Analysis of Synchronous Generator Internal Insulation Failures

E.B. Bakie, Student Member, IEEE, B. K. Johnson, Member, IEEE, H. L. Hess, Senior Member, IEEE,
and J. D. Law, Member, IEEE

Abstract-Winding insulation failures within a synchronous There are many resources regarding synchronous machines
generator cause extensive damage and are expensive to repair. and the transient effects associated with fault conditions [2].
The understanding of the machine behavior during internal However, these resources focus primarily on the effects of
failures is very important for the full protection of the machine. external fault conditions and little is written on the effects of
Within the literature, little is written relating to modeling, internal fault conditions. Most of the literature written on the
analysis, and experimental testing of synchronous generators
effects of internal faults of synchronous machines focuses on
during internal fault conditions. This paper provides a
summary of an experimental approach taken to capture this
modeling and analysis rather than experimentation.
phenomenon. These results will discuss real-time measurements This paper focuses on experimental measurement and
taken during controlled internal faults on a laboratory scale analysis of internal fault conditions of a 220-volt, 20 h.p.
generator. This data will also be compared to data recorded by laboratory synchronous generator. The internal faults
protective relays during an evolving internal insulation failure explored in the experimentation include controlled turn-to-
of a 145 MVA utility generator. turn, turn-to-ground, and turn-to-turn-to-ground. This paper
only discusses the data taken for turn-to-ground faults. The
I. INTRODUCTION results include turn-to-ground internal faults for two
Full protection of synchronous generators during fault generator-operating points. A commercial protective relay
conditions is an essential part of power system reliability. was used collect data during the fault conditions. The system
Failures of these machines result in lost revenues, expensive tested was high resistance grounded.
repair costs, losses in production capacity and possibly This paper also compares and analyzes data collected by
service to customers. There is an ongoing need for a better protective relays during an internal fault of a utility generator
understanding of the effects of internal faults within to the data collected during the experimentation. The utility
synchronous machines in order to provide better generator generator failure resulted initially from an internal turn-to-
protection. ground fault that evolved into a more severe fault condition.
Stator ground fault protection of synchronous generators is Only the time period where the turn-to-ground was present
very important because of the potential amount of damage will be compared to the experimental results in this paper.
caused by ground faults. The goal of the protection is to The primary goal of the experimentation is to analyze
detect the first ground fault quickly while the current is low internal faults and provide recommendations for updating
before the fault evolves into a double-line-to-ground fault, relay settings for 100% stator winding protection.
which will cause extensive damage and higher repair costs
[1]. The protection needs to be configured so an insulation II. EXPERIMENTAL SETUP
failure anywhere in the winding is detected and corrective The experimental setup consists of an induction machine
action taken to limit the damage to the generator. with pilot exciter, a microprocessor relay, circuit breakers,
Stator ground faults can cause extensive damage to the instrumentation transformers, and other auxiliary equipment.
stator winding insulation, the stator conductors as well as The general layout of the system is shown in Fig. 1. The
damage to the core structure. Mechanical damage could also synchronous generator tested is a 6 pole, 220V, 20-hp
occur. With properly configured generator protection, the machine driven by an induction motor in a motor-generator
damage caused by internal faults can be reduced. set.
One hundred percent stator ground fault protection can be
achieved on high resistance grounded synchronous
generators by implementing a differential over-voltage
scheme (59G) across the grounding resistor [1]. Both the 60
Hz and 180 Hz voltages across the grounding resistor are
monitored. Based on the percentage of the above quantities,
the protective relay can detect a ground fault condition along
the entire length of the stator windings.

E. Bakie, B. K. Johnson, H. L. Hess, and J. D. Law are with the

Department of Electrical and Computer Engineering, University of Idaho,
Moscow, Idaho 83843, U.S.A. (e-mail [email protected],
[email protected], [email protected] and [email protected])
This work was funded by the University of Idaho ECE Department.
Fig. 1. Overview of Experimental Setup.

