Mohammed, Zakaria Fadlalmoula (1996) An investigation of

turbogenerator dynamics and control. PhD thesis.

http://theses.gla.ac.uk/1727/

Copyright and moral rights for this thesis are retained by the author
A copy can be downloaded for personal non-commercial research or
study, without prior permission or charge
This thesis cannot be reproduced or quoted extensively from without first
obtaining permission in writing from the Author
The content must not be changed in any way or sold commercially in any
format or medium without the formal permission of the Author
When referring to this work, full bibliographic details including the
author, title, awarding institution and date of the thesis must be given

Glasgow Theses Service

http://theses.gla.ac.uk/
theses@gla.ac.uk

AN INVESTIGATION
DYNAMICS

OF TURBOGENERATOR
AND CONTROL

THESIS
SUBMITTED
TO THE DEPARTMENT
OF
ELECTRONICS
AND ELECTRICAL
ENGINEERING
OF GLASGOW UNIVERSITY

FOR THE DEGREE

DOCTOR OF PHILOSOPHY

BY
ZAKARIA

FADLALMOULA

MOHAMMED

ABSTRACT:
This thesis provides an investigation of the dynamics and control of
turbogenerators from a multivariable control viewpoint. The multivariable control
framework chosen -Individual Channel Analysis and Design- is particularly
dynamical
it
the
the
of
uncontrolled
characteristics
appropriate since encapsulates
for
limitations
to
the
subsequent
potential and
exposing
system with a view
is
The
the
thesis
a
closed-loop control.
main contribution of

complete new

insight into why excitation/governor control with Power System Stabilisers (PSS)
has been so successful for the control of turbogenerators connected to an infinite
bus is provided by the small-signal multivariable, analysis framework, Individual
Channel Analysis and Design. The multivariable analysis justifies treating the
turbogenerator system as a pseudo- Single-Input Single-Output, (SISO) system
SISO
is
first
loop
loop
is
treated
the
as a
exciter
closed and
where the governor
function
disturbances.
The
for
the
of
prime purpose of rejecting voltage
system
the PSS is identified as that of overcoming an awkward switch-back frequencydomain characteristic of the excitation channel so as to permit high-performance
bandwidths
10
be
to
that
rad/sec
otherwise could not
up
excitation channel
in
Thus,
addition to the control requirements of set point regulation of
obtained.
the terminal voltage and shaft speed, the PSS provides for a second control
disturbance
important
frequency
the
requirement of strong voltage
rejection over

0
10
The
PSS
is
to
rad/sec.
of
control
option
also assessedagainst other
range
Several
other results concerning stability robustnessto system
control options.
different
in
follow
from
the analysis in a
system configurations
uncertainties
transparentand immediate way.

Acknowledgments

During the course of this research, I have been assisted by many

individuals to whom I am personally, and intellectually greatly indebted.
In Glasgow, Professor John O'Reilly, my Supervisor, offered me
friendliness,
without which this work
unfailing guidance, encouragementand
have
been
difficult
to undertake.
very
would
In Sudan, I am indebted to the late Dr. Mohmoud Sharief, of the National

Electric Corporation, Sudan, who introduced me to the world of Power

Systems,by shifting my academicinterestsfrom the field of Electronics.
This research has also been greatly influenced by discussion with Dr.
William Leithead of the Industrial Control Centre, at Strathclyde University.

And lastly, and above all, I am greatly indebted to the SudanesePeople,

funds
for
(and
difficulties
hundreds
despite
like
their
provided
me
me)
who
through the Ministry of Education

Table of Contents
Page

Content

Introduction

Chapter I

1.1 Power system stability

1.2 The synchronousmachine

1.3 Damping and synchronising torques

1.4 Power system stabiliser (PSS)

1.4.1 Robust PSS

1.4.2 Fuzzy PSS

10

1.5 Researchof this thesis

12

Chapter 2

Review of ICAD

2.1 Introduction

15

2.2 Multivariable analysis using ICAD

16

2.3 Non-diagonal control

26

Chapter 3 Modelling and Multivariable AnalYsis of the turbogenerator

3.1 Introduction

29

3.2 Park's equations for synchronousmachines

31

3.3 Synchronousmachine model for small signals studies

36

3.4 Damper winding representation

42

3.5 Exciter and voltage regulatorsmodels

45

3.6 Governor and turbine models

46

3.7 Analysis of turbogeneratorsusing ICAD framework

47

3.8 Analysis of parametersaffecting model performance

53

3.8.1

the 3rd-order model

53

3.8.2

the 6th-order model

54

3.8.3

in
changes loading conditions

58

3.8.4

in
changes synchronousreactance

60

3.8.5

in
changes tie-line reactance

62

3.8.6

in
damping
constants
changes

63

3.8.7

inertia
in
constants
changes

65

Chapter 4

Control Options Available to Turbogenerator Systems

1 Introduction

69

4.2 Multivariable analysis of turbogeneratorregulation

71

4.3 Factors affecting the switch-back characteristics

76

4.3.1

Loading conditions

76

4.3.2

Synchronousreactance

77

4.3.3

Tie-line reactance

78

4.4 Control option 1: Excitation/govemor control without PSS

80

4.5 Control option 2: Excitation/govemor control with PSS

86

5.6 Robustnessof turbogeneratorsystemwith PSS

93

5.7 Control option 3: Ex/Go control with swappedinput/output pairings

95

Chapter 5 Analysis of some recent developments in turbogenerator control

5.1 Introduction

102

5.2 Recent studies involving the PSS

105
105

5.2.1 Augmented PSS

5.3 Recent methods not involving the PSS

115

5.4 Non-diagonal control

124

Conclusion

Chapter 6

6.1 Introduction

126

6.2 Primary result

126

6.3 Some other results

129

6.4 Future work

130

Appendices
" Appendix I:

References

132

" Appendix 2:

Nomenclature

142

" Appendix 3:

Derivation of the small signal third order model

145

* Appendix 4:

Derivation of the transfer-function matrix

151

* Appendix 5:

Third and sixth order linear models matrices

153

Chapter 1: Introduction
1.1- Power System Stability
Since the early days of the electric power industry, it was recognized that, the

have
be
to
the
electric energy supplied
consumermust
quality,
a reliable service, as
is
be
The
produced economically.
related to the voltage magnitude
quality
well as

level and its electric frequency; both must be within a narrow range with respect to
their rated values, so that consumers can expect that electromechanical devices and
loads
Reliability,
types
to
their
other
of electrical
perfonn according
specification.
on the other hand, is a concept associated with the availability of the service

be
it.
In
to
short, power supplied consumermust stable.
whenever consumersneed
Power system stability is a complex subjectthat has challengedpower system
for
engineers
many years, and early stability problems were associatedwith remote
hydraulic generating stations feeding into metropolitan load centers over longdistance transmission [1]. For economic reasons, such systems were operated close
to their steady-state stability limits, (a power system is in a steady-state operation

frequency
is
fixed
loads
it
to
and voltage).
applying power
consumer
at a
when
Steady-statestability, in the past, was normally deducedfrom consideration of the
[2].
ignores
both
This
damping,
procedure
positive and negative,
power angle curve
in the machine and consequentlydoesn't provide a complete picture or solution of
This
is
the stability problem.
particularly true for a machine with a voltage regulator,
in
dynarnic-stability
the
region with positive synchronizing torque,
when operating

because the movement of the rotor, subsequent to a small displacement, is largely

dependentupon the damping in the system. ff positive damping is present the rotor
is
damping
If
die
hunt,
if
there
the
machine
no
oscillation will
away.
will
and the
damping is negative the rotor oscillations will increase in amplitude and may
become
the
to
unstable.
eventually cause machine
With the continual growth in power systemsand in interconnections,there is
install
larger
larger
to
generating
and
a corresponding continual economic pressure

fact
in
in
that
the
tends
turn
to
there
this
process
result
units, and
was no avoiding
higher reactances, lower inertias being used; and that this tendency sometimes
[3].
Available
job
designing
difficult
the
a reliable and stable systemmore
makes
of
tools to the power system engineer for system analysis and planning or system
have
evolved through the years, and the methods of analysis and the
operation

dictated
by
in
developments
the art of computation and the
models used were
dynamic
Models
theory
stability
of
systems.
and analysis methods started very
(slide
As
simple
rules and mechanical calculators were used). power system evolved
interconnections
independent
found
be
between
to
and
systemswere
economically
increased.
The
the
the
stability problems
attractive,
complexity of
use of analogue
limited
boards
the
size networks were analyzed
computer
gave
means of study and
to solve power flows, fault currents and transient stability problems. For instance,

for
"first
the
models
synchronous
machine,
swing" under transient
using simplified
No
for
field
the
studied.
modeling
conditions was
or torque control was included,

because
the time constantsof these loop were recognized to be long, but
not only
limitations
due
imposed
by
hardware,
to
the
also
so that the implementation of field
loops
Therefore,
it was necessary to
torque
to
control
or
was not easy
accomplish.
keep the overall machine parameters of a power system within bounds, if a

degree
reasonable
of stability

is to be preserved, and Power Engineers turn to

(voltage
increase
limits.
turbine
to
the
regulators and
stability
control action
valves)
In the early 1950s,engineersbecameawareof the instabilities introduced by
the "then" modem voltage regulators, and stabilizing feedback circuits came into
[4].
1960s,
large
interconnected
In
the
action
systems experienced growing
it
inherently
damping
large
thought
that
the
oscillations, and was
of
weak natural
[5],
it
the
and weakly-coupled systems was
main cause
and was suggested that the

damping
be
by
introduced
through the
system
could
enhanced
artificial signals
[6].
Improvements
to system stability also came about by way of
excitation systems
faster-fault clearing and continuous-acting voltage regulators with no dead band,
benefits
high
degree
for
increasing
the
with
of excitation system with a
of response

been
1940s.
the
steady-statestability
recognizedsince
early
Other methods of improving power system stability known to engineersin the
1950s was the technique of replacing the system loads, when faults occur in the
transmission lines, by braking resistors, short-circuit current limiters, and fast
[7].
valving,

1.2- The synchronous machine

One of the main components of an electric power station is the synchronous
in
it
Usually
takes
the
energy
conversion
mechanical-electric
place.
machine,
which

is a 3-phase synchronous type, and is based on Faraday's induction law [8]. Its
essentialcomponentsare: (1) an armature winding, (2) a magnetic field, and (3) a mechanical energy input,
in the form of a force or a torque which causes a relative motion of the armature
field.
in
This
to
the
the armature, and
motion produces voltage
winding with respect

is
torque
generatedopposite to the mechanicaltorque of the prime mover
an electric
has
load.
dynamic
The
the
power system
problems are mainly
generator
a
when
those of the synchronous machines in power systems. For instance, the lowfrequency oscillations of a large electric power system are due to the mechanical
in
have
high
Therefore,
the
to
mode oscillations of
synchronous machine.
order
a

is
in
the
to
quality service,
a closed-loop or
synchronous machine made perform
feedback fonn Fig. 1.1.
,

Cqef

VTef

Exciter

Governor

Field
Turbine

mech. 4- -1-,
(0 .
torque

-4

Vt

Synchronous
machine
Sensor Grid

Fig. 1.1 Turbogenerator system

Traditionally, advantage is taken of the relative weak interaction between the

dc field current (which affects mainly the voltage magnitude at the machine
terminals) with respect to the prime mover (turbine) torque (whose direct effect is on
the electrical power output), when the generator is connected to an isolated electric

load or when it is acting in parallel with other generatorsin a system form. Small
deviations from the fixed referencesare correctedby governor and voltage regulator
In
controls.
a conventially-controlled turbogeneratorsystem, an automatic voltage
(AVR)
for
the exciter field current, to
regulator
provides continuous control action
keep the generator terminal voltage constant, and the governor positions the steam
inlet valves to control the prime mover (turbine) torque.
1.3- Damping and synchronizing torques
When a synchronous machine loses synchronism or "falls out of step" with

its
the
the rest of
system, rotor runs at a higher or lower speedthan that required to

frequency.
"slip"
The
between
field
generate voltages at system
rotating stator
(correspondingto system frequency) and the rotor field results in large fluctuations
in the machine power output, current, and voltage; this causes the protection system
to isolate the unstable machine from the system.

