Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

1

A Novel Flux- Based Method for Synchronous Generator Loss of


Excitation Protection

Hamid Yaghobi
(1, 2)
, Hashem Mortazavi
(2)
, Kourosh.Ansari
(1)
, Habib Rajabi Mashhadi
(1)
,
Member, IEEE, and Hassan Khorashadi zadeh
(3)
, Member, IEEE, and Hossein Borzoe
(2)

1-Department of Electrical Engineering, Ferdowsi University of Mashhad, Mashhad, Iran
2- Khorasan Regional Electric Company (KREC), Iran
3- Siemens Energy Inc, USA.


Keywords: Synchronous generator, loss of excitation protection, FEM, linkage flux, search coil




Abstract
Synchronous generator Loss of Excitation
(LOE) if dose not detected rapidly; not only
has a harmful affect on the generator, but also
has a devastating impact on power system
stability. This paper presents a novel method
for detecting Loss of Excitation in
synchronous generator based on change in the
magnetic flux linkage. Generator air gap flux
linkage is measured via search coils sensor
installed under the stator wedges. The search-
coil sensor is widely used in electrical
machine monitoring in recent years.
Theoretical approach based on Finite Element
Method (FEM) together with experimental
results derived from a 4-pole, 380V, 1500
rpm, 50 Hz, 50 KVA, 3-phase salient-pole
synchronous generator confirm the validity of
the proposed method. Extensive simulations
show that the proposed method has various
distinct features that are advantageous over
existing schemes. Speed and sensitivity are
the main points featured in the proposed
method. The effectiveness of the proposed
protection is demonstrated by comparing with
conventional LOE protection.

INTRODUCTION
The generators are the main and significant
parts of power system and their failure can
cause severe damage on the machine,
interruption of electrical supply, and
consequently economic loss.
A synchronous generator requires sufficient
DC voltage and current to its field winding to
retain synchronism with a power system. As a
rule, the generator field is adjusted so that
reactive power as well as real power is
delivered to power system. The excitation
may be completely or partially lost due to
accidental tripping the field breaker, field
open circuit, flashover of the slip rings,
regulator failure or loss of supply to the
excitation system. The behavior of a generator
in any case of field loss will depend on its
initial loading, machine and system
characteristics and the governor. If the field
loss occurs at full load the stator currents may
be as high as twice the full load current and
high currents may be induced in the rotor. An
overheating of the generator may occur
quickly under these circumstances. The
condition must be rapidly distinguished and to
10-E-CAP-1278

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
2

evade generator damage, the generator must
be isolated from the system. A LOE condition
which is not detected can also have a
demolishing impact on the power system by
causing both a loss of reactive power support
as well as creating a significant reactive
power drain. On large generators this
condition can cause an area wide system
voltage collapse [1], [2].
In 1949, a single phase offset Mho relay was
introduced for the fast detection of LOE in
synchronous generators [3]. The security for
LOE has since been a great concern to many
authors and utility under takings. The concern
was about the possible maloperation of the
relay during stable power swing under certain
conditions [1]. The conventional LOE
documented solution to this problem is to
intentionally delay the operation of LOE
relay. The method presented in [4] proposed
an adaptive LOE relay, augmented by suitably
selected time derivatives of impedance seen
by the relay, to ensure correct discrimination
between genuine loss of excitation and power
swing. Providing time delay to avoid false
operation on stable power swing is not
considered as an ideal solution and there
continues some user apprehension about the
relay performance [5]. This deliberately time
delay increases the operating time of the relay
which means that the MVARS drawn by the
generator persist for a longer time, making the
power system more vulnerable to severe
voltage dips. In [6], the authors present
evaluation of the performance of the adaptive
LOE protection based on the rate of change of
the apparent reactance accomplished by
considering the well known positive offset
method. In [7], on the base of simulation of
loss of excitation, the method of LOE
protection for generator based on artificial
neural networks is proposed. The presented
method can be considered as an accelerated
tripping factor and in conjunction with other
criteria, not as a sole LOE protective relay.
Most of previous work about the LOE
protection have been concentrated on
enhancing the security of offset Mho relay
and evaluating the effects of the power system
disturbances on these relays [1][5]. The
other new papers present the importance of
proper coordination of LOE relay with under
excitation limiter and generator capability
curve [8][10]. The need to improve
coordination between generator protection
and control has come to light after recent
misoperation of generator protection during
major system disturbances. Two significant
disturbances are the 1996 western area
disturbances and 2003 east coast blackout.
The method presented in [11] gives a new
algorithm based on measuring of the 3-phase
reactive power output of the machine, and
monitors the direction and magnitude of
leading reactive power at the generator output
terminal. This algorithm does not response
satisfactorily to partial loss of field conditions
as long as the AVR can increase the field
current enough to correct the situation.
On the other hand, fault monitoring of
rotating electrical machines using magnetic
flux measurements have been proposed as a
topical subject of investigation [12]- [18]. In
addition to above mentioned studies, in this
paper, it will be shown that generator air gap
flux is a useful variable for accurate detection
of LOE condition.
The objective of this paper is to present a
novel method for LOE protection based on
change in the magnetic flux linkage of the
synchronous generator. The method consists
in obtaining the magnetic flux linkage via a
search coil sensor. It will be shown that flux
based LOE protection is not only more
sensitive but also is faster than conventional
LOE relay. The main idea of proposed
method is validated by experimental results
and FEM analysis. The performance of the
proposed method has been studied for a
single-machine infinite bus system using
MATLAB software.

