GER-4004

Dynamic Testing of Generator Protection

Using a Model Generator Platform
 t h
56 Georgia Tech Protective Relaying Conference

DYNAMIC TESTING OF GENERATOR PROTECTION USING A

MODEL GENERATOR PLATFORM

Dale Finney Mark Adamiak Bogdan Kasztenny

[email protected] [email protected] [email protected]
(905) 201 2421 (610) 992 6967 (905) 201 2199

GE Power Management
215 Anderson Avenue
Markham, Ontario
Canada L6E 1B3

Atlanta, GA, May 1-3, 2002

 Dynamic Testing of Generator Protection Using a Model Generator Platform

1. Introduction
Modern, microprocessor-based generator protection relays integrate many functions into a
single package. Each protection element is designed to detect a specific abnormal condition in
the system and to initiate a particular tripping sequence. Static testing of individual functions in
many cases will not validate the ability of the entire package to respond correctly to different
system conditions. This is particularly true for generator protection.
EMTP-type simulators may not allow comprehensive dynamic testing due to limitations in
the models (simulation of true internal faults in machines and transformers, generation of natural
third harmonics by a generator, etc.). Therefore in some cases such as ungrounded distribution
networks, generator, motor and transformer protection, physical made-to-scale models of the
protected equipment, and actual field recordings are exceptionally useful.
This paper will demonstrate the advantages, methodology and sample results of dynamic re-
lay testing using a scaled model generator platform. Some of the of abnormal conditions applied
to the relay include:
Internal faults (true turn faults)
External faults with CT saturation
Loss of prime mover
Over/under excitation
Stable/unstable power swings
Inadvertent energization
Low frequency operation (static starting)
Loss of excitation
A new generator relay is tested on our unique physical model and state-of-the-art digital
simulator. Particular attention is paid to new algorithms, such as stator differential with extra
immunity to CT saturation conditions.
Sample test results are included.

2. Simulation and Testing Tools: Analog versus Digital

2.1. Generator Testing Needs
The model for in-depth testing of a modern generator protective relay should consist of the
following components:
A synchronous generator with electrical parameters consistent with those of a typical
high-voltage generator. The generator should produce a third harmonic component
with a voltage distribution that is typical of a high-voltage generator.
A prime mover with a control system capable of constant speed and/or constant real
power control and with matching gains and time constants.
An excitation system with a control system capable of constant voltage and/or con-
stant reactive power control. System gains and time constants should match the real
world equivalents.

Page 2 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

A high-voltage-interconnect system consisting of a generator grounding system, unit

transformer, circuit breaker, transmission line model, and equivalent grid system. The
transmission line model should allow for the simulation of both strong and weak sys-
tems.
The model should be capable of off-line and on-line operating modes. In on-line mode the
model should be capable of transferring real and reactive power to an equivalent grid system.
The model should allow for the application of phase-to-phase and phase-to-ground electrical
faults at the following points:
The neutral of the machine.
At various points on the stator winding.
At the machine terminals within the zone formed by the generator Current Transform-
ers (CTs).
At the machine terminals outside the zone formed by the generator CTs.
At various points on the transmission line model.
The model should allow for the application of the following abnormal conditions:
A loss of prime mover.
A loss of excitation.
Over/under-excitation.
Over/under-voltage.
Over/under-frequency.
An open phase or phase unbalance.
Accidental energization.
Generation rejection.
The model should permit for both wye and open delta relay Voltage Transformer (VT) con-
nections. In addition the model should allow for the simulation of a failure of one or more VT
fuses.
The Circuit Breaker (CB) should be controllable from the protective relay and the CB posi-
tion should be available as a status input to the relay under test.

2.2. Internal Faults

This is probably the single most important reason for building and using analog simulators. In
the case of rotating machines and power transformers, internal short circuits can occur between
different points of the same winding, between different windings, or between a winding and a
core, all within the complex magnetic core of the apparatus. In the case of rotating machinery, the
situation is even more complicated as part of the apparatus rotates.
Because modeling of such true inner faults is a complex task, testing for internal faults is
typically limited to faults at the machine/transformer terminals within the differential zone of
protection as outlined by the main CTs.
In the case of power transformers, several attempts have been made recently to model true
internal faults [1]. Some of the solutions have been incorporated into commercially available
simulation packages or as pre-processors to standard simulation tools such at ATP/EMTP.