With electric power systems,the change in electrical torque of a synchronous

following
into
be
two components:
a perturbation can resolved
machine
ATe = TsA6 + TD A(o

where
TsAb

is the component of torque change in phase with the rotor angle

Ab
is
perturbation
and

referred

to

as the synchronizing

torque

is
Ts
the synchronizing torque coefficient.
component;
TDAo) is the component of torque change in phase with the speed

deviation Aw and is referred to as the damping torque component; TD

is the damping torque coefficient.
System stability depends on the existence of both components of torque for each

Lack
in
the
torque
of
synchronous machines.
of sufficient synchronizing
results
instability through an aperio&c drift in rotor angle. On the other hand, lack of
damping
in
instability.
torque
results oscillatory
sufficient
1.4- Power systemstabilizer
The availability of thyristor exciters and the use of electrohydraluic
fast
for
the
valving,
with
provide
governors
achieving significant
means

improvements in turbogenerator control. The effects of high-gain, fast-acting

by
have
been
[91,
low-frequency
the
studied
many authors
excitation systems
with
due
large
beinar
to
the
the
creation of
electric power systems
cause
main
oscillations
in
for concern. Subsequent investigations revealed that automatic voltage regulation as

flows
dynamic
Various
heavy
the
stability problem.
power
were aggravating
well as
been
have
dampen
low
to
these
signal
suggested
oscillations.
additional stabilizing
In particular, a stabilizing signal derived from speederror [10] has been found to be
in
damping
have
been
system
effective
oscillations,
and
many
methods
very
design
"Power
System
Stabilizer,
by
this
to
suggested
signal
way of a so called
PSS". One common design approach to the PSS is based on eigenvalue analysis,
identify
to
the modes of oscillations which are close
whereby eigen patterns are used
to the region of instability in the s-domain and which are likely to be a source of

instability;

dynamics
PSS
through
new
poles and zeros are then
additional

introduced to shift those poles away from the danger zone. Other methods include
the root-locus, pole placement, adaptive control etc.
...
Most of the methods use the concept of synchronizing and damping torques

in the machine described above in Section (1.3). DeMeflo and Concordia [ 11] were
damping
by
first
the
to
to
theory
the
effect provided
analyze
use phasecompensation
Nowadays
lead-lag
for
damping
lead-lag
the
machine oscillations.
controller
a
for
damping
is
the
providing
most widely employed excitation controller
controller
AVR-PSS
The
to
conventional approach
characteristics of a synchronous machine.

design can basically be classified as a sequentialdesign consisting of two stages :

Firstly, the AVR is designed to meet the required voltage regulation performance.

Then, the PSS is designed to meet the required damping performance. Basicafly,
faster excitation with higher voltage ceilings provide increased synchronising
torques with improved transient performance, but at the expense of injecting
into
The
damping
PSS
the
system.
supplies a supplementary phasenegative
(power)
is
damping
to
to
signal
considered
additional
advanced speed
which
provide
the lightly damped machine. Stabilizers using shaft speed as an input signal were

disturbances
from
load
designed
direct
they
and applied, since
avoid
successfully
fluctuations. However, they are sensitive to local mode oscillations, which
A
design
their
their
and restrict
effectiveness. combination of power and
complicate
introduced
PSS.
Many
they
the
speed was
and
are now replacing
speed
more recent
(eg
[12]),
have
been
PSS
the
tuning
the
studies
published on various methods of
filter
[13],
in
the
additional use of a notch
exciter path
and the advantages or

[14],
[15].
[14],
Kundur
transient
otherwise of supplementaryexciter
gain reduction
Gain
(TGR)
in
Ontario
in
Transient
Reduction
is
PSS
Hydro,
the
analyze
which no
in
[15],
General
Co.
Larsen
Electric
the
the
exciter; while
of
advocatesthe use
used
Grondin
[16]
Hydro-Quebec,
TGR.
More
of
recently,
et.el.
of
replaces the
3-term
PSS
compensator.
conventional
with a
The above works are representative but by no means exhaustive. These

interest
in
indicate
for
PSS
that
turbogeneratorcontrol shows
the
the
use of
works

little sign of diminishing. In 1995, severalauthorssuggestnew approachesto design

implement;
PSS
to
suitable
which are robust and easy
and two of the most recent
methods are:(a) Robust PSSdesign [ 17,18]
The objective of designing a robust power system stabilizer (RPSS) is to make

the overall system stable both within the normal operating conditions and within
fault,
situation
as
some extreme
such
a
and at the same time maintain a certain
degree of system performance. In designing an RPSS, each individual parameter
K,

K6 of the synchronous machine, (see Chapter 3), is replaced by its

..........

correspondingparametericuncertainty.
Within the power system operating range of interest, the K's parameters are

boundedwith lower and upper values.

The control design model of a robust PSSis developedto fit into the standard
configuration of Hoo analysis and synthesisas shown in Fig. 1.2. In Fig. 1.2, P is the
interconnected system which consist of nominal plant Go, uncertainties weighting
function Ru, and perfonnance weighting function Wp. In 1.2, Wd represents the

disturbance
functions
combination of
of the mechanical torque and the
weighting
field voltage. A denotesthe uncertaintieswhich are unknown but bounded; K is the
BPSS controfler neededto be specified.

7
-4

Fig. 1.2 GeneralInterconnectionstructure

(b) Fuzzy logic power system stabiliser (FPSS) :[ 19,20]

A new type of PSSbasedon fuzzy set theory is proposedin reference [20] to

improve the dynamic performance of a power system by allowing the PSS to have

damping
good
characteristicover a wide rangeof operatingconditions.
The first step in designing a fuzzy controller is to decide which state variables
input
dynamic
be
taken
the
representativeof system
performancemust
as
signals to
the controller. In power systems, based on previous experience, generator speed

deviation (A(o) and acceleration(Aw) are chosento be the input signals of the fuzzy
PSS. The input to the excitation system would be the control variable which is
fuzzy
After
input
PSS.
the
the
the
of
choosing
actually
output
proper variables as
fuzzy
it
is
decide
linguistic
the
to
the
controller,
required
and output of
on
variables.
These variables transform the numerical values of the input of the fuzzy controller,

to fuzzy quantities. For power systems, seven linguistic variables for each of the

10

input and output variables are used to describethem. These are, LP (large positive),
M[P (medium positive), SP (small positive), VS (very small), SN (small negative),
MN (medium negative), and LN (large negative). In order to find the minimum of
for
different
initial
loop
is
inputs,
simulation
conditions
an open
stabiliser
(Aw)
find
The
to
the
minimum and maximum of
and
results are used
performed.
(Ae)). After specifying the fuzzy sets, it is required to determine the membership

functions for these sets, (e.g. bell-shapedfunction, triangle-shapedfunction etc.).

...
The degree of membership can be defined as functions when the control variable

is
amplitude continuous.
PSS output

A(o
A(o

LN

MN

vs

SP

MP

LP

LP

vs

SP

MEP

LP

LP

LP

LP

MP

SN

vs

SP

MEP

MEP

LP

LP

SP

MN

SN

vs

SP

SP

MEP

LP

vs

MN

MN

SN

vs

SP

MIP

MIP

SN

LN

MN

SN

SN

vs

SP

NIT

MN

LN

LN

MN

MN

SN

vs

SP

LN

LN

LN

LN

LN

MN

SN

vs

Table 1.1 Decision table for PSSoutput

11

A set of rules which defme the relation between the input and output of fuzzy
in
designing
PSS.
found
knowledge
be
the
the
areaof
available
using
controller can
These rules are defined using the linguistic variables. The two inputs, speed and
These
49
for
in
rules are shown
each synchronous machine.
rules
acceleration, result

in Table 1.1 above where all the symbols are defined in the basic fuzzy logic
I
has
following
Rule
A
the
typical
terminology.
rule
structure :
" If speed deviation is LP (large positive) AND acceleration is LN (large negative)
then VpSS(output offuzzy PSS) is VS (very small)".
Now it is required to find the fuzzy region for the output for each fuzzy rule

like
Minimum-Maximum,
Maximum-Product
Fuzzy
methods
or
etc.
rules
using
...
AND
AND
is
The
are connected using
operators.
operator
used to obtain the

between
input
functions.
between
Later,
the
this
minimum
membership
minimum
function
is
found.
Finally, the output membership
the
result and
output membership
function of a rule is calculated.This procedureis carried out for all the rules and for
fmd
is
To
function
the output
every rule an output membership
obtained.
membership function due to all of these rules, the maximum among all of these rules

is calculated.
Since a nonfuzzy signal is needed for the excitation system, by knowing the
function
fuzzy
its
determined.
be
the
of
controller
numerical value can
membership
There are different techniques for defuzzification of fuzzy quantities such as
Maximum Method, Height Method, and Centroid Method.