SYNCHRONOUS GENERATOR LOE
PROTECTION
When a loaded generator loses its field
current, the generator becomes an induction
generator until the prime mover is removed
after which it may become an induction motor

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
3

[1]. In a LOE condition, the generator
operates in the under excited region of its
capability curve. Such a LOE causes the
generator to accelerate rapidly because the
power into the generator from prime mover
can only now be transformed to energy stored
in the rotors angular velocity. Without
removing the power input from the prime
mover, the generator will soon be self-
destructed. The effects of such a LOE on the
power system includes: loss of reactive
support, considerable reactive drain, locale
voltage collapse, power swings, extreme
reactive power flow and probably
transmission line tripping.

Fig.1. Loss of field R-X diagram scheme 1 [10].

The most broadly applied method for
detecting a generator LOE situation is the use
of distance relays to sense the deviation of
impedance as viewed from the generator
terminals. This distance relay approach used
as a combination of exciter-initiated
protection (e.g. for inadvertent field breaker
opening) and mho-type relays. The
impedance relays may be combined with
directional units and under voltage elements
to reduce the likelihood of misoperation
during system disturbances.
There are two basic designs of LOE
protection. The first technique (scheme 1
Fig. 1) consists of two offset Mho units. An
impedance circle diameter equal to the
generator synchronous reactance and offset
downward by
2
1
of the generator transient
reactance is used for the Zone 2 distance
element. The offset of the relay characteristic
prevents operation of the protection for
external three-phase faults close to the
generator terminals. The operation of this
element is delayed approximately 30-45
cycles to prevent misoperation during a stable
transient swing. A second relay zone, set at an
impedance diameter of 1.0 per unit (on the
generator base), with the same offset of
2
1
of
the generator transient reactance is used also.
This Zone 1 element has a few cycles of delay
and more quickly detects severe under
excitation conditions. When synchronous
reactance is less than or equal to 1.0 per unit
(e.g. hydro generators) only the Zone 2 is
used and is set with the diameter equal to 1.0
per unit.

Fig.2. Loss of field R-X diagram scheme 2 [10].

The second relaying method (scheme 2 Fig.
2) consists of an under voltage unit, an
impedance unit and a directional unit. In this
case the generator synchronous and transient
reactances are used to determine the settings.
As with the first scheme, two elements are
used, one without significant delay (typically
0.25 second for the most severe condition)
and the other delayed to prevent misoperation.
For both schemes the relay settings are based
on CT and VT secondary quantities, thus the

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
4

impedances need to be calculated on the CT
and VT secondary basis.
Another method of LOE protection is offered
by Siemens in static and digital generator
protection [19]. Positive sequence admittance
seen by the relay is used in the numerical
algorithm to determine the synchronizing
power of the machine. The operating
characteristic in the admittance plane is a
combination of straight lines as shown in
Fig.3.
Fig.3. Under excitation characteristic in the
admittance plane [19].

The characteristic is adapted to the generator
capability curve. The protection trips
immediately if the admittance moves to the
left beyond line 3 which means that stability
has been lost permanently. An alarm or
several seconds delayed trip is initiated when
the impedance has crossed only line 2 of the
characteristic. The protection is also set to
operate below 60% to 70% of the no-load
excitation voltage and upon malfunction of
the voltage regulator or failure of the
excitation voltage; the machine can be
disconnected after a short delay e.g. 1.5
seconds.

Conventional LOE protection problems
The distinguish matter between LOE and
transient power swing during major system
disturbances, receives as one of the biggest
challenge of LOE protection relay [20], [21].
The accepted solution for this dilemma is
making deliberately delay in the operation of
LOE relay. LOE response time includes two
time delays. The first is the inherent time
delay of power system in response to loss of
field condition. The permissible time before
damage can occur in these protection
depends on the type of machine, type of
excitation loss, turbine governor
characteristics and system conditions. The
time can be as short as 10 seconds or as
long as many minutes [22]. In these
circumstances, the generator may be able to
run for several minutes without requiring to
be tripped. There may be sufficient time for
remedial action to restore the excitation, but
the reactive power demand of the machine
during the failure may severely depress the
power system voltage to an unacceptable
level. Second time delay is related to
protective relay detection speed and designer
setting. This deliberately time delay increases
the operating time of the relay which means
that the MVARS drawn by the generator
persist for a longer time, making the power
system more vulnerable to severe voltage
dips. In addition, for operation at high initial
power output, the rotor speed can rise to about
105% of rated speed, where there would be
low power output and where a high reactive
current of up to 2.0p.u. may be drawn from
the supply. Rapid automatic disconnection is
then required to protect the stator windings
from high current and to protect the rotor
from damage caused by induced slip
frequency currents. Although the detection
method upgraded to modern two-zone mho
relays to enhance protection, the relay time
delay operation is persisting.
The variation of machine impedance with the
rate of slip is another matter of concern [21]
on the sensitivity and security of LOE relay
operation. The effect of this variation during
asynchronous operation is to cause generator
reactance varying at slip speed. In
consequence, the loss of excitation impedance
locus does not settle at a single point, but it
continues to describe a small orbit about a
mean point. A protection scheme for loss of
excitation must operate decisively for this
condition, but its characteristic must not
inhibit stable operation of the generator.
Another matter of concern is the performance
of AVR when generator absorbs VARs (under
excitation mode) [20]. There is an

