may overstress the rotor winding insulation. Because of
all these potential hazards to the generator, the protective relay, usually called loss of
excitation or loss of field relay (device 40), is an instantaneous relay designed to quickly
detect the operation of the generator in the leading VAr direction. It is not unusual to oper-
ate a generator in the leading power factor region, to help control the voltage of a network,
for example, but reducing the excitation too far, increasing the leading VAr load, reduces
the synchronizing torque and risks the possibility of loss of synchronism. Generator
underexcited capability curves from the manufacturer may be used to coordinate the
protective relay settings with the generator capability.
13
Figure 22: Distance Relay Protection
Protective relays intended to protect for loss of excitation for generators in parallel with a
utility network have been traditionally based on directional distance relays. This type of
relay monitors voltage and current from a pt and a ct and measures the ratio of voltage to
current. Using a simple balanced beam analogy, the force V maintains the contact open for
all values of I up to the threshold current.
Figure 23: Balanced Beam Impedance Relay
As long as the current is exerting less force than the voltage, the relay is not tripped. When
the current exerts a force equal to or greater than voltage, the contact closes initiating a
breaker trip. This simple relay is responsive to impedance, with no directional characteris-
tic. In other words, increased current can be translated into reduced impedance because:
V
——— = Z
1
14
Therefore, as current increases and voltage remains constant, impedance decreases. The
tripping point of the relay may thus be defined as a value of impedance. Impedance trans-
lates into distance, even when the voltage is depressed by high fault current. As the im-
pedance of the line between the relay and a fault increases with distance to the fault, the
ratio of the voltage to the current is proportional to the distance. By adding a directional
characteristic to the relay, the ct and pt connections can be selected to cause the relay to
look toward the generator.
Figure 24: Impedance Relay Tripping Characteristic
The relay most commonly used is called an mho relay. It provides a circular characteristic
like the distance relay, but offset from the origin reflecting its directional nature. While this
relay is used for utility power plant applications, where this relay is very selective in detect-
ing loss of excitation vs. a normal swing of the generator through a leading power factor
transient condition, it is an expensive and complex-to-coordinate relay.
Figure 25: Directional Mho Relay Tripping Characteristic
15
For smaller generators whose contribution to the stability of a network is not significant or
for generators operating in parallel on an isolated load, two other alternatives are available.
If the normal operation of the generator will not require operation in leading power factor, a
reverse VAr sensing relay may be applied. Operating like the reverse power relay, the
reverse VAr relay can be set to trip at some level of leading reactive current flow or leading
reactive power flow, depending on the characteristic of the relay selected.
A third method monitors the actual flow of dc current in the field and provides a trip after a
time delay to ride over loss of load transients. This final method goes directly to the source
of underexcitation if the purpose of the relay is to detect a failure of the excitation system.
Loss of excitation power is relayed separately from AVR failure when the power supply to
the excitation is three phase. It is separate because loss of one of the phases, a result of a
fuse blowing or a rectifier that will not conduct, reduces forcing voltage from the excitation
system, but does not necessarily produce any detectable change in generator perfor-
mance under single unit or paralleled operation. The AVR can accommodate the loss of
forcing under most operating conditions, exceptions being a large motor start with a single
generator or a fault on the generator output.
Figure 26: Loss of Excitation Using dc Current Sensing
16
Figure 27: Negative Sequence Relay for Loss of Excitation Power (Fuse Failed) Protection
It is possible to detect the loss of a phase in the three phase supply using a negative
sequence voltage relay (device 47). The fuses, generally on primary and secondary of the
power potential transformer supplying the excitation system, may also be equipped with
fuse blow indicators. The action taken when the relay(s) detect a phase loss may be to
alarm or to trip the generator and excitation. Alarming may be selected in this case be-
cause the operation of the exciter is impaired but still functional for most operating condi-
tions. The election must be made by the operator of the plant, evaluating the potential of
fiscal loss due to tripping the generator off line compared with possible consequences of
loss of motor starting or fault clearing capability.
Figure 28: Field Ground Fault Illustration
17
Another failure mode is normally considered only for generators equipped with slip rings
and brushes. The rotating field (rotor) of the generator is insulated from the rotor lamina-
tions for a relatively low voltage. If the insulation should fail, or if brush dust should cause
tracking to the rotor shaft causing a single point ground in the field circuit, the excitation
system and the generator will generally continue to operate normally. If another ground
occurs in the field circuit, the result is likely to be arcing, overheating of the field conduc-
tors, and catastrophic damage to the rotor. Ground detecting relays are applied to these
machines to detect the first ground in the field and alarm or trip to prevent the occurrence
of a second ground in this circuit.
Figure 29: Field Ground Relay Protection Scheme
Of course brushless excited generators may also develop ground faults, but rotor ground
detection has not been used, perhaps due to the lack of a suitable relay to detect the
ground without adding some brushes and slip rings. Ground fault protection could be
applied to the exciter field that is stationary. Certainly ground faults can occur due to wind-
ing overheated or abrasion of the insulation, but this practice is not common.
18
Figure 30: Brushless Exciter with Fused Rectifiers
Unique to the brushless excited generator is the potential for failure of a rotating rectifier.
Because the rectifier is rotating, direct indication of the failure is not possible. The rotating
diodes may be fused to protect the exciter windings from the potentially high fault current
resulting from a diode failing shorted, or more commonly, fusing is not supplied with the
exciter. Thus, two problems may occur on the exciter. A diode may fail shorted without
fusing, requiring fast action to save the exciter from winding damage, or a diode may fail
open or have a fuse open. With the latter failure, the only danger to the exciter is the com-
pensating increase in field current that the AVR will supply in order to compensate for the
loss in excitation current to the generator field. This danger is slower to develop, however,
allowing us some time to detect the problem.
19
Figure 31: Typical Rectifier Diode Waveform
To protect the exciter, it is possible to detect the failure of a diode by monitoring the ripple
current in the exciter field. The harmonic content of the field current changes when a diode
fails, so we can monitor the harmonic content of the field, calibrating the relay for the
normal harmonic content when all diodes are working correctly. If a diode shorts, the
harmonic content changes substantially, and the relay can quickly provide a trip signal to
the generator. If a diode opens, the harmonic content change is more subtle, but using the
time available to verify that harmonic content is truly abnormal and not just a transient
condition, the exciter diode failure relay reliably detects the open diode and gives an alarm
or trip signal to the generator.
With the relaying methods reviewed above, the most common excitation system-related
problems can be detected, and effective action may be taken to address the problem with
actions to alarm, trip, transfer to manual, or transfer to backup AVR.
20
