You are on page 1of 4

Control for Stability

in Interconnected Power Systems


Graham J. Rogers
ABSTRACT: The increase in size and complexity of interconnected power systems,
coupled with industlys commitment to maximum security at minimum cost, has led to
the development of many special control devices. These control devices ensure that the
system is able to operate, without instability,
under a wide range of system conditions.
This paper describes the development of a
number of stabilizing controls. The robustness of the overall power system is emphasized along with modifications to the basic
controls necessary to achieve this robustness. Analytical tools used in the design of
practical power system controls are described. The importance of validation of
modeling and simulation methods by planned
system tests and by the analysis of naturally
occurring faults is stressed. Likely future
system developments are reviewed, including the implication on the type of controls
that may be necessary.

Introduction
In recent years, power systems, worldwide, have grown markedly in size and complexity. In order to maximize efficiency of
generation and distribution of electric power,
the interconnections between individual utilities have increased and the generators have
been required to operate at maximum limits
for extensive periods of time. In addition,
the most economic sites for generation plants
are often remote from load centers and the
power must be transmitted over long distances. The majority of power system interconnections are made through AC transmission lines and the interconnected generators
run synchronously. In a large interconnected
system, such as that in North America, there
may be thousands of synchronous generators
in service to supply the load. Each generator
normally has separate controls that are used
to regulate the real and reactive power supplied by the generator to the system.
Transients in power systems are analyzed
using many levels of modeling detail. At one
extreme is the study of electromagnetic tranDr. Graham J. Rogers is with Ontario Hydro, 700
University Avenue, Toronto, Ontario M5G 1x6,
Canada.

sients initiated by steep wavefront pulses


(such as lightning strikes); at the other extreme is the study of long-term transients,
with periods of several minutes or more, involving the interaction between slow automatic controls and manual control by system
operators. And between is the study of electromechanical oscillations between the synchronous generators within the system. Control for the stability of the electromechanical
oscillations is the subject of this paper. The
time period of concern is from 1 to 40 sec
following a disturbance, and the frequency
range is from 0.1 to 2.0 Hz. Because of the
separation between the frequency of the three
types of transients, each can be studied using
simplified dynamic models. In electromechanical oscillation studies, detailed models
are used for machines, including their excitation and governing systems, but the highfrequency network transients are ignored as
are the low-frequency steam turbine boiler
dynamics and the slow system controls (such
as on-load tap changers). The resulting system is governed by nonlinear differential
equations, which describe the interchange of
electromechanical energy between the generators through the transmission network.
Because of their essential nonlinearity, the
stability of power systems depends on the
severity of the applied disturbances. Criteria
for power system design specify the types of
fault the system must be able to withstand
without major loss of synchronism and consequent breakup. It is also critical that the
power system remains stable while operating
with no faults. Power system analysts refer
to these separate, but related, stability problems as transient stability and small-signal
stability, respectively. In general, the system
operating conditions are restricted most by
the need to maintain transient stability. In
recent years, however, as power systems
have been operated with higher power transfer levels to meet economic constraints,
small-signal stability problems have become
apparent. In order to achieve the required
high transfers of power, the controls associated with the generators have become critical.
In special cases, asynchronous DC transmission is utilized with controlled rectification at one end and controlled inversion at

the other. In association with DC links, and


in order to achieve a uniform voltage distribution through the power system, thyristorcontrolled capacitors and reactors (static var
compensators) have been found necessary.
These devices provide additional local controls that have significant effects on power
system stability.
The power system control designer must
ensure that the power system is stable locally
and globally. Global control is achieved
through correct design and coordination of
the local controls of the individual components of the power system and by restricting
the allowable operating conditions of the
system. In this paper, a number of stability
problems and their solutions are described.

Local Generator Stability


The two most important controllers on
modern synchronous generators are the speed
governor and the automatic voltage regulator
(AVR). In an interconnected system, neither
fulfill their nominal function alone: the speed
governor also controls the active power supplied by the generator to the system; the AVR
controls the reactive power supplied by the
generator to the system.
Automatic Voltage Regulator

The automatic voltage regulator plays an


essential role in keeping the generator in
synchronism with other generators on the
system. In order to do this efficiently, it must
be fast-acting. On a new plant in Ontario
Hydro, high-gain electronic controls, with a
controlled thyristor power output stage, are
used to supply the field of the generator, proportionally to the difference between a reference signal and the generators terminal
voltage magnitude. The result is to produce
a highly oscillatory, potentially unstable,
mode of electromechanical energy interchange between the local generator and the
rest of the power system. The mode may be
stabilized by feeding an additional signal into
the AVR input. Most commonly, the signal
is derived from the speed of the generator
rotor, although the generator power and frequency also may be used [ 11, [ 2 ] .
A dynamic compensator is used to modify

