PROTECTION OF

SYNCHRONOUS GENERATORS
Sture Lindahl

ABB Network Partner AB

II Protection of Synchronous Generators

GENPC000.DOC 2007-05-22 15.23

 III

Preface
Generally, the power stations represent about 50% of the capital cost in a
power supply system. The power generator itself plays an important role in
the energy conversion process in the power station. A generator has more
failure modes than any other component in the power system. The
generator protection system must detect the faults rapidly. Otherwise, there
is a risk that the power generating unit will suffer a protracted forced
outage. In some nuclear power stations, there are a spare rotor, a spare
stator or both. Usually, there is no spare generator. This means that a
generator failure will cause a forced outage of the entire power generating
unit. Such outages will cause a substantial increase of the costs for power
generation.
The ambition, to protect the generator against all faults, results in ad-
vanced protection and monitoring systems. There is, however, considerable
divergence in opinion on the extent of generator protection systems. It is
important that the protection system detects fault that may hurt humans and
damage equipment. Most power systems tolerate the disconnection of one
generating unit without running into serious problems. A fault on another
power system component may cause the generator protection system to op-
erate non-selectively. Such an unwanted operation may cause a blackout of
the power system or the disconnection of customers.
In this document, we discuss generator faults and abnormal
conditions. We also describe several aspects of the generator protection
system.
My intent has been to compile a document that can serve as an intro-
duction and a background to further studies. I have presupposed that the
reader has some knowledge in power system engineering. I plan to use the
document in further education courses. The intended readers are power sys-
tem operators and protection engineers. I hope that design engineers in the
power industry will find the document useful as a background.
As a former teacher in automatic control, I am aware of the
importance of feedback. I encourage all readers to let me know of errors,
misunderstandings and suggestions for improvement.

GENPC000.DOC 2007-05-22 15.23

IV Preface

GENPC000.DOC 2007-05-22 15.23

 V

Contents
1. Introduction
2. Synchronous Machines
3. Generator Faults
4. Fault Statistics
5. Background
6. Differential Protection
7. Underimpedance Protection
8. Overcurrent Protection
9. Interturn Fault Protection
10. Protection Against Open Circuits
11. Stator Earth-Fault Protection
12. Field Earth-Fault Protection
13. Underexcitation Protection
14. Overvoltage Protection
15. Reverse Power Protection
16. Unbalance Protection
17. Out of Step Protection
18. Abnormal Frequency Protection
19. Inadvertent Energising Protection
20. Bearing Current Protection
21. Breaker Failure Protection
22. Operational Experience

GENPC000.DOC 2007-05-22 15.23

VI Contents

GENPC000.DOC 2007-05-22 15.23

 1.1

1. INTRODUCTION
Generally, the power stations represent about 50% of the capital cost in a
power supply system. The power generator itself plays an important role in
the energy conversion process in the power station. A generator has more
failure modes than any other component in the power system. The
generator protection system must detect the faults rapidly. Otherwise, there
is a risk that the power generating unit will suffer a protracted forced
outage. In some nuclear power stations, there are a spare rotor, a spare
stator or both. Usually, there is no spare generator. This means that a
generator failure will cause a forced outage of the entire power generating
unit. Such outages will cause a substantial increase of the costs for power
generation. It is very important that the protection system detects all faults
that may hurt humans and damage equipment. This means that the
generator protection system must have a high degree of dependability. It is
high if the probability of not having a failure to operate is high.
Most power systems tolerate the disconnection of one generating
unit, one power transformer, one power line or one busbar section without
running into serious problems. A fault on adjacent power system
component may cause the generator protection system to operate non-
selectively. Such an unwanted operation may cause a blackout of the power
system or the disconnection of customers. The generator protection system
must have a high degree of security. It is high if the ability of not having
an unwanted operation is high.
All protection systems must have a high degree of dependability and
a high degree of security. We say that a protection system has a high degree
of reliability if the ability of not having an incorrect operation is high. The
reliability is the combined ability of not having a failure to operate and of
not having an unwanted operation.
The ambition to protect the generator against all faults results in
advanced protection and monitoring systems. There is, however,
considerable divergence in opinion on the extent of generator protection
systems.
To reduce the operating costs for the power generation, utilities
reduced their staff and many power stations are unmanned. To avoid
increased risks for damage to such unmanned power station it is necessary
to equip these power stations with advanced protection systems and remote
control systems. Depending on the size (cost) and importance for the power
supply we place different demands on the protection system. In hydro
power stations with only one generating unit and in nuclear power stations,
the protection systems have to meet the most stringent requirements. In

GENPC010.DOC 2007-05-22 15.24

1.2 Introduction

hydro power stations with several small generating units, we may simplify
the protection systems. Such protection systems have to satisfy the legal
requirements but no other requirements.
Protection systems in pumped storage hydro plants require many
special considerations because of the many modes of operation. They
include (1) generation, (2) pumping, (3) synchronous compensation
prepared for generation and (4) synchronous compensation prepared for
pumping. One special problem is to maintain the correct current direction
in reversible pumped storage units.
In the mid 1960's, many utilities introduced electronic control
equipment in power stations. After accumulating sufficient experience
concerning reliability, maintainability and disturbance immunity, they also
introduced static protection equipment. Electronic protection equipment has
shorter operate times, better earthquake safety and does not need as much
panel space as older electromechanical protection equipment. The
introduction of electronic protection equipment made it possible to
introduce more sophisticated protection functions. Now, we seldom install
new electromechanical protections.
Microcomputer based measurement and control equipment has been
used for several years. In the early 1980's, computer based protections
became available. In the late 1980's, complete integrated protection system
using digital signal processing techniques become commercially available.

1.1 Voltage Classes

It is convenient to use abbreviations for intervals of system voltages. The
abbreviations LV, MV, HV, EHV and UHV are common. To avoid
misunderstandings, we give our definition in Table 1.
Table 1 Classes of system voltages and their designation.
Description Designation System voltage
kV
Low Voltage LV -1
Medium Voltage MV 1-99
High Voltage HV 100-344
Extra High Voltage EHV 345-999
Ultra High Voltage UHV 1 000-

GENPC010.DOC 2007-05-22 15.24

Introduction 1.3

1.2 General Demands

The general demands on the protection system are based upon some risk
analysis, experience and tradition. The legal requirements give the
minimum demands. Further demands must be motivated by reduced
disturbance and repair costs. Commonly, it is required that earth faults,
short circuits and other severe faults must be detected by two independent
protections. The fault must be cleared even if one switching device fails to
operate. The generator protection system must also provide adequate back-
up protection for (1) the associated buswork, (2) the station auxiliary
transformer, (3) the magnetisation transformer, (4) the generator step up
transformer, (5) the busbar on the high voltage side of the step up
transformer and (6) the power lines that start at the power station. A
fundamental requirement is that the operate time must be as short as
possible. However, we have to consider the risk for unwanted operation,
the cost for the protection system and the costs associated with the
damages.
This document describes several protections that may belong to the
complete generator protection system. The descriptions are brief and
tutorial. To simplify further studies, we have included references to
relevant original publications. Many manufacturers can provide more
detailed information concerning their products.
Often, the generator and the step up transformer are protected by
common equipment. This means that is can be difficult to give a sharp
distinction between the generator protections and the transformer
protections. This document contains some aspects on transformer
protection and back-up protection. There are many mechanical and thermal
protection devices that help to avoid damage to the power generating unit.
We have not included these devices in this document. Instead, we have
concentrated on the protection equipment with electrical energising signals.

GENPC010.DOC 2007-05-22 15.24

1.4 Introduction

GENPC010.DOC 2007-05-22 15.24

 2.1

2. SYNCHRONOUS MACHINES
A synchronous machine is not a simple device. The armature, or stator,
winding is arranged in three symmetrical phase belts in slots in the stator
surface. The magnetic field intensity can be controlled via the DC current
in the rotor, or field, winding. A synchronous machine can operate as a
generator or as a motor. In our case, generator operation is the most
common operating mode. When the synchronous machine operates as a
generator, a prime mover drives the rotor. The prime mover can be a
diesel engine, a gas turbine, a hydro turbine or a steam turbine. A
synchronous generator may operate alone with a single load or in parallel
with other generators on a large power system.

2.1 The Development of the Synchronous Generator

The frequency of the EMF in a synchronous generator is:

np (1)
f =
60

where
n = the rotational speed in revolutions per minute [rpm], and
p = the number of pole pairs.

The three-phase currents in the stator winding generate a rotating magnetic

field. During steady-state conditions, this field is stationary to the rotor and
its field. This explains the term synchronous machine.
Some synchronous machines have a round or cylindrical rotors.
Other machines have salient pole rotors. In the round rotor case, the field
winding lies in slots cut axially along the rotor the rotor length. The diameter
of a round rotor is small and usually in the order of 100 centimetres. Such a
round rotor is suitable for operation at high speeds driven by a gas turbine or a
steam turbine. Therefore, it is known as the turbo-generator. When the
prime mover operates at a low speed, the rotor may have a larger diameter.

GENPC020.DOC 2007-05-22 15.24

2.2 Synchronous Machines

The capacity of a synchronous generator is:

S ≈ n D2 LBA (2)

where
n = the rotational speed [rpm],
D = the diameter of the rotor [m],
L = the length of the rotor [m],
B = the magnetic flux density in the air-gap [T], and
A = the linear current density on the rotor [A/m].

The rotational speed is given by ( 1 ). In a 60 Hz system the highest rotational

speed is 3 600 rpm and in a 50 Hz system the highest speed is 3 000 rpm.
The diameter of the rotor is limited by the centrifugal forces and the
stresses in the retaining ring. Despite of the rapid development of materials, it
is not possible to increase the diameter beyond 1.3 m for a generator with a
rated speed of 3 000 rpm. Corresponding figure is 1.2 metres for a generator
with a rated speed of 3 600 rpm. It is possible to increase the diameter to
1.8 metres for generators with four poles.
The length of the rotor is limited by the critical speed and the risk for
high vibration amplitudes. Usually, the ratio L/D must not exceed 6-7.
There is no hope that it should be possible to a higher magnetic flux
density in the air-gap than 1.2 T.
Our remaining hope is the linear current density. Better cooling of
the rotor winding and the stator winding makes it possible to increase the
capacity of synchronous generators.

2.2 The Development of the Hydro-Generator

The synchronous machine possesses a wide range of characteristics and
special features. They make complete protection difficult. We will review
these conditions before we discuss the detailed application of protection
equipment.
Figure 1 shows the maximum capacity in MVA of hydro-generators
in Sweden.
In 1938, the maximum capacity was 40 MVA and increased to
500 MVA in 1980. During these 42 years, the average rate of increase is
more than 6% per year.

GENPC020.DOC 2007-05-22 15.24

Synchronous Machines 2.3

HYDRO-GENERATORS IN SWEDEN

Rated Apparent Power [MVA]

600
Harsprånget G5
500

400 Juktan G1
Ritsem G1
300 Porjus G11
Seitevare G1
200 Stornorrfors G1
Harsprånget G1
Hjälta G1
100 Torpshammar G1
Hojum G1
Stadsforsen G1
0
1900 1920 1940 1960 1980 2000
Year

Figure 1 Maximum capacity of hydro-generators in Sweden.

2.3 The Development of the Turbo-Generator

In 1901, C.E. Brown patented the turbo-generator with a cylindrical rotor.
The first generator of this type had a rated voltage of 2 kV, a rated power of
250 kVA and a rated speed of 3 900 rpm that corresponding to a rated fre-
quency of 65 Hz. The increasing demand of electricity has led to a very rapid
increase of the maximum rated capacity of power generating units. Now, the
rated power of the largest turbo-generators for 50 Hz is 1185 MVA for a ma-
chine with two-poles and 1640 MVA for a machine with four poles. Figure 2
shows how the maximum capacity depends on the cooling method.

GENPC020.DOC 2007-05-22 15.24

2.4 Synchronous Machines

Turbo-Generator Size (50 Hz)

2000
Influence of Cooling Method

1640
MVA

1500

1200
850
1000

500
200

0
Air Hydrogen Water/Hydrogen Water

Figure 2 Maximum rated capacity of turbo-generators.

Figure 3 shows the maximum size of turbo-generating units in Sweden.

1250
*)
Net active power [MW]

Forsmark 3 & Oskarshamn 3

1000
Ringhals 3 & 4
Ringhals 2
750
Oskarshamn 2
500
Oskarshamn 1
Karlshamn 1, 2 & 3
Stenungsund 3 & 4
250
Stenungsund 1 & 2

0
1900 1920 1940 1960 1980 2000
Year
*) Increase of maximum capacity in existing power plants

Figure 3 The development of turbo-generating units in Sweden.

In 1959, Swedish State Power Board commissioned the turbo-generating

unit Stenungsund 1 with two radial steam turbines and two axial steam tur-

GENPC020.DOC 2007-05-22 15.24

Synchronous Machines 2.5

bines. This means that the unit has two generators with a rated capacity of
25.3 MVA and two generators with a rated capacity of 68.8 MVA. The
total rated capacity is 188.2 MVA. In 1985, Oskarshamns Kraftgrupp AB
commissioned the nuclear unit Oskarshamn 3 with one turbo-generator. It
has a rated capacity of 1 294 MVA. During these 26 years, the average rate
of increase is about 12% per year.

2.4 Power Station Configuration

The use of uniform and standardised generator protection is one way of
achieving a high degree of reliability. By doing so, we can reduce the
design errors. A uniform design of the protection system eases the testing,
management and maintenance.
Varying power station configurations obstruct the use of a uniform
and standardised generator protection system. The most important factor is
the varying power station configuration. Figures 2.2-2.5 show the most
common configurations of power stations.

Figure 4 Unit-connected generators without generator breakers.

GENPC020.DOC 2007-05-22 15.24

2.6 Synchronous Machines

Figure 5 Unit-connected generators with generator breakers.

Figure 6 Generators connected to common step-up transformers.

GENPC020.DOC 2007-05-22 15.24

Synchronous Machines 2.7

Figure 7 Generators connected to a common generator busbar.

Besides the varying power station configuration, following factors

influence the design of the generator protection system:
• generator circuit-breaker or not
• earthing of the neutral
• location of voltage transformers
• location of current transformers

2.5 Asynchronous Operation

Synchronous operation is the desired mode of operation for most synchro-
nous machines. There are, however, incidents that may cause sustained
loss of synchronism or transient loss of synchronism. Generally, loss of
synchronism is an abnormal mode of operation for synchronous machines.
Consider a synchronous generator connected to a strong power sys-
tem. Now, let us assume that the field circuit-breaker receives an opening
impulse. Then, the breaker will disconnect the exciter from the field wind-
ing. The inductance of the field winding will cause a gradual decrease of
the field current. It will fall as the discharge resistor absorbs the energy
stored in the field winding. Eventually, the loss of excitation will cause a
sustained loss of synchronism.
After this, the synchronous machine will operate as an induction ma-
chine and it will run above synchronous speed. It will continue to generate
power and the setting of the turbine governor will determine the amount of

GENPC020.DOC 2007-05-22 15.24

2.8 Synchronous Machines

power generated. Slip frequency current will flow in several paths in the ro-
tor. The field winding may form such a path. Another path is the damper
winding of a salient pole machine. The current will also flow in slot wedges
and the solid rotor body.
When the machine operates as an induction generator, the external
power network will provide the necessary excitation. This means that the
machine will absorb much reactive power. The reactive current may ap-
proach or even exceed the rated current of the machine.
In [ 1 ], Mason and co-authors describe a series of asynchronous
running tests. The test object was a 588 MVA turbo-generator equipped
with special temperature-measuring devices on both the stator and rotor.
Below, we reproduce some results presented in [ 1 ].

Table 1 Ratings of generators referred to in [ 1 ].

Generator Sn Pn Vn In
MVA MW kV kA
A 558 500 22.0 15.437
B 37.5 30 11.8 1.835
C 75 60 11.8 3.67
D 75 60 11.8 3.67
E 150 120 13.8 6.27
F 150 120 13.8 6.27
G 133 120 13.8 5.585
H 75 60 11.8 3.67
I 37.5 30 11 1.97
J 56 45 11.8 2.75

GENPC020.DOC 2007-05-22 15.24

Synchronous Machines 2.9

ASYNCHRONOUS OPERATION
0.6

0.5

Slip, s [%] 0.4

0.3

0.2

0.1

0.0
0.0 0.2 0.4 0.6 0.8 1.0
Active Power, P/S [pu]

Figure 8 Slip during asynchronous operation.

ASYNCHRONOUS OPERATION
1.2
Quadrature Coponent of
Stator Current [pu]

1.0

0.8
0.6

0.4

0.2

0.0
0.0 0.2 0.4 0.6 0.8 1.0
In-Phase Component of Stator Current [pu]

Figure 9 Reactive current during asynchronous operation.

GENPC020.DOC 2007-05-22 15.24

2.10 Synchronous Machines

Reference

[1] Mason, T.H., Fairney.W., Arnold, J.J. & Thelwell, M.J.:

"Asynchronous operation of turbo-generators", Report 11-02,
CIGRE-Session, Paris, 1972.

GENPC020.DOC 2007-05-22 15.24

 3.1

3. GENERATOR FAULTS
The protection of synchronous generators involves the consideration of
more possible faults and abnormal conditions than the protection of any
other power system component. When designing the protection system we
have to consider (1) stator faults, (2) rotor faults and (3) abnormal
operating conditions.

3.1 Stator Faults

Damage to the stator winding itself or its insulation may cause stator short-
circuits or stator earth-faults.

3.1.1 Stator Short-circuits

Ageing, overvoltage, overcurrent or loss of cooling may cause stator short-

circuits. External short-circuits, improper synchronisation and loss of
synchronism may cause large currents. These currents cause high forces
that may displace the stator winding and by that an internal short-circuit.
An external short-circuit is accompanied by very large fault-currents.
The electromechanical forces increase considerably when the size of the
generator increases. The size of the electromechanical forces may amount
to more than 100 N/cm at sudden short-circuits. Generally, the utilities
require that synchronous generators shall withstand, without damage, all
types of short-circuit on the generator terminals.
Short-circuits clear of earth are less common faults. They may occur
on the end portion of the stator coils. They may also occur in the slots if
there are two coils in the same slot. In the latter case the fault will involve
earth in a very short time. The short-circuit currents do not depend on the
generator neutral earthing principle.
Generally, thermal power units commissioned during the last 30
years have phase-segregated generator buswork. Such a design reduces
considerably the risk for two-phase and three-phase short-circuits close to
the generator terminals.

3.1.2 Stator Earth-faults

Pohl has investigated how the earth-fault current damages the sheets of a
synchronous machine. In [ 2 ], he describes the result of tests with three
different insulating materials between the stator sheets, two different values
of the earth-fault current and four values of the fault clearance times.

GENPC030.DOC 2007-05-22 15.25

3.2 Synchronous Machines

Figure 1 shows the number of stator sheets that melt together because of an
earth-fault. We can see how the damage caused by the earth fault current
may increase with the fault clearance time. Table 1 shows how to interpret
the legend in the figure.

Sheets Melted Together

150

Legend
# SM 2 A
SM 5 A
100
CM 2 A
CM 5 A
M 2A

50 M 5A

0
4 sec 1 min 10 min 60 min
Fault Clearance Time

Figure 1 Damage caused by the earth fault current.

Table 1 Insulation material and fault clearance time.

Legend Insulation Material Fault Current
A
SM 2 A Schellackmikafolium 2
SM 5 A Schellackmikafolium 5
CM 2 A Kompoundmikafolium 2
CM 5 A Kompoundmikafolium 5
M 2A Mikanit in Plattenform 2
M 5A Mikanit in Plattenform 5

3.2 Rotor Faults

The field circuit of a synchronous generator consists of the rotor winding
proper and associated circuits. These may include the slip rings and
brushes, the field circuit-breaker, the armature of a rotating exciter or the
rectifier and the secondary winding of the magnetisation transformer in a
static exciter. This circuit is an isolated DC circuit and it is not necessary to

GENPC030.DOC 2007-05-22 15.25

Synchronous Machines 3.3

earth it. Generally, the field circuit of modern generators is operated

unearthed. Earth-faults, interturn faults and open circuits may occur in the
field circuit. Overexcitation or unbalanced loading may overheat the rotor
itself.