Page 3 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

Some attempts at modeling true internal faults in rotating machines, generators in particular,
have been made as well, but the industry is far from seeing the models validated and incorporated
in standard simulation packages.
Analog simulation using made-to-scale machines solves the problem to a great extent. Typi-
cally an appropriate machine is selected (or build) and hand-wound. A number of artificial termi-
nals are created along the windings and brought out so that short-circuits involving those termi-
nals can be created.
The major problem with this approach is that the magnetic structure of the machine is altered
when making extra external connections to the taps along the windings. Arcing faults or faults
that truly evolve cannot be accurately modeled. With all the limitations, however, analog models
are the most realistic and accurate way to simulate internal faults in machines and transformers.
Such simulators, though, are quite expensive and are rarely available.

2.3. External Faults

Saturation of main and relay CTs and switch-off transients are the main concerns when test-
ing the stator differential protection.
Particular attention must be paid to long-lasting external faults of small current magnitude
but very long d.c. time constant. Small magnitude currents create small restraints. This, in turn,
places an operating point on a dual-slope differential characteristic in the region of lower slope,
where the relay is very vulnerable to misoperation. If a given generator relay is based on percent
differential characteristic alone, it may misoperate on external faults of low current magnitude
and long d.c. time constant.
Switch-off transients associated with clearing an external fault is yet another test condition to
be considered. An external fault saturating a CT to the extent of bringing the differential-
restraining point close to the operate / no-operate boundary needs to be simulated. Next, the fault
has to be cleared. As the algorithm filtered magnitudes ramp-down, an unfavorable transient may
be generated within the relay causing misoperation.
Given the testing requirements and the data / physical models available, the tester must
choose the most appropriate tool (digital or analog simulator) to meet the requirements.

2.4. Stator Ground Faults and Third Harmonic Methods

The amount of third harmonic and proportions between the third harmonic in the neutral
voltage and zero-sequence terminal voltage, as well as fundamental frequency components and
exported power during stator ground faults and normal conditions are described by quite complex
relations. In addition rotor asymmetry and natural noise in the signals may play significant role
for stator ground protection methods based on the third harmonic.
Because of the above reasons analog models are much more suitable for overall testing of
stator ground fault protection relays and algorithms.
Testing at off-nominal frequencies using an analog model may create a problem, though.
Keeping the third harmonic content in proportion requires fine-tuning depending on actual fre-
quency of a grounding resistor of the made-to-scale machine. At the same time, the response of
the relay shall be tested at ramping-up or down frequencies (in order to simulate ground faults
during machine start up).

Page 4 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

In general every method that is based on harmonics should be tested at off-nominal and
ramping frequencies in order to validate performance of the frequency tracking system of the re-
lay. If the frequency tracking does not work correctly, a certain amount of the fundamental fre-
quency signals leaks into measurements of the other harmonics and may cause misoperation.

2.5. Power Swings

Digital tools can be used for testing out of step tripping and power swing blocking functions.,
and a digital simulator can handle a variety of system configurations and fault conditions. How-
ever, when a response to internal faults under power swings and off-nominal frequencies is con-
sidered, analog models are more capable because they can handle true internal faults in machines.

2.6. Loss of Excitation

An analog model with actual field winding and associated d.c. source, controls, and decay
time constant proved to be a better way of testing the loss of field protection. This is particularly
true if one considers short circuits in the field winding as a potential source of loss of excitation.
An analog model interfaces better with actual sources of excitation and real controllers.
However, as long as short-circuits and similar abnormalities are left out of testing, digital closed-
loop simulators can be used for this purpose. However, the interdependency of the other machine
functions are only as accurate as the digital model.

2.7. Overexcitation
Overexcitation protection can be tested using both analog and digital models. An analog
model is typically limited to one particular magnetizing characteristic but most detection meth-
ods are based on a volts-per-hertz measurement, which is independent from the magnetic core
that is being protected.
An interaction between generator controls must be taken into account when testing the over-
excitation protection.
Attention must also be paid to the operating principle of the V/Hz element. The core being
protected from overheating may be one piece of iron (machine alone) or more (machine plus
transformer). The relay, though, measures and responds to three phase-to-ground or phase-to-
phase voltages. Different vendors treat the three-phase voltages differently. Phase-A voltage, av-
erage, or maximum voltage can be used. A global thermal history, or three per-phase thermal
counters may be implemented. With three counters, the average may be derived, or the first one
to reach the operate threshold would operate. Significantly different results can be obtained when
testing V/Hz relays under unbalanced conditions.
Response time of the V/Hz element should be approached with care as well. While the volt-
age (numerator) can increase abruptly causing a step change of the V/Hz ratio, the frequency (de-
nominator) cannot. Some relays apply certain amount of post-filtering to measured frequency in
order to bring more security while still covering the fastest possible frequency ramp (say
10Hz/sec). With post-filtering the relay would smooth out any step changes in frequency if such
changes are applied in laboratory conditions. If this is the case, the start mark for the testing
timer becomes fuzzy. Generally, the frequency shall be assumed constant: for any given fre-
quency and required V/Hz ratio, a corresponding voltage shall be calculated. When applying the