12

Researchof this thesis

The function of excitation/govemor control with PSS appears to be well
[101,
[211,
[22],
from
following
the
a power systems analysis perspective
understood
Concordia
].
[
by
DeMello
11
and
celebrated paper

From a control engineeringpoint of view, as opposedto a power systemspoint

is
function
PSS
less
indeed
the
the
of
clear and
raises a number of
of view,
fundamental issues; not the least of which is if the supposed role of the PSS is to
is
it
damping,
(damping)
that
phase compensation
provide system
why
cannot be
indirectly
directly
by
PSS.
to
the
than
the
excitation system rather
applied
way of

While a number of control-oriented studies have appeared[23-27], they

design
to
themselves
control
of various types and provide supporting
confine
improved
evidence of

in
damping
terms of the conventional
system
often

One
damping
described
torque
synchronising and
analysis
above.
notable
exception is the very interesting work by Hamdan and Hughes [281 which uses the

[29]
frequency-response
MacFarlane
Rosenbrock
and
multivariable
methods of
[301 and pole-zero analysis to demonstratewhy the PSS is effective. There will be
for
it
to
to
this
occasion
refer
essentially complements the control
work again
herein
is
different
from
that
explanation reported
which, as we shall see, very
darnping
by
In
torque
synchronising
and
conventional
analysis. short, this
offered
PSS,
in
[31],
[32],
is
ingenious
demonstrate
the
that
originally conceived
an
work

13

device
which obviates an extremely awkward excitation control
control
desired
to
control specification.
characteristic,at small price, so as meet an overall
The key to uncovering the function of the PSS and what control specification it
is designed to meet is to recognise from the outset that a single turbogenerator
A
infinite
bus
is
2-input,
2-output
to
multivariable system. thorough
a
connected an

and readily understandablemultivariable analysisof the turbogeneratorsystem and

the function of the PSS is provided by the multivariable control framework known

[33],
[34],
[35],
[36].
Channel
Analysis
Design
Individual
and
as
The thesis is organised as follows. After introducing the Individual Channel
Analysis and Design (ICAD) in Chapter 2, Chapter 3 provides an in-depth
Further
turbogenerator.
the
multivariable analysis of
multivariable analysis of
turbogenerator regulation in Chapter 4 identifies the fundamental role played by
the PSS in facilitating single-loop control of the multivariable system to meet a

certain control specification and analysesthe stability robustnessof the combined

is
The
PSS
to
control scheme
system uncertainties.
control option
also assessed
Chapter
5
two
two
against
other
control
options.
control
methods
analyses
recent
"a

for stabilising a turbognerator connected to an infinite bus, Conclusions are

in
Chapter
6.
presented

14

2:

ter
2.1-

Channel

Review of Individual

Design
and

Analysis

Introduction
In this chapter a concise review of the Individual

Analysis
basic

(ICAD)
Design
and

framework

system,

can

which

multivariable

be extended

(multi-input
systems

In a typical

input
system

then

2-input
of a

to

the

multi-output

to derive

closed-loop

stability

can

be

easily

transient

rejection,

methodology;

achieved

multi-input

cases).

using

multi-output

actuator

certain

protection,

In SISO systems
the

which

: steady state response,

response,

and robustness.
by

case of

general

system output,

which

design goals need to be met. These include

disturbance

2-output

task, once it is specified

design

control

is intended

is presented. It contains the

in terms

framework
the
aspects of

Channel

Nyquist/Bode

classical

(MIMO)

these goals

systems,

on the

hampers
the
which

hand,
have
the problem
other

of cross-coupling

direct application

design
ICAD,
methods.
analysis and

new application
Leithead

of classical
oriented

[33], enables the application

design to multivariable
such application
output

framework

which

developed

by O'Reilly

of Nyquist/Bode

by
system
overcoming

problems

and

analysis and
preventing

-nature
from
the
arise
very
of m-input

determination
the
such
as:
systems,

15

influence
the
of

m-

of the

the application

on the controller,

plant structure

margins to multivariable

design
individual
the
systems, and
of each

it
depends
individual
though
on all other
element even

controller

in the controller

elements

matrix.

Analysis

2.2- Multivariable

ICAD
using

Consider the multivriable

feedback control

configuration

of

2.1.

Fig.

Controller
Matrix
W(T
-1

Reference
Vector
-

U(

Plant
Matrix
-G-s)
L
(

Input

G(s)
is
the matrix
where
plant,

is the controller

K(s)
input

vector

the customer
individual

input-output

attention

the

diagonal

on

controller;

transfer

input

pairing

2-input

function

function
output

is
paired
ui
is called

2-output

then Fig.

control problem

transfer

is
Y(s)
the plant
and

specification,

Output
Vector
y

Fig. 2.1 Standard multivariable

plant

of phase and gain

vector.

with

a channel.

multivariable

the

U(s)

is the

In accord

with

matrix,

2.1 can be redrawn

16

representing

output

yi; this

Focusing
problem
as in Fig.

our
with
2.2

diagonal
ki
the
are
where

diagonal
the
elements of
controller

matrix

K(s), and gij are the elements of the plant matrix G(s) for i, j=1,2.

r --------------It1

t2

Fig. 2.2

The 2-input

2-output

multivariable

control

problem

with

diagonal feedback

Consider
reference

r,

transmission
the other
observe
from

the

via

to

the forward

its

follows
921(S),

in
Fig.
also
second

signal

associated

output

2.2 that

reference

yl.

feedback

the forward
r2 to

17

yj

subsystem
cross-signal

is via

the

the first

forward

This

paths: one directly

two parallel
the bottom

from

transmission

signal

through
and

g1l(s);

912(S)-

We

transmission

bottom

feedback

subsystem
from

These
signal
and 912(S)

r2 to yj can, after

little
a

from

transmissions

block

algebra,

in
[33],
described
in
be
represented as
are

r, to

the details

yj and

of which

Fig. 2.3.

Ri

Ref. 2

Fig. 2.3 Individual

In other

transmission
output
signal

912922-1h2

topological

words,

Channel

1 together

at

structure

Fig.

2.3.

C, between

with

r2

its
associated
and

design
structure
channel

the individual

reference

the additive
output

depicts

Channel
C,
:

1 and its

cross-reference

1. Likewise,

of Fig. 2.2., Channel

by

signal

associated

'disturbance'

symmetry

of

C2 between reference

the
2

2,
together with a unity negative feedback
output

18

control

'disturbance'

loop and a cross-reference

is
signal 921911-1h1r,

2.4.
Fig.
by
represented

Ref. 1

Fig. 2.4 Individual

design
structure
channel

[331 that the two individual

Thus, we have established

feedback

control

structurally
control

problems

equivalent

problem

921(s) are retained

Fig.
of

to

Fig.

of
the

2.2;

original
all

signal

and no structural

From Fig. 2.3, Channel

Channel
C2
:

2.3

and

2-input

2.4,

C, has the forward

together

are

2-output

interactions

information

channel

via

feedback
912(s) and

is lost.

path single-input

single-

function
(SISO)
transfer
output
klgll(l-

(2.1)

1412)

19

where
912921

,Y(S) =

(2.2)

911922

h2
and
_

k2922

(2.3)

1+ k2.922

Similarly,

from

Fig.

C2 has the forward

2.4, Channel

path

SISO transfer function

C2

k2922(l

(2.4)

-, yh)

where
klgll
1+klgll

(2.5)

In particular,
channels

the multivariable

C, and

multivariable
magnitude,

signal

case, the two

nature

[331 by the

C2 is described

structure

function

interaction

channels

behave

y(s).

into

two equivalent

open-loop

2-output

y(s)

almost

like

two

interaction
multivariable

is

is low

small

(in which

independent

between

in

loops);

the loops

is

is
decomposed
system

set of SISO system. Each SISO system is the

channel transmittance

the feedback

When

SISO

frequency

complex

between the loops

is
large
in
y(s)
magnitude,
when
high. Thus, the 2-input

of the turbogenerator

between input i and output

loop between output

20

i and input

i, with

i open but the other

loop

feedback

closed.

is not

systems

multivariable

channel transmittances
the plant

such
ICAD

new;

individual
the
which

SISO

ensures

structure

channel transmittances
of the original

structure

multivariable

by-way

systems

of

loss
information.
no
of

function

with

Channel

C, to reference

that

this

structure

function

are

y(s) to
The

related.

is maintained

The closed-loop

in the
structure

multivariable

is
described
ri and rj

SISO

the role of

are simply

plant

of

these

aspects of the plant

in the scalar multivariable

encapsulated

decomposition

to make explicit

are reformulated

The significant

structure.

equivalent

Although

response

of

by

(2.6)

(s)
(s)r2(s)
(s)ri
(s)
di
tij
+
yj
=
where
ci
tii (S)=1+Ci

(2.7)

di(s) =ij

(2.8)

+Ci

gj

Assuming
rejection

performance

characterised

entirely

* for stable reference

provided

y: #O, the transient

of

the

output

by the transmittance

dynamic
response

21

(S)
Yi

can

be

Ci(s); namely,

inputs and disturbance

ci
is
stable.
1+Ci

and disturbance

inputs,

is
stable
yi(s)

* for

input

reference

ri(s), the tracking

performance

of

is

yi(s)

Ci

by

indicated

for

1+ Ci

input

disturbance

is indicated

performance

Hence given a diagonal

manner

with

frequencies
crossover
usual significance
Assuming
transmittances,

disturbance

rejection

on

yi(s)

1
1+Ci

by

controller,

dynamic
the
to
assess
used
the usual

di(s),

the transmittance

performance

such

domain

Ci(s)
bandwidths
of
and

in

by the controller

achieved

frequency

Ci(s), can be

as the

parameters

of tii(s) retaining

their

[34].

no pole/zero

cancellations

the pole-zero

structure

occur

within

the

of the channels

channel

is specified

by Table 2.1 below

Zeros

Channel

C,

----------------------

Channel

C2

Zeros

Poles

(1-'Yh2)
of

-----------------------

Zeros

Poles

of

911,912,921,

h2

-----------------------------------

(1-,
of
yhl)

Table 2.1 Open-loop

22

Poles

of

922,912,921,

channel structure

h,

SISO systems, the possible

As in classical
dynamical

performance

is
adversely
case

output
From

individual
the
of

Table

channel

2.1,

the number

Ci is the number

in
the 2-input
channels

by the presence

affected

half

of right

closed-loop

RHPZs.

of channel

plane

(RHPZs)

zeros

2-

(1-,
is
determined
RHPZs
of
of
yhj) and

of
from

Result 2.1 as follows

Result

2.1

Suppose

that

the Nyquist

plot

function

the (1,0)

yhj encircles

direction

than in an anti-clockwise

the

of

point

multivariable

N times

direction.

more

structure

in a clockwise

Then Z, the number

of

RHPZs of (1-, yhj) is given by

N=Z-P

P
RHPPs
is
the
where
Of yhj
number of
Channel

Whether

the structure
control

RHPZ's

of yhj

kj(s)
gain

of

Ci is minimum

phase depends on yhj. When

and y are the same and the magnitude

is large,

the RHPZ's

(1of
yhj) are essentially

(1- y) which are the 2-input

From
a result
zeros.
the encirclement

similar

of the point

2-output

23

by the Nyquist

the

plant RHP system

to Result 2.1, the latter

(1,0)

of the

plot

correspond
of y(s).

to
The

from each other and are different

different
channel zeros are

from

the plant system zeros.

In a very direct

its Nyquist

particularly

plant associated
indicates

manner the multivariable

plot,

indicates

with

nature.

of the cross coupling.

the strength

(1,0)
it
indicates
to
the
point
approach
is

ill- conditioned.

The

function

transfer

of

presence

Similarly,

its Nyquist

plot,

are those

RHPP's

number

multivariable
actual

which

indicated
are

(1,0). Whether
the

indicates

it

By its closeness

of

structure

and

RHPZ's

the

Channel
or not
of

function

closed-loop

of

functions

performance

The
yhj.

the

achieved

The

elements and

presence

these

of

of the point

phase, is indicated

point

(1,0).

important
yhj are

24

of

number

yhj, particularly

of encirclements

Ci is minimum

encirclements

structure

of

by the number

The
of y.

ChannelCi.
aspects of

structural

are RHPP's

the

of

plot of y(s).