0272-170818910100-0019 $01 00 G 1989 IEEE

January 1989

19

the stabilizing signal to the AVR in such a


way that the damping of the electromechanical mode is increased. The device is often
referred to as a power system stabilizer. Although apparently straightforward, a number
of practical difficulties with such devices
have occurred.
One of the most serious difficulties experienced with early power system stabilizers
fitted to steam-turbine-driven generators was
their interaction with the turbine-shaft dynamics and consequent instability of the first
torsional mode. The reason for the interaction was twofold. First, speed measurement
at the generator rotor contains a strong component of this torsional mode. Second, the
stabilizer compensation is essentially a phaselead circuit, which increases the high-frequency gain of the stabilizer. A change in
the location of the speed measurement transducers to a node of the lowest torsional together with tuned torsional filters was the
first solution to this problem [ 2 ] .This, however, led to additional exciter modes,
which restricted the stabilizer gain and,
hence, the achievable damping of the electromechanical mode. The most recent stabilizers achieve rejection of the torsional
modes by using a combination of generator
power and speed (the Delta PIOmega stabilizer) [3]. With this device, the additional
exciter mode introduced by the stabilizer is
far less sensitive to the stabilizer gain and,
thus, higher values of electromechanical
mode damping may be obtained.
Under transient conditions following a severe fault, the action of a speed input power
system stabilizer is often opposite to that required. It is important that stabilizer output
be restricted both in the positive and negative
directions. The negative limit is the most
critical. It is normally set to between -5 and
-10 percent of the rated terminal voltage
setting. The positive limit is normally set to
20 percent of the rated terminal voltage setting.
Practically, the stabilizer acts by modulating the voltage in the vicinity of the generator in such a way that the generator electrical torque has a component in phase with
the generator rotor speed change. During severe swings, the voltage may be forced to
dangerously high levels if uncontrolled. A
terminal voltage limit signal is thus required,
which opposes the stabilizer signal for terminal voltages higher than the maximum safe
level. In general, the limiter takes the form
of a high-gain terminal voltage control loop,
biased off for normal voltage. When operating, torsional modes present in the terminal
voltage signal may become unstable; it is
important to limit the high-frequency loop

20

gain of the voltage limit circuit to avoid instability.


Governor

The governors time response is slow


compared to the frequency of the local electromechanical oscillation, and its dynamics
have little effect on this modes stability.
Modern steam turbines, with very fast electrohydraulic governor valve drives, are now
being used, which have been found to give
rise to torsional instabilities similar to those
caused by early power system stabilizers.
Torsional filters have proved effective in
controlling this type of instability [4].
With hydraulic generators, the governor
auxiliary controls, which set the transient
droop and reset time, require careful tuning
to assure stability of the generator both when
running in isolation from the rest of the system and when synchronized to the system
[ 5 ] . With the introduction of electrohydraulic gate controls in modern hydraulic
turbine governors, there is scope for the introduction of more radical control design.

Interarea Stability
Following both small and large disturbances, a power system experiences low-frequency oscillations, which are associated
with groups of synchronous generators
swinging against other groups of synchronous generators through weak transmission
connections. The frequencies of these interarea modes are lower than those of the local
modes. In general, interarea modal frequencies lie in the range of 0.1 to 0.8 Hz, whereas
the local modal frequency range is from 0.8
to 2.0 Hz. The lower modal frequency and
the fact that many machines participate in
the mode make interarea oscillations more
difficult to control than local oscillations.
Nonetheless, it has not yet been found necessary to use centralized controls to stabilize
these modes.
Power system stabilizers on large generating units can be designed to help damp
interarea modes in which the generators are
significant participants [6]. Thus, it is important that the dynamic compensator associated with each power system stabilizer is
designed to ensure a positive contribution to
damping of all modes having frequencies between 0.1 and 2.0 Hz. This may entail a
slight reduction in the damping of the higherfrequency local modes.
At the lower end of the interarea mode
frequency range, both hydraulic and steam
turbine dynamics can affect the damping of
the mode. The nonminimum-phase characteristic of the hydraulic turbine can cause the

turbine torque to increase with increasing


speed rather than decrease, as required to
damp system oscillations. In steam turbines,
the reheater stage time constant is the critical
element, which may introduce a phase lag
and reduce the low-frequency damping. In
both types of turbine, a simple phase-lead
compensator may be used in the governor to
ensure that the turbine characteristics do not
increase the risk of low-frequency, interarea
instability [7], [8].