Page 5 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

calculated voltage, a well-defined start mark is created allowing for precise measurement of the
response time of the V/Hz element under test.

2.8. Overall Dynamic Testing

Overall dynamic testing should be the last step for building an appropriate comfort level for
new relay designs protecting a-few-hundred-MW generators. In the case of a rotating machine
there are many complex interactions between the electromagnetic and mechanical portions of the
system as well as between the machine, its excitation system and all the relevant control systems.
These interactions becomes even more complex once problems with the equipment itself, either
the generator, its controls, or the excitation system, are to be tested.
A microprocessor-based relay itself is a complex device as well. The relay would track fre-
quency, respond in a certain way to low-frequency currents and voltages (response of the input
magnetics), etc. This calls for a number of test scenarios that involve overlapping events that
need to be considered (stator ground fault following a VT fuse fail, during a ramping frequency,
for example). Typically, scenarios like that can be tested only on a complete made-to-scale ana-
log model.

3. Analog Generator Model

A power generator, being a rotating piece of equipment, has complex dynamic interactions
between the prime mover, generator controls, and the power system over a wide range of operat-
ing conditions. For completeness, dynamic testing of generator relays requires an accurate model
on which to exercise the relay over the ensemble of possible operating conditions.
Physical scale models of two 907MVA and a 144MVA cross compound generator and inter-
facing transformer and line model were developed in the early 1970s in a joint venture between
the MIT Electric Power Systems Engineering Laboratory and the Laboratory and the American
Electric Power Service Corporation [2]. The models were originally built due to concerns over
lack of field data in this area as well as a need to verify computer models used in large system
studies. The system has demonstrable characteristics and controls that are similar to actual field
systems. The 1444MVA model was acquired for the purposes of dynamic testing.
The 1444 MVA model was built to one-millionth scale and as such, each half of the cross
compound unit is rated for about 700VA with a 221V output rating. On a scaled basis, the model
is electrically correct carefully modeling the per-unit reactance, actual machine time con-
stants, and rotor inertia. Included in the overall control system are operational amplifier models
of a supercritical generator with three time constants and appropriate transfer functions and gains
to simulate governor dynamics and droop.
The machine is hand-wound and gives access to all stator terminals as well as access to in-
termediate points on the stator windings. For synchronized operation, the unit can be paralleled
with the local utility. As the design is physically robust, it can be subjected to a wide range of
faults and system operating scenarios with relative impunity. Various fault types, fault locations,
and fault initiation angles can be effected and consistently repeated. Since the full load currents
are about 1.8 A 2.5:5 current transformer were installed to bring the steady state current closer to
typical field conditions. Figure 1 shows a one line diagram of the test system as implemented.

Page 6 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

The system can test for virtually any generator abnormality, for example, Stator Differential,
Stator Ground, Loss of Excitation, Overexcitation, Over/Under Frequency, Reverse Power, etc.
One of the more uniquely modeled aspects of the systems is the excitation system and field
winding. On an actual generator, the L/R ratio of the field winding is quite large yielding a time
constant on the order of several seconds on Loss of Excitation. As the rotor of the models electri-
cally falls short of this number, equivalent time constants are achieved by electronically provid-
ing a negative resistance in the field supply. In the actual implementation, the time constant is
adjustable over a wide range of times. As such, Loss of Field protection performance can accu-
rately be tested for a range of machine time constants.
2.5:5 2.5:5

LINE LINE BUILDING

G 52
MODEL MODEL SERVICE

715 VA
221 V

240:120

G60

Figure 1. Diagram of the analog generator model.

Figure 2. D.C. motor (left) acting as a prime mover for the generator (right). Hand-wound generator brings
out taps along the windings for short-circuit tests.