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
5

apprehension that the regulator will
undershoot while trying to maintain the
limit and thereby cause a momentary
excursion of the apparent impedance into the
relay characteristics. The last but not the least
dilemma of LOE relay is its setting and
coordinating with under excitation limiters of
AVR and generator capability curve.
Transient stability analysis for each power
plant recommended for proper setting of LOE
relay [10].

PROPOSED TECHNIQUE
Impedance method uses of voltage and
current as relay inputs. Because CTs and PTs
are usually installed in the machine for
different purposes as default, the protection
schemes that depend on the study of these
inputs are mostly being notified. Unlike the
impedance methods, the proposed technique
use flux linkage as an input for protection. In
this technique, search coil is used for
measuring the air gap flux linkage from inside
the machine. The main idea of this paper was
analyzed by FEM modeling of a 4-pole, 1500
rpm, 50 Hz, salient-pole synchronous
generator. Then measurement of the machine
flux linkage is explained in section III.B. In
section IV, for an isolated generator,
experimental results on the above mentioned
machine are presented. Then by simulation
with MATLAB software, it will be shown
that the presented method is also valid for a
synchronous generator connected to power
system. The comparison of proposed method
with conventional protection presented in
section V. Finally, Section VI concludes the
paper.

Modeling of LOE with 3-D Finite-Element
In many applications, a 2-D finite element
analysis gives quite precise prediction of the
magnetic field distribution and machine
performance. Some limitations of 2-D
analysis imply that the use of 3-D analysis is
inevitable if an accurate analysis is needed.
These limitations are presented in [23]. In this
research, the magnetic field distribution is
calculated by 3D-FEM using ANSYS
workbench software.



Fig. 4. FEM model for the analysis of salient pole
synchronous generator.

Salient-pole synchronous generator is used to
investigate different kinds of machine faults.
Excluding loss of excitation, some other faults
such as stator turn-to-turn short circuit have
been investigated [24]. The nonlinear
relationship between the flux linkage with the
excitation current and phase current has made
the modeling and analysis of salient-pole
synchronous generator very complicated. In
order to improve the overall accuracy,
saliency and slotting, magnetic saturation
effects, and the position of the winding inside
the stator are included in the FEM modeling.
Slotting and the position of the winding can
be seen in Fig.4. In order to prevent
confusion, only excitation and phase A
windings are shown in Fig.4. Also, the effect
of magnetic saturation was included in the
FEM software by using B-H curves of the
rotor and stator cores.
Although the complete model of salient pole
synchronous generator was considered for
different normal and abnormal operation
situation for both experimental and simulation
analysis, this paper concentrated on loss of
excitation condition.

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
6

Fig. 5. Distribution of magnetic flux density in
salient-pole synchronous generator under healthy
operation (front view).

Figs.5 and 6 show the typical magnetic flux
distribution under normal and loss of
excitation of studied generator as obtained by
the FEM. In a salient pole synchronous
machine, the air gap is nonuniform and the
reluctance of the magnetic flux path is a
function of time. Therefore the magnetic
fluxes along the d-axis and the q-axis are
varying. Under d-axis (under poles), the flux
linkage between the stator and rotor windings
is MAX due to the minimum reluctance
between the stator and rotor in these areas.
Under q-axis, the flux linkage between the
stator and rotor windings is MIN due to the
large air gap between the stator and rotor in
these areas.
In Fig.5 load current is 20% of the rated load.
In this figure magnetic flux distribution varies
from 0.177 Tesla (q-axis) to 1.595 Tesla (d-
axis). Fig.6 illustrates flux distribution in
LOE condition. In this figure magnetic flux
distribution varies from 0.008 Tesla to 0.07
Tesla. According to these figures, when LOE
occurs, the machine flux linkage declines to
near zero. Therefore, magnetic flux
distribution could be applied for detection of
LOE in synchronous generators.
In fact, in electrical machines air-gap
magnetic field distribution in no-load and on-
load performance under healthy conditions is
symmetrical (neglecting the insignificant
inherent asymmetry in the magnetic field
distribution due to the differences on
mechanical structures). When LOE occurs,
the magnetic field distribution of machine is
still symmetrical but its magnitude decreases
to near zero.

Fig. 6. Distribution of magnetic flux density in
salient-pole synchronous generator under loss of
excitation (front view).