RHPP's and RHPZ's of the transfer function

significant
hj(s)

the multivariable

By its magnitude

by

(1,0) by the Nyquist

of the point

encirclements

indicated

are

aspects of the

RHPP's
are

are those which

RHPP's

y,

the extent to which the plant

RHPP's

significant

elements

these

structure

those structural

its multivariable

function

Thus,
indicators

by the controller.

by
the
of

Hence

diagonal
given a
assess the

structural

aspects

2-input
of

2-output

Two

the channel.

of

the role of single-loop

concerning
by ICAD

Table 2.1 and Result 2.1 can be used to

controller,

main

in multivariable

subsystem

results
control

[
33]:
systems are

Result 2.2
have
Nyquist-Bode
systems

Multivariable

SISO signal

with each individual

associated

gain and phase margins

transmission

channel

Result 2.3
The gain

Ci = kigii(l-,
measures

and phase

ij = 1,2, (i # j)

yhj

provided

function
structure

the scalar

2-input

and

diagram

with

(2.4)
of

of the structure
gain

depends on the control

to establish
the closed-loop

Channel

2-output

the polar

on the control

is
required
result
which stabilises

(2.1)

of

the influence

Ci transmittance

structure

that

associated

are
the

robustness

multivariable

(1,0)
to
the
-ijdoes not come close
point.

Although,

itself

margins

Channel
each

ki(s)

of the scalar

is readily

kj(s),
gain

channel

assessed,

i: * j. An existence

that there exists a diagonal

controller

system and to relate the structure

Ci transmittances

plant.

25

this

to the structure

of

of the original

[34] gives a set of results on the existence of diagonal

ICAD,

structure

Ci
to the structure
of

2.3- Non-diagonal
Plants

amend this

By

poor

way

control

specification

structure

attempts

ICAD

shows

or elsewhere

-compensation

and robustness.
structure

In general,

(meeting

by feedforward

structural

performance

plants,

the

however,

use

by
of

is that

between the gain and phase margins

it is

preferable

to

either excessive phase sensitivity

or

whenever

sensitivity

in

to

restriction

is not alleviated

restriction

multivariable

Some plants may exhibits

[341 restriction

is not a significant

fundamental

with

pre(post)-

output to the plant

that

lack
that
of robustness
and

the

are made to

namely

approaches;

of the controller

its presence weakens the link

excessive

readily

Normally

by two

analysis,

control.

plant

from

identified

in the face of plant uncertainties)

non-diagonal

improve

are

functions.

in ICAD

2-output

2-input

:-

structure

of

the

system and relate

of the original

feedforward
and

on performance;

pre(post)

poor

structure

compensation

diagonal

Control

with

multivariable

output.

the closed-loop

stabilises

which

controllers

possible,

control.

both.
These sensitivities
or

26

do arise

Nyquist
the
when

plots

(1,0)

of feedback

and so cause the stability

it

Sometimes,

different

cost weighting

matrices,

interest
range of

the frequency

function

the multivariable

to be more

control

be anticipated

might

to use non-

far
from
the point
plots

to shift the Nyquist

compensation

robust.

yhi(s), go too

is
it
is
this
the
tempting
case
when

(1,0)
the
point;
near
diagonal

functions,

of the multivariable

design
a

that

the system in

diagonalises
which

displaces
and so

the Nyquist

y(s) from the vicinity

with

plot

of

(1,0),

of the point

lack
the
of
might remove
due
to excessive sensitivities.
robustness
Unfortunately,

[361,

O'Reilly

the

this is not always

the case. From Leithead

on

feedback

utility

following
by
the
summarised
(I) it cannot improve

significant
(ii)

it

attempting

bandwidth)
individual)

(i. e. with

to achieve high-

when attempting

channel

bandwidth

than all

greater

dynamics).
system
no

achieves

assignment

control

is

performance

control

performance

non-diagonal

and

of
to

in

more

inputs
achieve

the

transfer

to

than

that

outputs

function

of

diagonal

with

low-performance

presence

achieved

control

structural

elements,

27

by

swapping
control

(i. e. low

difficulties

gij(s) are minimum

the
when

channel

when

the

phase; but

it

may

make

individual
(iii)

function
no

achieves

assignment
attempting
(iv)

possible

transfer

it

of

(vi)

to

than

when

gij(s) are non-minimum

that

outputs

performance

by

achieved

with

the

phase.

swapping

diagonal

lack of robustness

alleviate

or excessive

(v) it can improve

plant

uncertain

elements,

more

inputs

in

control

the

when

to improve robustness.

it cannot

sensitivity

improvements

structural

the ability

family
or a

it can imProve

due to excessive

phase

sensitivity.

fixed
of a

controller

of plants with uncertain

the ability

of a controller

to cater for an
structure.

to achieve

system

framework,

is used

integrity.

Finally,
to

ICAD as an applic ation- oriented

investigate,

techniques
helicopter

submergence

to

and

analyse
various

control

recommend

system

[38], and gas turbine

number

such

as single

submarine

operating

applications,
[37],

[331.

28

design

of
main
at

rotor

shallow

Chapter

3: Modelling

and Multivariable

Turbogenerator
3.1-

system

Introduction
Turbogenerator

operating

broadly
may

system stability

equilibrium

under normal

operating

as

conditions

Since power systems rely on synchronous

disturbance.
generation

electrical

of

power,

a necessary

is that all synchronous

system operation

and to

being
after
subjected to a

regain an acceptable state of equilibrium

satisfactory

be defined

it
in
the
that
to
of
system
a state of
enables
remain

the property

for

Analysis of the

machines

condition

for

machines

in a

system remain

in synchronism.

A synchronous

field
has
the
two
essential elements:
machine

power

and the armature.

armature
current.

Normally,

is on the stator.

The field

is on the rotor

winding

is excited

and the
by direct

When the rotor is driven by a prime mover (turbine),

field

field
the
of

rotating

magnetic

voltages

in the three-phase

frequency

the field

armature

induced
the
alternating
of

winding

windings

alternating

of the stator.

The

voltages and of the resulting

currents

that flow in the stator windings

depends

on the speed of the rotor.

29

induces

the

load
is
when a
connected

The frequency

of the stator

electrical

is thus synchronised

quantities

hence
designation
the
speed:
This chapter
each component

or group

the given

components

objectives

of a given

describe

The

modelling

always

will

of a power

can be

or models,

representations,

best describe

be

system

of the system performance

possible

study;

How
:can

questions

of components

of

the system to meet the

expressions

and what mathematical

it
that
of each system element so

the characteristics

be
quantified
can

machine".

following
the
addresses

of the several

made; which

will

"synchronous

so that an analysis

represented

with the rotor mechanical

for the analysis.

and

analysis

been a challenge.

of

the

The problem

synchronous

has

machine
intensely
on

was worked

in

the 1920s and 1930s [39] and has been the subject of several more
recent

investigations
synchronous

possible

[40-42].
machine

There

are

configurations

in [22], it is neither adequate nor practical

model

for

all

study conducted
the USA,
modelling"

power

number

published

"Effects
named

to devise a "universal"

problems.

Nevertheless,

of synchronous

in 1971, concludes

in stability

that "in

Council

in

machine

general,

studies to use accurate machine

30

of

But as mentioned
.

by the North East Power Co-ordinated

in a report

more important

system

dynamic

infinite

it is
data

than to use a more elaborate machine model"

support this. One finding
models

developed

models,

provide

with

by Anderson & Fouad [21] state that "the

(F, D, and Q) windings,

with

improvement
no

practically

F&Q

only

[21]. Other findings

Another,

windings".

over a simpler

the synchronous,

conclusion

is that,

the third

for
dynamic stability
adequate
the 8th order

model

order

simulation

in matching

model which

and subtransient

in both d, and q axis; and a third

the generator
The

transient,

linear

circuits

order linear

model

model

by
made

is a comparison

E. L. Busby et. el, [431 between an eight-order

includes

i. e. high order

of

model.

is in most

cases

is
and
nearly as close as

the test result.

Also

Dandeno

[44]
"when
data,
the
el.
remark
et.
using
standard machine
a more
complex

model

more accurate

results

analysis

operating

machine

in the stability

further
as we shall see,

offers,

3.2-

of a synchronous

of the y-function

insight

did not always

studies performed".

model

orders

ICAD

through

and comparison,

for different

give

the

at the same

conditions.

Park's

Equations for a synchronous

machine --

Fig. 3.1 shows the schematic of the cross section of a threephase synchronous

field
poles.
machine with one pair of

31

Axis

of phase c
Fig 3.1

The machine

winding
current
voltages

through

theory

consists

a magnetic

and produces
in

the

armature

The field
field

The

winding.

winding

of

to this

view,

process,

problem

the power
operating

design

32

alternating
of

at minimum

large
is
cost

be
to
realized
gains

are enormous.

From

system

is a very

in

constantly

the field

direct
carries

induces

which

and the economic

problem,

the solution

multivariable

elements:

essential

system to ensure stable operation

complex

point

machine

of two

and the armature winding.

interconnected
a very

Synchronous

a control

high-order
changing

the system,

it is essential

analyze specific

The linearised

model

with

a good grasp of the characteristics

its
individual
those
as of
of

equations

of

elements.

for the dynamic

behavior
proposed

of the
by Park

have appeared in the last 50 years which use

varying

assumptions

One or more of the following

models.

of

and to

assumptions

based
are
upon the equations

machine

[8]. Many publications

Park's

and complexity

using the right degree of detail of system

problems

system as well

synchronous

to make simplifying

this requires

representation.
the overall

dimensionality
high
the
of

Because

environment.

to derive

turbogenerator
have been used

assumptions

[451 :
behind
transient
voltage
(2) derivatives

of axis currents

time are negligible

stator

flux
and

in comparison

resistance

sinusoidally

is
reactance
constant.

distributed

linkage

with rotational

is negligible

with

respect to

terms.

and the stator

windings

are

round the air-gap.

(4) effect of change in speed in the voltage equations is neglected.

(5) magnetic

hysteresis

is negligible.

magnetic

saturation

effects are negligible.

On the other hand, the mechanical

are very

well

established

[46],

equations of the machine

and three basic assumptions

33

are

in
deriving
made

these equations

not vary

from

rotational
(3) shaft
generation

greatly

(1)
:
machine rotor

synchronous

(1
speed
pu).

power losses due to windage

is constant

power
action.

except

Assumption

friction
and

as a result

1 allows

power of the machine

(P,, ) as supplied
The machine

is the difference

of speed power
power

circuit

Rotation

q-axis

to be

2, the acceleration

between the shaft power

by the prime mover and the electrical

electrical

(2) machine
are ignored.

per unit

equated with per unit torque. From Assumption

does
speed

power

(Pe)-

is shown in Fig. 3.2.

d-axis

Wr
rb

...........

..............................

axi.

................
6 Ph4Se CL

40
c

Fig. 3.2 Machine electric

The stator circuit

carrying

alternating

damper windings.

currents.
The field

consist

circuit

3-phase
armature
of

The rotor circuits

winding

34

windings

field
comprise

is connected

and

to a source of

direct current.