4. Digital Generator Model

The Real Time Digital Simulator (RTDS) [3] has become a de-facto standard for closed-loop
testing of protection and control equipment.

Page 7 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

Figure 3 presents configuration of the simulator used for relay testing, while Figure 4 shows
photos of the physical equipment. Three sets of amplifiers are available for testing three-terminal
line applications. For generator testing, two sets are actually required in order to feed the relay
under test with neutral and terminal-side currents. The third set may sometimes be needed if an
overall generator-transformer differential protection is to be tested.
The main disadvantage of any off-the-shelf simulation package is lack of ability to simulate
true internal faults in electrical apparatus such as motors, generators or transformers as described
in section 2.
The RTDS is a general-purpose high-accuracy simulation system. As such it suits better cer-
tain generator testing needs. For example, the simulator is a better source of external fault cur-
rents as compared with analog simulator because it more easily allows different system configu-
rations, different parameters of CTs, and it is more repeatable (establishing residual magnetism in
the CT cores, for example), etc.
Some other tests, however, must be performed on a physical model. Stator ground fault pro-
tection is a good example. While the RTDS can be used as a very accurate and stable source of
the third harmonic for testing accuracy of the relay and quality of filtering, the physical model
needs to be used to test the overall performance of the protection system.

5. Generator Relay Under Test

5.1. Hardware Overview
The new protective relay [4] is built on a Universal Relay platform [5]. The platform allows
for modular hardware that can be ordered and configured to suit variety of application needs. In
particular, the relay can be configured with three sets of current inputs allowing applications that
involve current-based functions on the third set of currents, such as step-up transformer tertiary
winding monitoring or protection.
The user-selectable a.c. inputs are configured using the mechanism of sources. The relay may
be ordered with different combinations of CT and VT inputs. These inputs are configured indi-
vidually (ratios, rated secondary values, connection, etc.). Sources are next combined from
available phase and auxiliary voltages, and phase and ground currents. Protection and control
elements are subsequently configured to respond to selected sources. This mechanism is
equivalent to virtual a.c. wiring and becomes a very powerful configuration tool. In particular it
allows configuring, Time Overcurrent (TOC) protection or Distance Backup functions to respond
to neutral- or terminal-side currents as per user preferences. Moreover, the source setting for each
protection or control element is under multiple setting group control: a given TOC or other ele-
ment can be dynamically switched from terminal- to neutral-side set of currents depending on
user-programmable conditions.
Current and voltage signals are sampled at 64 samples per cycle. Second order Butterworth
filter is used for antialiasing.
Transducer modules are available that could be configured to respond to dcmA or resistive
temperature sensors for thermal protection and other protection and/or control functions.
Digital Inputs and Output Contacts are provided on a modular basis as well allowing config-
uring hardware components (a.c. inputs, digital inputs, contact outputs, transducers) per require-
ments of a given application.

Page 8 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

The relay supports multiple protocols over an Ethernet port and TCP/IP including TFTP,
MMS / UCA 2.0, etc. as well as an embedded web server. This allows cost-effective applica-
tions for distributed generation [6].

Figure 3. Architecture of the digital simulator.

Page 9 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

Figure 4. Digital simulator used in testing.

5.2. Phasor Estimation and Frequency Tracking

The relay samples currents and voltages at 64 s/c. Current signals are pre-filtered using an
optimized MIMIC filter for rejection of d.c. components as well as off-nominal frequency oscil-
latory signal distortions. The filter outperforms classical numerical MIMIC filters in terms of fil-
tering quality. A different pre-filter optimized for voltages and their distortions is applied to volt-
ages.
Current and voltage phasors (magnitudes and angles) are filtered using full- and half-cycle
Fourier filters, respectively. Third harmonic in the voltages required for stator ground fault
protection is estimated using the full-cycle Fourier working on raw voltage samples.
Frequency tracking algorithm is based on zero-crossing detection of a composite voltage sig-
nal. The selected signal is a Clarke transform equivalent to the positive-sequence voltage. Owing
to the Clarke transformation, the frequency is tracked correctly during VT fuse fail conditions
and other abnormal situations. The relay starts tracking frequency from 3 Hz using the voltage
signal as low as 3V secondary. The frequency tracking and metering algorithms are fast enough
to keep with a 10Hz/sec ramp in frequency. At the same time a robust post-filtering is applied
that guards the relay from frequency tracking errors resulting from phase shifts and other abnor-
malities.
True RMS values are also calculated based on raw samples and can be used as effective in-
puts to certain thermal protection elements. Symmetrical components, power and energy meter-
ing and many other functions are incorporated.