Measurement of the Machine Flux Linkage
with Search Coil
The search coil sensor (see Fig.9) is widely
used for flux measurement in electrical
machine monitoring in recent years [13], [25].
The theory behind the search-coil senor is
Faradays law of induction. By Faraday's law,
the voltage induced in single turn search coil
wound round a stator tooth is
(1)
L B v e
ind
). ( =
Where
ind
e is the effective value of voltage
induced by the air-gap flux, B is the air-gap
flux, v is the component of the wire velocity
perpendicular to the direction of the magnetic
flux density, andL is the effective length of
the search coil.
If the magnetic flux through a coiled
conductor alters, a voltage proportional to the
rate of change of the flux is generated
between its leads. The flux through the coil
will change if the coil is in a magnetic field
that varies with time. The signal detected by a
search-coil sensor depends on the
permeability of the area of the coil, the core
material, the number of turns, and the rate of
change of the magnetic flux through the coil
[26].
This type of sensor is primarily used in harsh
environments where high reliability sensing

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
7

can be afforded [26]. Also, this sensor has
more advantages such as insensitivity to
external conditions (humidity, temperature,
etc.), no mechanical disassembling /
reassembling, and easy to remove. The
disadvantage of this sensor is that these coils
have to be placed in the stator slots by the
manufacturer or have to be placed in overhaul
of the machine.
These inductive sensors (search coils) observe
the real distribution of the air-gap flux
density. This indicates that all damping
effects coming from the saturation and the
parallel current branches of the stator winding
are taken into account. This sensor is readily
accessible on the market and its cost is very
low comparing with capacitive air-gap
monitoring system. Easiness and low price of
the used inductive sensors allow the
installation of a high number of sensors [27].

EXPERIMENTAL RESULTS
To demonstrate the performance of presented
protection via search coil sensor, a series of
actual LOE tests to a salient-pole synchronous
generator have been fulfilled. It will be shown
that for an isolated synchronous generator
operation, when LOE happened the induced
voltage in search coil (which is completely
proportional to air gap flux) decline to zero.

Experimental Test Setup, and Numerical
Measurement
The machine used in this study is a 50 KVA,
380V, 4-pole, 1500 rpm, 50 Hz, salient-pole
synchronous generator, and machine having
48 stator slots. The stator of this generator has
a 3-phase, one layer, lap winding, and four
parallel branches in each phase. The structure
of the testing laboratory and experimental test
setup is shown in Fig.7. It consists of a
synchronous generator connected to a three-
phase load. This generator has no damper
winding, and is driven by an induction
machine.

Fig. 7. Testing laboratory and experimental test
setup.

In the experiments performed in this work,
search coils and designed electronic-
microcontroller board are used for measuring
the flux linkage from inside the machine.
Electronic-microcontroller board has one
master and forty eight slaves. Electric
diagram of the board is shown in Fig. 8,
where two slaves (of 48 slaves) of this system
can be observed.
Two-wire Serial Interface (TWI) is used in
this system. The TWI protocol allows the
system designer to interconnect up to 128
different devices using only two bi-directional
bus lines, one for clock (SCL) and one for
data (SDA). All devices connected to the bus
have individual addresses.
Sampled voltage data taken from search coils
are directly proportional to the rate of change
of the flux. Forty eight single turn search coils
were installed along the grooves of the stator
teeth with a pitch of
o
15 with the aim of being
able to determine how the flux linkage
distribution in the generator changes when an
abnormal operational condition is present.
Schematic and experiment view of these
search coils are illustrated in Fig.9.
Serial port interface has been used for
connecting electronic-microcontroller board
to computer. The induced voltages in the
search coils are communicated to serial port
of the computer by means of electronic-
microcontroller board. This structure is shown
in Figs.7 and 10.