Transient Stability
Following large system disturbances, some
synchronous generators may swing sufficiently to lose synchronism with the system.
This is prevented, for a wide range of specified, severe system faults by the provision
of an adequate transmission system, with
rapid fault clearance facilities, and by setting
system operating limits. The system design
and its operating limits are based on extensive simulation of the nonlinear system electromechanical dynamics. The stability of the
system following a severe fault can be aided
significantly by the installation of fast-acting
AVRs on major generating units. As noted
previously, fast-acting AVRs generally require power system stabilizers to give adequate small-disturbance stability. The power
system stabilizers may not produce the correct control input to the AVR following a
large system disturbance, and additional
ovemding nonlinear controls may be necessary. In Ontario, the Transient System Excitation Control (TSEC) [9] is used to force
up the voltage at the terminals for generators
accelerating following a fault. The maximum voltage is limited by a special fastacting, bang-bang voltage limiter system.
TSEC is operative for only the first swing of
the system oscillation following the fault.

Direct Current Transmission


Thus far, synchronous generator controls
and their influence on different aspects of the
stability of power systems have been discussed. As a result of various system constraints, some technical and some economic,
DC transmission is being used more and
more.
The characteristics of the firing angle controls (pole controls) at the rectifier and inverter interfaces between the DC transmission and the AC transmission can have a
significant impact on both local and global
stability. The pole controls can affect the stability of nearby generators; it is usual to supply supplementary controls to prevent instability. The pole controls generally have a

IEEE Control Systems Magazine

high bandwidth and interaction between the


controls, and the shaft torsional dynamics of
nearby steam turbines have produced torsional instability that was stabilized by modifying the high-frequency characteristics of
the current control loop and adding a notch
filter to the power modulation control [ 101.
Because the DC transmission can control
significant amounts of power, its effect on
the damping of interarea modes can be considerable. The need for an AC system whose
voltage is relatively insensitive to changes in
the power transmitted through the DC system has led to additional voltage control devices, such as static var compensators and
synchronous condensers, being placed in the
AC system close to the DC transmission terminals.
Thus far, the controls for each DC transmission system have been designed individually. There are insufficient DC links in service to draw generic conclusions about the
type of control needed to ensure global as
well as local stability. As the transmission
capability of DC links is growing, there is
little doubt that their interaction with the system is likely to be more severe in future systems and that special consideration will have
to be given to the effect that their control
design has on the global stability of the system [ l l ] .

Analysis Tools
Because of the size of power system
models being used in stability analysis, specially developed computer programs are
used. Step-by-step integration of nonlinear
equations of the system is used in the study
of transient stability. For small-signal stability studies, equations are linearized about
an operating point, and eigenvalue and eigenvector techniques are used to perform
modal analysis of the system. Ontario Hydro
currently uses programs in which systems
having up to 12,000 AC network nodes and
1500 synchronous machines may be represented. Up to 1000 of the synchronous machines may be modeled, in detail, with AVR
and governor, leading to, potentially, 15,000
dynamic states [ 121. Clearly, the consistency
of data for such system models is as much a
concern as the mathematical techniques of
modeling and analysis [131.
Fortunately, the stability problems encountered in regular power system design are
often more local in nature, and reduced-order models can be determined that adequately represent the system for their study.
In particular, the design of AVRs, governors, and power system stabilizers for smallsignal stability often can be performed using

January 1989

a model of a single generator connected via


a transmission line to a constant voltage
source (infinite bus). Large models of the
system are, however, still necessary to check
these locally designed controllers for their
effect on the global stability of the power
system [6]. For transmission system design
and the determination of operating limits to
ensure transient stability in the first few seconds following a fault, a reduced-order system may be simulated that retains detail in
an area close to the fault, with distant generators represented by aggregate models [ 141,
[ 151. Low-frequency interarea oscillations
excited by severe disturbances may lead to
groups of generators losing synchronism after
several periods. Accurate simulation of this
phenomenon requires extensive system modeling beyond the immediate vicinity of the
fault. Although it may be possible to produce
reduced-order models, the concepts of close
and distant areas may not be valid. Additional work is required on system reduction
techniques to allow the retention of the lowfrequency modes, accurate in both frequency
and damping, and which maintain the basic
structure of the original power system model.
Validation of the simulation models is a
continuing process. Field tests, which can be
camed out with no risk to system performance, are used in Ontario Hydro to improve the detailed modeling of this particularly important plant. This leaves in question
the accuracy of simulations following major
system disturbances. Transients following
naturally occumng faults are monitored and
compared with simulations of the same
events to provide pointers to the need for
dynamic model refinement.