Page 10 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

5.3. Stator Differential Protection

Stator differential protection is based on a dual-slope dual break-point differential character-
istic. In addition, the element incorporates CT saturation detection. The saturation detector re-
sponds to any external fault even without actual CT saturation. The detector will not trigger on
any internal fault even if the CTs saturate. When saturation is declared, the relay would check
phase relation between the neutral- and terminal-side currents: if both currents are significant, the
angle between them must be less than 90 degrees to issue permission to trip. If only one current
is significant, permission is granted without checking the angle. This solution is based on bus dif-
ferential protection [7] and ensures superior performance on external faults saturating the main
(or relays) CTs.
The response time to internal faults is between half and full power system cycle.

5.4. Stator Ground Protection

Third harmonic neutral undervoltage, or differential third harmonic elements provide stator
ground fault protection. The latter compares third harmonic components in the neutral of the ma-
chine and in the zero-sequence voltage at the terminals of the machine. Power supervision (win-
dow) is incorporated allowing sensitive settings on machines that have their third harmonic con-
tent varying significantly as the exported power changes.

5.5. Other Function

A standard set of generator protection features has been implemented. This includes acci-
dental energization, loss of excitation, current unbalance, overexcitation, etc. These functions
have been tested as described below.

6. Sample Test Results

6.1. Stator Phase Faults
The model generator is placed on-line. An internal phase-to-phase fault is applied at various
locations on the stator. The machine load is also varied. The operating time of the relay is meas-
ured. The location of the fault is specified from the neutral of the machine and is recorded as a
percentage of the total stator impedance (i.e. in test A: 14% of winding A to 6% of winding B).

Table 1. Sample Test Results for Stator Phase Fault Protection.

Test Faulted Phases Location Load (W) Operate Time (ms)
A a-b 14%-6% 300 8

B b-c 6%-29% 400 11

C c-a 29%-14% 500 9

D a-b 50%-50% 500 9

E b-c 50%-50% 400 9

F c-a 50%-50% 300 11

G a-b 100%-100% 300 9

Page 11 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

H b-c 100%-100% 300 9

I c-a 100%-100% 300 10

Figure 5 presents an internal fault example showing a sub-cycle operating time of the stator
differential protection element.

80

72

64

56

48

40

32

24
Secondary Current (amps)

16

16

24

32

40

48

56

64

72

80
0.8 0.805 0.81 0.815 0.82 0.825 0.83 0.835 0.84 0.845 0.85 0.855 0.86 0.865 0.87 0.875 0.88 0.885 0.89 0.895 0.9
Time (sec.)
IaN
IaS
G60 Trip
Figure 5. Sample stator phase fault.

Page 12 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

6.2. Stator Ground Faults

In this test, the model generator is placed on-line. A ground fault is applied at various points
on the stator of the machine. The operation of the fundamental overvoltage (59N) and third har-
monic undervoltage (27TN) elements are verified.

Table 2. Relay Response to Stator Ground Faults.

Test Faulted Phase Location Operate Time (s) Elements Operated

A A 14% 1.0 59N, 27TN

B B 6.4% 1.0 59N, 27TN,

C C 29% 1.0 59N

The model generator is placed on-line. A fault is applied at the neutral point (100% stator
ground) at various levels of power. Pickup is set just below the available 3rd harmonic voltage.
The power-blocking window is set at 140-210 watts.

Table 3. Sample Test Results for Stator Ground Fault Protection.

Test Pickup Power (W) Operate Time(s) Elements Operated

A 0.09 75 5.0 27TN

B 0.09 175 NA None

C 0.09 245 5.0 27TN

6.3. Loss of Excitation

The model generator is placed on-line. In tests A-E, the excitation is switched off. The load-
ing of the machine will determine the path of the impedance locus into the relay characteristic re-
sulting in a zone 1 or zone 2 operation (see Figure 6 for an example). In test F, the AVR is
switched to manual and the excitation is reduced until an operation occurs.

Table 4. Sample Test Results for Loss of Excitation Protection.