0-7803-8987-5/05/$20.00 2005 IEEE. 106

A. Control Setup no load and full load operating profile of the test generator
The microprocessor relay serves as the main controller from 0 to 100% of the length of the stator winding.
of the experimental setup. The relay is connected to a
personal computer (PC) via a serial port. The entire system
can be controlled through this interface. The relay is
configured to control auxiliary equipment that can start/stop
the prime mover, disconnect the load from the system, and
interrupt the field current of the synchronous generator. The
relay is also configured to alarm when a loss-of-potential
(LOP) from a voltage transformer occurs. In the future, the
microprocessor relay will be interfaced with a
communications processor so the experiments can be
performed from a remote location.
The faults themselves will be applied by circuitry that Fig. 3. 180 Hz neutral voltage profile of the laboratory generator.
allow the fault type and time duration to be controlled. C. Generator Stator Winding Configuration
B. Synchronous Generator Model The stator windings of the synchronous generator are
The model parameters for the synchronous generator tested composed of two parallel paths per phase. Each parallel path
were also determined by experimentation. Applying single- consists of nine series coils, with each coil having ten turns.
line-to-ground (SLG) faults at the terminals of the generator Each turn produces 0.71 volts. During a rewind of the
tested the accuracy of the generator model. The measured synchronous generator, multiple taps were inserted
fault magnitudes were compared to calculated values. The throughout the lower turns of the stator winding. The taps
calculated value about a 5% larger than the resulting were inserted on Phases A and B and are labeled SCXX
measured values. This is good from a protection viewpoint (where XX is a number identifying a location). Phase C did
because the relay will be more sensitive. The parameters not have any taps inserted within the winding.
used in the calculations were the subtransient values, with the The neutral wire and the tap terminals have been brought
negative and zero sequence parameters approximated based out of the machine into a common distribution box. With
on typical machine constants [3]. Fig. 2 shows plots from the this distribution box, a variety of internal faults can be
captured event record of a SLG fault at the terminals of the applied to the stator winding for different numbers of turns
machine and the system response time to the fault. The close to the neutral of the machine. The number of turns
protection system clears the fault in about 3 cycles and the between each tap is shown in Table 1. The winding
terminal voltage decay due to loss of excitation ends after 15 configuration and tap location for each phase are shown in
cycles. Fig. 4.

Table I
Stator Winding Tap Locations and Number of Turns from the Neutral
Terminal.

SUBCOIL PHASE NUMBER OF TURNS

SC11 A 1
SC21 A 1
SC31 A 1
SC41 A 2
SC51 A 5
SC12 B 1
SC22 B 1
SC32 B 1
SC42 B 2
SC52 B 5

Fig. 2. SLG at the generator terminals.

Also needed to provide 100% stator ground fault

protection on the machine is the third harmonic profile on the
machine at different operating conditions. Fig. 3 shows the

107
 VA in Fig. 6. Fig. 6 also shows the fault current and the
phase-B and C voltages during the fault, as well as the
neutral-to-ground voltage. During this fault the tripping of
the generator excitation was delayed as shown. The tripping
of the field was delayed to collect more data with the fault
was on the stator. This is shown in Fig. 6 with the SELogic 3
line at the bottom of the figure. The line gets thicker when
the relay detects the fault and the T indicates when the
relay sent the trip command. The dotted vertical line
indicates when the relay event record triggered. The relay is
set to record eight cycles of pre-fault data.

Fig. 4. Stator Tap Locations.

D. Synchronous Generator Protection Scheme

The protection schemes used during the testing include
instantaneous overcurrent (50) and stator ground fault (64G)
[1]. These numbers are ANSI/IEEE standards used in
protective relaying. Only the 60 Hz element was used during
this phase of the testing and the 180 Hz (59G) will be added
later. The 50 element is used as a backup and for event
triggering purposes. Initially, recording the data during
internal failures was of interest, to determine what can be
Fig. 6. Internal fault 10% up from the neutral with no load on the generator
detected using only these two elements in the system under and excitation tripping delayed.
test. Later, these results will be used to implement better The relay detected the presence of the fault on the low end
protection schemes using the 180 Hz element (59G). The of the winding using the 60 Hz neutral voltage element. The
59G element consists of measures 60 Hz and 180 Hz voltage magnitude of the current shown is 100-times the actual fault
across the grounding resistor. The system instrumentation is current. This is done to because the relay was designed for
shown in Fig. 5. larger systems with higher current transformer ratios.
Harmonic analysis of the neutral voltage is shown in Fig. 7.
Synchronous
Generator Note there is no 180 Hz present in the neutral-to-ground
A
voltage throughout the duration of this fault. This is
B primarily due to the pole pitch of 7/9 of the synchronous
C
generator. This may cause problems in detecting a low-end
winding fault on the 180 Hz element for this operating
`
condition.
VN IN IC IB IA
VC VB VA

OUT101 OUT102 Protective Relay

S/G Excitation Induction

CB Machine
CB

Fig. 5. System instrumentation and relay inputs.

III. EXPERIMENTAL RESULTS

The first test of an internal fault condition on the
laboratory synchronous generator was done with the machine
running at rated speed and with the load open circuited. This
testing was also done at 50% rated voltage (110 VLL) to
reduce the chance of damaging the machine during the fault.
The first fault applied to the generator was a single-line-to- Fig. 7. Harmonic analysis of the grounding resistor voltage during an internal
ground (SLG) fault on tap SC51. This is about 10% from the fault with no load on the generator.
neutral end of the phase-A stator winding. The voltage-to-
ground at that point was recorded at 8.0 volts and is labeled

108
 Second, a light load was added to the generator and the and a circuit breaker was added in the fault path so the
fault was repeated at the same location and terminal voltage internal fault could be removed. During this fault the
of 110VLL. Fig. 8 shows the data captured during the fault. machine operated under the same conditions as the first case.
IA in Fig. 8 is the measured fault current; VA is the voltage Fig. 10 shows the data captured during this condition. Fig.
at the fault point. Again, the tripping of the generator 10 shows the fault current magnitude to be the same as Fig.
excitation was delayed. Comparing Fig. 8 to Fig. 6, the fault 6. However, looking at the SELogic 3 line at the bottom of
current has a higher magnitude and the neutral-to-ground the figure the relay detected the fault much quicker. Also
voltage contains more harmonics. notice that the fault current goes to zero, indicating that the
protection system actually cleared the fault. Note, once the
generator excitation is tripped, the fault current begins to
decay.