General Comments
The degree of stability of power systems
is less important than in many other control
problems. What is required is for the power
system to remain stable over a wide range
of operating conditions. Oscillations of about
1 Hz in frequency with a damping ratio of
0.05 are commonly encountered associated
with synchronous machines having slow excitation systems and no power system stabilizers, and which give rise to no stability
problems. There is little need, therefore, for
optimal design of controllers in order to
maximize damping. Indeed, because of the
nonlinear nature of the system dynamics, the
robustness of optimal controls, based on linear analysis, is often suspect [16]. Any design based on linear analysis of a reduced
system should be checked extensively by
nonlinear simulation of the full interconnected system [6].

The modes of oscillation involving the


system as a whole, the interarea electromechanical oscillations, normally can be stabilized by decentralized controllers placed at
those generating units that participate significantly in the mode. However, the advent of
multiterminal DC links imbedded within the
AC system may well require centralized DC
pole controls to ensure global system
stability.
Robustness of power system control design is important but has been approached in
a very practical sense thus far. Lack of robustness quickly shows when commissioning and operating new plants and immediate
steps are necessary to rectify problems encountered. In some cases, quick solutions to
local robustness problems have led to global
stability problems as the power system has
developed. For example, power system stabilizers designed for correct compensation at
a single local mode natural frequency can
destabilize lower-frequency interarea modes.
Another extreme case could be that special
controls designed for the protection of expensive plants could cause the plant to be
tripped from the system to the detriment of
the stability of the overall power system. Coordination of control design is clearly necessary.
For the future, there is a need for continued work in the dynamic simulation of very
large systems. The development of efficient
techniques that utilize the fundamental physical properties of the power system, such as
singular manifolds [171, are required to ease
the computational burden of transient and
small-disturbance stability studies. Methods
for more systematic ways to ensure robustness, applicable at the design stage, would
also be useful. However, as with all evolving systems, there is always the possibility
of new methods of control interacting in an
unexpected way to introduce potentially
unstable modes. Such developments are difficult to forecast and indeed may be hidden
by oversimplified modeling in the initial design stages. Therefore, it is important that
any assumptions made in the power system
model must be realistic. Oversimplified
models should be treated with extreme caution and used to provide only qualitative results of a general nature.

Conclusions
Control is vital to maintain the stability of
modem interconnected power systems. Although a vast amount of experience has been
accumulated in the last 20 years, the changing nature of the power system continues to
provide challenges to the system designer.

21

Simulation and analysis methods are under


constant revision and improvement to enable
the designer to ensure that the system is robust and may be operated in an economic
manner. There remains considerable scope
for new approaches that will reduce the computational burden involved in the simulation
of large power systems.

References
[I] P. L. Dandeno, A. K. Karas, K. R. McClymont, and W. Watson, Effect of HighSpeed Rectifier Excitation Systems on Generator Stability Limits, IEEE Trans. PAS,
vol. 87, pp. 190-201, 1968.
[2] P. Kundur, D. C. Lee, and H. M. Zein ElDin, Power System Stabilizers for Thermal Units; Analytical Techniques and OnSite Validation, IEEE Trans. PAS, vol.
100, pp. 81-95, 1981.
[3] D. C. Lee, R. E. Beaulieu, and J . A. R .
Service, A Power System Stabilizer Using
Speed and Electrical Power Inputs-Design
and Field Experience, IEEE Trans. PAS,
vol. 100, pp. 41514157, 1981.
141 D. C. Lee, R. E. Beaulieu, and G. J . Rogers, Effects of Governor Characteristicson
Turbogenerator Shaft Torsionals, IEEE
Trans. PAS, vol. 104, pp. 1255-1261,
1985.
[5] P. L. Dandeno, P. Kundur, and J . P. Bayne,
Hydraulic Unit Dynamic Performance
Under Normal and Isolated ConditionsAnalysis and Validation, IEEE Trans.
PAS, vol. 97, pp. 2134-2143, 1978.
[6] P. Kundur, M. Klein, G. J . Rogers, and
M. S. Zywno, Application of Power System Stabilizers for Enhancement of Overall
System Stability, to be presented at IEEE
PES Summer Meeting, 1988.