Test Generator Load Zone Operate Time (s)

A 100 2 14.121
B 200 2 8.41
C 300 2 6.577
D 400 1 4.934
E 500 1 3.673
F 300 2 17.59

Page 13 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

60

54

48

42

36

30

24

18

12

6
X

12

18

24

30

36

42

48

54

60
60 54 48 42 36 30 24 18 12 6 0 6 12 18 24 30 36 42 48 54 60
R
Impedance Locus
Zone 1
.
Zone 2
.
Figure 6. Sample impedance locus during loss of excitation conditions.

6.4. Current Unbalance

In this test the model generator is placed online. Placing an open-phase at a point between the
generator and the system creates an unbalance. As the open phase is moved closer to the genera-
tor the percentage of the negative-sequence current (I2) increases accordingly. The reset column
specifies the time elapsed from the previous fault. For reset times less than 240 seconds, the ele-
ment will maintain thermal memory and the tripping time will be reduced accordingly.

Page 14 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

Table 5. Sample Test Results for Current Unbalance Protection.

Test I2 (pu) Reset Pickup TDM Operate Time (s)

A 0.1 240.0 0.068 21.63 984.0

B 0.08 240.0 0.068 2.163 150.0

C 0.08 58.0 0.068 2.163 35.8

D 0.04 240.0 0.025 15.38 678.0

E 0.04 123.0 0.025 15.38 345.0

F 0.138 240.0 0.025 15.38 56.0

G 0.138 178.0 0.025 15.38 42.0

6.5. Over-excitation
In this test the model generator was placed off-line. Alternately increasing the system voltage
or decreasing the system frequency creates an over-excitation condition. The reset column speci-
fies the time elapsed from the previous fault. For reset times less than 30 seconds, the element
will maintain thermal memory and the tripping time will be reduced accordingly.

Table 6. Sample Test Results for Over-Excitation Protection.

Test Voltage Frequency Reset Operate Time (s)

A 70 40 30 13.2

B 70 30 15 13.8

C 80 60 30 114

D 90 60 7.5 5.4

E 100 60 30 10.3

6.6. Abnormal Frequency

The model generator is placed off-line and the frequency is alternately increased and de-
creased from nominal.

Table 7. Sample Test Results for Abnormal Frequency Protection.

Test Frequency (Hz) Operate Time (s)

A 56 0.115

B 57 0.50

C 58 10.0

D 59 50.0

E 61 50.0

F 62 10.0

Page 15 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

G 63 0.50

H 64 0.110

6.7. Abnormal Voltage

The model generator is placed off-line and the voltage is alternately increased and decreased
from nominal. The overvoltage element has a definite time characteristic. The undervoltage ele-
ment has an inverse time characteristic. Correct operation of these elements is verified.

Table 8. Sample Test Results for Abnormal Voltage Protection.

Test Voltage (V) Operate Time (s)

A 85 30.1

B 50 105

C 45 53.6

D 40 36.2

E 35 27.3

F 30 21.8

G 25 18.2

6.8. External Faults with CT Saturation

Immunity of stator differential protection with respect to CT saturation and switch-off tran-
sients, as well as the combination of the two, was tested.
The relay performed exceptionally well due to its dedicated CT saturation and directional
principles. Figure 7 illustrates a typical sequence of events under deep CT saturation: an external
fault is detected first and signaled by setting the STATOR DIFF SAT flag; directions of the neu-
tral and terminal currents are sensed as opposite and the directional principle resets its permission
to trip by resetting the STATOR DIFF DIR flag; when the CTs start to saturate, differential cur-
rent appears jeopardizing security of the relay (the STATOR DIFF PKP flag is set meaning the
operating characteristic is entered). A typical differential protection (without any dedicated
countermeasures) would misoperate at this point. The relay under test does not misoperate owing
to its operating logic (OPERATE = PKP AND (DIR OR NOT(SAT)).

7. Conclusions
Generator protective relays are one of the most critical protection applications. A relay that
could inadvertently reject, say 800MW generation, can cause a lot of damage to the power sys-
tem. Criticality of generator protection could be only compared to protection of very critical
lines, EHV busbars, and large step-up transformers.
At the same time a generator relay includes a number of protection functions. These protec-
tion elements can be tested individually using a 50/60Hz injection from a test set.

Page 16 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

It is important, however, to validate new relay designs by testing complete relaying applica-
tion dynamically either on an analog made-to-scale machine or a digital simulator. Only in this
way complex interactions within the machine and its control system, and within the protective
relay could be fully tested in a variety of operating conditions.
Comprehensive dynamic testing is also the best environment for comparative testing between
legacy devices and alternatives from various vendors.
This paper discusses major test requirements for generator relay testing and presents sample
test results for a new generator protective relay. The new solution has been tested using both a
unique analog simulator and state-of-the-art digital simulator.