A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
8


PA0/ADC0 40
PA1/ADC1 39
PA2/ADC2 38
PA3/ADC3 37
PA4/ADC4 36
PA5/ADC5 35
PA6/ADC6 34
PB0/XCK/T0 1
PB1/T1 2
PB2/INT2/AIN0 3
PB3/OC0/AIN1 4
PB4/SS 5
PB5/MOSI 6
PB6/MISO 7
PB7/SCK 8
PA7/ADC7 33
RESET 9
XTAL1 13
XTAL2 12
PC0/SCL 22
PC1/SDA 23
PC2/TCK 24
PC3/TMS 25
PC4/TDO 26
PC5/TDI 27
PC6/TOSC1 28
PC7/TOSC2 29
PD0/RXD 14
PD1/TXD 15
PD2/INT0 16
PD3/INT1 17
PD4/OC1B 18
PD5/OC1A 19
PD6/ICP 20
PD7/OC2 21
AVCC 30
AREF 32
MASTER
ATMEGA16
PA0/ADC0 40
PA1/ADC1 39
PA2/ADC2 38
PA3/ADC3 37
PA4/ADC4 36
PA5/ADC5 35
PA6/ADC6 34
PB0/XCK/T0 1
PB1/T1 2
PB2/INT2/AIN0 3
PB3/OC0/AIN1 4
PB4/SS 5
PB5/MOSI 6
PB6/MISO 7
PB7/SCK 8
PA7/ADC7 33
RESET 9
XTAL1 13
XTAL2 12
PC0/SCL 22
PC1/SDA 23
PC2/TCK 24
PC3/TMS 25
PC4/TDO 26
PC5/TDI 27
PC6/TOSC1 28
PC7/TOSC2 29
PD0/RXD 14
PD1/TXD 15
PD2/INT0 16
PD3/INT1 17
PD4/OC1B 18
PD5/OC1A 19
PD6/ICP 20
PD7/OC2 21
AVCC 30
AREF 32
SLAVE1
ATMEGA16
PA0/ADC0 40
PA1/ADC1 39
PA2/ADC2 38
PA3/ADC3 37
PA4/ADC4 36
PA5/ADC5 35
PA6/ADC6 34
PB0/XCK/T0 1
PB1/T1 2
PB2/INT2/AIN0 3
PB3/OC0/AIN1 4
PB4/SS 5
PB5/MOSI 6
PB6/MISO 7
PB7/SCK 8
PA7/ADC7 33
RESET 9
XTAL1 13
XTAL2 12
PC0/SCL 22
PC1/SDA 23
PC2/TCK 24
PC3/TMS 25
PC4/TDO 26
PC5/TDI 27
PC6/TOSC1 28
PC7/TOSC2 29
PD0/RXD 14
PD1/TXD 15
PD2/INT0 16
PD3/INT1 17
PD4/OC1B 18
PD5/OC1A 19
PD6/ICP 20
PD7/OC2 21
AVCC 30
AREF 32
SLAVE2
ATMEGA16
D
7
14
D
6
13
D
5
12
D
4
11
D
3
10
D
2
9
D
1
8
D
0
7
E
6
RW
5
R
S
4
V
SS
1
V
D
D
2
V
EE
3
LCD1
LM016L
R1
10k
R2
10k
R3
10k
R4
10k
R5
10k
R7
10k
RE-READ-2
R9
10k
VCC-10
GND-11
GND-31
VCC-10
GND-11
GND31
VCC-10
GND-11
GND-31
POT-10K
1
2
3
U3:A
4093
1
2
5V-1W
zener
R8
1Meg
1
2
5V-1W
zener
R10
1Meg
5-VOLT
5-VOLT
1
2
3
U1:A
4093
SENSOR
10-VOLT
TERMINAL
TERMINAL
C3
10u-50V
C4
100n
10U
10U
C5
100n
C6
100n
C7
100n
C2
10u-50V
10U
C8
100n
C9
100n
C10
10u-50V
5-VOLT
5-VOLT
C12
4.7U-16V
R12
100
R13
100
C1
4.7U-16V
LED
DIODE
6
5
4
1
2
P521-XX817
OPTOCOUPLER
5-VOLT
R6
1k
R11
10K
Fig. 8. Electric diagram of the electronic-microcontroller
designed board to obtain the flux linkage.


(b) (a)
Fig. 9. (a) Schematic view of search coil. (b) Search
coils placed along the stator teeth in experiment.



Fig.10. Waveform samples are transferred to a
standard personal computer (PC).

The voltage induced in search coils was
sampled by electronic- microcontroller board.
In other words, electronic-microcontroller
board has been designed for measuring
simultaneously sampling data of the flux
linkage in cross-section of mentioned
generator. In fact, the forty eight search coils
are sampled simultaneously by electronic-
microcontroller board.

Induced Voltage in Search coils under
Healthy and LOE Operations
Figs.11 and 12 show the measured voltage
induced in 31
st
search coil, under healthy and
LOE operations at no-load respectively. As
illustrated in Fig.11, under LOE operation,
induced voltage in search coil is declined to
near zero. It must be noted that the
experimental setup did not connect to a power
system, so when LOE happens the induced
voltage in search coil decline to zero rapidly.
In the next section, it will be shown that if the
generator connected to power system, in LOE
condition the induced voltage in search coils
decline to near zero with a time delay.


Fig. 11. Measured voltage under healthy operation
(no- load).

Fig.13 shows induced voltages in 31
st
search
coil, under on-load condition in the normal
operation. According to Figs.11 and 13, due to
the effect of armature reaction, under on-load
conditions, induced voltage in the search coils
slightly leave fromthe induced voltage in the
search coils under no-load conditions. In fact, in
a p poles, healthy, symmetrical machine, the
magnetic axis of each pole is located at 360/p
geometrical degrees [28]. Under load
conditions, the pole axis is not an axis of
symmetry whereas in no-load conditions the
pole axis is axis of symmetry. When a generator
is delivering power to the load, the axes of
symmetry of the magnetic field deviates from
the polar and interpolar axes [29].


Fig. 12. Measured voltage under LOE operation
(no-load).
Electronic-microcontroller
board
Machine under test

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
9


Fig.13. Induced voltage in the 31
st
search coil under
on-load condition in the normal operation.

Fig.14 shows the forty eight voltage
waveform (induced in the search coils) in the
healthy operation. In this figure, symmetry in
the magnetic field distribution under healthy
operation is presented (also, see Fig.5).
Fig.15 shows the forty eight voltage
waveform (induced in the search coils) in the
LOE operation. This figure show that the
magnetic field distribution is lost and induced
voltage in search coils decline to zero (also,
see Fig.6). As demonstrated in Fig.6, Fig.15
validates the FEM modeling of machine for
LOE condition. As stated earlier (in section
III.B), the magnitude of induced voltage in
search coils are completely proportional to
generator air gap flux linkage [25], [26]. So in
the next sections instead of induced voltage in
search coil, linkage flux will be used.