[7] F. R. Schleif, G . E. Martin. and R. R. Angell, Damping of System Oscillations with


a Hydrogenerating Unit, IEEE Trans.
PAS, vol. 86, pp. 438-442, 1967.
[8] F. M. Hughes, Improvement of Turbogenerator Transient Performance by Control
Means, Proc. IEE, vol. 120, pp. 233-240,
1973.
[9] J. P. Bayne, P. Kundur, and W. Watson,
Static Exciter Control to Improve Transient Stability, IEEE Trans. PAS, vol. 94,
pp. 1141-1 146, 1975.
[IO] M. Buhrman, E. V. Larsen, R. J . Puvko,
and H. S . Patel, Experience with HVDCTurbine-Generator Torsional Interaction at
Square Bute, IEEE Trans. PAS, vol. 99,
pp. 966-975, 1980.
[ l l ] R. L. Cresap and J. F. Hauer, Emergence
of a New Swing Mode in the Western Power
System, IEEE Trans. PAS, vol. 100, pp.
2037-2043, 1981.
[I21 D. Y. Wong, G . J . Rogers, B. Porretta, and
P. Kundur, Eigenvalue Analysis of Very
Large Power Systems, IEEE Winter Power
Meeting, New Orleans, LA, 1987.
[13] P. L. Dandeno, P. Kundur, A. T. Poray,
and M. E. Coultes, Validation of Turbogenerator Stability Models by Comparisons
with Power System Tests, IEEE Trans.
PAS, vol. 100, pp. 1637-1643, 1981.
141 R. Podmore, A Comprehensive Program
for Computing Coherency Based Dynamic
Equivalents, PICA-79 Conf: Proc., May
1979.
151 J. S. Lawler and R. A. Schlueter, Computational Algorithms for Constructing
Modal-Coherent Dynamic Equivalents,
IEEE Trans. PAS, vol. 101, 1982, pp.
1070-1078, 1982.
1161 J. H. Chow and J . J . Sanchez-Gasca, FRquency Response Evaluation of State Space
Designed Controllers for Systems with

Lightly Damped Oscillatory Modes-A


Power System Stabilizer Example, Proc.
26th Con$ on Decision and Control, Los
Angeles, CA, Dec. 1987.
[I71 B. D. Riedle and P. V. Kokotovic, Integral Manifolds of Slow Adaptation, IEEE
Trans. Automat. Contr., vol. 31, pp. 316324, 1986.

Graham Rogers was


born in Birmingham, England, in 1933. After
completing an engineering apprenticeship and
national service in the
Royal Air Force, he attended Southampton University where he graduated in 1961 with firstclass honors in electrical
engineering. From 1961
to 1964, he was a Consultant Mathematician with AEI (Rugby) Ltd.
From 1964 to 1978, he was Lecturer in electrical
engineering at Southampton University where he
taught Control Theory and did research into the
dynamics and control of electrical machines. Since
1978, when he immigrated to Canada, he has been
employed by Ontario Hydro, where he is currently
System Design Engineer, Specialist-Controls in the
System Planning Division. His responsibilities include the development of techniques for system
stability analysis and their application to specialized power system problems. He also holds the
appointment of Associate Professor (part time) at
McMaster University. He is a Fellow of The Institute of Mathematics and its Applications and a
Registered Professional Engineer in the Province
of Ontario.

4 989 Conference on Neural Networks


The IEEE Control Systems Society is
among the sponsors of the third IEEE International Conference on Neural Networks.
The conference will be held in Washington,
D.C., at the Washington Sheraton Hotel on
June 19-22, 1989. The conference will include exhibits of the latest neurocomputers,
neural network software, and applications
presented by some 40 companies and organizations. Papers of eight pages or less are
solicited on the following areas: optical and
electronic neurocomputers, combinatorial
optimization, network architectures, neural

22

network theory, neurobiological connections, knowledge processing, learning algorithms, and novel applications including vision, robotics, self-organization, communications, control, and speech recognition
and synthesis.
The Conference Chair is Shun-Ichi Amari,
the International Chair is Rolf Eckmiller, and
the Program Committee Chair is Robert
Hecht-Nielsen. The Organizing Committee
Chairs are Wesley Snyder and Allen Stubberud.
Prospective authors should contact Nomi

Feldman, Conference Coordinator, at the


address below. February 1, 1989, is the
deadline for final copy of manuscript typed
in standard IEEE conference proceedings
format on IEEE mats. For further details or
to request an IEEE Authors Kit, call or
write:
Nomi Feldman
ICNN-89 Conference Coordinator
3770 Tansy Street
San Diego, C A 92121 USA
Phone: (619) 453-6222

l E E Control Systems Magazine

You might also like