3.3 Abnormal Operating Conditions

Let us consider a single synchronous generator or a group of synchronous
generators connected to a power system by a single-circuit power line. The
line may have several sections and intermediate substations. Now, we
assume that one line circuit-breaker, either in the power station or in one of
the substations, opens inadvertently. We have to consider the risk that the
automatic voltage regulator (AVR) does not operate correctly. The
excitation may also be under manual control. In such cases there is a risk
that we subject our customers to both overfrequency and overvoltage.
There may be legal requirements that make it necessary to disconnect the
generator in such cases.
A failure in the automatic voltage regulator (AVR) or the excitation
system may reduce or even interrupt the excitation current. We say that
loss-of-excitation has occurred. When a synchronous generator loses
excitation it operates as an induction generator running above synchronous
speed. The generator starts to draw reactive power from the network. The
wattless current produces the main flux in the machine. The machine will
continue to generate active power. The load setting and speed droop of the
turbine governor determine the value of the power generation. Turbo-
generators are not suited for such operation because they do not have
damper windings and will quickly overheat from the induced currents in
the rotor iron.
Let us consider a synchronous generator that is operating in
synchronism with a power system. Now, we assume that the generator loses
its driving force, e.g. due to an inadvertent closing of the stop valves in a
steam turbine or the wicket gates in a hydro turbine. The generator remains
in synchronism with the power system and continues to run as a
synchronous motor. The synchronous machine draws sufficient power
from the power system to drive the prime mover. Such motoring does not
damage the generator but may damage the prime mover. Especially steam
turbines may be damaged by overheating if the steam flow ceases.
Unsymmetrical faults may produce more severe heating of the
synchronous generator than symmetrical faults or balanced loading.
External unsymmetrical faults may be either series faults or shunt faults.
Such faults cause unbalanced loading of the synchronous generator. The

GENPC030.DOC 2007-05-22 15.25

3.4 Synchronous Machines

negative sequence currents that flow during these unbalanced faults induce
voltages in the rotor of the generator. The frequency of these voltages is 2fn
where fn is the power frequency. The induced voltages cause rotor currents
that tend to flow on the surface of the rotor forging and in the non-magnetic
rotor wedges and retaining rings. The resulting losses quickly raise the
temperature. If the abnormal condition persists, the metal will melt,
damaging the rotor structure.
There are [ 1 ] four potential sources of shaft voltages on generator
shafts (1) the generation of a shaft voltage by an asymmetrical generator
airgap or stator field which links with the rotor shaft, (2) the capacitive cou-
pling of the excitation control system with the shaft, (3) the shaft voltages
generated by charge separation in the latter stages of the steam turbine and
(4) the stresses of the turbine stationary and rotating blades/nozzles or
buckets cause the blades to become magnetised by magnetostriction, which
on rotation, generates a small AC voltage across the turbine. It is common
practice to earth the shaft at one location. The contact between the hydro
turbine and the water column prevents charging of the shaft. Usually,
turbo-generators have earthing brushes at the turbine end of the generator.
It is common practice to earth the shaft at only one point to avoid
circulating currents. The remote bearings are insulated from earth. If this
insulation deteriorates, induced shaft voltages may be high enough to
penetrate the oil film in the bearing and some current starts to flow. The
current causes pitting of the bearing surfaces. Essentially, pitting continues
until the bearing loses its low coefficient of friction, the friction losses
increase, the bearing surface breaks up and the surface is wiped. Generally,
a bearing will be damaged within second if the shaft current is higher than
2 A.

References

[1] Buckley, G.W., Corkins, R.J. & Stephens, R.N.: "The

Importance of Grounding Brushes to the Safe Operation of Large
Turbine Generators", IEEE Trans. on Energy Conversion, vol. 3,
no. 3, pp. 607-612, September, 1988.
[2] Pohl, R.: "Eisenverbrennung durch Lichtbögen niedriger Strom-
stärke", AEG Mitteilungen, vol. 20, no. 1, pp. 36-41, January,
1930.

GENPC030.DOC 2007-05-22 15.25

 4.1

4. FAULT STATISTICS
It is an advantage to know the failure rate, the type of failures and the
causes of failures when we design the protection system for a power
system. Sometimes, it is a necessary condition for the design of a cost-
effective protection system to have such fault statistics available. There is,
however, limited information on generator faults.
This section contains a review of the available sources. It also con-
tains a compilation of failure rates for synchronous machines. The rate of
external faults and system abnormalities depends on the power system. We
do not have the ambition to discuss these rates here. It is believed that the
material allows us to carry out simple probabilistic risk analyses. We have
to be careful when using the results of the analysis. One should not assume
that the calculated results have better accuracy than a factor of two.

4.1 Concepts and Definitions

Vetter [ 7 ] discuses the definitions that RWE in Germany used to collect
fault data. Unfortunately the paper does not contain numerical values of the
failure rate for synchronous generators.
In 1979, CIGRE Study Committee 11 (Rotating Machines) formed a
Working Group, WG 11.08 Reliability (of rotating machines). The goals of
the working group included: 1) to examine, review and recommend
reliability definitions, 2) to examine data collection systems, 3) to collect
statistical operating data on reliability, 4) to recommend ways to improve
the collection and analysis of reliability data.
Manufacturers have expressed their concern that the reports give in-
sufficient details and that the data often have poor quality. The manufactur-
ers have also claimed that the data have a bias towards the users. The data
do not recognise the contributions of operator errors. Working Group 11-08
has concluded [ 1 ] that manufacturers cannot expect to obtain enough de-
tails concerning the root causes from operating statistics to permit them to
improve specific design features. However, there should be sufficient infor-
mation for them to identify weak components and frequent failure modes.
It is a fact that the average unavailability and forced outage rate of
the generators from year to year, even when a large sample size is involved.
Also, a few of the infrequent but serious faults dominate the statistics. For
these reasons, lifetime statistics or ten year averages are considered more
valuable than annual reports.
CIGRE WG38-03 has compiled an Application Guide [ 2 ] on Power
System Reliability Analysis. It contains a wealth of information about reli-

GENPC040.DOC 2007-05-22 15.25

4.2 Fault Statistics

ability concepts and methods. There are also some examples of reliability
data in the Application Guide [ 2 ]. These data do, however, not allow us to
estimate the failure rate of synchronous generators and, especially not, how
often generator protection systems have to operate.

4.2 The failure rate of turbo-generators

This section contains some data about failure rates for turbo-generators.

4.2.1 CIGRE-Questionnaires 1976-1985

In reference [ 1 ], Jeffreys summarises the results of several CIGRE-ques-

tionnaires. Figure 1 reproduces some of his reliability data.

MTTF for Turbo-Generators

50
CIGRÉ 1989

40

30
Hours*1000

20

10

0
Canada France India Sweden UK USA

Figure 1 Mean Time To Failure for turbo-generators.

We can estimate the failure rate at 42 faults/(generators, years) for the

turbo-generators in France, UK and USA.
Table 1 reproduces a breakdown [ 1 ] of the number of faults and the
repair time for turbo-generators by major components.
About 50% of the faults and the repair time are faults that the genera-
tor protection system has to detect. This means that the failure rate is close
to 20 faults/(100 generators, year) when we consider only electrical faults.

GENPC040.DOC 2007-05-22 15.25

Fault Statistics 4.3

Table 1 Breakdown of turbo-generator reliability.

Component Repair Time Number of Events
% %
Field winding 21 6
Rotor and attachments 17 4
Stator winding 19 5
Stator core and frame 5 1
Auxiliary systems 23 38
Excitation and brushgear 7 32
Miscellaneous 8 14
Total 100 100

4.2.2 BBC

Kramer and Reinhard [ 3 ] describe the system that BBC used to collect
fault data for steam turbines and turbo-generators. The paper contains only
limited information on failure rates for turbo-generators. Table 2 gives the
availability and forced outage rate (FOR) for turbo-generators. The data
represents five calendar years, 1968-72.

Table 2 Failure statistics for turbo-generators, 1968-72.

Rated Power Service Time Availability FOR
MVA h % %
175 207 569 97.9 0.3
350 109 918 97.3 0.08

Table 3 gives availability for turbo-generators. The data base contains

faults from the commissioning year and up to and including 1972. No data
are older than 15 years.

Table 3 Failure statistics for turbo-generators, -1972.

Rated Power Service Time Availability
MVA h %
175 544 417 96.3
350 116 626 96.5

The total exposure corresponds to a service time of 75 generator-years.

GENPC040.DOC 2007-05-22 15.25

4.4 Fault Statistics

4.2.3 Sweden

A working group [ 11 ] in Sweden has compiled some failure data for

turbo-generators in nuclear power stations. The data base comprises data
from four nuclear power stations: Barsebäck, Forsmark, Oskarshamn and
Ringhals. Table 4 comprises a summary of the data base.
Table 4 Faults on turbo-generators in Sweden.
Unit From To Ng Ny Nf λ
year year # gen.y # %/y
Barsebäck 1 1975 1990 1 16 1 6.25
Barsebäck 2 1977 1990 1 14 2 14.29
Forsmark 1 1980 1990 2 22 0 0.00
Forsmark 2 1980 1990 2 22 1 4.55
Forsmark 3 1985 1990 1 6 0 0.00
Oskarshamn 1 1972 1990 2 38 3 7.89
Oskarshamn 2 1974 1990 1 17 0 0.00
Oskarshamn 3 1985 1990 1 6 0 0.00
Ringhals 1 1977 1990 2 28 3 10.71
Ringhals 2 1977 1990 2 28 4 14.29
Ringhals 3 1981 1990 2 20 0 0.00
Ringhals 4 1983 1990 2 16 0 0.00
All Units 1972 1990 19 233 14 6.01

Table 4 shows that the total exposure is 19 turbo-generators, 233 generator-

years. The longest individual exposure is from 1972 to 1990. The data base
contains 14 generator faults. This means that the average failure rate is
14x100/233 = 6 faults/(100 generators, year).
The working group used the data base to find out the most common
type of generator fault. Figure 2 shows the relative frequency (%) of turbo-
generator faults in the 12 Swedish nuclear units.
The number of earth-faults in the stator is high, 28.6% of all turbo-
generator faults. This motivates a reliable and sensitive earth-fault
protection system that trips the generator with only a short time-delay.
It is surprising that the number of faults involving loss-of-field is so
high. In Sweden, the risk for voltage collapse sets the transfer limits of the
Trunkline network. The reactive absorption will increase to more than 50%
of the rated capacity of the synchronous generator after the loss-of-field.
Such an increase of the reactive loading may be a severe contingency when

GENPC040.DOC 2007-05-22 15.25

Fault Statistics 4.5

operating close to the voltage collapse limit. These facts and the high rela-
tive failure rate underline the importance of protection against loss-of-field.

Generator faults 1972-1990

50 Sweden, 233 generator-years
%

35.7
40

28.6
30

14.3

14.3
20

7.1
10

0
S. short-c. S. earth-f. R. short-c. R. earth-f. Loss-of-f.

Figure 2 Types of turbo-generator faults in Sweden.

4.2.4 Allianz Versicherungs-AG

References [ 4 ] and [ 5 ] contain data that are useful when analysing the
risks associated with turbo-generator faults. The author presented the mate-
rial at two forums in München arranged by the insurance company Allianz
Versicherungs-AG.
Unfortunately, references [ 4 ] and [ 5 ] contain any data on absolute
failure rates. According to [ 4 ] the number of rotor faults is 80% higher
than the number of stator faults. The costs associated with rotor damages
are 70% higher than the costs associated with stator damages.
Figure 3 shows a breakdown [ 5 ] of the stator damages according to
the cause and the type of faults. The hatched bars show the percentage of
the costs associated with the fault type. The full bars show the percentage
of the number of stator damages.
Figure 3 shows that the costs associated with electrical faults on the
stator winding are considerably larger than the relative number of faults.
Reference [ 5 ] does not give the distribution of earth-faults and interturn
faults. The data show the importance of fast and reliable protection against
stator earth-faults and stator short-circuits.

GENPC040.DOC 2007-05-22 15.25

4.6 Fault Statistics

Stator Damages
60 Allianz Versicherungs-AG, 1969-75
%
Legend
50
Cost

40 Number

30

20

10

0
A B C D

A) Earth-faults and Interturn Faults B) Abnormal Current Forces

C) Testing of Windings D) Miscellaneous

Figure 3 Relative cost and number of stator damage.

Reference [ 5 ] does also contain data on rotor damages. Figure 4 shows a

breakdown by causes and types of damages. The hatched bars show the
percentage of the costs, associated with the faults that the insurance
company has to compensate for. The full bars show the percentage of the
number of rotor damages that the owner has reported to the insurance
company.

Rotor Damages
60 Allianz Versicherungs-AG, 1969-1975
%
50
Legend
40 Cost
Number
30

20

10

0
A B C D

A) Interturn Faults, Winding Breaks and Double Earth-Faults B) Retaining

Ring and End Regions C) Series Faults D) Miscellaneous

Figure 4 Types of rotor damages.

GENPC040.DOC 2007-05-22 15.25

Fault Statistics 4.7

Figure 4 shows that the costs associated with electrical faults are consider-
able higher than the relative number of faults. It is not very surprising that
the retaining rings and the end regions of the rotor represent a considerable
share of the costs associated with the rotor damages. The data emphasises
the importance of fast and reliable rotor earth-fault protections and loss-of-
excitation protections.

4.3 The Failure Rate of Hydro-Generators

This section contains some data about failure rates for turbo-generators.

4.3.1 CIGRE-Questionnaires 1976-1985

In reference [ 1 ], Jeffreys summarises the results of several CIGRE-ques-

tionnaires. Table 5 and Table 6 reproduce some of his reliability data.

Table 5 Reliability records for hydro-generators.

Country Sn Exposur Forced Sche- Mean Time Mean
e Outage duled Between Time To
Time Outage Failures Repair
Time
MVA Unit- h h h h
years
Brazil 10-100 118 2,316 21
-"- >100 224 3,592 45
1
Canada All 3,054 52 11,217 67
-"- All 732 52 14 3,970 82
-"- 10-100 54 25 54
-"- >100 12 40 36
Germany >100 144 144 154 113,880 1,880
Japan A 12 7,111 26
-"- B 80 16,926 11
-"- D 1,012 50,211 12
-"- E 60 26,364 4
-"- G 192 3,756 41
1) Canada has only scheduled outage time in second line in summary.
2) Only major outages.

GENPC040.DOC 2007-05-22 15.25

4.8 Fault Statistics

Table 6 Reliability records for hydro-generators (contd.).

Country Sn Exposur Forced Sche- Mean Time Mean
e Outage duled Between Time To
Time Outage Failures Repair
Time
MVA Unit- h h h h
years
Norway 10-100 1,615 16 92,467 166
-"- >100 210 58 39,991 263
Sweden 10-100 888 11 40,941 53
-"- >100 181 63 16,868 121
Switzerland 10-100 48 12 608 20,148 72
USA All 1,353 51 189 13,253 78

From these data we can estimate the failure rate at 60 faults/(100 genera-
tors, year).
Table 7 shows the distribution of number of faults by components.

Table 7 Number of faults for hydro-generators.

Canada Norway Sweden USA
All Units 10-100 >100 10-100 >100 All
Units
MVA MVA MVA MVA
Stator exclud. sta- 4 8 4 6 5 0
tor winding
Stator winding 9 4 15 25 16
Rotor 7 11 15 10 14 6
Bearing 21 6 13 10 4 18
Cooling 6 30 35 27 21 18
Excitation 59 11 24 5 1 35
Other 3 25 5 0 0 7
Components
Total 100 100 100 73 70 100

GENPC040.DOC 2007-05-22 15.25

Fault Statistics 4.9

Table 8 shows the distribution of forced downtime for hydro-generators by

components.
Table 8 Forced downtime for hydro-generators.
Can- Norway Sweden USA
ada
All 10- >100 10- >100 All
Units 100 100 Units
MVA MVA MVA MVA
Stator exclud. 45 19 6 15 0 5
stator winding
Stator winding 39 15 57 75 66
Rotor 8 14 56 50 18 11
Bearing 23 8 17 10 4 10
Cooling 2 2 2 5 4 4
Excitation 20 12 4 5 1 4
Other Compo- 2 6 0 0 0 2
nents
Total 100 100 100 142 102 102

4.3.2 Nordel

Nordel compiles failure statistics from Denmark, Finland, Norway and

Sweden on faults that cause disturbances on the power systems with a rated
voltage from 40 to 400 kV. The latest available report [ 9 ] comprises
statistics for the period 1982-91.
Figure 5 shows the failure rate [faults/(100 generators, year)] for hy-
dro-generators with a rated capacity, S greater than 5 MVA. At the end of
1991, the number of hydro-generators in Finland was 87, in Norway 672
and in Sweden 167.

GENPC040.DOC 2007-05-22 15.25

4.10 Fault Statistics

Hydro-Power Generators
50
Nordel 1982-91 ( S > 5 MVA )

35.6
40
Faults/(100 gen., year)

30

18.3
15.1
20
5.7

10

0
Finland Norway Sweden Nordel

Figure 5 Failure rate for hydro-generators, Nordel 1982-91.

We can see that the failure rate corresponds to a mean time between
failures of 5 years. There is considerable variation in the failure rate from
one country to another. Figure 6 shows the cause to the failures of the
hydro-generators mentioned above.

Hydro-Power Generators
60
Nordel 1982-91 ( S > 5 MVA )

%
45

40
26

21

20
6

2
1

0
Ligthning Nature Sabotage Staff Equipment Other

Figure 6 Causes to hydro-generator faults, Nordel 1982-91.

GENPC040.DOC 2007-05-22 15.25

Fault Statistics 4.11

We can see that failure of the technical equipment is the most common
cause. It is also obvious that the power houses provide ample protection of
the hydro-generators against lightning, nature and third party.

4.3.3 Sweden

Reference [ 8 ] contains some data on failure rates for hydro-generators in

Sweden. The data base contains faults that occurred during the period
1964-1978 and the average failure rates are:

λp = 7.9 faults/(100 generators, year) for permanent faults,

λt = 6.7 faults/(100 generators, year) for temporary faults

λa = 14.6 faults/(100 generators, years) for all faults.

In reference [ 8 ], it is not possible to find the distinction between

permanent and temporary faults. The failure rate, in reference [ 8 ], includes
both electrical faults and other faults.
According to Siljeholm [ 6 ], the failure rate is 10.7 faults/(100 gene-
rators, year) for all hydro-generators owned by Sydkraft.

4.3.4 Norway

Samkjöringen in Norway compiles [ 10 ] disturbance and fault statistics for

the Norwegian main power system. This power system comprises power
lines with a rated voltage equal to 40 kV or more. It also comprises genera-
tors with a rated capacity of 5 MVA or more. Finally, it comprises shunt
capacitors with a rated capacity of 10 Mvar or more. The average failure
rate for hydro-generators during the period 1980-84 is 15.8 faults/(100
generators, years). For the same period, the average failure rate for
synchronous condensers is 26.3 faults/(100 machines, years).

4.4 Summary
The probability of an electrical failure of a given synchronous generator is
low. One can relate the failure rate of a synchronous generator to the failure
rate of 400 kV overhead line. Let us assume that the failure rate of 400 kV
overhead lines is 0.5 faults/(100 km, year). We assume that the failure rate
of a synchronous machine is 15 faults/(100 generators, year). The failure
rate of the synchronous machine then corresponds to the failure rate of a
15/0.5 = 30 km long 400 kV overhead line. It is common practice to use
GENPC040.DOC 2007-05-22 15.25
4.12 Fault Statistics

circuit-local back-up protection on such transmission lines. Few protection

engineers would argue that the failure rate is so low that we can
compromise on the demand to use one main protection and one back-up
protection.

References

[1] Jeffreys, R.: "Reliability of rotating machines", ELECTRA, no.

126, pp. 34-53, October, 1989.
[2] Kelley-Régnier, L. (ed.): "Power System Reliability Analysis
Application Guide", Report, Working Group 38-03, CIGRE,
1987.
[3] Kramer, H. & Reinhard, K.: "Schaden- und Zeitferfügbarkeits-
Statistik an Dampfturbinen und Gereratoren", BBC-Nachrichten,
vol. 56, no. 3, pp. 83-91, March, 1974.
[4] Kugler, H.: "Schäden an Turbogeneratoren", Der
Maschinenschaden, vol. 45, no. 5, pp. 179-188, May, 1972.
[5] Kugler, H.: "Schäden an Turbogeneratoren", Der
Maschinenschaden, vol. 49, no. 6, pp. 221-235, June, 1976.
[6] Siljeholm, Ö.: "Driftstörningar på Sydkrafts huvudsystem 1983",
Rapport DN-8412-22, Sydkraft, 1984-12-21.
[7] Vetter, H.: "Verfügbarkeit großer Maschineneinheiten", Elektriz-
itätswirtschaft, vol. 66, no. 24, pp. 744-752, 1967.
[8] "Elkraftteknisk handbok. Vol. 4, Elkraftsystem", Esselte
Studium, Uppsala, 1984.
[9] "Nordel Driftstörningsstatistik - Fault Statistics", Nordel, 1991.