T h e e le m e n t
does not
o p e ra te

T h e d ire c tio n a l fla g

d ro p s o u t

T h e s ta to r d iffe re n tia l
p ro te c tio n e le m e n t p ic k s
u p d u e to C T s a tu ra tio n

T h e s a tu ra tio n fla g R elay B operates

is s e t s a fe ly b e fo re
th e p ic k u p fla g

Figure 7. Sample external fault case with CT saturation caused by extremely long d.c. time constant.
The relay under test is secure (STATOR DIFF OP flag). Typical protection tested in parallel misoperates
(RELAY B TRIP ON flag).

Page 17 of 18
 Dynamic Testing of Generator Protection Using a Model Generator Platform

8. References
[1] KASZTENNY B., KEZUNOVIC M., GALIJASEVIC Z., WILLIAMS D., A New ATP Add-On
For Modeling Internal Faults In Power Transformers, Proceedings of the 62nd Annual American
Power Conference, Chicago, April 10-12, 2000.
[2] ADAMIAK M., DAS DHURBA, GARDELL J., PATEL S., VIERS D., Performance Assess-
ment of a New Digital Subsystem for Generator Protection, Proceedings of the 20th Annual
Western Protective Relay Conference, Spokane, WA, October 19-21, 1993.
[3] Real Time Digital Simulator. Instruction Manual, RTDS Inc. Publication, 1995.
[4] G60 Generator Management Relay. Instruction Manual, General Electric Publication No.
GEK-106228B, Markham, Canada, 2001.
[5] POZZUOLI M.: Meeting The Challenges Of The New Millennium: The Universal Relay, Pro-
ceedings of the 52nd Annual Conference for Protective Relay Engineers, College Station, Texas,
April 5-8, 1999.
[6] FINNEY D., KASZTENNY B., ADAMIAK M., Generator Protection Needs in a DG Environ-
ment, Proceedings of The Power Systems Conference 2002: Impact Of Distributed Generation,
March 13-15, 2002, Clemson University, Clemson, SC.
[7] KASZTENNY B., SEVOV L., BRUNELLO G., Digital Low-Impedance Busbar Protection
Review of Principles and Approaches, Proceedings of the 54th Annual Conference for
Protective Relay Engineers, College Station, TX, April 3-5, 2001. Also presented at the 55th
Annual Georgia Tech Protective Relaying, Atlanta, GA, May 2-5, 2001.

Dale Finney received his Bachelor of Engineering degree from Lakehead University in 1988. Dale began
his career with Ontario Hydro as a Protection & Control Engineer. Currently, Mr. Finney is employed as
an Applications Engineer with General Electric Power Management and is a member of the IEEE.
Mark Adamiak received his Bachelor of Science and Master of Engineering degrees from Cornell Uni-
versity in Electrical Engineering and an MS-EE degree from the Polytechnic Institute of New York. From
1976 through 1990, Mark worked for American Electric Power (AEP) in the System Protection and
Control section where his assignments included R&D in Digital Protection and Control, relay and fault
analysis, and system responsibility for Power Line Carrier and Fault Recorders. In 1990, Mark joined
General Electric where his activities have ranged from algorithm development, simulator design, substa-
tion automation, and advanced technology development. In addition, Mr. Adamiak has been actively in-
volved in developing the framework for the implementation of the MMS/Ethernet peer-to-peer communi-
cation solutions for next generation relay communications. In 1986, Mark was the winner of the Eta
Kappa Nu (HKN) society's "Outstanding Young Electrical Engineer" award. Mark is a member of HKN,
a Senior Member of IEEE, past Chairman of the IEEE Relay Communication Sub Committee, a US
member on IEC TC57 - Working Group 11 on Substation Communication, and a registered Professional
Engineer in the State of Ohio.
Bogdan Kasztenny received his M.Sc. and Ph.D. degrees from the Wroclaw University of Technology
(WUT), Poland. He joined the Department of Electrical Engineering of WUT after his graduation. Later
he was with the Southern Illinois and Texas A&M Universities. Currently, Dr. Kasztenny works for GE
Power Management as a Chief Application Engineer and a Project Leader for the Universal Relay family
of protective relays. Bogdan is a Senior Member of IEEE, has published more than 100 papers and holds
several patents on protection and control.

Page 18 of 18