There are three armature phase windings

on the stator

of the machine,

equivalent

armature

phase windings:

kd on the d-axis

and

winding

which

by two

d-axis,
the
on
on the rotor,

are permanently

winding

fd

in pu may be written

as

field
is
there
also a
and

is dc-excited.

which

The linearised

follows

ad

kq on the q-axis,

are available,

d-axis,
the
on

have been replaced

Some
damper
the
on
q-axis.
windings

and aq winding

short-circuited

which

b,
and c
a,

Park's

equations

[471: -

Avd = -AVq

AVq --A'4fd

(3.2)

ATd ""Xad''Ifd + Xad'lkd - XdAId

(3.3)

ATq "'XaqAlkq - XqAIq

(3.4)

AI
AI
AVkd "X
+X
''I
-X
d
kkd
kd
fd
ad
ad

(3.5)

AI
AWkq "' X
kq -X
kkq

aq

AI

(3.6)

A*kd + O)R
kd

A,

0
kd '-"

(3.7)

A*kq

Al

0
kq --

(3.8)

+ O)R kq

A*fd
AE fd "2
+
o)Rfd
'
AlVfd

X
---2 ffd

AI

A,

(3.9)

fd

(3.10)

A,
AI
+X
-X
d
kd
fd
ad
ad

35

Where V denotes voltage in volts, Ia current in amperes,

a resistance
electrical
3.3-

in ohms, iV a flux linkage in webers, and (o a speed in

radians per second.

Synchronous

for

machine

model

shown

in Fig.

low-frequency

oscillation

studies:
The circuit
simplification
studying

of a multigenerator

the

stability

performance

Thus the external

system.

as a

from
the viewpoint
system

of

in

the

represent

the

of only

reactance

system as seen from the terminals

3.3 may be considered

one machine

infinite-bus
and

of the machine studied.

Gov/Turb.
Torque
S ynch.
Machinel

I R+jX

Exciter
Voltage

Fig. 3.3

Single generator connected to an infinite

The machine contains

through

the synchronous

the terminal
through

voltage;

the machine,

bus

two channels, one from the exciter

it
is
machine, and
usually associated with

the other

is from

the prime

mover,

again

So,
it
is
linked
there
to
the
and
shaft speed.

36

are two

system

(w);
speed

while

the exciter

(Eref)

(T

the

ref

).

All

synchronous

and the reference

machine

in

and

position

input

ICAD

to

of the prime mover

take

place

within

the

the

problem

of

context,

be addressed considering

system should

dynamics

the structural

and the shaft

are the reference

interactions

multivariable

(Vt)

voltage

the two system inputs

for the whole

stability
only

the terminal

outputs,

machine

and the

of the synchronous

machine

of the synchronous

customer specifications.
The dynamical
its

in both
Park's

equations

describe

which

through

transferred.

(w)
and the electric
speed

in a damper winding

induced

damper

winding

modelling.

As for

oscillating

frequency

can

the d-axis

affect
simply

described

completely

and q-axis

is extremely

lov-frequency

the

oscillations,

(kd, kq) is negligibly

be

ignored
windings,

equations.

37

and hence

the

dynamics
(Tel)

torque

the

current

small;

hence

in

system

their

high, their eignmodes

oscillations,

by algebraic

between

its
and
mechanical

low-frequency

During

interaction

the

of the generator

the turbogenerator

from

aspects are obtained

and mechanical

properties

electrical

the

electric

characteristics

they

natural

will

not

be
can

What is left is the field

(fd) circuit

windings

of the machine

which

must be described,

because
its
low
only
of
eignmodes frequency,
is directly

connected

differential

equation

in the model.
linear

This

configuration
order)

and Philips

equations

involving

Park's

subsystems

[49]

representing

derived

none of the assumptions

models

more simplified

using

one

space

formulation

synthesis
general

more

of

have emphasized

authors

[ K,

in

......

machine,

in Section

later
were

the

machine

turbogenerator

purpose
listed

developed

equations,

turbogenerators

the performance

38

in which

but
were used,

methods

control

an

Shackshaft

by different

because of the wide spread use of digital

equations

and exciter

authors

[501, [511, [52].

assumptions,

on

and when

model,

(3.2)

the need of Matrix

describing

based

K6]

is
system
obtained.

turbogenerator

a general

or

linearised
a set of

the prime mover (turbine)

to the synchronous

are added

interacting

overall

parameters

machine,

[221.
studies

oscillations

have presented

special

low-order

the synchronous

to describe the synchronous

equations

the equations

[48],

the basic

constitutes

representing

the torque

be
included
must

machine

be
for
low-frequency
the
and could
used
Hefferon

but also because it

system. Finally,

of the synchronous

(third

model

to the exciter

not

These

and state-

analysis

and

computer.

The

of a synchronous

machine
based

no damper windings

with
H-P

on

Appendix

model,

are

(3), the linear

and with negligible

in

given

equations

(3).

Appendix

the

representing

resistance
From

the system are

(MS2+
Dsb = ATm - K2ATfd - KlA5

AVfd

K3

K4

(3.12)

AEfd A6
- 1+ K3Td'os
1+ K3Td'os

(3.13)

AVt = K6AVfd + K5A6

where

s=j(o

torque input,

is the

complex

mechanical

AEfd is the field voltage input, Aw is the rotor speed

output,

A6 is the rotor

output,

M is the inertia

general term allowing,

prime-mover

ATm is the

frequency,

damping

AVt
is
the terminal
angle and
The coefficient

constant.

for
damper
windings,
not only
and variations

in load torque

voltage

D is a rather

but also for

with

changes

in speed. In general, D is not a constant and varies with speed and

power

flow.

However,

for any particular

speed and power when

be
D
to
a
mean
arise,
can
assumed
about
value
changes
only small
remain

constant.

schematically

The block

diagram

these equations.

39

Fig.
of

3.4 below

shows

ATm

AEf(

Fig. 3.4

Third

linearised
order

The constants
torque

expressions;

circuit

equations;

terminal
parameters
conditions

voltage.
Kj,

model of the synchronous

K1 and K2 are derived

K3 and K4 are derived

K5 and

K2,

K4,

of the machine.

K5,

K6

This model

40

the electric

from the field

K6 are derived

K3 - which

Except

from

machine

from

the generator

is an impedance
depend
constitutes

on

winding

the

factor-

the

operating

the basic model

for synchronous

it
is
From
third
of
order.
equations
machines, and

(3.10 - 3.13), the transfer-matrix

form, is shown in Fig. 3.5.

Aco

A
G(s)
AEfd

Avt

30

Fig. 3.5

Transfer

function

matrix model of synchronous

machine

where
G(s)

-gll(s)

912(S)-

(3.14)

922(S)_
-921(S)

and
(3.15)

WAT
(S)AEfd
+
912
911
m,

(3.16)

AVt " 922 (S)AEfd + 921WAT

m

where the numerator

function

between Ae) and AT,

: transfer

function

between Aw and

AEfd-

: transfer

function

between AVt and AT.

-

(s) : transfer
912(S)

921(S)

922(S)

polynomials

: transfer

function

between

AVt and

AEfd

by
:are given
(3.17)

g, 1(s) = s(l + sK3Tdo)

(3.18)

912(s) = -K2K3s

41

0.

921(s) = K5(1+sK37vdo)(Oo -K3K4K6(Oo

2+
Ds) + K3 (KlK6 - K2K5)0)o
922(s) = K3K6(Ms

and the open-loop

system characteristic

(3.20)

is given by

polynomial

2+
A(S) = (Ms
Ds)(1 + sK3Tao) + K, (I + sK3Tao)(oo - K2K3K4(Oo

3.4-

Damper

windings

This

developed
DeMello
basis

function

transfer

by

representation

Hefferon

for

the

.-

block

diagram

and Philips

Concordia
and

[11],

explanation

of

and popularised

has, for many years, provided

the

fundamental

into basic dynamic behaviour,

the

damper

of

power

assumption
stability

parameters

requirements.

windings;

only

(quaderature

be
acceptable
may not
analysis.

Therefore,

are required

Quadrature-axis
for that purpose.

generator

the direct-axis

currents

to more accurately

reactances and open circuit

parameters

produce currents

predominate).

for disturbances

additional

This

considered

synchronous

in

machine

behaviour.
the
model

defined
constants are

The 3rd order model may be improved

42

giving

does not account for

basis
the
that short circuits
on
factor

dynamic

The model, while

insight
a good

were considered

3.4, first

Fig.

by

and control

low

model

[481

characteristics

generator

(3.21)

one step

by taking

reactance 9d9
X/
value

Xd.

Of

(transient,

involve

effect,

initial
from
its
time
with

The variation

the effect

adding

Hughes

Another

dampers
of
which

described by Tdo
improvement

This,
state.

extends the 3-rd order model of the synchronous

function

To represent those additions,

block

[531 which

will

during

predominate

that is, the subtransient

conditions,

to a 5-th order model.

transfer

constant).

transient

value to a steady state

be
an exponential
will

time

open circuit

fast changing

direct-axis
the
of

of the variation

account

in

machine

an extended

diagram model is presented by Saidy and

takes into account

the damper windings

on both

the d- and q- axes. The block diagram is shown in Fig. 3.6 below

Fig. 3.6

Fifth-order

linear model of synchronous

machine with

damper windings
From

which

the numerator

the
synchronous
of
model
i

polynomials

machine are given by

43

5-th
the
of

order

gli(s) =s

(3.22)

912(s) " -sk2gl(s)

(3.23)
(3.24)

k692(s)
k6d93(s))
+
(Oo(k5 )]
MS2
k2k5)
k6k2d
+ (oo(klk6 +
9'2'2(s)"`91(s)[k6
(Oog3(s)(k2k6d
-

(3.25)

loop
is
by
open
system
charateristic
and
polynomial given
Y(S) =

2+
Ms

(kl
(S)
(s))
k292
k2d93
o)0
-

(3.26)

where
91

a(l + TIS*3
--,
(1
+ aTdosXl+ k3TLS)

92 = -k4d

(3.27)

(i + CT.S)

(3.28)

(1 + aTd'OsXl + k3T dos)

(3.29)

k4q
93 =
(1 bTos)
+
"en

k6d
k2d
0,
for
damper
and
windings, and so
=0
no
93 "'

k3
and gl(s) =-(I
+ k3Tdos)

(3.30)

k4

(3.31)

(S)
--(I
2
+ k3rdos)

we end up with
model

[481

the third-order

model

Hefferon
of

(Tao and TO are the sub-transient

.

constants)

44

and Philips

open circuit

time

3.5-

Exciter

and voltage regulator

The original
to provide
winding;

function

of the exciter

an adequate excitation
in addition,

the power system by controlling

field

The control

current.

and reactive

power

machine,

performs

the field

functions

of system

system,

block

diagram

form

for

and other

in
transient
use

stability

and small-signal

stability.

equipment

variety

of

These are intended

for

studies.

wide

The one used in

configuration

Vr5,,,

(i + STC + STCJ)
(1 +

Fig. 3.7 IEEE

Xi
XI

STB

+ STBO

type STlA

are

following
has
the
model which

this study is the IEEE type STlA

of the

12 model structures

the

representing
in use [55].

systems currently

excitation

of

of voltage

limits

ensure that the capability

excitation

and

performance

include the control

[54].
has
The
IEEE
not exceeded.
standardised
in

control

voltage and thereby the

and the enhancement

functions

The protective
synchronous

flow

system

to the satisfactory

essential

is

and voltage regulator

to the synchronous machine field

the excitation

functions

protective

models

Excitation

45

Ke

Efd

(1+ STO

system model

in
which
a small-signal

be
study can
represented simply by

Vref

Efd

Ke
(1+ sTe)

Fig. 3.8 IEEE

3.6-

type ST1A Excitation

Governor

and turbine models

delivered

Power
the mechanical

by a generator
"prime
of a

power

of

electrical

convert

sources

thermal

derived
energy
energy

by synchronous

(kinetic

energy

from fossil

that is, in turn,

by controlling

such as steam, water

mover"

flow.
steam or water

valves regulating

mechanical

is controlled

is
diesel
This
or
engine.
usually

gas turbines
closing

(simplified)
system model

done by opening or
The prime movers
energy

of

to electrical

to

machines.
least
at

in the system must respond to this change in order

power

maintain

balance.

its

The

way

imbalance

is

frequency.