Fig.14. Forty eight measured voltage under healthy
operation.


Fig.15. Forty eight measured voltage under LOE
operation.

Different normal and abnormal operation
conditions are tested. The linkage flux of
generator measured for different loading
condition such as low and high PF loading,
small and heavy generator loading. In each
situation the test repeated for normal and loss
of field condition. From Fig.15 it can be
clearly seen that in LOE situation the air gap
flux of generator has declined rapidly to zero.
It worth mentioning that for evaluating the
precise of the method for every slot a search
coil installed. For practical application it can
be recommended that only one search coil is
adequate for detecting LOE condition.

Fig.16. The system studied.

ANALYSIS OF THE PROPOSED
METHOD
Simulation Result
Experimental results limited to analyze only
isolated generator operation. So for extending
the result of proposed method for a generator
connected to an infinite bus, simulation was
done with MATLAB software. Different
normal and abnormal operation conditions

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
10

were simulated on a salient synchronous
generator connected to infinite bus via a three
phase step-up transformer. The simulations
were done by MATLAB software. Effects of
AVRs and governors were included in
simulations. The structure of the simulated
system is shown in Fig.16. The data for the
system is given in Appendix. The generator
air gap linkage flux and impedance seen by
LOE relay recorded for different operating
regime, especially for power swing.
In Fig. 17, curves a and b show a general
shape of the measured flux and apparent
impedance associated with LOE for light load
and heavy load respectively, whereas curve
c illustrates the traverse of apparent
impedance and flux change on a three phase
short circuit at generator terminal. Curved
and e demonstrate the reaction of generator
flux and impedance for stable normal and
severe power swing. In all the cases, initial
point for different faults is the 4
th
second.
It is apparent that a severe power swing
results in an unnecessary tripping of
conventional LOE relay, while the propose
scheme is more stable for this condition.
As shown in Fig.17 it can be easily seen that
the air gap flux for various operating
situations has a distinctive change. In Fig.17
(a) and (b) it was shown that if LOE happens
in different load conditions then the flux will
decline to about zero rapidly. So we can
define a novel protection for detecting loss of
field of synchronous generator based on the
flux variation.

Comparison with Conventional LOE
Protection
Any proposed LOE protection should be
immune for responding to normal system
operations or system faults that are in the
process of being used for clearing by other
protection devices. Failing to do so may
results in further system instability.
Obviously, it is desirable that the protective
relay isolates a trouble zone as rapidly as
possible. The LOE detection times, using the
two techniques were compared for the various
case scenarios in Table I. In impedance
scheme (Fig.1), the time taken by impedance
seen by the relay traversing to inner circle is
considered (instantaneous trip of relay).
Extensive tests show that if measured flux
dropped more than 30 percent of nominal
flux, the setting will be acceptable (flux
setting<=0.7 nominal flux). In LOE condition
the traverse time from the initial load point to
the relay characteristic of conventional
impedance loci will be between 2 to 7
seconds [20], [22]. In addition, to avoid false
tripping during stable power swing, there is
an intentional time delay.



(a) Normal LOE (light Load).



(b) Normal LOE (heavy Load).


(c) Three phase short circuit at generator terminal.

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
11


(d) Power swing.

(e) Severe power swing.
Fig.17. All the incidents happened at 4
th
second.

The simulation studies show that the proposed
relay is faster for detecting of LOE
conditions. Fig.17 (a) and (b) show the
simulation results of the LOE that occurs at
time instant t=4s, while the load varies from
10% (light) to 100% (heavy). From Table.I it
can be seen that for light load LOE, the
impedance moves into the inner circle in
2.1019s, while it takes 0.802s for LOE
detection by flux method. For heavy load
LOE the operation time is 2.5215s and 1.373
respectively.
According to [10], [20]-[22], one of the main
concerns about LOE protection is false
tripping during a stable transient power
swing. The simulation results also show that
the proposed relay is more effective for
discriminating between the power swings and
LOE. Fig.17 (d) and (e) show the reaction of
two protection methods to normal and severe
power swing. From Fig.17 and Table I it can
be clearly seen that the flux method has a
more accurate response than conventional
impedance method though for the worst case
of simulated power swing.
As stated earlier, linkage flux is measured via
a search coil without solving any especial
impedance calculation equations. This makes
the application of the method simple and
straightforward.
The generator protection must detect short
circuit at generator terminal very fast and trip
the breaker. According to Table. I the flux
method has a maloperation in discriminating
between LOE and three phase short circuit at
generator terminal. From the selectivity point
of view of protective relay, if the operation of
this relay is delayed approximately 25-30
cycles it would prevent maloperation of flux
method during a three phase short circuit at
generator terminal.
Extensive tests show that the proposed
method has various distinction features that
are advantageous over existing schemes. The
first is that the proposed scheme is able to
detect LOE during various kinds of loading
and power swings, even if the power swings
are severe. The second distinction feature for
the method intends to keep it independent of
machine impedance variation resulting the
rate of slip during asynchronous operation. In
addition, the effects of the step up
transformer, power system impedance, shunt
load, and transmission line conditions are also
analyzed, and it is found that the proposed
protection can still perform satisfactorily
under these effects. And the last but not the
least features of the flux method is that it is
faster than conventional protection.