[ 10 ] "Statistikk over feil under driftforstyrrelser i det norske

hovednettet 1984", Report, Samkjöringen av kraftverkene i
Norge, Oslo, May, 1985.
[ 11 ] "Värmekraftaggregatens Reläskyddssystem", The Swedish
Trunklinecommittee, Operations Committee, The Working
Group for Protection Systems, 21 October, 1991.

GENPC040.DOC 2007-05-22 15.25

 5.1

5. BACKGROUND
This section contains a short historical background to the protection of syn-
chronous generators. This section is entirely based on [ 1 ]. Electrical ma-
chines for use in power systems were being produced in the final decades
of the last century. Then, the only electrical protective devices available
were fuses. They were suitable for use with some small motors. Apparently
fuses could not provide adequate protection of generators. They were
particularly unsuitable for use with three-phase machines because the
operation of a single fuse in case of a phase-to-earth fault on a machine
would have caused it to carry unacceptably high negative sequence
currents. In addition the operation of fuses in the three-phase circuits could
not have initiated the opening of field-winding circuits.
In 1902 H.W. Clothier stated in a paper that he presented to the IEE
that for large AC generators it is feasible to do without fuses or other
automatic devices. This view was also expressed by W.B. Woodhouse
during a paper by H.L. Riesley that he presented to the IEE in 1903.
In spite of these views, there were those who felt that protection re-
lays should be used. Reverse-current relays were referred to in a paper by
C.H. Merz and McLellan entitled "Power station design" that was presented
to the IEE in 1904. These relays were directional relays and their task was
to detect current fed from the busbars into faulted machines. The authors
stated that these relays would not detect short circuits within machines be-
cause the voltages would be too low for the relays to function correctly.
The authors nevertheless felt that automatic protective devices were
required. In the discussion on this paper L. Andrew stated that he had
demonstrated a reverse-current device in 1898 but recognised that it still
was not entirely satisfactory. Reverse-current relays, developed by Brown
Boveri, were referred to in a paper that was presented to the British
Association in 1903. Five years later C.C. Garrard referred to the use of
reverse-current relays with 0.5 s time lags to allow for transient effects in a
paper presented to the IEE. Clearly most machines were operating
without effective protective equipment thus far.
In 1910 K. Faye-Hansen and G Harlow presented a paper "Merz-
Price protective gear and other discriminative apparatus for alternating-cur-
rent circuits" to the IEE. They advocated the use of circulating-current
protective schemes (differential relays) on alternators and showed ar-
rangements suitable for delta- and star-connected machines. Even then sev-
eral speakers in the following discussion expressed the view that such
schemes were unreliable and not suitable for applications to alternators.
One speaker, A.E. McKenzie, stated that two 4 MW, three-phase machines
GENPC050.DOC 2007-05-22 15.25
5.2 Background

in Manchester had then been protected by Merz-Price schemes for two

years and that similar schemes were being applied to new 6 MW machines.
He also expressed the view that devices should be provided to detect loss of
excitation. Eventually it became the standard to apply current-differen-
tial protective systems to the main winding of all machines.

References

[1] Wright, A. & Christopoulos, C.: "Electrical Power System

Protection", Chapman & Hall, London, 1993.

GENPC050.DOC 2007-05-22 15.25

 6.1

6. DIFFERENTIAL PROTECTION
A short-circuit between two (or three) phases in a synchronous machine is a
very severe fault. Within a short time, such a fault may cause much
damage. There is a risk that a short-circuit that is not cleared properly may
cause a fire in the power house. All generators must have a short-circuit
protection that detects internal short-circuits quickly, immediately trips the
HV unit circuit-breaker (or the MV generator circuit-breaker), the field
circuit-breaker, and de-energises the fault location. The short-circuit
protection should also shut down the prime mover, turn on CO2 if provided
and give an alarm. The short-circuit protection may also initiate the transfer
of station auxiliaries from the normal supply from the generator terminals
to the reserve supply from the network.
According to Mason [ 3 ], it is the standardised practice of manufac-
tures to recommend differential protection for generators rated at 1 MVA or
higher. In 1948, Harder and Marter [ 2 ] stated that most of the generators
have such differential protection. Above 10 MVA, it is almost universally
the practice to use differential relays [ 5 ].
A rotating machine provides a classical application of differential
protection. Usually, all equipment, the CTs and the circuit-breakers are near
each other. This minimises the possible error due to long cable runs. In ad-
dition, there is only one voltage level involved. This means that the CT
ratio and types can be the same. The CTs may have matched characteristics.
A generator differential protection should have dedicated CT cores and cir-
cuits. Such cores should not be used with any other protections, metres, in-
struments or auxiliary transformers without a careful check on the effect on
CT performance.
The CTs used for the generator differential protections is almost in-
variably located in the buses and leads immediately next to the generator
winding. This is done to limit the zone of protection so a fault in the
generator is immediately identifiable for quick assessment of damage,
repair and restoration of service. The generator buswork is usually included
in some overall differential protection.
It is very important the differential protection does not operate in
case of external short-circuits. This is important in power stations with
several large generating units. Most power systems are not designed to
withstand a line fault and a simultaneous loss of a generating unit (several
generating units). There is a risk that the generator differential protection
maloperates if a CT saturates. Close to power stations, the time constant, of
the DC component of the short-circuit current, may be very long and in the
order of 100 to 150 ms. External short-circuits with fully developed DC
GENPC060.DOC 2007-05-22 15.26
6.2 Differential Protection

component puts severe demands on CT and differential protection. Walker

[ 4 ] states that the differential protections for the generators in the Snowy
Mountain complex are stable even if the through fault current amounts to
10 times the rated current of the generators.
The generator differential protection can be very rapid and sensitive.
During many years, it has been possible to obtain a sensitivity of 5 to 10%
of the rated current of the generator and an operating time of 100 to
200 ms. The long operating times were necessary to avoid unwanted
operation at external short-circuits when the DC component has a long time
constant. According to [ 1 ], modern generator differential protection may
have a sensitivity of 2 or 3% of the rated current of the generator. The
generator differential protection RADSG has an operate time of 1 to 3 ms.
There may be more than one differential protection in a power
station. In such cases, one differential protection is associated with the
generator and one with the generator step up transformer. Several
transformers may feed power to the auxiliaries. Such transformers may
have their own differential protections. Sometimes an overall differential
protection covers the generator, the buswork and the step up transformer.
The sensitivity of the generator step up transformer differential may be
about 20% of the rated current of the transformer. This differential
protection must not trip when we energise the generator step up transformer
and a high inrush current flows from the network to the transformer. After
synchronisation, a small inrush current flows through the generator
differential protection but it does not place severe demands on the
differential protection. We have to use interposing current transformers in
the overall differential protection and in the transformer differential
protection. Some digital protections, which are available on the market,
may not require external interposing transformers. We never delay the
generator differential protection and the transformer differential protection.
The overall differential protection provides back-up protection. The
sensitivity of the overall differential protection is equal to the sensitivity of
the transformer differential protection. We observe that the overall differen-
tial protection may operate non-selectively if we do not delay its tripping
signal. This is the case when two generators share the same step up trans-
former and a short-circuit hits one of the two generators. Such an non-
selective function may prevent tripping to houseload. The failure to trip to
houseload may delay the restart of nuclear units. We may delay tripping
signal from the overall differential protection. With this, we may avoid
non-selective operations. The delay makes it easier for the operating
personnel to interpret the relay indications and locate the fault. The fault
current on the low voltage side of the step up transformer may be very high.

GENPC060.DOC 2007-05-22 15.26

Differential Protection 6.3

Faults on the generator buswork or the generator circuit-breaker may

subject personnel and equipment to very high stresses. The delay of the
overall differential protection may increase those stresses. Generally, we
recommend that the overall differential protection trips the unit circuit-
breaker without any delay. The selectivity and the easier fault location are
not as important as the reduced stresses.

GENPC060.DOC 2007-05-22 15.26

6.4 Differential Protection

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[2] Harder, E.L. & Marter, W.E.: "Principles and Practices of Relay-
ing in the United States", AIEE Trans., vol. 67, pp. 1075-1081,
1948.
[3] Mason, C.R.: "The Art and Science of Protective Relaying",
John Wiley and Sons, New York, 1956.
[4] Walker, C.W.: "Relay Protection in Hydro-Electric Power
Stations of the Snowy Mountains Authority", The Institution of
Engineers, Australia, Electrical Engineering Transactions, vol.
EET-5, no. 2, pp. 311-316, September, 1969.
[5] "Relay Protection of A-C Generators", AIEE Committee Report,
AIEE Trans., vol. 70, pt. I, pp. 275-282, 1951.

GENPC060.DOC 2007-05-22 15.26

 7.1

7. UNDERIMPEDANCE PROTECTION
The line protections shall detect all shunt faults on the transmission
network and to trip associated circuit-breakers. The back-up protection
system must operate when a line protection fails to operate or when a
circuit-breaker fails to interrupt the fault current. Often, the generator
underimpedance protection is one part of this back-up protection system.
Sometimes, there is no overall differential protection and we need another
type of back-up protection when the main protection fails to operate.
Table 1 shows the steady-state short-circuit currents from generators
at a three-phase short circuit on the generator terminals. We assume that the
excitation voltage is constant.

Table 1 Steady-state short-circuit currents from generators.

Generator Steady-state short-circuit current [pu]
No-load excitation Full-load excitation
Turbo-generators 0.5 1.5
Hydro-generators 1.0 2.0

Nowadays, most synchronous machines have static excitation systems. It is

normal practice to feed the excitation system from a dedicated excitation
transformer connected to the generator buswork. The generator voltage
falls to about 25% of nominal voltage on a three-phase short-circuit on the
high voltage side of the step-up transformer. In such cases, the excitation
system cannot maintain the excitation voltage. This means that the fault
current may fall far below the values given above. Therefore, it is seldom
possible to use a simple overcurrent protection as back-up protection.
We can attain reliable back-up protection by using underimpedance
protections. The sensitivity of the underimpedance protection must be high
because it has to detect short circuits at the remote end of all transmission
lines that stars at the power station. According to [2], one should delay the
underimpedance protection at least 0.8 seconds, perhaps 2 to 3 seconds. We
may compare the tripping from the underimpedance protection with the
tripping from Zone 3 of distance protections.

GENPC070.DOC 2007-05-22 15.26

7.2 Differential Protection

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.

GENPC070.DOC 2007-05-22 15.26

 8.1

8. OVERCURRENT PROTECTION
The Swedish electrical safety regulations demand that all generators have
an overcurrent protection in each phase. These regulations also demand that
all generators have switching devices that disconnect all phases of the
generator from the rest of the power system. A unit connected generator
and transformer may use common protection and switching devices.
Many old synchronous generators have rotating exciters. Often, these
generators have simple overcurrent protections [ 1 ]. The task of this over-
current protection is to provide back-up protection for internal short-
circuits and for external shunt faults. Generally, the setting is 1.5 In where In
is the rated current of the generator. Often, the time delay is 1.5 seconds.
The introduction of static exciters made it necessary to replace the
overcurrent protections by underimpedance protections. Short-circuits close
to the power station may reduce the generator terminal voltage. The
reduced terminal voltage reduces the excitation current. The fault current
may fall below the rated current of the generator and the overcurrent
protection may fail to operate. Underimpedance protections may operate
even if the fault current is lower than the rated current of the generator.
Large generators are often equipped with differential protection that
serves as main short-circuit protection. It is good engineering practice to
equip the generator with short-circuit back-up protection. The overall
differential protection is the best back-up protection. All generators do not
have such back-up protections. Generators without overall differential
protection must have underimpedance protection or overcurrent protection
as back-up protection.
When the excitation current depends on the generator terminal volt-
age, a simple overcurrent protection may not be reliable enough. As de-
scribed in Section 7, underimpedance protections may provide reliable
back-up protection. The combination of overcurrent and undervoltage
relays may provide better short-circuit back-up protection than simple
overcurrent protections. We can use voltage-controlled overcurrent
protections where the overcurrent relay is disabled until the generator
terminal voltage drops below the set level of the undervoltage relay. We
can also use voltage-restrained overcurrent protections where the pickup
value of the overcurrent relay is proportional to the generator terminal
voltage.
Persisting overcurrents in the interval [1.0In,1.4In], where In is the
rated current of the generator, are not detected by overcurrent protections or
by underimpedance protections. Such overload must be detected by dedi-
cated thermal overload protections or by over-temperature sensors. Modern

GENPC080.DOC 2007-05-22 15.26

8.2 Overcurrent Protection

overload protections may have an adjustable time constant that can be ad-
justed to the thermal properties of the generator.

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.

GENPC080.DOC 2007-05-22 15.26

 9.1

9. INTERTURN FAULT PROTECTION

An interturn fault is a short circuit between two points on the same phase
winding in an electrical machine. Figure 1 shows an interturn fault on phase
C in a synchronous generator.

INTERTURN FAULT

Phase Windings
A

N B

Figure 1 Interturn fault in a synchronous generator.

Interturn faults are quite rare but they have occurred and will continue to
occur. Buttrey, Hay and Weatherhall [ 1 ] have provided some information
concerning the failure rate of interturn faults. They point out that it is diffi-
cult to identify faults that were initiated by interturn faults. Usually, the
source of initiation of a stator winding fault is destroyed by the subsequent
damage. During the period 1970-74, thirteen major dielectric breakdowns
have occurred on stator windings of machines rated at 200 MW or above.
The number of exposed generators are not given. We have estimated the
number of exposed generators at 50. This means that the failure rate of
interturn fault may be in the order of 5.2 interturn faults/(100 generators,
year).
Buttrey, Hay and Weatherhall [ 1 ] have also provided some informa-
tion concerning the magnitude of the fault current. They state that it may be
in the order of 100 kA on a 500 MW generator. Let us assume that the
power factor (cos(ϕ)) is 0.8 and that the rated voltage of the generator is
20 kV. Then, the fault current is about 5 per unit. Such high local fault cur-
rents can cause severe damage to the iron core. The fault clearance time
must be as short as possible.

GENPC090.DOC 2007-05-22 15.26

9.2 Interturn Fault Protection

Modern medium size and large size turbo-generators have stator

windings with only one turn per phase per slot. According to Sarma [ 4 ]
and Warrington [ 8 ], such generators do not need interturn protection
because interturn faults cannot occur without involving earth.
The longitudinal differential relay described in Section 6 will not de-
tect interturn faults because there is no difference in the currents at the ends
of a winding with shorted turns. An interturn fault would have to burn
through the major insulation to earth or to another phase before the longitu-
dinal differential relay can detect the fault. Many utilities have disregarded
interturn faults on the basis that if they occur they will quickly develop into
earth faults [ 9 ]. This is probably true if the fault is in the slot portion of
the winding. It will, however, take a little longer in the end portion of the
winding.
An approach of this kind is never attractive and may be entirely un-
justified. There is a possibility of the machine being seriously damaged be-
fore the fault evolves so that can be detected by the longitudinal differential
protection or by the earth fault protection. An instantaneous and sensitive
interturn fault protection would contribute to the reduction of the resulting
damage to the windings and the iron core. An interturn protection will also
provide backup protection for some phase-to-phase faults.
We may use various protection schemes to detect interturn faults.
Usually, they require additional bushings and terminals. Turbo-generators
have physical restrictions that make it difficult to bring out additional
terminals. Because of the general use of hydrogen cooling on large turbo-
generators, it is not practical to bring out additional terminals from the
windings for installing current balance schemes to detect interturn faults.
Some hydro-generators have stator windings with only one turn per
phase per slot. Many large hydro-generators have phase windings with two
or more parallel circuits. According to Mason [ 3 ], many hydro-generators
in Canada have interturn protection as described in [ 5 ].
Wang, Zhang, Wang and Yu have documented [ 7 ] an analysis of in-
ternal faults in a synchronous generator. They have analysed a machine
with windings with six parallel circuits per phase. There are six current
transformers per phase at the neutral end of the phase winding and one
current transformer that senses the current to the neutral point resistor.
There is one current transformer per phase at the line end of the phase
windings. There are therefore 22 current transformers.

GENPC090.DOC 2007-05-22 15.26

Interturn Fault Protection 9.3

9.1 Transverse Differential Protection

Sometimes each phase winding comprises two identical circuits connected
in parallel. A transverse differential protection can detect interturn faults in
such machines, if the windings are brought out separately. Figure 2 shows a
transverse differential protection that uses current transformers connected
at the line end of the phase-windings. A transverse differential protection
can also use current transformers connected at the neutral end of the phase-
windings.

TRANSVERSE DIFFERENTIAL PROTECTION

R R R

A B C

Figure 2 Transverse differential protection.

Balanced current in the two windings produces a circulation of current in

the current transformer secondary circuit. An interturn fault will result in a
circulation of current between the windings. This will produce a current in
the operate winding of the transverse differential protection. The interturn
protection will also detect some phase-to-phase short circuits. It will,
however, not always provide as good protection as the longitudinal
differential protection. One can use a dependent time overcurrent relay as
the measuring relay R. Time delay reduces the main advantage of the
interturn protection, namely the instantaneous tripping.
It is also possible to use an instantaneous differential relay. Here a bi-
ased system should always be used, as it is not possible to guarantee in ad-
vance that exact current sharing between the windings will take place. A

GENPC090.DOC 2007-05-22 15.26

9.4 Interturn Fault Protection

small error in this sharing would cause unwanted operation of the

transverse differential protection.
All generators in the Snowy Mountain complex have windings with
two parallel circuits [ 6 ]. Each half of each phase winding is brought out
separately at the neutral end. The necessary current transformers, with
cross-connected secondaries, are situated ahead of the neutral point. In the
earlier stations, Snowy Mountains Hydro-Electric Authority (SMHEA)
installed biased-differential relays for this application. The reason was that
the restraint feature would take care of any unbalance that might be present
between the two halves of the winding. This proved to be a false premise at
light load, however, due to the then unappreciated nature of the harmonic
currents that circulate in any generator with split-phase winding. An
interesting property of the interturn protection was discovered at Tumut 1
when an operator inadvertently tripped the field breaker of a generator. The
generator was running at no-load while disconnected from the rest of the
power system. This action resulted in unwanted operation of the interturn
protection on all three phases. Subsequent tests showed that, immediately
following the field breaker trip, the circulating currents did increase to a
value above the operate current. At that time the operate value was 5% of
the rated phase current. The rapid rate of change of the field voltage during
the field suppression process caused circulating currents in the split-phase
winding. SMHEA increased the operate value of the interturn protection to
7.5% on all machines at Tumut 1 and later at Tumut 2. The incident
showed very clearly that the previously supposed advantage of employing a
biased-differential relay for interturn protection was far from real. The
small amount of restraint at light load is negated by the summation of the
circulating currents in the operating coil of the protection. Therefore,
SMHEA installed high-impedance transverse differential protection at
Murray 1 and subsequent stations.
Bär, Grau and Kienast [ 2 ] describe the protection system for a
pump storage plant with 4 units. Each unit has a rated apparent power of
290 MVA. Each unit has, among other protections, sensitive and percent-
age-stabilised transversal differential protections. The operate current is 5%
of rated phase current. The operate time is 10 ms.

9.2 Zero Sequence Voltage Protection

For generators with windings without parallel circuits we cannot use the
transverse differential protection described above. Reference [ 10 ] docu-
ments an interturn protection for windings with only one circuit per phase.
The protection uses three single-phase voltage transformers connected to

GENPC090.DOC 2007-05-22 15.26

Interturn Fault Protection 9.5

the line end of each winding. One must connect the other primary terminal
of the voltage transformers to the neutral point of the generator. The
protection measures the zero sequence voltage across the machine.
Normally, no zero sequence voltage should exist but a short circuit of one
or more turns on one phase will cause the generated EMF to contain such a
component. Figure 3 shows such a zero sequence protection.