If there is excess generation,

through

effect

frequency
the
tend to speed up and
there

is

into

energy

Since the load of a power system is always changing,

one generator

water,

fuels and nuclear fission)

converted

or

deficiency
a

of

on

will

generation,

46

we

this

power

speed

and/or

sense

generator

the generator

sets will

hand,
if
the
rise; on
other
the generator

speed

and

frequency

will

frequency

are used as control

drop. These deviations

automatically.

which

deviation
the
senses

action

[56].

signals to cause appropriate

be
can
reduced

as with

valve

device
a

appropriate

exciters,

to any desired

The Governor/Turbine

satisfactory.

it into

and converts

For speed governors,

which

speed and/or

This is done by a speed governor,

action

model

from nominal

valve

a composite
is the most

level

[571,
in
this
model used
study

is shown in Fig. 3.9 below

(1)

1+ ST9

)ref

Fig. 3.9

3.7-

Analysis

In block

1+

Governor/Turbine

of turbogenerator

diagram,

sTt

Linear Model

ICAD
system using

the turbo generator

Fig. 3.10 below

47

Tm

technique

is
represented
system

by

Vref
10

L---------------

------------L

Fig. 3.10 Turbogenerator

The synchronous
modelled
loop

machine

3rd
higher
as a
or

between

the rotor

((o)
speed
output,

Neglecting
may or may not exist.
moment, the governor/turbine
they

do

on

multivariable

the

that of the synchronous

simply
analysis

of the synchronous

answer two questions

to encounter

input
(PSS)

the presence of the PSS for the

dynamics,

loops;

don't

y(s) of equation

(2.2)

lying

affect

which

as
the

is

Under
the physical
machine model.

machine, we need to

(1) what types of system configuration

likely

and the exciter

Power System Stabiliser

feedback

function

could

be

and a cross-coupling

and exciter/AVR

principal

structure

before-as said

order model,

in the form of a so-called

(Eref),

system

dynamic instability

48

and loading conditions

are

?. (2) under what situations

the

can

synchronous

itself,

machine

without

control

effects,

produce negative damping ?

Embarking

turbogenerator
O'Reilly
of

control

now

on

a multivariable

analysis

of

the

depicted
in
Fig. 3.11 along the lines
system

Leithead
and

[331 and the review

Chapter
of

2; we

consider

first

signal transmission

from the governor

reference

(AUg) to its associated turbogenerator

(Aw)
speed output

the forward

in Fig. 3.11.

---------------

ATm
kl(s)

lu

Governor
Control i

Ao)

gil
+

921

912

Ref

AEfd

AVref

g22

k2(s) 'I

Avt

1+

___

--1

Excitation

Control

Fig. 3.11

+
I,

Synchronous

Block diagram of a 2-input

control

Generator

2-output

system analysis

49

turbogenerator

for

This

forward

directly

signal

through

feedback

g1l(s); the

(AVref)

is via

the

signal

transmissions

after

bottom

little
a

921(s), the

paths:

bottom

from
algebra,

[33], be represented

from the generator

to the turbogenerator

exciter

block

via

transmission

cross-signal

voltage reference

other

two parallel

one

exciter

and 912(s). We observe also in Fig. 3.11 that

subsystem

the forward

follows

transmission

feedback

subsystem

AUg to Aco and from

the details

of which

terminal

(AW
speed output
and

912(s). These

AVrd to

Aw can,

described
are

in
3.12.
Fig.
as

Channel
-----------Reference
+
Aug
+

C,
output
A-*

911(1-'Yh2)

Aw

dl

Governor
Control
-------------------

912 h2
922

Vref
j6,

Fig. 3.12 Individual

design
channel
structure
channel

50

: governor/speed

in

In other words,
Channel

transmission
associated

generator

reference

(AVt), together
loop

(A(o).

and

represented

governor

912922

signal

Likewise,

a unity

at the generator

of the topological

terminal

feedback

'disturbance'

terminal

voltage

excitation

signal

output

control

AU,
-'h,
g,, g,,

by, Fig. 3.13.

Channel
ReferenEe
AVref

the additive

C2 between the generator

negative

a cross-reference

with

-'h2AV,

signal

(AUg) and its

reference

by symmetry

(AVref) and its associated

with

the individual

(Aco) together

speed output

3.11,
Channel
Fig.
of

structure
voltage

C, between

'disturbance'

cross-reference
speed output

Fig. 3.12 depicts

------------------k2(s)
lExciter

C2
922(1-'11)

1+

y+d2 output

Control

AVt

L-------------------

921h,
911

Aug

Fig. 3.13 Individual

design
channel
structure
channel

51

: exciter/voltage

is

Thus, we have established

channel feedback control
together

structurally

problems of Fig. 3.12 and Fig. 3.13, are

to the original

equivalent

feedback

turbogenerator

[33] that the two individual

control

problem

via

information

is lost. From Fig. 3.12, Channel

path single-input

C, = klgll(l Similarly,

transfer
C2

(SISO) transfer

C, has the forward

function

(3.32)
Fig.

3.13,

Channel

C2 has the forward

SISO
path

function
(3.33)

- -Yhl)

(3.22) to (3.26),

From equations
function

y(s)

of

this

2-input

assumes the explicit

-K2[K,

(1 + sKJ)-

2-output

parameterised
K 3K4

when

represented

using

the extended

structure

parameterised

3rd-order

form given by

K2K.

model

(3.34)

)O ]

with

the damper

windings

(5-th order model) :

(k5
-k2

k6MS

the multivariable

]O)o

K6

,Y(S)= (1
(KlK6D)s
+
+ sK3T)[K6(MS+
or,

and no structural

1412)

from

(1
k2922

model

921(s) are retained

single-output

2-output

3.11;
Fig.
of
all signal

interactions

912(s) and

2-input

+ k692 + k6dg3)oo

+oo(klk6-k2k5)+oo(k2k6d+k6k2d)

52

(3.35)

This explicit
function

form for the multivariable

parameterised

is
y(s)
extremely

in analysing

useful

the effects

change in system parameters on the performance

instance,

related

K, related

to the system damping

contributed

3.8-

to the natural

by the exciter,

Analysis

In

state-space

represented

frequency

of any

of the system; for

K5

of oscillation,

K6
and
related

to the phase lag

[4], [58].

of parameters

(i) Model complexity

structure

affecting

model performance

:-

:form

the

by the following

synchronous

machine

models

are

equation

i= Ax+ Bu
(3.36)
Cx

The 3rd order model :With

equivalent

(dampers)

amortisseurs

circuits

relating

winding

neglected,

the machine flux linkages

3.14
below
in
Fig.
are as shown

53

the

and currents

L,
j4

. 00

'01

Lfd
R fd

Tq

Laq

Taq

Efd

Fig. 3.14 Synchronous

machine equivalent

circuit

3rd-order
:

model
Using equation

(3.36), the states of the 3rd order model, [x], are

to direct axis]. And in a

[ rotor angle, speed, voltage proportional

form
:
matrix
Ab

I
F
ATm
LAEfd I

[A6 Aet]
and y=

LAE'q' j

where
-0

0).
A=

[ a21
a31

001

0a23 1

a22
0

B=

a33

(The matrices

b21

01 1

C=
[C21
and

NF1000

b32

elements are written

in Appendix

C33

5)

The 6th order model :When machine

(the model

amortisseurs

d-axis
has
one
now

(damper)
damper

54

windings

included,
are

and two q-axis

damper

windings),

the equivalent

circuits

machine linkages

relating

and

3.15
below
in
Fig.
is shown

currents

Lfd

La

Vaq
qI

R fd

Llq

L2q

Rlq

Efd

Fig. 3.15 Synchronous

In equation
[x],

[rotor
are

direction,
winding

first

machine equivalent

(3.36),

angle,

winding,
in

winding

X=l

AVId

ATfd I
U

AE-M

6th-order
:

[A6 Ae
and y=
t

IAVlq
AV2q

where

55

damper winding

q-direction,

And in a amtrix

the 6-order model is shown below :

ATfd

circuit

model

the states of the 6-th order linear model,

field
speed,

damper

in q-direction].

R2q

form,

second
equation

in ddamper

(3.36)

for

A=i

wo

a2l

a22

a23

a24

a25

a26

b2l

a3l

a33

a34

a35

a36

b32

B=

a4l

a43

a44

a45

a46

a5l

a53

a54

a55

a56

a63

a64

a65

a66

_a6l
-010000

and

C23

C24

C25

C26

-C21

(the matrices elements are given in Appendix

Since
dynamic

of the simplifications

response

the

comparing

function

structure

polar

loading

synchronous

condition

reactance

plots

of

the

multivariable

In particular,

Q=0.3,

Vt=l)

1.76
Pu, xq--:
pu)

Fig. 3.16.

56

response;

or any order

polar

model orders described

(P=0.9,

(xd1.81

and sixth

the

can be shown using frequency

for
both
y(s)
models.

for
different
these
two
of -y(s)
heavy

of third

dynamic
behaviour
model on
by

ways

demonstrating
of

of a system is via its frequency

characteristics

the influence

linear

one of the effective

5 ).

plots

above at a

and a typical
are shown

in

-1

......................

go-2
E

-3

..........

.........

..........

. ............................

........

-4

........

...........

...

.........

...

.........

.......................

..........

-5
-3

-2

0
Real A)ds

-1

20
0 .......
0 -20

.......

......

-40 .......

.................
.... ............

.........

........
........

........................

.............

...........

...........

10-1

10

......

10

FREQUENCY

....................

-90

;--;.;

-180

:.

..................

-2

10-1

10

:- - Z. ; Z. ................

10

,....

Z.
... . .,

101

FREQUENCY

(---

3rd-order

Fig. 3.16. Nyquist

These polar
provides

further

plots

model;

(-

---)

6th-order

model

& Bode plots of y(s) for third and sixth

order models

are not significantly

justification

different,

and this

for the very common use [59], [60]

57

[611, [62],

3-rd
the
of
order model

of the synchronous

machine

shown in Fig. 3.4.

(H) Changes

in loading

Secondly,

reactive
fixed

factor

on the multivariable

synchronous
Three

(P=0.736,

leading
and

changes
conditions
reactance
factor),

power

from

different

445
---l.

loading

(P=0.912,

[53].

The initial

29S50 to

41.720 to

loading

xq--0.959 pu)
lagging

conditions,

Q=0.392)

power

Q=0.16),

in
are chosen as

angle of the machine

rotor

58.850 for

loading

the three

follows:
change as

does
in
increase
the
real power, so
with

these different

pu,

K5

is

of

K3 does not change at all (impedance

the rest of the parameters

(sometimes

for
y(s)
a

The machine is connected to a tie-line

While
x, =0.2 pu.