Fig. 8. Internal fault 10% up from the neutral with light load on the generator
Fig. 10. Internal Fault 10% up from the neutral with no load on the
and excitation tripping delayed.
generator and excitation tripping not delayed.

IV. UTILITY GENERATOR FAILURE

The fault was picked up by the 60 Hz element as it did
with the no load case. The harmonics present in the neutral- The neutral-to-ground voltage recorded during an internal
to-ground voltage is shown in Fig. 9. Notice that by fault of the 145MVA utility generator is shown in Fig. 11.
changing the operating condition of the generator, the amount The change in the 64G1T trace shows that a stator ground
of 180 Hz in the neutral-to-ground voltage increases from fault was detected on the generator by the 60 Hz element.
zero percent to about seventeen percent. This change can be The magnitudes of the fault currents suggest that this is a
used for protection purposes. low-end winding single phase-to-ground fault. This machine
was also high resistance grounded. The amount of third
harmonic voltage present is 85.2% of the fundamental.
According to relay event records, the generator excitation had
been removed shortly after the protective relays detected the
fault. Note, in this case the fault currents do not decay after
the generator field breaker was supposedly opened.

Fig. 9 Harmonic analysis of the grounding resistor voltage during an internal

fault with light load on the generator.

For the third case, the tripping of the field was not delayed

109
 REFERENCES
[1] J. L. Blackburn, Protective Relaying: Principles and Applications, 2nd
ed, New York: Marcel Dekker, 1998, pp. 231-274.
[2] P. P. Reichmeider, D. Querry, C. A. Gross, D. Novosel, S. Salon,
"Partitioning of Synchronous Machine Windings for Internal Fault
Analysis," IEEE Trans. on Energy Conversion, vol 3, Dec. 2000, p. 1.
[3] P. M. Anderson, Analysis of Faulted Power Systems, Iowa State
University Press, 1973, p. 202.

APPENDIX
Table II
Generator Technical Data
Rated Voltage 220V
Rated Current 56 A
Power 20 h.p.
Fig.11: Initial stator ground fault detected by protection relays on a 145- Frequency 60 Hz
MVA utility generator. Number of Phases 3
Number of Poles 6
The initial stator ground fault on the utility generator Speed 1200 RPM
evolved into a more severe phase-to-phase-to-ground fault Number of Slots/Phase/Pole 6
condition and later a three-phase fault with current Turns/Coil 10
Turns/Slot 20
magnitudes reaching values above 13kA. This fault led to Pitch 7/9
the entire machine needing to be rewound. Fig. 12 shows the Number of Damper Bars/Pole 5
current waveforms captured during the later part of the Number of Slots 54
internal fault. Fig. 12 shows the fault currents beginning to Xd 2.8 ohm
decay, as did the laboratory generator when the relay Xq 2.7 ohm
Rf 16 ohm
detected the fault. This case is different because the currents
Lf 3H
begin to rise again rather than continue decaying as the case Tdo 186 msec
shown in Fig. 6.

BIOGRAPHIES

Eric B. Bakie received the B.S. degree in electrical from the University of
Idaho, Moscow, ID in 2003. He is presently pursuing his Master of Science
in electrical engineering at the University of Idaho. His research interests
include power system protection, short circuit studies, and electrical
machines. Mr. Bakie is also a certified EIT in the state of Idaho.

Brian K. Johnson (M92) received the Ph.D. degree in electrical

engineering from the University of Wisconsin-Madison in 1992. He is
currently a professor in the department of electrical and computer
engineering at the University of Idaho, Moscow, ID. His interests include
HVDC transmission, FACTS, custom power technologies, energy storage,
Fig. 12: Utility generator fault currents after the stator ground fault evolved. and utility applications of superconductivity, power system protection and
electromagnetic transients in power systems. Dr. Johnson is a member of
CIGRE and is a registered professional engineer in Wisconsin and Idaho.
V. CONCLUSIONS
Little information is available regarding internal faults of Herbert L. Hess (SM92) received the Ph.D. degree in electrical
synchronous generators. Several papers exist relating engineering from the University of Wisconsin-Madison in 1993. He is
currently an associate professor in the department of electrical and computer
primarily to the modeling and simulation of internal engineering at the University of Idaho, Moscow, ID. His research interests
failures, but few contain actual experimental results or include power electronics, electric machines and drives and power quality.
results analyzed from relay event records. This paper
Joseph D. Law (M91) received the Ph.D. degree in electrical
discusses the methods taken to experimentally document engineering from the University of Wisconsin-Madison in 1991. He is
internal fault behaviors for a laboratory synchronous currently an associate professor in the department of electrical and computer
generator and compare them to data taken from relay event engineering at the University of Idaho, Moscow, ID. His research interests
include electric machines, power electronics and electrical disturbances in
records recorded during similar failures on a much larger power systems.
machine.

110