TABLE I
TWO METHOD OPERATION COMPARISONS
Case
Operation time (s)
Flux
method
Impedance
method
LOE-Light Load 0.802 2.1019
LOE- Heavy Load 1.373 2.5215
3 Ph. Short Circuit at
Generator Terminal
False
operation
Not operated
Usual Power Swing
Not
operated
Not operated
Severe Power Swing
Not
operated
False
operation

CONCLUSIONS
The paper proposed a novel flux-based
relaying technique for synchronous generator
loss of excitation protection. Based on air gap
flux measurement by installing search coils, a
new LOE protection has been developed from
which loss of field can be evaluated readily.

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
12

The main idea of proposed method is
validated by experimental results and FEM
analysis. The performance of the proposed
method has been studied for a single-machine
infinite bus system using MATLAB software.
The technique was tested under various fault
and no-fault conditions. Although the
proposed method has a simpler technical and
theoretical basis, it was perfectly
discriminated between generator loss of
excitation and other operation conditions. The
experiment and simulation tests show that the
proposed technique is indeed immune to the
transient power swings. The speed and
sensitivity are the main points featured in the
proposed technique. At the time being, in
rewinding of an old 12 MW synchronous
generator the search coils were installed.

APPENDIX
Generator Data:
=200MVA; V=13.8KV; PF=0.95; S
305 . 1 =
d
X p.u.;. 296 . 0
'
=
d
X p.u.; 252 . 0
' '
=
d
X p.u.;
474 . 0 =
q
X p.u.; 243 . 0
' '
=
q
X p.u.; 18 . 0 =
l
X p.u.; 01 . 1
'
=
d
T s;
0053
' '
=
d
T s; 1 . 0
' '
=
q
T s; 3 8544 . 2 = e R
s
p.u.; H = 3.2 s;.
pole=32.
Hydraulic Turbine and Governor: PID
Governor System
07 . 0 =
a
T s; 33 . 3 =
a
K ;droop 05 . 0 =
p
R ; 163 . 1 =
p
K ; 105 . 0 =
i
K
; 0 =
d
K ; 01 . 0 =
d
T s; beta=0; 67 . 2 =
w
T s; Initial mechanical
power= 0.75 p.u.
Excitation System: IEEE Type 1
3 20 = e T
r
s; 300 =
a
K ; 001 . 0 =
a
T s; 001 . 0 =
f
K ; 1 . 0 =
f
T
s;
Transformer:
S= 210MVA; Y; V=13,8/230 KV; 0027 .
1
= R p.u.; 08 . 0
1
= L p.u.;
Transmission Line:
V = 230 kV; ) 0014 . 0 0529 . 0 (
2 1
j Z Z + = =
/km; ) 0061 . 0 61 . 1 (
0
j Z + = /km; l = 100 km.

ACKNOWLEDGMENT
This work is partly supported by the Mashhad
Power Station. Authors wish to thank Mr.
Dehnavi, Dr. Hashemian and Mr. Pordeli in
providing the experimental test setup and
testing laboratory.

REFERENCES
[1] ANSI/IEEE C37.102-2006, IEEE Guide
for AC Generator Protection, New York,
NY Feb 2007: IEEE PES.
[2] IEEE Committee report, Survey of
Experience With Generator Protection and
Prospects for Improvement Using Digital
Computers, IEEE Trans. Power Del.,
vol. 3, no. 4, pp.1511-1522, Oct.1988.
[3] C.R.Mason, A New Loss of Excitation
Relay for Synchronous Generators, AIEE
Trans., part II, vol. 68, pp.1240-1245,
1949.
[4] S.R.Tambay and Y.G. Paithanka , A new
adaptive loss of excitation relay
augmented by rate of change of
reactance, IEEE, Power Engineering
Society General Meeting, pp.1831-1835,
vol.2, 2005.
[5] C.J .Mozina, Upgrading the Protection
and grounding of generators at Petroleum
and Chemical Facilities, Petroleum and
Chem. Ind. Tech. Conf, 51st Annual Conf.
Proc. Sept. 2004.
[6] A. P. Morais et al., Performance
Evaluation of the Adaptive Loss of Field
Protection in Synchronous Generators by
means of the Positive Offset Method,
IEEE LATIN America Transactions, vol.
7, no. 6, Dec.2009.
[7] F.Bo, L.Xiaoquan, X.Peng and L.J unjie,
The research UL-P of loss of excitation
protection for generator based on the
artificial neural networks, Power and
energy engineering conference, APPEEC
2009. Asia-Pacific, pp.1- 4, 2009.
[8] G.R.Brub and Les M.Hajagos,
Coordination of Under Excitation
Limiters and Loss of Excitation Relays
with Generator Capability, 2009.
[9] C.J .Mozina, Coordinating generator
protection with transmission protection
and generator control-NERC standards
and pending requirements, 2010.
[10] Working Group J -5 of the Rotating
Machinery Subcommittee, Power System
Relay Committee, Coordination of
generator protection with generator
excitation control and generator