K6 decreases

function

(P=0.812,
factor
unity power

factor

respectively.

conditions

(real and

structure

(xd:

reactance

Q=0.752),

Saidy
and

Hughes

in
loading
of changes

the effect

power)

considered.

conditions:

K2

changes sign).

conditions

and

K4;

K, increases

while

Polar plots

K5

and

of y(s) for

in
3.17,
it
Fig.
are shown
where

is observed that y(s) moves towards the (1,0) point for increasing
(loading
real power
to

decrease

frequency

at

condition),

1 rad/sec,

it

and although
increases

the interaction

again

in

the

start

important

(4-10)
(see
Chapter
4)
From
Result
2.3
rad/sec.
range
of

58

Chapter

2, this

indicates

for increasing

model uncertainties

....................

-0.5

...........

...................

.............

...........

r ..............

........

.............................

.............

-2
-1.5

robustness

to

order models

at

real power.

.........

Ae

-1.5

a decrease in stability

...

-1

......

..............

................

.......

..........
0
-0.5
Real Axis

.............

.....

0.5

20
0

...............

(9-20

. ........

..............................

.........................

-40

L-60 2
10

.............

.....

0
10
FREQUENCY

10-1

.................

1
10

LU
CO)

...................

........................

-90

m
CL

...................

-180
-2

10-1

10

10

1
10

FREQUENCY

(-)
Fig.

low
3.17.

-----'(
Nyquist

) medium, and ( ----

) high

Bode
and
plots of y(s) for third
different
operating conditions

59

(iii)

Changes in synchronous
Consider

model

Three

considered;
synchronous

y(s)

moves

higher
point

pu[65].

values
of view

synchronous

choices of a particular
loading

Xd=--Xq=

reactance
[64],

pu
plots

reactance
1

the

are
[631,

pu
.6

synchronous

and

of

(P=0.9,

condition

synchronous

Polar

multivariable

y(s) are shown in Fig. 3.18 where it is observed

towards

reactance.

the

(1,0)

point

of synchronous

model

reactance

in
practice
used

less than 2.0 pu for which

On the other hand,

reactance.

of more power generation

there is no stability

increasing

robustness to model uncertainties

model synchronous

reactance

for

Again from Result 2.3 of Chapter 2, this

decrease
in
a
stability

for increasing

of

Xd=--Xq=2.0

reactance

synchronous
indicates

values

synchronous

Xd--=Xq=2.6

structure
that

different

function

at a fixed

reactance

namely

reactance

different
of

next the effect

synchronous

Q=0.3).

reactance:

desirable
are

from

[66].

values of

Typical

the

[67],

however,

are always

our multivariable

analysis

shows that

robustness

problem

point).

60

( y(s) nowhere

near the

0
oo

.............

-0.5

...........

-1

-1.5
-2

................

... ......... ............. .............

.................

....

9-2.5
.

..............

': * **

4e :

..
.....................................

-4

.... ....

...

.............

.....

....................

..........

....................................

.....

-3
-3.5

....................................

I ...............

...............

.......................

::,

*"**..........

ol................:

I ..........

0
Real A)ds
20
0

................

. ...

...............

...............

:Z,. r L;:-.
-,

:::..

-20

......

....

............

....

.......................

0
10

10-1

:.

.-..::

....................

... ... ... ...... .............................

-40

10

...............

..........

1
10

FREQUENCY

LU
.............

2
3: _90

.................

CL

-180
10 -2

10-1

10

1
10

FREQUENCY

()

high reactance,

(-

---

medium,
-)

()

low

Fig. 3.18 Nyquist and Bode plots of y(s) for third order models
different
synchronous reactance
with

61

in tie-line

changes

Taking

now the effect of tie-line

of the generator;
(strong

the tie-line
line),

transmission

transmission

line).

factor

jight

current

flowing

synchronous
pu;

and

reactances

the

multivariable

reactance

is
from
0.2
xe
changed
pu,

to 0.4 pu (medium)

loading
into

The

conditions.
the grid

system

infinite-bus

voltage

and

are considered

constant.

The

reactance

of the generator

terminal

Vt=1.0

structure

voltage
function

is

1
Xd=--

pu.

-1

-2

...........

...............

..........................

.................

................

...............

.........

Co
..............

-g

445 pu, Xq=0.959

.

Polar

plots

y(s) are shown in Fig. 3.19.

0001,

............

0.8
(weak
and
pu

The generator is operated under lagging power

.....

on the performance

variation

-4

-5
0
Real Axis

23

62

of

the

20

..........

.............
(D-20

.................................

-40

..........

....................

...............

..........................................

-60'
10

0
10

10-1

1
10

FREQUENCY

LU

C/) -90

. ...........

.....................

(L

...............

-180
10-2

10-1

10

10

FREQUENCY

()

(strong
----)

medium

()

line
trans.
weak

Fig. 3.19 Nyquist and Bode plots of -y(s) for third order models
3
different
lines.
to
transmission
connected
It

is

that

observed

y(s)

increasing

model

tie-line

throughout

the

frequency

increase

even

more

after

Again

moves
(weaker
range

from

weaker.

indicates

in
decrease
stability
a
model tie-line

the

transmission
(0-4)

4 rad/sec

becomes

for increasing

towards

Result

(1,0)

point

lines)

reactance

read/sec,
as the
2.3

of

and

start

transmission
Chapter

2,

for

to
line
this

robustness to model uncertainties

reactance.

63

Effect

of the damping

Here,

the

loading

synchronous
constant
to

D=6

structure

0
-0.2

of

effect

reactance

constant

is considered.

(P=0.9,

condition

Q=0.3,

Vt=l)

pu,

and to

function

D=10

Polar

pu.

plots

of

in
Fig. 3.20.
y(s) are shown

................

........

....

............

..........

.................

........

-0.4
-0.6

..........

.............
...............
...........

...................................
.............

.x

.....

..........

..........

-0.8

............................................................

-1.2
-1.4

..................

..............

..........

................

-1.6

..................

............

..............................

le

S,
..............

and a typical

(xd-*--':
1.81 Pu, xq--1.76 pu) and an inertia

...........

...........

on the

The machine is at a

.............

variation

of M=7. The damping constant is changed from D=3 pu,

..............................

.............

damping

of the generator

performance
heavy

constant

...........................

...........................

-0.5

0.5

Real Axis

64

the

multivariable

20
0

...............

(9-20

.............................

...............

.................

.....................................

-40
L-60 2
10

0
10
FREQUENCY

10-1

1
10

-90
(0
< -180
3c

.....

.........

............................

CL

10 -2

10

10-1

1
10

FREQUENCY

) high damping

(
medium, -------

low, ()

(------)

Fig. 3.20 Nyquist and Bode plots of y(s) for third order models
damping
3
different
constant value
with
As

it

is

clear

that

the

important

of

y(s)

idicator)

at the

Again
4).
(see
Chpater
(5-15)
rad/sec.
range

from Result 2.3 of Chapter 2, this indicates

robustness

interaction

damping
constant espacially
model

increases for decreasing

frequency

gain

(the

in
decrease
stability
a

for decreasing

to model uncertainties

damping
model

constant.
(vi)

Effect

Finally,
performance

of changing

the

inertia

effect

of

of the generator

constant

inertia

constant

is considered.

65

variation

The machine

on

the

is at a

heavy

loading

synchronous

(P=0.9,

condition
reactance

(xd`1.81

Vt=l)

and a typical

1.76
The
damping
Pu, xq--:
pu).

is D=3 pu. The inertia

constant

Q=0.3,

is
from
M=3 to
constant
changed

M=7 pu, and to M=10 pu. Polar plots of the multivariable

function

in
3.21
Fig.
y(s) are shown

............

...........

-1

12-2
.x

....

....

M
ad

.....
.............
........
....
.....
....

/*

..........

..........

-4

-3

Z'.......

\:
...........

... ......

..........

-2

....................

.....:...............

e: ........

.......

..........

...............

023
Real A)ds

66

structure

20
0

..............

..............

.............
...........

(9-20

................

....................................
..............

-40
L--60
10 -2

10-1

0
10
FREQUENCY

1
10

0
10

1
10

Co
< -180
Z

...

...................

EL

Qan
-W
10

10-1

FREQUENCY

()

high,

(- ---

-) medium,

()

low

Fig. 3.21 Nyquist and Bode plots of y(s) for third order models
inertia
3
different
constant
value of
with

It is observed
increase
This

that the highest

frequency

the gain of y(s) at a lower

indicates

uncertainties

decrease

forincreasing

in

inertia
the
value of

stability

than smaller
robustness

inertia
constant.
model

67

constant

to

values.
model

Final

The

remark:

synchronous
turbogenerator

effect

machine

of

models

changes

of

some parameters

on the stability

to model uncertainties

robustness

of

of the

3.1
in
Table
are summarised

below

Parameter

Effect on stability

Loading

decrease with the increase of real power

conditions

Synchronous

reactance

robustness

decrease with the increase of reactances

Damping constant

increase with the increase of damping

Inertia

decrease with the increase of inertia

Tie-line

constant
reactance

Table 3.1 Effect

decrease with the increase of reactance

of parameters change on the stability

68

Chapter

4:

Control

Options Available

to Turbogenerator

Systems

4.1- Introduction
Traditionally,

the problem

into the two separate problems

the design

control;
control

of

of power

of excitation

excitation

control

independently.
systems are carried out

is designed assuming constant flux linkage,

set point regulation

is split

systems control

control

and governor

systems

and governor

The turbine governor

is
and
used to achieve

4.1.
in
Fig.
the
of
shaft speed, as
Synchronous

U1
9
+

generator without
terminal voltage
regulation

Turbine/

(0

-Governor

Fig. 4.1

Subsequently,
constant
regulation

mechanical

machine

speed regulation

the excitation/AVR

torque input,

of the generator

is designed, assuming

is
and
used to achieve set point

terminal

voltage,

4.2.
in
Fig.
as

(1)

69

Synchronous
Vref

Exciter/
b-AVR

Fig. 4.2 terminal

consist

is typically

required

effective

Voltage

signals.

frequency
over a

of the governor

subjection

an alternative

turbogenerator

of a single two-input,
structure

is typically

regulation

two-output

of a one-machine

interpretation

adequate

turbogenerator

system, connected

control

together with a supplementary

system stabiliser

(PSS), is as depicted

in Appendix

to consist

feedback system where the overall

turbine/governor

is as defined

to be

is possible;

be
can
considered

and

bus

to higher-

required

incorporating

infinite

to

feedback systems. Speed

to be effective

regulation

is considered

rejection.

However,
specifically,

regulation

0
10
to
rad/sec, providing
range

over a frequency

disturbance
voltage

voltage regulation

single-output

0
1
to
rad/sec, avoiding
range
frequency

--

turbogenerator

words,

of two single-input,

regulation

Vt

PSS

------------------------------

In other

generator with
governor speed
regulation

control,

excitation/AVR

speed power

in Fig. 3.10. (Nomenclature

2). In the latter

70

to an

multivariable

context,

three questions
loop

(a)
to what extent is the singlearise:

immediately

regulation

approach

appropriate

(b)
is
the role
;
what

of the

PSS ; (c) are there other more suitable approaches to the regulation
These questions

in
the discussions
are answered

3
respectively
and
4.2-

(P=0.9,

(xd

1,2

in this Chapter.