ZERO SEQUENCE VOLTAGE PROTECTION

A B C

Figure 3 Zero Sequence Voltage Protection.

The voltage transformers have a broken-delta connected secondary winding

that energises a relay. Therefore, the relay receives a quantity that is
proportional to the zero sequence component of the voltage across the
machine.
An earth fault will also produce a zero sequence voltage on the gen-
erator terminal. Most of the voltage will be expended on the earthing resis-
tor. The zero sequence voltage drop in the machine will be small. The zero
sequence component will be limited to 1 or 2% [51]. Therefore, it is prefer-
able to measure the zero sequence drop across the generator windings,
rather than the zero sequence voltage to earth at the line terminals. This is
done by a voltage transformer connected to the line terminals, with the neu-
tral point of the primary winding connected to the generator neutral, above
the earthing resistor.
The third harmonic component of the EMF will appear as a zero se-
quence voltage. It is likely that the amount of the third harmonic will
exceed the required operate voltage. We can introduce a filter that removes
GENPC090.DOC 2007-05-22 15.26
9.6 Interturn Fault Protection

most of the harmonic content in the secondary voltage from the broken-
delta connected winding. We can also provide a filter to extract the third
harmonic component from the broken-delta connected winding and apply it
as a relay bias [51].

GENPC090.DOC 2007-05-22 15.26

Interturn Fault Protection 9.7

References

[1] Buttrey, M, Hay, D. & Weatherall, P.M.: "Generator Interturn

Protection", First International Conference on Developments in
Power System Protection, 11-13 March, 1975, IEE Conference
Publication Number 125, pp. 42-49, London, 1975.
[2] Bär, G., Grau, H.-W. & Kienast, L.: "Der Generatorschutz in
elektronischer Bauweise im Pumpspeicherwerk Wehr", Elektriz-
itätswirtschaft, vol. 78, no. 5, pp. 147-154, 1979.
[3] Mason, C.R.: "The Art and Science of Protective Relaying",
John Wiley and Sons, New York, 1956.
[4] Sarma, M.S.: "Synchronous Machines (Their Theory, Stability,
and Excitation Systems)", Gordon and Breach, New York, 1979.
[5] Sills, H.R. & McKeever, J.L.: "Characteristics of Split-Phase
Currents As a Source of Generator Protection", AIEE Trans.,
vol. 72, pt. III (Power Apparatus and Systems), pp. 1005-1016,
1953.
[6] Walker, C.W.: "Relay Protection in Hydro-Electric Power
Stations of the Snowy Mountains Authority", The Institution of
Engineers, Australia, Electrical Engineering Transactions, vol.
EET-5, no. 2, pp. 311-316, September, 1969.
[7] Wang, X.H., Zhang, L.Z., Wang, W.J. & Yu, Z.H.: "Research
and Application of Protection Relay Schemes for Internal Faults
in Stator Windings of a Large Hydro-Generator with Multi-
Branch and Distributed Arrangements", Fourth International
Conference on Developments in Power System Protection,
Edinburgh, 11 - 13 April, 1989, IEE Conference Publication
Number 302, pp. 51-55, IEE, London, 1989.
[8] Warrington, A.R. Van C.: "Protective Relays. Their Theory and
Practice. Volume One", 2nd ed., Chapman and Hall, London and
New York, 1968.
[9] "Protective Relays Application Guide", 2nd ed., GEC Measure-
ment, 1975, 5th printing, October, 1983.
[ 10 ] "Reläskydd - HANDBOK FÖR KRAFTFÖRETAG", VAST,
November, 1982.

GENPC090.DOC 2007-05-22 15.26

 10.1

10. PROTECTION AGAINST OPEN CIRCUITS

An open circuit in the windings of a modern generator is not likely
because of the size of the conductors [ 6 ]. Some stator windings with large
cross-sectional area have several parallel circuits. Such a design eases the
construction and reduces the losses caused by the skin effect. It is difficult
to detect an open circuit in one of these circuits and to detect high
resistance in soldering-joints. Such faults may cause considerable damage
before it is detected and cleared. The longitudinal differential protection
cannot detect such faults. The negative sequence protection can detect such
faults.
Some generators have a double-winding design. Each phase has two
separate windings that form two star-connected windings. According to
Tideström [ 5 ], it is possible to protect such generators with a sensitive
protection against open circuits. The neutral point of both windings must be
brought out. Figure 1 shows such a neutral point differential protection.

NEUTRAL POINT DIFFERENTIAL PROTECTION

A B C

Figure 1 Neutral point differential protection.

One current transformer can replace the two cross-connected current trans-
formers if the neutral point is unearthed. This single current transformer
measures the current that flows between the two neutral points. Figure 2
shows such a neutral current unbalance protection.

GENPC100.DOC 2007-05-22 15.27

10.2 Protection Against Open Circuits

NEUTRAL POINT UNBALANCE PROTECTION

A B C

Figure 2 Neutral point unbalance protection.

The neutral point unbalance protection must not operate if harmonic

currents flow from one neutral point to the other. Sometimes the protection
will only give an alarm. The faults that the protection can detect cause,
according to Tideström [ 5 ] no immediate danger to the generator.
Let us now assume that we want to equip a generator with an
interturn protection and protection against open circuits. This means that
each phase must have four current transformers at the neutral end. Open
circuits are, according to Mason [ 4 ], most unlikely in well-constructed
machines.
Generally, utilities in the United States do not install protection
against open circuits [ 6 ]. According to Evenson [ 1 ] protection against
open circuits belongs to the standard protection for generators in Sweden
but such protection is uncommon outside Sweden. In [ 7 ] it is stated that
modern generators in Sweden do not have protection against open circuits.
In Canada, there are generators with very sophisticated interturn pro-
tection and protection against open circuits. Gurney [ 2 ] and Handel [ 3 ]
describe the new hydro power plant Revelstoke. Now, there are four gen-
erating units with a rated apparent power of 485 MVA each. In the future,
there will be six generating units. Each phase winding of the generators has
eight parallel circuits. There are twelve transverse differential protections
that compare the current in two neighbouring circuits. At the neutral end,
two current transformers feed a longitudinal differential protection. At the

GENPC100.DOC 2007-05-22 15.27

Protection Against Open Circuits 10.3

line end there are four current transformers per phase. To detect phase-to-
phase short circuits, there are four three-phase differential protections.

References

[1] Evenson, O.: "Reläteknik för högspänningsanläggningar", Kom-

pendium, Lidingö, 1961.
[2] Gurney, J.H.: "Control and Protection Design of the Revelstoke
Hydroelectric Project", IEEE Trans. on Power Apparatus and
Systems, vol. PAS-104, no. 8, pp. 1987-1997, August, 1985.
[3] Handel, R.D.: "Electrical Design of the Revelstoke Hydroelectric
Project", IEEE Trans. on Power Apparatus and Systems, vol.
PAS-104, no. 8, pp. 2012-2019, August, 1985.
[4] Mason, C.R.: "The Art and Science of Protective Relaying",
John Wiley and Sons, New York, 1956.
[5] Tideström, S. H:son (ed): "Ingenjörshandboken/Allmän Elektro-
teknik", 3rd ed., Nordisk Rotogravyr, Stockholm, 1959.
[6] "Relay Protection of A-C Generators", AIEE Committee Report,
AIEE Trans., vol. 70, pt. I, pp. 275-282, 1951.
[7] "Reläskydd - HANDBOK FÖR KRAFTFÖRETAG", VAST,
November, 1982.

GENPC100.DOC 2007-05-22 15.27

10.4 Protection Against Open Circuits

GENPC100.DOC 2007-05-22 15.27

 11.1

11. STATOR EARTH-FAULT PROTECTION

There are many methods for the detection of earth faults on synchronous
generators. The choice of method depends on the layout of the power
plant. Section 2 contains the single-line diagram of four power plants. To
summarise, there are units with the generator and the step-up transformer
connected as a unit. The generator may or may not have a generator circuit-
breaker. Two or more generators may share a common step-up transformer.
One may connect several small generators to a common generator busbar.
In such cases, the number of step-up transformers may be one or more. The
method for the detection of earth faults depends also on the system earth-
ing. Wilheim and Waters have written a book [ 8 ] on system earthing.

11.1 Earthing of Generating Units

Many small generators have, according to Tideström [ 6 ], solidly earthed
neutral. In such generators, the short circuit protection system can also de-
tect earth faults. A sensitive differential protection can [ 6 ] also detect
earth faults if the rated current of the generator is less than 500 A.
Other generators have high-impedance earthed neutral. This means
that they have unearthed neutral, high-resistance earthed neutral or resonant
earthed neutral. Most generators have high-resistance earthed neutral. One
can use a high voltage resistor and connect it directly to the neutral point of
the generator. It is also possible to use a low voltage resistor and connect it
on the secondary side of a single-phase distribution transformer. The
neutral point of the generator is connected to the primary side of the
distribution transformer. The main task of the neutral point resistor is to
limit the overvoltage on the windings and buswork of the generating units.
Overvoltages on the high voltage side of the step-up transformer may cause
such an overvoltage. The stray capacitances between the high voltage
winding and the low voltage winding of the step-up transformer determine
the magnitude of the overvoltage on the generator winding and associated
buswork. It is often necessary to install surge capacitors on the low voltage
side of the step-up transformer if there is a generator breaker. The highest
overvoltage on the buswork occurs while the generator breaker is open.
There is a rule-of-thumb for the selection of the neutral point equip-
ment. The effective resistance, RN [Ω] seen from the neutral point of the
generator should be equal to the capacitance to earth as in equation ( 1 ).
All capacitances are zero sequence capacitances ( = capacitances to earth
with all phase conductors connected to each other).

GENPC110.DOC 2007-05-22 15.27

11.2 Interturn Fault Protection

1 (1)
RN =
3ω (C w + Cb + C a + Ct )

where
Cw = the capacitance of the generator winding [F/phase],
Cb = the capacitance of the buswork [F/phase],
Ca = the capacitance of the auxiliary transformers [F/phase],
Ct = the capacitance of the step-up transformer [F/phase].

High earth fault currents may damage the iron core if the fault clearance
time is long. The risk for damage is small if the earth fault current is lower
than 15 A when there is an earth fault on one generator phase terminal.
According to Walker [ 7 ], the safe limit is 5 A.
Below, we will discuss only generators with high-impedance earthed
neutral. We describe earth fault protection systems for generators and trans-
formers connected to a unit. Figure 4 in Section 2 shows the single-line dia-
gram for such a power plant with generating units without a generator
breaker. Figure 5 in Section 2 shows the single-line diagram for generators
with a generator breaker. One can use similar earth fault protection systems
in power plants where two or more generators share a the same step-up
transformer. Figure 6 in Section 2 shows the single-line diagram for such a
power plant. Finally, we will describe earth fault protection systems for
generators connected to a common busbar. Figure 7 in Section 2 shows the
single line diagram for such a power plant.

11.2 The Earth Fault Protection System

The task of the earth fault protection system is to detect earth faults on the
winding of the generator, on the associated buswork, on the primary
winding of the auxiliary transformer and on the primary winding of the
step-up transformer. A single phase-to-earth fault will cause an increase of
the voltage on the other phases and on the neutral point. The voltage rise
depends on the fault location and on the fault resistance. The healthy phases
will assume full phase-to-phase voltage if an earth fault without fault
resistance hits the line terminal of one winding of the generator.
Simultaneously, the neutral point will assume full phase-to-neutral voltage.
The voltage rise will decrease when the fault resistance increase. The
voltage rise will be negligible if the earth fault occurs on the phase winding
close to the neutral point.

GENPC110.DOC 2007-05-22 15.27

Interturn Fault Protection 11.3

To detect an earth fault on the windings of a generating unit one may

use a neutral point overvoltage relay, a neutral point overcurrent relay, a
zero sequence overvoltage relay or a residual differential protection. These
protection schemes are simple and have served well during many years.
However, at best these simple schemes protect only 95% of the stator
winding. They leave 5% at the neutral end unprotected. Under
unfavourable conditions the blind zone may extend to 20% from the
neutral. There are several methods to detect an earth fault close to the
neutral point. Figure 1 illustrates some fundamental properties of some
types of earth fault protections. The intention is to illustrate general
methods and define some classes of earth fault protections.

TYPES OF EARTH FAULT PROTECTIONS

A
N
B

C
Line End Earth
Fault Protection
Neutral End Earth
Fault Protection
Combined Earth
Fault Protection
Total Earth
Fault Protection

Figure 1 Types of earth fault protections.

The line end earth fault protections can detect earth faults on almost the
entire generator winding but have a blind zone close to the neutral point.
The size of the blind zone may be 5-20%. The main task of the neutral end
earth fault protection is to detect an earth fault close to the neutral point.
Such protections may cover 20-40% of the winding. Sometimes these
protections can detect earth faults close to the line terminals. To cover the
entire winding, one can use a combined earth fault protection that com-
prises line end earth fault protection and a neutral end earth fault
protection. There are finally total earth fault protections based on a
method that makes it possible to detect earth faults anywhere along the
entire generator winding.
GENPC110.DOC 2007-05-22 15.27
11.4 Interturn Fault Protection

11.3 Line End Earth Fault Protection

Neutral point overvoltage protections, neutral point overcurrent protections,
zero sequence overvoltage protections and residual differential protections
are line end earth fault protections.

11.3.1 Unit Generator-Transformer Configuration

The neutral point overvoltage protection is a common earth fault protec-

tion for unit-connected generators. Figure 2 shows a neutral point overvolt-
age protection.

NEUTRAL POINT OVERVOLTAGE PROTECTION

A B C

N
U>

Figure 2 Neutral point overvoltage protection.

A single-phase voltage transformer connected to the generator neutral ener-

gises the neutral point overvoltage protection. Such protection detects earth
faults on the generator windings, on the buswork and on the primary wind-
ing of the auxiliary transformer. It can also detect earth faults on the
primary winding of the step-up transformer in units without generator
breaker and while the generator breaker is closed. The blind spot near the
neutral may be as small as 5%.
The neutral point overcurrent protection has similar properties as the
neutral point overcurrent protection. The overcurrent protection detects
earth faults on the generator windings, on the buswork, on the primary
winding of the auxiliary transformer and on the primary winding of the

GENPC110.DOC 2007-05-22 15.27

Interturn Fault Protection 11.5

step-up transformer. Figure 3 shows such a neutral point overcurrent

protection.

NEUTRAL POINT OVERCURRENT PROTECTION

A B C

I>

Figure 3 Neutral point overcurrent protection.

A single-phase current transformer energises the neutral point overcurrent

protection. The primary current is equal to the current that flows from the
generator neutral to earth. Evenson states [ 3 ] that the neutral point
overcurrent protection is inferior to the neutral point overvoltage
protection. The blind zone near the neutral may be 20-30%. A zero
sequence overvoltage protection can also detect faults on the generator
system. Figure 4 shows such zero sequence overvoltage protection.
Three single-phase voltage transformers energise the zero sequence
overvoltage protection. The primary winding of each voltage transformer is
connected to a phase conductor. A secondary winding on each voltage
transformer form a broken delta that energises the overvoltage relay.
Each voltage transformer has an amplitude error and a phase error.
This means that the secondary zero sequence voltage may not represent the
primary zero sequence voltage exactly. To avoid unwanted operation, the
zero sequence overvoltage setting must be higher that the neutral point
overvoltage setting.
Lohage and co-workers have documented [ 1 ] practice for the earth
fault protection in hydro power plants owned by Vattenfall (Swedish State
Power Board).

GENPC110.DOC 2007-05-22 15.27

11.6 Interturn Fault Protection

Generally, large hydro power units with generator breakers have one
neutral point overvoltage protection and one zero sequence overvoltage
protection. The neutral point overvoltage protection must cover at least
95% of the stator winding. In new plants a combined earth fault protection
that can detect earth faults anywhere along the generator winding replaces
the neutral point overvoltage protection. The delay for this sensitive
protection is usually 1.2 second. Often, the zero sequence overvoltage
protection covers about 80% of the generator winding. Normally, the delay
is 0.4 second.

ZERO SEQUENCE OVERVOLTAGE PROTECTION

U>

Figure 4 Zero sequence overvoltage protection.

The earth fault protections described above cannot detect earth faults on the
primary winding of the step-up transformer while the generator breaker is
open. To detect such faults, one may use a zero sequence overvoltage pro-
tection. It is connected to the primary winding of the step-up transformer.
The sensitivity is about 80% and the delay is equal to 0.8 second.

11.3.2 Several Generators Connected to a Common Busbar

Let us now consider a power plant with several generators connected to a

common busbar. The busbar has one or more transformer bays. Usually, the
busbar has no feeder bays. In such plants it is common practice that the
generators have unearthed neutral. Often, there is a requirement to limit the
overvoltage on the busbar while only one generator is in service. This case

GENPC110.DOC 2007-05-22 15.27

Interturn Fault Protection 11.7

determines the maximum size of the resistor connected to the neutral point
of the generator. When all generators with such resistors are in service, the
total earth fault current may become too high.
Some busbars may have a bay for an earthing transformer with a neu-
tral point resistor. In such cases, the system has a high-impedance earthed
neutral. Sydkraft uses a combined earthing transformer and station
auxiliary transformer. Some plants may have only a step-up transformer
with a Y- or Z-connected winding connected to the busbar. One can use
this neutral point to connect a neutral point resistor. It is not necessary to
install neutral point resistors at each generator if there is an earthing
transformer with a neutral point resistor of if the step-up transformer has a
neutral point resistor. It is also possible to avoid using neutral point
disconnecters otherwise necessary to limit the earth fault current.
Neutral point overvoltage protections, neutral point overcurrent pro-
tections and zero sequence overvoltage protections cannot select the faulty
generator if several generators are connected to one common busbar.
Figure 5 shows a residual differential protection that can select the
faulty generator when several generators are connected to a common
busbar.
One only needs three-phase current transformer if the neutral point of
the generator is unearthed. Unavoidable amplitude errors and phase errors
limit the sensitivity of the earth fault protection. On external short-circuits,
the fault current from the generator may be very high and may contain a
substantial DC component. The fault currents may cause a false secondary
zero sequence current. There is a risk that this false current will cause un-
wanted operation of the earth fault protection. To avoid such unwanted op-
erations, the short circuit protection may block the earth fault protection on
external short-circuits. The closing of the generator breaker may cause tran-
sient residual currents. These currents may limit the sensitivity of the resid-
ual differential protection.
Let us now assume that each generator has a neutral point resistor.
To obtain selective clearance of earth faults, it is necessary to use a residual
differential protection. Figure 5 shows such a protection that is energised
from three-phase current transformers and one neutral point current
transformer.
Generally, it is necessary to clear earth faults on the buswork and on
the primary winding of the step-up transformers. To detect such faults, one
may use a zero sequence overvoltage protection. Three single-phase
voltage transformers connected to the primary winding of the step-up
transformer may energise the zero sequence overvoltage protection.

GENPC110.DOC 2007-05-22 15.27

11.8 Interturn Fault Protection

RESIDUAL DIFFERENTIAL PROTECTION

A B C
R

Figure 5 Residual differential protection.

11.4 Neutral End Earth Fault Protection

An overvoltage (or overcurrent) generator earth fault protection is straight-
forward, secure, and dependable earth-fault protection. However, it suffers
from two disadvantages [ 4 ]. First, it will not detect earth-faults near the
generator neutral. Second, it is not self-monitoring. That is, an open circuit
anywhere in the relay, primary or secondary of the voltage transformer (the
current transformer) or an open neutral point resistor may not be detected
before a fault occurs.
The induced EMF in a synchronous generator contains harmonics. It
is possible to use the third harmonic to detect earth faults close to the
neutral point and in the neutral point equipment. The induced third
harmonic voltages cause a third harmonic current that flows through the
neutral point resistor.
An earth fault close to the neutral point will shunt the neutral point
resistor and the third harmonic voltage over the neutral point resistor. Ac-
cording to Schlake, Buckley and McPherson [ 5 ] such a third harmonic
protection can detect earth faults with a fault resistance less than 1 000 Ω. It
can detect such faults on 20% of the generator winding near the neutral
point.