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
13

capability", 62nd Annual Conference for
Protective Relay Engineers, pp.150 164,
2009.
[11] O.usta, m.h.musa, m.bayrak and
m.a.redfern ,A new relaying algorithm to
detect loss of excitation of synchronous
generators , Turk J Elec Engin, vol.15,
no.3, 2007.
[12] V.P.Bui, O.Chadebec, L.-L. Rouve, and
J .-L. Coulomb, Noninvasive fault
monitoring of electrical machines by
solvingthe steady-state magnetic inverse
problem, IEEE Trans. Magn., vol. 44,
no. 6, pp.1050 1053, J un.2008.
[13] M.D.Negrea, Electromagnetic flux
monitoring for detecting faults in
electrical machines, Ph.D.dissertation,
Helsinki University of Technology,
Laboratory of Electromechanics, Helsinki,
Finland, 2007. The dissertation can be
read at
http://lib.tkk.fi/Diss/isbn9512284774.
[14] H. Henao, C. Demian, G.-A. Capolino,
A frequency-domain detection of stator
winding faults in induction machines
using an external flux sensor, IEEE
Transactions on Industry Applications,
vol. 39, no. 5 pp. 1272- 1279, Sept.-Oct.
2003
[15] S. Liu, Z. Liu, and O. A. Mohammed,
FE-Based modeling of single-phase
distribution transformers with winding
short circuit faults, IEEE Trans. Magn.,
vol. 43, no. 4, pp.1841 1844, Apr.2007.
[16] Z. Liu, O. A. Mohammed, and S. Liu, A
practical method for building the FE-
based phase variable model of single
phase transformers for dynamic
simulations, IEEE Trans. Magn., vol. 43,
no. 4, pp. 17691772, Apr. 2007.
[17] O. A. Mohammed, Z. Liu, S. Liu, and N.
Y. Abed, Internal short circuit fault
diagnosis for pm machines using fe-based
phase variable model and wavelets
analysis, IEEE Trans. Magn., vol. 43, no.
4, pp. 1729-1732, Apr.2007.
[18] H. Wang, and K. L. Butler, Finite
element analysis of internal winding faults
in distribution transformers , IEEE
Trans. Power Del., vol.16, no. 3, pp. 422
428, J ul. 2001.
[19] H.J . Herrmann, Diego Gao, Under
excitation protection based on admittance
measurement excellent adaptation on
generator capability curves , The 1st
International Conference on Hydropower
Technology & Key Equipment 2006,Beijing,
China , Oct. 27-29, 2006.
[20] J .Berdy, Loss-of-excitation protection
for synchronous generators, IEEE
Transactions on PAS, vol.94, no.5, Part.1,
1975.
[21] AREVA T&D Automation & Information Systems,
Network Protection and Automation
Guide, 2007. See: http://www.areva-td.com
/scripts
/solutions/publigen/content/templates/Sho
w.asp?P=930&L=US.
[22] W.F. Mackenzie et al. , Loss- of field
relay operation during system
disturbances working group report- 1971,
IEEE Transactions on PAS, PAS-94,
no. 5, Sep/Oct.1975.
[23] T. F. Chan, W.Wang, and L. L. Lai,
Performance of an axial-flux permanent
magnet synchronous generator from 3-D
finite-element analysis, IEEE Trans.
Energy Convers., vol. -, no.99, pp.1-8,
2010.
[24] H.Yaghobi, K.Ansari, and
H.R.Mashhadi, Finite element analysis
and radial basis network modeling of
salient-pole synchronous generator with
internal faults based on magnetic flux
linkage (Unpublished work style),
unpublished, Revise requested in IEEE
Trans. Energy Convers.
[25] G.Loisos and A. J . Moses , Critical
Evaluation and Limitations of Localized
Flux Density Measurements in Electrical
Steels, IEEE Trans. Magnetics., vol. 37,
no. 4, jul. 2001.
[26] J .Lenz and A.S. Edelstein , Magnetic
Sensors and Their Applications, IEEE
Sensors J ournal, vol. 6, no. 3, J un.2006.
[27] M. T.Xuan, J .J . Simond, R. Wetter, S.
Keller, A novel air-gap monitoring
system for large low speed hydro-

A Novel Flux- Based Method for Synchronous Generator Loss of Excitation Protection

25
th
International Power System Conference
14

generators, IEEE, Power Engineering
Society General Meeting, 2006.
[28] M. F. Cabanas et al., A new on-line
method for the early detection of broken
rotor bars in asynchronous motors
working under arbitrary load conditions,
Industry Applications Conference,
Fourtieth IAS Annual Meeting, vol. 1,
pp.662 669, Oct. 2005.
[29] M.V.K. Chari , Nonlinear finite element
solution of electrical machines under no-
load and full-load conditions, IEEE
Trans. Magn., vol. 10, no. 3, pp. 686689,
Sep. 1974.

You might also like