GENPC110.DOC 2007-05-22 15.27

Interturn Fault Protection 11.9

11.5 Combined Earth Fault Protection

When combined to form a protection system, each relay covers the blind
zone of the other. Therefore, the combined protection system will detect
earth faults anywhere on the stator winding.
Griffin and Pope describe [ 4 ] the earth fault protections used by
Georgia Power Company. For over 30 years, Georgia Power Company has
grounded all system generators through a distribution transformer with a re-
sistance-loaded secondary. A current transformer is then connected in
series with the secondary resistor to supply current to one or more
overcurrent relays. When properly set, these relays will provide sensitive
protection for 90 to 95 percent of the generator stator winding, and will not
operate incorrectly for external faults. At the end of 1981, this system is
installed on 126 generating units, ranging in size from 15 to 900 MW. In
the past 25 years, nearly 20 earth faults have been cleared with minimal
equipment damage, and no incorrect operations have occurred.
In 1977, Georgia Power Company concluded that it would be
prudent to protect all large generators with an additional earth fault
protection system that was completely independent of the existing
overcurrent scheme, would give reliable protection to 100% of the
generator, and would continuously monitor the generator earthing system.
Two types of systems have been installed. One type injects a current at a
subharmonic current, and trips on an increase of current caused by the
reduction in generator capacitance that results from a single phase-to-earth
fault. The other type employs two overlapping voltage relays - an
overvoltage relay that protects the high voltage end of the machine, and an
undervoltage relay, tuned to respond to the third harmonic, which protects
the neutral. Griffin and Pope states [ 4 ] that both schemes have performed
extremely well, and that the combine overvoltage/undervoltage scheme has
already properly detected an earth fault. It should also be noted that in
1980, a Georgia Power Company generator on which a 100% earth fault
protection had not yet been installed, was badly damaged by a ground fault
that occurred very near the neutral and was not detected by the 90%
protection [ 4 ].

11.6 Total Earth Fault Protection

In 1936, Diesendorf and Groß [ 2 ] pointed out the need for an earth-fault
protection that detects earth faults on the entire stator winding. They ana-
lysed the method to inject a power frequency voltage at the neutral of the
generator.

GENPC110.DOC 2007-05-22 15.27

11.10 Interturn Fault Protection

There are schemes that inject a subharmonic voltage into the pro-
tected plant. An overcurrent relay monitors the subharmonic current that
flows to the protected plant. An earth fault anywhere on the stator winding
will increase the subharmonic current. This scheme provides total coverage
of the entire stator winding. However, the cost of the implementation tends
to be high due to the cost of the injection equipment.

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[2] Diesendorf, W. & Groß, E.: "Zur Theorie der Pohl'schen
Nullpunktsverlagerung für vollständigen Gehäuseschlußschutz",
E und M, vol. 54, no 22, pp. 253-256, 31 May, 1936.
[3] Evenson, O.: "Reläteknik för högspänningsanläggningar", Kom-
pendium, Lidingö, 1961.
[4] Griffin, C.H. & Pope, J. W: "Generator Ground Fault Protection
Using Overcurrent, Overvoltage, and Undervoltage Relays",
IEEE Trans. on Power Apparatus and Systems, vol. PAS-101,
no. 12, pp. 4490-4501, December, 1982.
[5] Schlake, R.L., Buckley, G.W. & McPherson, G.: "Performance
of Third Harmonic Ground Fault Protection Schemes for
Generator Stator Windings", IEEE Trans. on Power Apparatus
and Systems, vol. PAS-100, no. 7, pp. 3195-3202, July, 1981.
[6] Tideström, S. H:son (ed): "Ingenjörshandboken/Allmän Elektro-
teknik", 3rd ed., Nordisk Rotogravyr, Stockholm, 1959.
[7] Walker, C.W.: "Relay Protection in Hydro-Electric Power
Stations of the Snowy Mountains Authority", The Institution of
Engineers, Australia, Electrical Engineering Transactions, vol.
EET-5, no. 2, pp. 311-316, September, 1969.
[8] Wilheim, R. & Waters, M.: "Neutral Grounding in High Voltage
Transmission", Elsevier Publishing Co., New York, 1956.

GENPC110.DOC 2007-05-22 15.27

 12.1

12. FIELD EARTH-FAULT PROTECTION

The field circuit comprises the field winding of the generator and
associated circuits of the exciter. We may classify faults in the field circuit
as open circuits, earth faults and short circuits.
The oil film in the bearings provides some insulation between the ro-
tor and the stator in a generating unit. Charge separation in the later stages
of a steam turbine may cause the blades to pick up electrons and acquire a
negative charge. Most turbine generators have one or more shaft-earthing
brushes to prevent build up of electrical charge on the shaft. This means
that the metallic parts of a turbo-generator have contact with earth via the
shaft earthing brush. The metallic parts of a hydro-generator have contact
with earth via the turbine and the waterways.

12.1 Open Circuits

Open circuits in the field circuit may occur on any type of generator.
Experience has shown that they are more likely to occur on slower speed
hydro-generators [ 12 ]. An open field circuit may cause burning at the fault
location. Besides local damages, an open field circuit causes a complete
loss of excitation. Section 13 contains a discussion on loss of excitation.

12.2 Earth Faults

The field winding is always insulated from the metallic parts of the rotor.
The insulation resistance is high if the rotor is cooled by air or by
hydrogen. The insulation resistance is much lower if the rotor winding is
cooled by water. This is true even if the insulation is intact. A fault in the
insulation of the field circuit will result in a conducting path from the field
winding to earth. This means that the fault has caused a field earth fault.
The field circuit of a synchronous generator is normally unearthed.
Therefore, a single earth fault on the field winding will cause only a very
small fault current. The leakage resistance of the field winding will deter-
mine the size of this fault current. This means that the earth fault current is
larger if the generator has a water-cooled rotor. Even then, the earth fault
does not produce any damage in the generator. Furthermore, it will not
affect the operation of a generating unit in any way. Utilities have, in fact,
operated generators in this condition for considerable periods. However,
the existence of a single earth fault increases the electric stress at other
points in the field circuit. This means that the risk for a second earth fault at
another point on the field winding has increased considerably. A second

GENPC120.DOC 2007-05-22 15.27

12.2 Field Earth-fault Protection

earth fault will cause a field short-circuit with associated consequences

described below.
The safest practice is, however, to trip the generator immediately
when the first earth fault occurs. This practice should certainly be followed
in unattended power plants [ 6 ]. However, some utilities are prepared to
take the risk of a second earth fault and its possible consequences in an at-
tended plant. Most utilities alarm on the indication of the first earth fault
and prepare to remove the unit in an orderly shutdown at the first
opportunity [ 5 ]. If one permits a generator to operate with a single earth
fault on its field winding, it should at least have automatic equipment for
immediate tripping the generating unit at an abnormal amplitude of
vibration. Such vibrations will also occur on synchronising and on external
shunt faults. As usual, one has to strike a balance between the dependability
and the security. The vibration-detection equipment should be in service
continuously and not be put in service manually after the first earth fault [ 6
].

12.3 Short Circuits

Danger arises if a second earth fault occurs at another point on the field
winding. The second earth fault will cause a short-circuit in the field wind-
ing. An overcurrent relay in the field circuit cannot detect a short circuit if
only a few turns are involved or if one pole of a slow-speed hydro-
generator is short-circuited. Furthermore, it is highly desirable for the field
circuit not to be opened during external power system faults. Such faults
may cause high currents to flow in the field circuit. Therefore, field
overcurrent protections are uncommon [ 12 ].
The normal field current of a large generator is considerable. The
fault current caused by the short circuit may very rapidly cause serious
damage at the fault locations.
Still more damage may be caused mechanically. The magnetic flux
of one pole will decrease if some turns of the field winding on that pole
should become short-circuited. If a large portion of the field winding is
short-circuited, the excitation current will be diverted, in part at least, from
the intervening turns. This will redistribute the flux on the rotor.
Consequently, the attracting force is still normal on some poles. The short
circuit will, however, weaken the attracting force on other poles. The result
is an unbalanced force and the number of short-circuited turns determines
the size of this force. The effects of a short circuit are most evident for
turbo-generators with rotors that have few poles. In a large machine, this
force may be about 50 to 100 tons [ 11 ]. The force rotates with the rotor

GENPC120.DOC 2007-05-22 15.27

Field Earth-fault Protection 12.3

and produces violent vibrations. Vibration-monitors may detect the

condition shut down the generating unit or at least give an alarm.
Cases are on record where the resulting vibration has broken bearing
pedestals, allowing the rotor to grind against the stator. Such failures have
caused extensive damage that was costly to repair and that kept the ma-
chines out of service for a long time [ 6 ]. The outage of a large generating
unit may cost several hundred thousands US dollars per day.
A short circuit on the slip rings causes a high fault current that may
overload the exciter. Optical sensors may detect the resulting arc. A short
circuit will give a large signal from the sensor. The optical sensor may also
detect sparks from worn brushes. Now, the signal level will be lower. Usu-
ally, the fault detecting equipment has two levels. The high level should de-
tect short circuits on the slip rings and will trip the unit. The low level
should detect sparks and will give an alarm. Usually, one delays the signal
about 5 seconds to avoid alarm on temporary sparking.

12.4 Demands on Field Earth Fault Protection

It is desirable that an indication of an earth fault should be given so that the
operator may take remedial actions at the earliest convenient opportunity.
The consequences of a double earth fault in the field winding are so severe
that many utilities want to detect the first earth fault and trip the generator.
The large capacitances increase the difficulties to detect earth faults on the
field winding of large generating units[ 8 ].
According to Tideström [ 7 ], the earth fault detector should only
give an alarm. The earth fault detector for the generators in the Snowy
Mountain Complex gives an alarm only [ 9 ]. There is a requirement on the
operator to take the unit out of service orderly with the least delay. Lohage
and co-workers describe [ 1 ] the requirement on the earth fault protection
of generating units owned by Vattenfall. The sensitivity of the earth fault
detector must be higher than 1 to 2 kΩ. The earth fault protection shall trip
the generator. Bär, Grau and Kienast [ 3 ] state that the sensitivity of the
earth fault detector must be higher than 2 kΩ.
Earth fault detection for the exciter and the field winding are impor-
tant and usually supplied as part of that equipment rather than applied by
the user [ 2 ]. However, if not supplied, or if additional protection is
desired, several methods are available to detect an earth fault on the field
circuit.

GENPC120.DOC 2007-05-22 15.27

12.4 Field Earth-fault Protection

12.5 Earth Fault Detectors for Units with Brushes

A simple earth fault detector consists of two lamps connected in series be-
tween the supply connections [ 10 ] and [ 12 ]. There is an earth connection
at the midpoint. Under healthy conditions, both lamps have the same
voltage across them and they should therefore be equally bright. Should a
fault occur on the field winding at any point other than the centre,
unbalance will occur. It will cause one lamp to be brighter than the other,
because of the unequal voltages across them. This earth fault detector gives
a positive indication only for faults near the terminals of the field winding.
One can obtain a more sensitive earth fault detector by replacing the
lamps with resistors and connecting the midpoint to earth through a
sensitive DC-type overvoltage relay [ 12 ]. When an earth fault occurs on
the field winding or exciter circuits, a voltage appears across the relay. It
will operate if the voltage is high enough. This scheme has also the
disadvantage that it will not operate for an earth fault near the midpoint of
the field winding [ 10 ].
Alternative schemes have been developed to eliminate this limitation
[ 10 ]. One arrangement includes a varistor in series with one of a pair of
divider resistors. It changes this blind spot with variations of the field
voltage.
Large generators have more complex arrangements involving AC or
DC auxiliary supply. The AC scheme comprises an auxiliary supply trans-
former. Its secondary winding is connected between earth and one side of
the field circuit through an interposed capacitor and a relay coil [ 11 ]. The
field circuit is subjected to an alternating potential that has practically the
same level throughout. Therefore, an earth fault anywhere in the field
circuit will cause an AC current that is detected by the relay. The capacitor
limits the current and blocks the normal field voltage. It also prevents the
discharge of a large DC current through the auxiliary transformer.
One may avoid the capacitive currents associated with AC injection
by injecting a DC voltage through a resistor [ 11 ]. The injected DC voltage
is arranged to bias the positive side of the field circuit to a negative voltage
to earth. The negative side of the field circuit is at an even greater negative
voltage to earth. This means that an earth fault at any point on the field
winding will cause current to flow through the power unit and the DC-type
relay. The current is limited by including a high resistance in the circuit and
a sensitive relay is used to detect the current.

GENPC120.DOC 2007-05-22 15.27

Field Earth-fault Protection 12.5

12.6 Earth Fault Detectors for Units with Brushless

Exciters
The introduction of brushless exciters has increased the difficulties to
detect earth faults on the field winding [ 8 ].
Generators with brushless exciters are supplied with means to drop
pilot brushes on slip rings to measure the insulation level of the field
winding on a periodic basis [ 2 ].
Ungrad, Winkler and Wiszniewski describes [ 8 ] an advanced earth
fault protection system. Henninger, Knütter and Schmindel [ 4 ] have also
described this protection for generators with brushless exciters. To obtain
good monitoring of the field circuit, the earth fault detector uses pilot slip
rings continuously

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[2] Blackburn, J.L.: "Protective Relaying, Principles and Applica-
tions", Marcel Dekker, New York, 1987.
[3] Bär, G., Grau, H.-W. & Kienast, L.: "Der Generatorschutz in
elektronischer Bauweise im Pumpspeicherwerk Wehr", Elektriz-
itätswirtschaft, vol. 78, no. 5, pp. 147-154, 1979.
[4] Henninger, K., Knütter, E.-F. & Schmiedel, K.: "Two-Step Rotor
Earth-Fault Protection of High Sensitivity for Synchronous Ma-
chines", Siemens Power Engineering, vol. 2, no. 1, pp. 13-17,
January, 1980.
[5] Horowitz, S.H. & Phadke, A.G.: "Power System Relaying", Re-
search Studies Press and John Wiley & Sons, Taunton and New
York..., 1992.
[6] Mason, C.R.: "The Art and Science of Protective Relaying",
John Wiley and Sons, New York, 1956.
[7] Tideström, S. H:son (ed): "Ingenjörshandboken/Allmän Elektro-
teknik", 3rd ed., Nordisk Rotogravyr, Stockholm, 1959.

GENPC120.DOC 2007-05-22 15.27

12.6 Field Earth-fault Protection

[8] Ungrad, H., Winkler, W. & Wisniewski, A.: "Schutztechnik in

Elektroenergisystemen. Grundlagen, Stand der Technik,
Neuentwicklungen", Springer-Verlag, Berlin..., 1991.
[9] Walker, C.W.: "Relay Protection in Hydro-Electric Power
Stations of the Snowy Mountains Authority", The Institution of
Engineers, Australia, Electrical Engineering Transactions, vol.
EET-5, no. 2, pp. 311-316, September, 1969.
[ 10 ] Wright, A. & Christopoulos, C.: "Electrical Power System
Protection", Chapman & Hall, London, 1993.
[ 11 ] "Protective Relays Application Guide", 2nd ed., GEC Measure-
ment, 1975, 5th printing, October, 1983.
[ 12 ] "Relay Protection of A-C Generators", AIEE Committee Report,
AIEE Trans., vol. 70, pt. I, pp. 275-282, 1951.

GENPC120.DOC 2007-05-22 15.27

 13.1

13. UNDEREXCITATION PROTECTION

One can increase the production of reactive power from a synchronous ma-
chine by increasing the excitation current (the rotor current). Now, the ma-
chine acts like a shunt capacitor. A synchronous machine can consume re-
active power if the excitation current is low enough. In this state, the ma-
chine acts like a shunt reactor.
The acceptable limit for overexcitation (reactive production) depends
on the prevailing active power generation (consumption). The acceptable
limit for underexcitation (reactive consumption) may or may not depend on
the active generation (consumption). Figure 1 shows a simplified
capability diagram for a synchronous generator. Manufacturers can
provide accurate capability diagrams for their generators.

CAPABILITY DIAGRAM

Rotor Current Limit Q

Stator Current Limit P

Practical Stability
Limit

Theoretical Stability Limit

Figure 1 Capability diagram for a synchronous machine.

We have plotted the active power, P [MW] along the horizontal axis and
the reactive power, Q [Mvar] along the vertical axis. Here, we have
assumed that the terminal voltage is equal to the rated voltage. Many
capability diagrams have axes scaled in per unit, i.e., P/Sn and Q/Sn where
Sn [MVA] is the rated apparent power of the synchronous machine. The
rated apparent power, Sn, the rated voltage, Un, the synchronous reactance,
Xs, and the power factor (cosϕ) define the acceptable reactive limits in the
simplified capability diagram.
The rated apparent power and the rated voltage determine the rated
stator current. The stator current may only temporarily exceed the rated

GENPC130.DOC 2007-05-22 15.28

13.2 Underexcitation Protection

current. During steady state, the synchronous machine must not operate
outside a circle with a radius equal to 1 per unit.
Let us assume that the active power P is zero. Now, it is not possible
to increase the reactive power to the rated apparent power. One cannot
reach the limit Q = Sn, because the excitation current becomes too high. It
must, however, be possible to operate the machine at P = Sncos(ϕ) and
Q = Snsin(ϕ). This determines the requirement on the excitation system and
the thermal design of the rotor. The reactive limit given by the highest exci-
tation current is also a circle. It passes the point ( P = Sncos(ϕ),
Q = Snsin(ϕ)) and has its centre on the negative reactive axes, theoretically
at Q/Sn = -1/Xs

13.1 Underexcitation of Synchronous Machines

There are also limits for the underexcitation of a synchronous machine. A
reduction of the excitation current weakens the coupling between the rotor
and the external power system. The machine may lose the synchronism and
starts to operate like an induction machine. Then, the reactive consumption
will increase. Even if the machine does not lose synchronism it may not be
acceptable to operate in this state for a long time. The underexcitation in-
creases the generation of heat in the end region of the synchronous
machine. The local heating may damage the insulation of the stator winding
and even the iron core.
International standards do not specify limits for underexcitation [ 4 ].
In UK, it is required that a generator should have a full load under reactive
consumption that at the power factor, cos(ϕ) = 0.95. In the US all large
turbo-generators must operate with reactive consumption at a rated
apparent power up to a limit corresponding to a power factor,
cos(ϕ) = 0.95. At a power factor equal to zero, the reactive consumption
must be about 50% of the rated apparent power, i.e. Q = 0.5Sn. Practically
the same requirements are used in Japan. In France, turbo-generators must
consume about 35% of the rated apparent power at rated active power, i.e.
P = Sncos(ϕ) and Q = -0.35Sn.

13.2 Loss of Excitation

A fault in the Automatic Voltage Regulator (AVR) or in the excitation sys-
tem may cause a total loss of excitation. A short circuit on the slip rings
will reduce the excitation voltage down to zero. This will cause a gradual
reduction of the excitation current and eventually a loss of excitation. An
open circuit in the field circuit will also cause a total loss of excitation. The
GENPC130.DOC 2007-05-22 15.28
Underexcitation Protection 13.3

unwanted opening of the field breaker has caused several losses of

excitation in Sweden. When the field breaker is open, high voltage is
induced in field winding and there is a risk for damages to the discharge
resistor.

13.3 Underexcitation Protection

Undercurrent relays connected in the field circuit have often been used [ 6 ]
for underexcitation protection. According to Mason [ 5 ], the most selective
type of underexcitation relay is a MHO-relay excited from the AC current
and voltage at the main terminals of the generator.
Walker has described [ 8 ] the underexcitation protection for the hy-
dro-generators in the Snowy Mountain Complex. The smaller generators (4
x 80 MW, 4 x 70 MW, 10 x 95 MW and 4 x 138 MW) have a simple DC
undercurrent relay fed from a shunt in the main field circuit. This relay
gives an alarm after a short delay and it is left to the operator to act
appropriately. The larger generators (6 x 250 MW) have a static excitation
system while the smaller have a conventional rotating exciter. Here the
impedance relay arranged for tripping will supplement the DC alarm relay.
Born and Fischer have documented [ 2 ] the background for the de-
sign of an electronic underexcitation relay developed by Siemens. The
introduction of generators with brushless exciter made it necessary to
energise the underexcitation relay from the stator only. Siemens had to
abandon the older approach to use field voltage as one independent
criterion. Figure 2 shows the characteristics of the underexcitation relay.
The characteristics are symmetric around the Q-axes (Q = 0) to suit
generators in pumped storage plants. One can set the distances from the
origin and the slopes individually. The protection system has two tripping
criteria with an adjustable time delay. Bär, Grau and Kienast have
described [ 3 ] use of such a relay in the pumped storage plant Wehr.
The most reliable underexcitation protection is, according to Sarma
[ 7 ], either a MHO-relay or a directional impedance relay with its
characteristic in the negative reactance area. Figure 3 shows the tripping
characteristics for such a MHO-relay for underimpedance protection.
Lohage and co-workers have described [ 1 ] requirements on under-
excitation protection for hydro generating units owned by Vattenfall. They
require that all new hydro-generators must have an underexcitation relay.
Many generators use a directional overcurrent relay for underexcitation
protection. The underexcitation relay shall trip the generator breaker and
start the breaker failure protection. Underexcitation cannot occur while the
terminal voltage is low if the excitation system operates correctly.
GENPC130.DOC 2007-05-22 15.28
13.4 Underexcitation Protection

Therefore, the underexcitation protection should use an undervoltage

criterion. Underexcitation will not cause a low terminal voltage if the
generator is connected to a strong network or if there are several generating
units in the power plant. In such a case, the stator current of the faulty
generator will increase. Therefore, the underexcitation protection should
use an overcurrent relay. Two criteria must be fulfilled before the
underexcitation equipment may trip the generator. The first criterion is the
directional overcurrent relay has operated. The second criterion is that
either the undervoltage relay has operated or the (non-directional)
overcurrent protection has operated.

UNDEREXCITATION PROTECTION

Q/U

P/U

Figure 2 Underexcitation protection.

An AIEE working group has prepared and sent out a questionnaire about
practice and experience of underexcitation protection. The group has ana-
lysed and documented [ 9 ] answers from 63 utilities in the US, Canada and
Mexico. The answers represented 309 generators rated at 60 MW or more,
installed since 1949. Already 1958, 72% of these generators had underexci-
tation protection. Of all utilities, 86% reported that their new generators
had underexcitation protection. Most of the underexcitation protections
tripped the generator.
The underexcitation relay tripped many generators during the North-
east Power Interruption on the 9 November 1965 and during the PJM Sys-
tem Disturbance on the 5 June 1967. Many questions were raised concern-
ing the correctness of these relay operations and whether they caused or
aided the disturbance. The IEEE Power System Relay Committee formed a

GENPC130.DOC 2007-05-22 15.28

Underexcitation Protection 13.5

working group to investigate and report on the performance of underexcita-

tion protection during system disturbances. The report [ 10 ] outlines the re-
sults of the working group study and points out areas where further work
may be required.
During the first disturbance, underexcitation relays tripped 16 turbo-
generators. Out of 193 units, 102 had an underexcitation relays all installed
to trip. These represent 81.8% of the system gross capability. The 16 units
tripped by the underexcitation relays represent 16.1% of the system gross
capability. Only one of the 16 units tripped by the underexcitation had a
voltage regulator in service. During the second disturbance, the
underexcitation relays tripped a total of 12 units or 28.4% of the total
system capacity. Only four of the 12 units tripped by the underexcitation
relays had their voltage regulator in service. The working group concluded
that 52 relay operations out of 68 were correct. There were 4 unwanted
operations and 4 missing operations. The working group could not
determine if the other 8 operations were correct or not.

MHO-Relay Used for

Underexcitation Protection
X

Practical Stability Limit

R
Theoretical Stability Limit

Tripping Chatacteristic

Figure 3 MHO-relay for underexcitation protection.

GENPC130.DOC 2007-05-22 15.28

13.6 Underexcitation Protection

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[2] Born, E. & Fischer, A.: "Elektronischer Untererregungsschutz",
Siemens-Zeitschrift, vol. 46, no. 12, pp. 912-915, December,
1972.
[3] Bär, G., Grau, H.-W. & Kienast, L.: "Der Generatorschutz in
elektronischer Bauweise im Pumpspeicherwerk Wehr", Elektriz-
itätswirtschaft, vol. 78, no. 5, pp. 147-154, 1979.
[4] Glebov, I.A., Danilevich, J.B. & Mamikoniants, L.G.:
"Abnormal Operation Conditions of Large Synchronous
Generators and their Influence on Design and Performance in a
Power System", Report 11-07, CIGRE-Session, Paris, 1976.
[5] Mason, C.R.: "A New Loss-of-Excitation Relay for Synchronous
Generators", AIEE Trans., vol. 68, pt. II, pp. 1240-1245, 1949.
[6] Mason, C.R.: "The Art and Science of Protective Relaying",
John Wiley and Sons, New York, 1956.
[7] Sarma, M.S.: "Synchronous Machines (Their Theory, Stability,
and Excitation Systems)", Gordon and Breach, New York, 1979.
[8] Walker, C.W.: "Relay Protection in Hydro-Electric Power
Stations of the Snowy Mountains Authority", The Institution of
Engineers, Australia, Electrical Engineering Transactions, vol.
EET-5, no. 2, pp. 311-316, September, 1969.
[9] "A Partial Survey of Relay Protection of Steam-Driven A-C
Generators", AIEE Committee Report, AIEE Trans., vol. 81, pt.
III (Power Apparatus and Systems), pp. 954-957, February,
1962.
[ 10 ] "Loss-of-Field Operation During System Disturbances", IEEE
Committee Report, IEEE Trans. on Power Apparatus and Sys-
tems, vol. PAS-94, no. 5, pp. 1464-1472, September/October,
1975.#

GENPC130.DOC 2007-05-22 15.28

 14.1

14. OVERVOLTAGE PROTECTION

The excitation system must have a capacity that makes it possible to
operate the unit at full load. The associated excitation current is much
higher than the current needed to reach rated voltage at rated speed with an
unloaded generator. At full load, the excitation current may be 200 to 300%
of the excitation current at no load. To improve the response of the
excitation system many exciters can increase the field voltage to about
600% of the field voltage at no load.
Let us assume that the operator is starting a unit, the rotational speed
of the unit is higher than 90% of nominal speed. The operator has not yet
synchronised the generator to the network. Now, the automatic voltage
regulator (AVR) controls the terminal voltage of the unit. There are several
contingencies that may cause a dangerous voltage increase. Such
contingencies include a fault in the exciter, a fault in the AVR itself, a short
circuit in the secondary circuit that energises the AVR or an open circuit in
the secondary circuit that energises the AVR. Such contingencies may
increase the excitation far beyond the excitation current at no load. This
will cause a terminal voltage that is higher than acceptable.
After synchronisation, the external power system has a dominating
influence on the terminal voltage. Commonly, the capacity of a given
generator is small in comparison with the capacity of the external power
system. This means that a fault in the AVR cannot alter the terminal voltage
substantially while the generator is operating in synchronism with an intact
network.
The terminal voltage will start to increase rapidly if a fully loaded
generator becomes disconnected from the network. Usually, the AVR will
reduce the terminal voltage and it will eventually approach 100%. The
contingencies mentioned above may cause a terminal voltage that is higher
than acceptable. Some units do not have an AVR or the operator may have
switched over from automatic voltage regulation to manual excitation
control. All these contingencies may cause an unacceptable terminal
voltage. The overvoltage will be even higher if the generator feeds
unloaded transmission circuits. Long overhead lines and power cables, with
large capacitive generation, worsen the situation.
The terminal voltage will increase even more if the load rejection
causes an increase of the rotational speed of the unit. The speed of a turbo-
generator must not increase over 112 to 115%. In hydro-power plants, the
speed increase may be much higher. This may happen if the unit does not
have any turbine governor, if the turbine governor is slow or if the operator
has switched over to manual speed control. In the worst case, the speed of

GENPC140.DOC 2007-05-22 15.28

14.2 Overvoltage Protection

the hydro-generator may approach the runaway speed of the hydro turbine.
The runaway speed of a Francis turbine is about 150 to 200% of nominal
speed. For a Kaplan turbine, the runaway speed is about 200 to 300%. One
has to remember that the induced EMF is proportional to the product of the
rotational speed and airgap flux. The flux is approximately proportional to
the excitation current.
In EHV transmission systems, a load rejection at the receiving end of
a long transmission line may cause too high voltages even if the AVR and
the turbine governor at the sending end operate correctly.
The core material in modern power transformers will saturate if the
applied voltage is higher than 110 to 125% at rated frequency. A saturating
transformer will reduce the overvoltage because the excitation current to
the transformer will increase. The saturated magnetisation impedance may
decrease to 0.1% of the unsaturated value. Unfortunately, the flux in a
saturated transformer may reach magnetic objects outside the iron core.
Eddy current losses may very rapidly increase the temperature of such
objects. Such hot objects may also increase the temperature of the oil in the
transformer tank and increase the risk for dielectric failures. Such a
situation may occur during start-up and shutdown of unit-connected
generators. There is a risk that the operator switches over to automatic
voltage regulation before the rotational speed has reached 90% of rated
speed. He may also forget to switch over to manual excitation control
during a shutdown of the unit.
Tideström recommends [ 4 ] a dependent time overvoltage relay for
overvoltage protection. Such a relay has a short delay if the overvoltage is
very high. The delay increases with decreasing overvoltage.
The overvoltage protection system for the generators in the Snowy
Mountain Complex comprises, according to Walker [ 5 ] an overvoltage
relay and a timer. The overvoltage relay picks up if the voltage is higher
than 130% of rated voltage. The time delay is 2 s.
The overvoltage protection system should, according to Sarma [ 3 ],
have two steps if the generator does not have an AVR. In thermal power
plants the first step should pick up if the voltage exceeds 125%. For hydro
power plants, the corresponding figure is 140%. The first step should trip
without delay. The second step should pick up if the voltage is higher than
110% and should have a dependant time characteristic.
The synchronous machines in the pumped storage plant Wehr have,
according to Bär, Grau and Kienast [ 2 ], an overvoltage protection system
comprising a frequency compensated overvoltage relay and a timer. The
overvoltage relay picks up if the voltage is higher than 120% of rated
power. The time delay is 3 s.

GENPC140.DOC 2007-05-22 15.28

Overvoltage Protection 14.3

Hydro-power units owned by Vattenfall should, according to Lohage

and co-workers [ 1 ] have an overvoltage protection system. It should
comprise an overvoltage relay and a timer. The overvoltage relay should
pick up if the voltage is higher than 120% of rated power. The time delay
should be about 1 second.
Most generators have two sets of voltage transformers. One set
comprises three single-phase transformers and it energises the protection
equipment and the AVR. The second set may comprise only two single-
phase voltage transformers. This set energises the overvoltage protection
system. It is necessary to use two sets of voltage transformers because
otherwise the overvoltage protection system and the AVR do not have
independent input sources. Usually, a phase-to-phase voltage energises the
overvoltage protection system to avoid unwanted operation at single phase-
to-earth faults.
One may supplement the overvoltage protection system by an
overexcitation protection system. The energising quantity of the
overexcitation relay is the quantity U/f [V/Hz]. Unit-connected turbo-
generators need such a relay to protect the step-up transformer. Incidents
have occurred during start-up and shutdown. Occasionally, the AVR was in
service while the rotational speed was well below rated speed. This caused
overexcitation of the step-up transformer and sometimes damage to it. Once
the overall differential protection detected the abnormal situation. The step-
up transformer became saturated and the differential relay operated because
of the excitation current of the transformer.

GENPC140.DOC 2007-05-22 15.28

14.4 Overvoltage Protection

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[2] Bär, G., Grau, H.-W. & Kienast, L.: "Der Generatorschutz in
elektronischer Bauweise im Pumpspeicherwerk Wehr", Elektriz-
itätswirtschaft, vol. 78, no. 5, pp. 147-154, 1979.
[3] Sarma, M.S.: "Synchronous Machines (Their Theory, Stability,
and Excitation Systems)", Gordon and Breach, New York, 1979.
[4] Tideström, S. H:son (ed): "Ingenjörshandboken/Allmän Elektro-
teknik", 3rd ed., Nordisk Rotogravyr, Stockholm, 1959.
[5] Walker, C.W.: "Relay Protection in Hydro-Electric Power
Stations of the Snowy Mountains Authority", The Institution of
Engineers, Australia, Electrical Engineering Transactions, vol.
EET-5, no. 2, pp. 311-316, September, 1969.

GENPC140.DOC 2007-05-22 15.28

 15.1

15. REVERSE POWER PROTECTION

The task of a generator in a power plant is to convert mechanical energy
available as a torque on a rotating shaft to electric energy. One exception is
a reversible unit in a pumped storage plant. During light load conditions,
they consume electric power and store it as potential energy. During peak
load conditions, those generators operate as normal hydro power units.
Sometimes, the mechanical power from a prime mover may decrease
so much that it does not cover bearing losses and ventilation losses. Then,
the synchronous generator becomes a synchronous motor and starts to take
electric power from the rest of the power system. This operating state,
where individual synchronous machines operate as motors, implies no risk
for the machine itself. If the generator under consideration is very large and
if it consumes lots of electric power, it may be desirable to disconnect it to
ease the task for the rest of the power system.
Often, the motoring condition may imply that the turbine is in a
very dangerous state. The task of the reverse power protection is to protect
the turbine and not to protect the generator itself. Generally, AC current
and voltage energise the reverse power protection system and it trips the
generator breaker. Therefore, one includes the reverse power protection in
the generator protection.
Steam turbines easily become overheated if the steam flow becomes
too low or if the steam ceases to flow through the turbine. Therefore, turbo-
generators should have reverse power protection. Neugebauer [ 3 ]
mentions several contingencies that may cause reverse power. One is a
break of a main steam pipe. A second is a damage to one or more blades in
the steam turbine. The third is an inadvertent closing of the main stop
valves. In the last case, it is highly desirable to have a reliable reverse
power protection. It may prevent damage to an otherwise undamaged plant.
During the routine shutdown of many thermal power units, the
reverse power protection gives the tripping impulse to the generator breaker
(the unit breaker). By doing so, one prevents the disconnection of the unit
before the mechanical power has become zero. Earlier disconnection would
cause an acceleration of the turbine generator at all routine shutdowns. This
should have caused overspeed and high centrifugal stresses.
When the steam ceases to flow through a turbine, the cooling of the
turbine blades will disappear. Now, it is not possible to remove all heat
generated by the windage losses. Instead, the heat will increase the
temperature in the steam turbine and especially of the blades. When a steam
turbine rotates without steam supply, the electric power consumption will [
6 ] be about 2% of rated power. Even if the turbine rotates in vacuum, it
GENPC150.DOC 2007-05-22 15.29
15.2 Reverse Power Protection

will soon become overheated and damaged. The turbine overheats within
minutes if the turbine loses the vacuum. Steam turbines may eventually,
according to Mason [ 2 ], become overheated if the steam flow is so low
that the electrical power generation falls below 10%.
The critical time to overheating of a steam turbine varies, according
to Mason [ 2 ], from about 0.5 to 30 minutes depending on the type of
turbine. A high pressure turbine with small and thin blades will become
overheated more easily than a low pressure turbine with long and heavy
blades. The conditions vary from turbine to turbine and it is necessary to
ask the turbine manufacturer in each case. It is also prudent to measure the
reverse power during commissioning of new units.
Power to the power plant auxiliaries may come from a station service
transformer connected to the primary side of the step-up transformer.
Power may also come from a start-up service transformer connected to the
external network. One has to design the reverse power protection so that it
can detect reverse power independent of the flow of power to the power
plant auxiliaries.
Hydro turbines tolerate reverse power much better than steam tur-
bines do. According to Lohage and co-workers [ 1 ], only Kaplan turbine
and bulb turbines may suffer from reverse power. There is a risk that the
turbine runner moves axially and touches stationary parts. They are not al-
ways strong enough to withstand the associated stresses.
Ice and snow may block the intake when the outdoor temperature
falls far below zero. Branches and leaves may also block the trash gates. A
complete blockage of the intake may, according to Mason [ 2 ] cause
cavitation. The risk for damages to hydro turbines can justify reverse power
protection in unattended plants. According to GEC [ 7 ], geared units may
require reverse power protection. This is because the design of the gears is
for driving in one direction only.
A hydro turbine that rotates in water with closed wicket gates will
draw electric power from the rest of the power system. This power will be
about 10% of the rated power. If there is only air in the hydro turbine, the
power demand will fall to about 3%. It is prudent to measure these values
during the commissioning.
Diesel engines should, according to Mason [ 2 ] have reverse power
protection. The generator will take about 15% of its rated power or more
from the system. According to GEC [ 7 ], a stiff engine may require
perhaps 25% of the rated power to motor it. An engine that is well run in
might need no more than 5%. According to Sarma [ 4 ], diesel engine units
usually require reverse power protection with a setting of 15 to 25%. It is

GENPC150.DOC 2007-05-22 15.29

Reverse Power Protection 15.3

necessary to obtain information from the engine manufacturer and to

measure the reverse power during commissioning.
Gas turbines usually do not require reverse power protection. The
power required to motor a gas turbine varies, according to Mason [ 2 ],
from 10 to 50% of rated power. It depends on the turbine design and
whether it has a separate load turbine.
A reverse power relay and a timer can provide adequate reverse
power protection. The reverse power relay can be a directional undercurrent
relay that measures the current that flows from the generator to the network
(an underpower relay). It may also be a directional overcurrent relay that
measures current that flows from the network to the generator (an over-
power relay). There are few sensitive current relays that also can withstand
the normal load current continuously. The relay must also temporarily
withstand the fault current that may flow through the relay. The setting
range of the timer should be from a few seconds to some minutes.
Figure 1 illustrates the properties of reverse power protection with
underpower relay and with an overpower relay. The underpower relay
gives a higher margin and should provide better dependability. On the other
hand, the risk for unwanted operation immediately after synchronisation
may be higher. One should set the underpower relay to trip if the active
power from the generator is less than about 2%. One should set the
overpower relay to trip if the power flow from the network to the generator
is higher than 1%.

Underpower Relay Overpower Relay

Q Q
Operate Operate
Line Line

Margin Margin
P P

Operating point Operating point

without without
turbine torque turbine torque

Figure 1 Characteristics of the reverse power protection.

GENPC150.DOC 2007-05-22 15.29

15.4 Reverse Power Protection

There is a risk that the generator breaker does not open on command from
the reverse power relay. Now, we are in danger. It is futile to start the
breaker failure protection from the reverse power relay. The operate value
of the breaker failure protection is usually higher than the operate value of
the reverse power relay. There is only a remote possibility that the breaker
failure relay will detect that the generator breaker has failed to open.
Let us consider a unit with a generator breaker on the primary side of
the step-up transformer. It has a unit breaker on the secondary side of the
step-up transformer. We can increase the dependability of the reverse
power protection by installing two reverse power relays. Two different sets
of instrument transformers should energise the reverse power relays. The
two reverse power relays should have approximately the same pick up
value. One reverse power relay should have a shorter delay than the other.
The first reverse power relay should trip the generator breaker and the
second relay should trip the unit breaker. With this, we can avoid necessary
interruption of the supply of auxiliary power through the step-up
transformer.
The demands on the reverse power protection are increasing. An
AIEE report [ 8 ] says that the reverse power relays may only give an alarm
immediately when power flows from the network to the generator. The re-
verse power relay should trip if the power reversal persists long enough to
cause damage to the turbine from overheating. A delay of at least one
minute will be permissible between the time when the power reversal starts
and tripping should occur.
An IEEE report [ 8 ] does not recommend any dedicated reverse
power protection. The interlocking system can provide certain reverse
power protection.
GEC points out [ 7 ] that the reverse power relay must detect active
power. The reactive power will not change very much when the active
power reverses. The reactive power seldom exceeds 60% of rated power.
This means that the phase angle between current and voltage may be as
high as 85o and may be either inductive or capacitive. This illustrates the
difficulties when designing a suitable reverse power relay. GEC also
mentions that there are sensitive three-phase power relays with a sensitivity
of about 0.5%.
According to Lohage and co-workers [ 1 ] the reverse power relay
must pick up if the active power from the network to the generator exceeds
2% of the rated power. They say that, the delay of should be from five to 10
seconds, depending on the type of turbine.

GENPC150.DOC 2007-05-22 15.29

Reverse Power Protection 15.5

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[2] Mason, C.R.: "The Art and Science of Protective Relaying",
John Wiley and Sons, New York, 1956.
[3] Neugebauer, H.: "Selektivschutz", Springer-Verlag, Berlin,
1955.
[4] Sarma, M.S.: "Synchronous Machines (Their Theory, Stability,
and Excitation Systems)", Gordon and Breach, New York, 1979.
[5] "Minimum Recommended Protection, Interlocking and Control
for Fossil Fuel Unit-Connected Steam Station. I - Overall
Protection", IEEE Committee Report, IEEE Trans. on Power
Apparatus and Systems, vol. PAS-92, no. 1, pp. 374-380,
January/February, 1973.
[6] "Power System Protection", vol. 1, 2 and 3, Application, Edited
by the Electricity Council, Peter Peregrinus, Stevenage, 1981.
[7] "Protective Relays Application Guide", 2nd ed., GEC Measure-
ment, 1975, 5th printing, October, 1983.
[8] "Relay Protection of A-C Generators", AIEE Committee Report,
AIEE Trans., vol. 70, pt. I, pp. 275-282, 1951.

GENPC150.DOC 2007-05-22 15.29

15.6 Reverse Power Protection

GENPC150.DOC 2007-05-22 15.29

 16.1

16. UNBALANCE PROTECTION

Single-phase loads, series faults and unsymmetrical faults may cause con-
tinuous and temporary unbalance loading of a synchronous machine. If the
unbalanced loading is too high and persist too long, the rotor of the
machine will overheat and become damaged. A three-phase generator
carries a balanced load if the stator phase currents are:

0π (1)
I a = I sin (2πf - )
3
2π
I b = I sin (2πf - )
3
4π
I c = I sin(2πf - )
3

and the phase-to-earth stator voltages are:

0π (2)
V a = V sin (2πf - + ϕ)
3
2π
V b = V sin (2πf - + ϕ)
3
4π
V c = V sin (2πf - + ϕ)
3

16.1 Unbalanced loading

The generator carries an unbalanced load if equation ( 1 ) and ( 2 ) do not
hold true. Unbalanced loading is not a very well defined concept. It is also
difficult to describe how unbalanced loading will increase the stresses that
the generator has to endure. A better description is the relative amount of
negative sequence current, I2/In that the generator carries. The negative
sequence current, I2 is:

I a + a2 I b + a I c (3)
I2 =
3

and In is the rated stator current of the machine. The relative negative se-
quence current is therefore well defined and one can measure it.

GENPC160.DOC 2007-05-22 15.30

16.2 Unbalance Protection

16.1.1 Causes of Unbalanced Loading

Unbalanced loading may produce more severe heating than balanced three-
phase operation. Series faults close to the generator will cause negative se-
quence currents. Unsymmetrical faults may produce more severe heating in
three-phase synchronous machines than symmetrical faults. Typical condi-
tions and incidents that can cause unbalanced loading are:
• Single-phase loads close to the power plant.
• Untransposed transmission circuits.
• Unbalanced step-up transformers.
• Series faults in the transmission network.
• Series faults on the secondary side of the step-up transformer.
• Series faults on the primary side of the step-up transformer.
• Pole discrepancy in the generator breaker.
• Unbalanced shunt faults close to the power plant.
• Unbalanced shunt faults on the generator buswork.

16.1.2 Consequences for the Generators

A synchronous machine overheats quite rapidly if it carries even low nega-

tive sequence currents. The negative sequence current generates a stator-
MMF that rotates with the same speed as the rotor but in the opposite direc-
tion. Seen from the rotor, this MMF has a frequency 2fn, where fn is the
power frequency (50 or 60 Hz). This MMF induces voltages with a fre-
quency 2fn in the rotor and its windings.
These voltages cause currents to flow in the rotor and associated
windings. Due to the skin effect, these currents flow close to the surface of
metallic objects in the rotor. The penetration depth in magnetic steel is less
than one millimetre at 50 Hz. These currents heat will quickly the rotor
body, the slot wedges, the retaining ring and the damper winding if there is
one. These components are normally already under great stress in large
turbo-generators. If the negative sequence current persists, the metal will
melt and damage the rotor structure.
The amount of negative sequence current that the machine can
tolerate depends on the design of the generator.

16.1.3 Turbo-Generators

Generators without damper winding do not have well-defined paths for the
induced double frequency currents. The electromagnetic and thermal
GENPC160.DOC 2007-05-22 15.30
Unbalance Protection 16.3

utilisation increases steadily. This is especially true for the rotors in turbo-
generators. This means that the turbo-generators are very sensitive for
unbalanced loading. Much development work has been done and will be
done to improve the unbalance loading capability. One has to pay special
attention to the material of the slot wedges. When the conductivity of the
slot wedges is higher than the conductivity of the rotor body, most of the
induced double frequency currents will flow in the slot wedges. In several
modern generators, the slot wedges do not have electrical contact with each
other. This means that the current will flow from slot wedges to the rotor
and back again. This will cause severe local heating. The general opinion is
that it is important to secure good electrical contact between the slot
wedges and the rotor body.

16.1.4 Hydro-Generators

Salient pole generators with Damper windings (hydro-generators) have

well-defined paths for the induced double frequency currents. The currents
flow mainly in the damper windings. Generally, hydro-generators have
strong damper windings and they can withstand higher negative sequence
currents than turbo-generators can. There are very few hydro-generators
without damper windings.

16.2 Continuous I2-capability

Because unbalanced loading may continue for long periods, each machine
is assigned a continuous negative sequence current capability
(continuous I2-capability). Usually it is expressed in percent of the rated
stator current.
Table 1 shows typical continuous negative sequence current
capability for generators with different forms of cooling [ 7 ].
Most countries support [ 3 ] the American suggestion concerning
continuous I2-capability. Table 2 contains the suggested capability.
The machine manufacturers can provide more accurate values.

GENPC160.DOC 2007-05-22 15.30

16.4 Unbalance Protection

Table 1 Negative sequence current capability.

Type of machine Type of cooling (I2)max K
and cooling medium % s
Turbo generator Direct hydrogen 30 lb/in2 10 7
Turbo generator Conventional hydrogen 30 lb/in2 15 12
Turbo generator Conventional hydrogen 15 lb/in2 15 15
Turbo generator Conventional hydrogen 0.5 15 20
lb/in2
Typical salient pole machine Conventional air 40 60

Table 2 Suggested continuous I2-capability.

Cooling Method Sn I2,max
MVA %
Indirect cooled 10
Direct cooled
-"- <960 8
-"- 961-1200 6
-"- >1200 5

16.3 Temporary I2-capability

The synchronous machine may carry large negative sequence currents
during fault conditions. Because normal fault clearance times are short,
only little heat is lost by the machine while the fault currents are present. In
addition, it is the heating caused by these currents that may cause damage
and therefore it is the input energy that must be limited. This means that we
have to define a temporary negative sequence current capability
(temporary I2-capability). The length of time, T [s] that a machine can
operate with negative sequence current without danger of being damaged
can be expressed in the form:
T
∫ 0 i 2 (t) dt = K (4)
2

where:
i22(t) = the instantaneous negative sequence current as a
function of time; the current is expressed in per unit
based and

GENPC160.DOC 2007-05-22 15.30

Unbalance Protection 16.5

K = a constant.

The generator can withstand a constant negative sequence current, I2 [A],

during a period, T [s], given by the equation:
2
⎛ In⎞ (5)
T = K⎜ ⎟
⎝ I2⎠

where
In = the rated current of the generator [A] and
K = a constant [s] that is typical for the type of generator.

The constant K tells us how many seconds the machine can withstand a
negative sequence current equal to the rated current of the generator. The
constant K depends on the size of the generator and the method of cooling.
For most generators the value is from 5 to 30 s, but for some hydro-genera-
tor it may be as high as 60 s.
There are industry standards that determine the permissible
unbalance to which a generator is designed [ 4 ]. For turbo-generators with
Sn ≤ 800 MVA the requirement is:

K ≥ 10 (6)

and for Sn>800 MVA, the requirement is:

K ≥ 10 - 0.00625(800 - Sn ) (7)

Figure 1 shows these requirements in graphical form.

For example, a 500 MVA generator would have K = 10 s and a
1 600 MVA generator would have K = 5 s.
Figure 2 shows data on the negative sequence current capability.

GENPC160.DOC 2007-05-22 15.30

16.6 Unbalance Protection

NEGATIVE SEQUENCE CURRENT CAPABILITY

15

K [s]

10

K = 10 - 0.00625*( S - 800 )
0
0 500 1000 1500 2000
Rated Apparent Power, S [MVA]

Figure 1 Negative sequence current withstand capability.

NEGATIVE SEQUENCE CURRENT CAPABILITY

10000
Low I2max

High I2max
Maximum Time [ s ]

1000

100 High K

10
Low K
1
0.01 0.10 1.00 10.0
Negative Sequence Current [ pu ]

Figure 2 Maximum time for a given negative sequence current.

GENPC160.DOC 2007-05-22 15.30

Unbalance Protection 16.7

The ability of large generators to stand negative sequence current is becom-

ing progressively less because their specific rating is still increasing
although their size has almost reached the limit of present material.

16.4 Unbalance Protection

Generally, the unbalanced protection consists of a dependent time overcur-
rent relay. A negative sequence filter energises the overcurrent relay. On a
log-log scale, the time characteristic is a straight line and can be set to
match the machine characteristic.
In 1953, Barkle and Glassburn suggested [ 2 ] that synchronous gen-
erators should have a negative sequence protection with dependent time
characteristic. They documented the fact that the heating caused by a phase-
to-phase short circuit on the generator terminals was nine times higher than
the heating caused by a three-phase short circuit on the generator terminals.
The report contains a description of a new negative sequence current relay .
It is suitable for generators with K in the interval 30 to 90 s.
Synchronous generators should, according to Mason [ 5 ], have a
negative sequence current relay with dependent time characteristic. The
negative sequence current relay should trip the generator. Mason points out
that the relay will seldom operate on external unbalanced faults. The unbal-
ance protection can be justified as a protection against series faults. It can
also be justified as aback up protection against external unbalanced shunt
faults.
An AIEE working group has prepared and sent out a questionnaire
about practice and experience of underexcitation protection. The group has
analysed and documented [ 6 ] answers from 63 utilities in the US, Canada
and Mexico. The answers represented 309 generators rated at 60 MW or
more, installed since 1949. In 1958, 40% of these generators have negative
sequence protection. Then, 70% of the utilities installed negative sequence
protection on new generators.
The new hydro power stations owned by Vattenfall will, according to
Lohage and co-workers [ 1 ], have a negative sequence current relay with
dependent time characteristic. They point out that the relay can detect pole
discrepancy in the generator breaker.
It is not uncommon, according to Horowitz and Phadke [ 4 ], to
apply negative sequence relays on the generator to alarm first, alerting the
operator to the abnormal situation and allowing corrective action to be
taken before removing the machine from service.

GENPC160.DOC 2007-05-22 15.30

16.8 Unbalance Protection

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[2] Barkle, J.E. & Glassburn, W.E.: "Protection of Generators
Against Unbalanced Currents", AIEE Trans., vol. 72, pt. III
(Power Apparatus and Systems), pp. 282-286, April, 1953.
[3] Glebov, I.A., Danilevich, J.B. & Mamikoniants, L.G.:
"Abnormal Operation Conditions of Large Synchronous
Generators and their Influence on Design and Performance in a
Power System", Report 11-07, CIGRÉ-Session, Paris, 1976.
[4] Horowitz, S.H. & Phadke, A.G.: "Power System Relaying", Re-
search Studies Press and John Wiley & Sons, Taunton and New
York..., 1992.
[5] Mason, C.R.: "The Art and Science of Protective Relaying",
John Wiley and Sons, New York, 1956.
[6] "A Partial Survey of Relay Protection of Steam-Driven A-C
Generators", AIEE Committee Report, AIEE Trans., vol. 81, pt.
III (Power Apparatus and Systems), pp. 954-957, February,
1962.
[7] "Protective Relays Application Guide", 2nd ed., GEC Measure-
ment, 1975, 5th printing, October, 1983.

GENPC160.DOC 2007-05-22 15.30

 17.1

17. OUT OF STEP PROTECTION

Many utilities have adopted reliability principles for the planning and
operation of their bulk power system. They use the principles when making
decisions on investments in power plants and transmission systems. The
system operators use the system when deciding on transfer limits and when
planning outages. The design principles often prescribe the tolerable
consequence of a set of contingencies but they may vary from utility to
utility. The set of contingencies may include:
• A loss of a large generating unit.
• An inadvertent tripping of a circuit breaker.
• A shunt fault on a transmission line.
• A shunt fault on a busbar

Often, the utilities do not tolerate loss of synchronism because of such con-
tingencies. Sometimes a utility may decide to delay investments and tempo-
rarily deviate from the reliability principles.
Severe stresses may result when a generator loses the synchronism
without having lost the excitation. Pole slipping is associated with high cur-
rent pulses and violent oscillations of the air gap torque in the generator.
Mason says [ 3 ] that it is not likely to lose synchronism with other
generators in the same station unless it loses excitation. Many units have
underexcitation protection. He concluded that it is not the usual practice to
provide out of step protection.
Synchronism can, according to GEC [ 4 ], be regained if the load is
sufficiently reduced, but if this does not occur within a few seconds it is
necessary to isolate the generator and then re-synchronise.
During the planning of the power plant Abwinden-Asten on the
River Danube, it was necessary to consider the long fault clearance time in
the connecting network. Each generating unit has a bulb turbine and the
turbine generator units have a low inertia constant. The low inertia
constants and the long fault clearance times made it necessary to install an
out of step protection. ELIN has, according to Dorfmeister [ 2 ], developed
an out of step protection. He does, however, not describe the method for
detection of pole slipping clearly. Static and dynamic estimation of the
rotor angel is a key task of the protection.
Hydro-generators owned by Vattenfall must not have out of step pro-
tection [ 1 ]. Some nuclear units in Sweden have a simple out of step pro-
tection. It comprises an overcurrent relay and some associated logic. After
three consecutive overcurrent pulses, it trips the unit.

GENPC170.DOC 2007-05-22 15.30

17.2 Out of Step Protection

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[2] Dorfmeister, K.: "Betrieb und Erfahrung mit einem neuartigen
Stabilitätsschutz in der Mikroprozessortechnik", ELIN-
Zeitschrift, vol. 34, no. 3/4, pp. 58-61, 1982.
[3] Mason, C.R.: "The Art and Science of Protective Relaying",
John Wiley and Sons, New York, 1956.
[4] "Protective Relays Application Guide", 2nd ed., GEC Measure-
ment, 1975, 5th printing, October, 1983.

GENPC170.DOC 2007-05-22 15.30

 18.1

18. ABNORMAL FREQUENCY PROTECTION

Abnormal frequency is related to abnormal speed of the generating units. A
single failure in the power plants will usually not cause abnormal speed de-
viations if the power system is a large and stable one. However, if the sys-
tem is a small isolated one with large generating units, the loss of one unit
may cause a big power deficit. Then, the system frequency begins to fall
and may reach so low values that it becomes necessary to disconnect other
units. Faults in the network may occasionally split a large power system
into small electrical islands. Such islands may have a big power surplus or
a big power deficit. Here, there is a risk for abnormal frequencies. There are
cases on record where a subsystem established a balance at 50% of normal
system frequency and 50% of normal network voltage. Now, the on-line tap
changers started to move and eventually they reached their extreme
positions. Then, the system operator closed a circuit-breaker and
reconnected the subsystem. The voltage level on the power network
became normal. The consequence of the disturbances was that customer
equipment became damaged by overvoltage during the service restoration
process.

18.1 Aims of the Abnormal Frequency Protection

The frequency control system shall provide the first line of defence against
abnormal frequency. A load shedding system provides the second line of
defence. Then, the abnormal frequency protection provides a last defence
line. One task is to prevent operation of customer loads at abnormal
frequencies. The Swedish Electrical Safety Regulations require that the
utility shall prevent changes in network voltage and system frequency that
may present danger to the customers. Another task of the abnormal
frequency protection is to prevent damage to power plant equipment.
Some utilities use an underfrequency relay to trip generating units to
houseload when a widespread blackout has occurred. The idea is to have
the generating units in a hot standby state and ready for synchronisation.
When the network operator has voltage restored the transmission system,
new generation will become available with a minimum of time delay. Such
actions aim at reducing the customer interruption times instead of
preventing loss of voltage.
Automatic switching of circuit breakers on abnormal frequency may
ease the service restoration. Some utilities believe that the service restora-
tion process will run faster and smoother if they can energise several dead
subsystems instead of synchronising an electrical island.

GENPC180.DOC 2007-05-22 15.30

18.2 Abnormal Frequency Protection

There are two major considerations associated with operating power

plants at abnormal system frequency:
• Protection of equipment from damage.
• Prevention of cascading trips that lead to a complete blackout

18.2 Effects of Abnormal Frequency

There are no standards for abnormal frequency operation of synchronous
generators. Reduced frequency results in reduced ventilation. Therefore,
operation at reduced frequency should be at reduced apparent power. The
underfrequency limitations on the generator, however, are usually less
restrictive than the limitations on steam turbines. Overfrequency is usually
the result of a sudden load reduction. Therefore overfrequency is associated
with light-load or no-load operation of the generator. Operation within the
allowable overfrequency limits of the turbine will not produce generator
overheating as long as operation is within rated apparent power and 105%
of rated voltage.
Abnormal frequency present hazards to other parts of the plant such
as:
• Steam turbine vibrations and increased stresses on blades.
• Reduced capacity of auxiliary equipment
• High temperatures caused by increased excitation current
• Overexcitation of transformers

18.3 Co-ordination with Load Shedding

The aim of the frequency controlled load shedding is to re-establish a bal-
ance between available generation and load by disconnecting load when the
system frequency drops below certain levels. The amount of load shedding
varies from country to country and from region to region. Typical values
range from 20 to 60% of system load. Experience has shown that the load
shedding system may not always prevent underfrequency. Then, one needs
the abnormal frequency protection to prevent damage to power plant equip-
ment and especially steam turbines.

18.4 Realisation of Abnormal Frequency Protection

All prime movers should have overspeed protection. The overspeed device
may be a mechanical centrifugal device or an overfrequency relay. Varia-
tions in generator voltage must not adversely affect the overfrequency
GENPC180.DOC 2007-05-22 15.30
Abnormal Frequency Protection 18.3

relay. Mechanical centrifugal devices may be difficult to adjust and it is

common practice to test them every year after maintenance outage periods.
It is easier to test overfrequency relays and they may reduce the start-up
time after maintenance outage.
The setting of the overspeed device may be 108 to 115% for turbo-
generators and 140 to 160% for hydro turbines.
The overspeed protection may be a part of the prime mover, or of its
speed governing system, or of the generator protection. The overspeed ele-
ment should operate the main stop valve to shut down the prime mover. It
should also trip the generator circuit breaker and the auxiliary breaker
where auxiliary power comes from the generator buswork. By doing so, it
is possible to prevent overfrequency operation of customer loads and power
plant auxiliaries.
The overspeed element should usually operate at 3 to 5% above full-
load rejection speed.
The underfrequency protections often have two steps. One frequency
relay trips the unit breaker without delay if the system frequency falls
below 95%. Another frequency relay trips the unit if the frequency does not
recover.
The aim of the frequency controlled load shedding is to cause the
system frequency to return to normal before it becomes necessary to trip
the generating units.

GENPC180.DOC 2007-05-22 15.30

18.4 Abnormal Frequency Protection

GENPC180.DOC 2007-05-22 15.30

 19.1

19. INADVERTENT ENERGISING PROTEC-

TION
Operating errors, breaker head flashovers, control circuit malfunctions, or a
combination of these have resulted in inadvertent or accidental energising
of off-line generators [ 2 ]. Many large generators have been severely dam-
aged, sometimes, beyond repair [ 1 ]. This section is entirely based on
reference [ 1 ] and [ 2 ].

19.1 Inadvertent Energising

Operating Errors: The use of more complex breaker patterns has resulted
in more frequent operating errors. Even with extensive interlocks between
unit breakers and disconnecters, there has been an increase in the number of
documented cases in which off-line units have been inadvertently energised
through the high voltage switch.

Breaker Head Flashover: The extreme dielectric stress in breakers and the
small contact gap spacing associated with their high-speed interrupting re-
quirement can lead to contact flashover. The risk of a flashover is higher
just before synchronisation or just after the unit is removed from service.

19.2 Generator Response to Inadvertent Energising

Three-Phase Energising: When a generator is accidentally energized with
three-phase system voltage while at low speed, it behaves like an induction
motor. If the generator is connected to a strong network, the stator current
will be about 3 to 4 times rated current.

Single-Phase Energising: Single-phase energising of a generator from the

high voltage system while at low speed subjects the generator to a signifi-
cant unbalance current. There will be no significant accelerating torque if
the voltage applied to the generator is single-phase. Breaker head flashover
is the most frequent cause of single-phase inadvertent energising.

19.3 Damage Caused by Inadvertent Energising

Turbo-Generator Damage: The initial effect of inadvertent energising of
a generator from standstill is rapid heating in iron parts near the rotor
surface. Slot wedges have little clamping load at standstill, resulting in

GENPC190.DOC 2007-05-22 15.30

19.2 Inadvertent Energising Protection

arcing between them and the rotor iron. The arc heating begins to melt the
metals.

Steam Turbine Damage: During an inadvertent energising incident, the

generator acts as a motor to drive the turbine. The generator starts to accel-
erate the turbine and the exciter. As it comes up to speed, the unit passes
through its natural torsional frequencies. Vibration, blade distortion and
rubbing may cause turbine damage if the energising source is not removed
soon enough. The blades in the steam turbine may become overheated if the
turbine continues to rotate at high speed without any steam flow. Bearing
failure due to insufficient lubrication can occur.

Hydro Unit Damage: Heating of the damper windings and the rotor mate-
rial, combined with the lack of proper ventilation, will create damage
quickly.

19.4 Systems to Detect Inadvertent Energising

Although the normal generator protection may detect some cases of
inadvertent generator energising, dedicated protection systems are
recommended to detect inadvertent energising. Unlike conventional
protection systems that provide protection when equipment is in service,
these new schemes provide protection when equipment is out of service.
The most widely used dedicated protection systems are:

Frequency Supervised Overcurrent Relays: This system uses a

frequency relay to supervise the trip output of sensitively set instantaneous
overcurrent relays. The overcurrent relays are automatically armed by the
frequency relay as the unit is taken off-line and they remain armed while
the unit is shut down.

Voltage Supervised Overcurrent Relays: This system utilises undervolt-

age relays to supervise the trip output of high-speed instantaneous overcur-
rent relays. The overcurrent relays are automatically armed by separate un-
dervoltage relays when the unit's field is de-energised and they remain
armed while the unit is shut down.

Auxiliary Contact Enabled Overcurrent Relays: This system uses a

combination of auxiliary contacts on breakers and switches to enable and
disable high-speed instantaneous phase overcurrent relays.

GENPC190.DOC 2007-05-22 15.30

Inadvertent Energising Protection 19.3

Distance Relays: This scheme uses a distance relay located in the high-
voltage switchyard that is set to look into the machine. The relay should be
set to detect the sum of the generator step-up transformer and machine
transient reactance with an appropriate margin.

Directional Overcurrent Relays: This scheme uses three directional phase

overcurrent relays with very inverse time characteristics. Voltage sensing is
from the generator such that the overcurrent relays will pick up on current
into the generator.

19.5 Schemes to Detect Breaker Head Flashover

Some dedicated schemes, described above for inadvertent energising ,can
detect breaker head flashover. However, the following schemes are widely
used to detect breaker head flashover in the generator breaker or in the unit
breaker.

Modified Breaker Failure Scheme: An instantaneous overcurrent relay is

connected in the neutral of the step-up transformer. The relay is set to re-
spond to a breaker pole flashover. The breaker gets a starting impulse when
the relay while an auxiliary contact indicates that the circuit breaker is
open.

Breaker Pole Disagreement: A current relay augments the conventional

breaker pole disagreement scheme. This relay senses whether any phase is
below a certain low threshold level (indicating an open breaker pole) while
that another phase current is above a substantially higher threshold level
(indicating a closed or flashed-over pole). Operation of the disagreement
circuitry initiates breaker failure tripping.

19.6 Conclusion
American utilities have observed that inadvertent energising of large
generators has significantly increased in recent years as generating stations
have become more complex. Operating errors, breaker head flashover,
control circuit malfunctions or a combination of these causes have resulted
in generators becoming accidentally energized. Major US turbine-generator
manufacturers have recommended, and many utilities are installing,
dedicated inadvertent energising protection systems.

GENPC190.DOC 2007-05-22 15.30

19.4 Inadvertent Energising Protection

References

[1] "Inadvertent Energizing Protection of Synchronous Generators",

IEEE Committee Report, IEEE Trans. on Power Delivery, vol. 4,
no. 2, pp. 965-977, April, 1989.
[2] "Impact of HV and EHV Transmission on Generator Protection",
IEEE Committee Report, IEEE Trans. on Power Delivery, vol. 8,
no. 3, pp. 962-974, July, 1993.

GENPC190.DOC 2007-05-22 15.30

 20.1

20. BEARING CURRENT PROTECTION

The oil film of the bearings in a generating unit provides some insulation
between the rotor and the stator. There is a risk that shaft voltages cause an
electrical breakdown of these oil-films. Then, bearing currents start to flow
and their sizes depend on the shaft voltage and the loop impedance. An
electric current that flows through the oil film of a bearing in rotating ma-
chines can cause serious damage.

20.1 Shaft Voltages

There are several sources that may cause voltages in the shafts of turbine
generator units. Both the waveform and the size of the voltage depend on
the type and the size of the machine and they also vary with the loading.
The most important shaft voltage sources are [ 1 ] and [ 3 ]:

Magnetic Unsymmetries: The generator itself creates the most important

shaft voltage that may cause damage. One cause of this phenomenon is due
to asymmetries in the magnetic circuit of the stator. The net dissymmetry
links the rotor shaft and it induces a shaft voltage. It is rich in harmonics.
Further causes are eccentrically mounted rotors and rotor sag. These
sources are strong because they have low source impedance and can drive
high currents that are limited only by the circuit impedance.

Axial Shaft Flux: This source is due to the possibility that the turbine
cylinder/shell combination acts like a permanent magnetic generator. Both
the stationary and rotating turbine blades/buckets can be magnetised by
magnetostriction and can form an additional source of voltage. This is an
AC source rich of harmonics, but it is also a weak source. It is not very
important in turbine generators, although it is very important in high-speed
axial-flow compressors used in the petrochemical industry.

Charge Separation: Charge separation in the later stages of a steam

turbine may cause the blades to pick up electrons and get a negative charge.
The rate of charge separation and therefore potential is directly proportional
to steam flow-rate and hence loading.

External Field Voltages: Potentials can also be induced by capacitive

coupling between the rotor field winding and the shaft. Thyristor-controlled
excitation systems contain high frequency components in their output volt-

GENPC200.DOC 2007-05-22 15.31

20.2 Bearing Current Protection

age waveform. This source is also weak and incapable of driving large cur-
rents, but again it can cause a breakthrough in the oil film.

20.2 Hazard Caused by Shaft Voltages

Practical tests on an unearthed rotor of a turbo-generator provide informa-
tion about the size of the shaft voltages. There was a DC voltage of about
45 V due to electrostatic charge in the turbine and to dissymmetrical insula-
tion resistances. There was an induced AC component of about 2.5 V at the
fundamental frequency, due to eccentricity of the rotor shaft. There was a
15 V rectangular wave AC component due to the static excitation system.
The oil film of the bearings provides some insulation between the ro-
tor and the stator in a generating unit. If the bearing pedestals at each side
of a turbo-generator are earthed, the shaft voltage will be impressed across
the oil film of the bearings.
Most turbo-generators have one or several shaft-earthing brushes to
prevent build up of electrical charge on the shaft. This means that the
metallic parts of a turbo-generator have contact with earth via the shaft
earthing brush. The metallic parts of a hydro-generator have contact with
earth via the turbine and the waterways. At least the DC voltage will
disappear when the shaft is earthed at one point.
The main outboard bearing of large horizontal turbo-generators is,
according to [ 4 ], insulted from the bedplate. Usually, no shaft current is
flowing because one end of the generator shaft is insulated from earth.
When failure of this insulation has occurred or has been short-circuited, the
shaft voltage will be impressed on the oil film of the bearing. A bearing
current will start to flow if an electrical breakthrough occurs in the oil film
of the bearing. The size of the bearing current is about 100 to 200 A. The
bearing or other vital components such as the combination equipment on
Kaplan turbines will be damaged. If the bearing current is higher than about
2 A and lasts longer than a few seconds, it will, according to Lohage and
co-workers [ 2 ], cause damage.

GENPC200.DOC 2007-05-22 15.31

Bearing Current Protection 20.3

20.3 Realisation of Bearing Current Protection

It is difficult to detect an electrical breakdown of the oil film in a turbine
generator bearing. Utilities have tried to use the shaft voltage for bearing
protection. Experience has shown that the bearing current gives more reli-
able protection. Now, it is assumed that the current in the generator shaft is
equal to the bearing current. One uses a shaft current transformer to
measure the shaft current. The shaft current transformer is similar to a
normal window-type current transformer. The shaft itself is the primary
winding. The core of the shaft-current transformer encompasses the shaft.
The secondary winding of the shaft current transformer energises an
overcurrent relay. This relay should operate if the secondary current is
higher than 1 to 2 mA. Furthermore, it should have very low power
consumption. The shaft current relay energises a timer that delays the alarm
and trip. The minimum primary operate current increases with the diameter
of the shaft. It is about 0.25 A for a shaft with a diameter of 0.2 metres and
about 0.75 A for a shaft with a diameter of 2.8 meters.
In 1951 only two utilities out of 25 used, according to an AIEE
report [ 4 ], shaft current relays to detect failure of bearing pedestal
insulation in turbo-generators. Only machines rated above 50 MVA had
such features. An overcurrent relay can detect a bearing pedestal insulating
failure. The relay has one terminal connected to the bearing pedestal and
the other to an insulated brush making contact with the generator shaft. The
oil film in the bearing prevents the relay coil from being short-circuited.
This is the only known technical information on shaft current protection in
the international relay literature.
Hydro-generators owned by Vattenfall have, according to Lohage
and co-workers [ 2 ], shaft current protection. The shaft current relay trips
the unit.

GENPC200.DOC 2007-05-22 15.31

20.4 Bearing Current Protection

References

[1] Ammann, C., Reicher, K., Joho, R. & Posedel, Z.: "Shaft
Voltages in Generators with Static Excitation Systems -
Problems and Solutions", IEEE Trans. on Energy Conversion,
vol. 3, no. 2, pp. 409-419, June, 1988.
[2] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[3] Buckley, G.W., Corkins, R.J. & Stephens, R.N.: "The
Importance of Grounding Brushes to the Safe Operation of Large
Turbine Generators", IEEE Trans. on Energy Conversion, vol. 3,
no. 3, pp. 607-612, September, 1988.
[4] "Relay Protection of A-C Generators", AIEE Committee Report,
AIEE Trans., vol. 70, pt. I, pp. 275-282, 1951.

GENPC200.DOC 2007-05-22 15.31

 21.1

21. BREAKER FAILURE PROTECTION

It is prudent to require that frequent faults with serious consequences shall
be cleared even if one relay or one switching device fails to operate when
required. Many utilities design their system so that it can withstand,
without damage, the clearance from a back-up relay or when one switching
device fails to operate.
The previous sections have documented main protection and backup
protection for the generator and related power system components. This
section will discuss the consequences of a failure to operate of a switching
device. It is sufficient to note that the probability of a failure to open is low.
According to a CIGRE questionnaire, it may be as low as 0.0001.
According to operational experience from CEGB it may be in the order of
0.001. Operational experience from Sydkraft shows that the circuit breakers
maloperate in about 3 percent of the disturbances. Here, it is sufficient to
note that the probability is greater than zero. Few protection engineers are
prepared to recommend their companies to neglect the risk of breaker
failure.
Sometimes it is possible to rely on remote backup protection. By in-
troducing breaker failure protection for the generator breaker and the unit
breaker it is possible to reduce the time for back-up fault clearance. Often it
is less expensive to install local back up protection and breaker failure pro-
tection than to reinforce the power system for longer fault clearance times.
It is common practice to use breaker failure protection with three or
four overcurrent relays and a timer. The breaker failure protection checks if
all phase currents, and sometimes also the residual current, fall to zero. If
this occurs within a certain time after the protection system has given a trip-
ping impulse to the circuit breaker. Usually one sets the timer to operate af-
ter 150 to 200 ms. If all current relays then have dropped out, the switching
device has interrupted the fault current and no further action is necessary. If
this is not true, the breaker failure protection trips adjacent circuit breakers.
Lohage and co-workers [ 1 ] discuss the setting of the current relays
in the breaker failure protection. Their conclusion is that the current relays
must have the same sensitivity as the relays that start the breaker failure
protection. For hydro-generators, they recommend that a setting of about 30
to 50 percent of the rated current of the generator. For turbo-generators, it
may be necessary to use even lower settings depending on the number of
the reverse power relays. It is hardly impossible to find overcurrent relays
that can always detect reverse power and simultaneously withstand normal
load current and fault current.

GENPC210.DOC 2007-05-22 15.31

21.2 Breaker Failure Protection

A local back-up protection system has the advantage that it is the cir-
cuit breaker in the same station that carries out the back-up protection. This
is an advantage for the system operators.
An AIEE report [ 3 ] documents contemporary practice for back-up
protection in thermal power plants. Then, it was obvious that the big invest-
ments in the power plants motivate the existence of a back-up protection
system. Then, the equipment was more sturdy and the fault currents were
lower. It is not a surprise that the utilities then accepted remote back-up
protection systems.
An AIEE working group has prepared and sent out a questionnaire
about practice and experience of underexcitation protection. The group has
analysed and documented [ 2 ] answers from 63 utilities in the US, Canada
and Mexico. The answers represented 309 generators rated at 60 MW or
more, installed since 1949. In 1958, 19 utilities said that they had no back-
up protection for short circuits in the generator. On the other hand, 18 utili-
ties required back-up protection for short circuits in the generator, 15 utili-
ties required breaker back-up for short circuits in the generator, 20 utilities
required back up protection for the high voltage busbar, 14 utilities
required a breaker back up for short-circuits on the busbar, 29 utilities
required the tripping of the generator when the line protection failed to
operate. The working group recommended installation of back-up
protection to obtain the same level of reliability for generator faults as for
line faults.
The unit breaker and the generator breaker (if it exists) for hydro-
generators owned by Vattenfall shall, according to Lohage and co-workers
[ 1 ], have breaker failure protection using three or four current relays.

References

[1] Andersson, B., Broman, H., Eriksson, P.- A., Fredriksson, S. &
Lohage, L.: "Generatorskydd i Vattenkraftstationer", Rapport,
Vattenfall, November, 1982.
[2] "A Partial Survey of Relay Protection of Steam-Driven A-C
Generators", AIEE Committee Report, AIEE Trans., vol. 81, pt.
III (Power Apparatus and Systems), pp. 954-957, February,
1962.
[3] "Relay Protection of A-C Generators", AIEE Committee Report,
AIEE Trans., vol. 70, pt. I, pp. 275-282, 1951.

GENPC210.DOC 2007-05-22 15.31

 22.1

22. OPERATIONAL EXPERIENCE

This section documents some operational experience of protection systems.

22.1 Experience from Belgium

Reference [ 1 ] documents a blackout that occurred in Belgium on 4 August
1982. The disturbance started in a nuclear power station with tree units.
The initiating event was the tripping of Doel 3 with a rated active power of
900 MW. At the time of the disturbance, the output of that unit was
700 MW and 449 Mvar. Unit 1 in Doel operated at 388 MW and 222 Mvar
while unit 2 operated at 382 MW and 236 Mvar. These two units operated
close to their reactive limits. There was no reactive reserve available in the
power station to cover the loss of the reactive output from unit 3.
The loss of Doel 3 caused a voltage depression on the 380 kV trans-
mission network. The automatic voltage regulators (AVRs) on the
remaining units increased the excitation current and tried to increase the
voltage level. In some remaining units, the overexcitation limiter in the
AVR allowed the field current to increase above the pick up level of the
overcurrent protection for the rotor. Then, the overcurrent relays tripped
these units. In summary, the tripping of one unit caused cascading trips of
several other units. This resulted in a widespread blackout. The restoration
of the network and connection of the customers took six hours.

22.2 Experience from France

Reference [ 3 ] documents a disturbance that occurred in France on 12
January 1987. The tripping of three big thermal units initiated the
disturbance. Independent faults caused these disconnections that occurred
within 42 minutes. The losses caused a voltage depression on the 400 kV
transmission network. The automatic voltage regulators (AVRs) in
remaining units increased the excitation current and tried to increase the
voltage level. This caused the tripping of 11 units with a total rated active
power of 8 500 MW. The rotor overcurrent relay tripped these. The tripping
of the generating units caused a widespread blackout in south-western
France. The network restoration and connection of customers took about
12 hours.

GENPC220.DOC 2007-05-22 15.31

22.2 Operational Experience

22.3 Experience compiled by IEEE

Reference [ 2 ] documents a survey of practices and experiences of genera-
tor back-up protection. The report contains an analysis of time-delayed
back-up protection. They are associated with unit-connected generators,
predominantly rated at 100 MVA and higher. In addition, it discusses appli-
cation and setting considerations. The questionnaire covered the following
types of back-up protection systems:

51V-C voltage-controlled phase time-overcurrent (disabled until

voltage drops below set level)
51V-R voltage-restrained phase time-overcurrent (current pickup pro-
portional to voltage)
51 phase time-overcurrent
21 phase distance (facing into the external power system)
46 negative-sequence time overcurrent
51N neutral time-overcurrent (in the high side of step-up
transformer)

The responses represented an average of 6 698 unit-years of experience

with major unit-connected generators. The largest reported unit of a
particular user ranged from 120 to 1 355 MVA, with an average of
850 MVA. The range of the smallest units was 75 to 678 MVA, with an
average of 140 MVA. The survey showed the flowing:

1. The majority apply all three classes of back-up protection, phase,

negative-sequence and high-side neutral overcurrent.
2. Distance relaying predominates versus overcurrent for phase backup
protection.
3. The use of backup, for sensing transmission faults, is less of a moti-
vation than sensing low-side and local switch-yard faults.
4. Almost all retrofit activity is confined to the negative-sequence
class, where it is presumed that static types are being applied.
5. The major objectives for use of negative-sequence overcurrent
relaying are to detect: (a) series unbalance and (b) local faults versus
transmission faults.
6. The major objective for use of neutral overcurrent relays is to
protect for unit-transformer or switchyard fault.
7. The respondents reported 26 correct and 19 incorrect operations.
The neutral overcurrent class had the most favourable ratio of
correct to incorrect operations.

GENPC220.DOC 2007-05-22 15.31

Operational Experience 22.3

a) Series unbalance accounted for all of the 8 correct negative-

sequence operations.
b) Explanations were included for 6 of the 10 neutral overcurrent
correct operations. Of the six, two for open breaker poles and
two where the relay operated after the differential relaying had
tripped the unit.
c) Causes for the 19 incorrect operations were:
9 faulty or maladjusted relays
3 wiring errors
3 incorrect settings
3 open potential contacts
1 personnel error

8. Three users reported cases of extensive damage where back-up

protection was not applied for the conditions encountered.

A discussion of application and setting considerations is included. Three

subjects not covered by the survey are also presented:
• Unit auxiliaries transformer circuit backup
• 100% turn-turn fault detection
• Stator-ground fault back-up

22.4 Conclusions
The two blackouts described above show the importance of co-ordinating
control system functions and protection system functions. It is not good
enough to co-ordinate the protection for the generator and the step-up trans-
former. In the setting of rotor current limiters and overcurrent protection it
is important not to restrict the capability of the unit by selecting typical set-
tings.
The IEEE survey shows that utilities in the US prefer to use distance
protection as phase back-up protection in favour of overcurrent protection.
It also shows that the neutral point overcurrent relay performs well as
backup protection. Another result is that it is difficult to achieve the desired
level of security. When a back-up protection maloperates during a power
system disturbance, there is a risk that it disconnects several units and
causes a very severe contingency. Few transmission systems can withstand
such contingencies.

GENPC220.DOC 2007-05-22 15.31

22.4 Operational Experience

References

[1] Van den Damme, R.: "The incident of August 4th 1982 of the
Belgian electricity system", Intercom, 12 September, 1983.
[2] "A survey of generator back-up protection practices", IEEE
Committee Report, IEEE Trans. on Power Delivery, vol. 5, no.
2, pp. 575-584, April, 1990.
[3] "Die Störung vom 12. Januar 1987, 11.42 Uhr", Unpublished
translation of an internal report from EdF, Mouvements
d'Energie, 2 February, 1987.

GENPC220.DOC 2007-05-22 15.31