Multivariable
The 3-rd

Options
of

of turbogenerator

analysis

order

Q=0.3)

model

with

[24]

is chosen

typical

regulation
at the operating

synchronous

point

reactance

of

0.65
121
1.76
0.3
pu.
PupXq =
pu and xa =
pu) with xe =
"=
Substituting

K4=1.4187,
2-input

K5=-0.1463,

2-output

-912

K6=0.4168

K2=0.8649,

K3=0.323,

[24]
into
Appendix
of

function
transfermachine

at these conditions,
911

Kl=0.7643,

the values

matrix of

(4), the
Fig. 3.11

is given by

974168s+ 377

922_

-10532
0942s +0.424s+54-2]

(4.1)

_-142520s-12716

_921

A(s) = 18.088S3+ 14.7452S2+ 7475566s+ 138.724

The governor

control,

k,

is
3.11
by
Fig.
given
of
,

kg
ki (s) = 41

(4.2)

)(1
)
+ sTg + ST,
,

71

T
is
the governor
where 9
is the steam turbine

value actuator time constant (0.2 sec), Tt

time constant (0.3 sec), and kg is the governor

loop gain (1). The excitation

by the fast

thyristor

response

k2, in Fig. 3.11 is provided

control,

excitation

system

the transfer

with

function
k2 (S) =

ke

(4.3)

(1
+ STJ

(200)
is
Te
the excitation
system gain
and

ke
is
the regulator
where

(0.01
time
constant
sec).
system
The

governor
the

regulate

kl(s)

control
speed,

machine

round the turbine/governor

loop subsystem
Subsequent
the

regulate

frequency
shown

channel
frequency

(2.4)

equation

4.1;

subsystem

that

is,

is applied

is closed

the loop

k1g, l(s) to yield

to

the closed-

hl(s) defined by equation (2.5).

of the exciter

application

terminal

machine

voltage

response of the excitation

in Fig.

equation

fig.

of

(4.2)

4.3.

A simple

is however

is relatively
response

k2(s) of equation
is

channel

approximation

follows.
as

the turbine/governor

72

the

through

C2 of equation

(3.33)

of the channel

C2 Of

Since the turbine/governor

bandwidth
slow with a
of

examined

(4.3) to

I
rad/sec, the
of about
subsystem

hl(s)

of

(2.5)

equation

have rolled

will

the faster

excitation

excitation

channel

k2922

in
frequency
the
off
C2 in equation

channel

C2 in equation

in the typical

(2.4)

(2.4);

that

to

is, the

be
can
approximated

channel frequency

excitation

interest
of

by

1
10
to
range of

in
Fig. 4.3.
as shown

rad/sec

In effect,

this is a formal

confirmation

loop has little

that the slow governor

loop response and so the performance

is largely

system

function

transfer

determined
922(S) with

fact
the
of
observed

effect

of the overall

by that
input

faster
the
exciter
on

field

turbogenerator

of the electrical

subsystem

and output

voltage

terminal

voltage.
40

L.L , Li

........................

.... .........

. .....................

.........

CO 20

...

....................................

.........

......

. ... .

-20
10-2

10

10-1

10

10

FREQUENCY
0

-90
LU
03

-180

......................

L-27010-2

..............

10-1

............

10,10
FREQUENCY

Fig. 4.3 Bode plots of excitation

excitation

............

(--)
channel

channel approximation

73

10

(-

and the
--

-)

Focusing

observed

in Fig.

extremely

awkward

resonance

(lightly

its

4.3 that

turbogenerator

by an anti-resonance

(LHP)

hand plane

but the close

lightly
the
of

apply

similar

rad/sec

to

because

LHP
the
of

implies

dynamics

the

need

(recall

Figs. 3.14-3.19).
k2922(1-'10
or
4.3. Furthermore,
with

active

control,

the loading

zero pair

for

Indeed,
would

closely

following
precise

with

choice

the closed-loop

condition,

[221 per se
damped

LHP

at by

Any attempt to
7

be problematic

the LHP pole pair

knowledge

of

of model observed

for
either
response

the

in

k2922

from
be
Fig.
observed
as may

be unstable

the frequencies

left-

including
range

922(S) would

unrealistically

the variation

analysis.

frequency
a

over

of a

is also arrived

conclusion

subsystem

electrical

damped

damped

lightly
the
of

Hamdan and Hughes [281 using pole-zero

direct

at

problem that besets the control

to these poles

interesting
an

zero pair;

(lightly

pole pair or lack of system damping

proximity

of a

7.58
at
rad/sec.

not the presence

set;

7 rad/sec

round

plane pole pair -0.31+j6.41)

plane zero pair -0.21+j7.58)

This is the fundamental

922(S) ,

it is

possesses the

response

characteristic

damped left-hand

followed

subsystem

frequency

switch-back

6.42 rad/sec closely

left-hand

on the electrical

therefore

of these pole-zero

choice

74

of reactances

pairs change

etc.,
.........

(see

Figs. 4.5,4.6,4.7),

direct
making
control

than one loading condition

Taking

electrical

the

impossible.

adverse

among the various

control

options
present

power

without

PSS

engineering
and

from the point

practice

excitation/governor

control

the

control

is

option

only

with

two

of view of past and

are excitation/governor

third

new

of

that may be considered,

excitation/governor

characteristic

response in Fig. 4.3 into account,

options

that demand scrutiny

completeness,

switch-back

frequency

subsystem

that might cater for more

control
PSS.

with

also

For

considered,
input/output

swapped

pairings.
Before

examining

the high-frequency

Result
phase

4.3 of

here, see Chapter

the higher
determined

limit

channel

two options,

observe that, since

of y(s), shown in Fig. 4.4 is less than one,

[681 establishes

stabilising

excitation

the first

controllers

for

the existence
the two

channels.

2, there is a separation
and governor

bandwidth

channel,

essentially

byg22(S)-

channel,
namely

75

of stable
Also,

bandwidths
of
the controller,

the excitation

minimum
when

as

between
k2(s), for

channel,

is

-1

-2
F-3

-4

-5

03
Real Axis

-2

Fig. 4.4 Nyquist

4.3-

Factors

(i) Loading

affecting
conditions

Two different
(P=0.9,

Q=0.4)

plot of y(s) of the model of equation (4.1)

the switch-back
:

loading conditions,

conditions

(loading

[691. The

6
reactance xd=--l. pu, and xq=1.55 pu.

Polar plots of the electrical

switch-back

low (P=0.6, Q=0.4), and high

are chosen as in Ahson and Nicholson

has
a synchronous
machine

loading

characteristics

for
different
these
subsystemg22(S)

in
4.5
it
is
Fig.
observed that a
are shown
where

characteristic

is created

condition).

76

with

increasing

real

power

m -20

.......

.......

...............

-40

An
L--. -2
10

0
10

10-1

1
10

FREQUENCY

LU
...........

_90
CL
-180

........................

..............

10-2

10-1

10

1
10

FREQUENCY

Fig.

4.5 Bode plot of electrical

for different

Two

different

[69],
and synchronous
pu
The machine

plots

of

values

(
high
and

synchronous

reactance

are

reactance xd-"=1.6 pu, and xq=1.55

[28].
pu
reactance xd=2.6 pu, and xq=2.6

has a low operating

strong transmission

function

reactance :

namely synchronous

considered;

4.6 where

low (---),

loading conditions,

GO Synchronous

Polar

subsystem

922(S) transfer

conditions

is
connected
and

to a

line.

of the electrical

it is observed

subsystem

922(s) are shown

that the switch-back

77

characteristic

in Fig.

gets

for
increasing
worse

in
the sense that
reactance,

model synchronous

higher gain and phase advance are required to overcome its effect.
0

-20

...................
......................
. ........

2--40

.......
..............
....................
.........

...................

-60
10-1

10-2

10

10

FREQUENCY

LU

-90

2
M
(L

....

.......................

-180

10

10-1

10-2

1
10

FREQUENCY

Fig.

4.6 Bode plot of electrical

for different

synchronous

(iii)

reactance:

Tie-line
Finally,

the tie-line

0.8 pu [51]. The generator

light
flowing

loading
into

synchronous

conditions.
the

grid

reactance

terminal
the
pu; and

subsystem

(small, and
--)

()
reactance,

reactance

function

922(S) transfer

large

0.2
from
is
pu, to
changed
xe

is operated under lagging power factor,

The infinite-bus
system

are

of the generator

voltage

Vt=1.0

78

voltage

considered
is

1.445
Xd":

pu, with

and current
constant.
pu,

The

0-959
Xq-":

damper
winding
a

D=3 pu. Polar

plots

different

two

values

where

it is observed

lower

frequency

of the electrical
of tie-line

subsystem

reactance

that the switch-back

in the weak tie-line

for these

922(S)

in Fig.

are shown
characteristic

case with

4.7

occurs

at

more gain and

phase advance required to overcome its effect.

0
m
-20

..........

..........

-40

....................

... .......

L-60
10 -2

0
10

10-1

1
10

FREQUENCY

............

-90

..............

U)
Z:
a- -180

..............

............

..........

-270

..............

..........

10 -2

10

10-1

1
10

FREQUENCY

(-------)

) weak,
Fig.

strong

4.7 Bode plot of electrical

for 2 different
Like
uncertainties,

having higher
windings,

the

decrease

the switch-back

synchronous

working

subsystem

922(S) transfer

function

tie line connections

in

stability

characteristic

robustness
worsen

with

to

model

machines

damper
reactances, and small number of

higher
operating
at

79

conditions

(high real power),

while

switch-back

governor

account,

control

excitation

control

control

options

this

in

available

(4.1)

(4.2)

of

to

adverse

frequency

subsystem

the model

and using

and

the

respectively;

Taking

of the electrical

characteristic
into

response

lines.

transmission

to weak

connected

and the
(4.3)

and

turbogenerator

described
below
systems are
.

4.4- Control

option

I:

One

way

is

to

frequency)

the

restrict

is achieved

k2(s) defined

system

ke = 200 and

where
reduction

TI=0.7542

T2=0.3367.
and

is

as

given

before.

turbine/governor

The

channel

the

channel

by using

function

cascade

with

m(l+sTl)/(I+sT2)
The governor
resulting

C1,

given

80

7
response at
gain

3
say

response excitation
of equation,

(4.3)

transient

gain

the
where

kl(s)

control

frequency

crossover
low,

to be relatively

a slower

PSS

switch-back

problematic

(open-loop

bandwidth

Te = 0.01, in
by

without

control

frequency
subsystem

by the transfer

[11]

(4.2)

avoiding

of the excitation

This

rad/sec.

of

of the electrical

characteristic
rad/sec

Excitation1governor

m=0.06533,
of equation

responses

by equation

(2.1),

of

the

and the

excitation

channel

C2, given by equation

(2.4),

in
Fig.
are shown

4.8 and Fig. 4.9 respectively.

50

.............
2

-50
10-2

..................

. ... .

10-1

10

10

10

FREQUENCY

0
7

-180

........

7-

.........

....

(n
x

CL

-360

.......

..............

-540
10-2

............

10-1

10

. .............

..........

........

10

10

FREQUENCY

C k, &, (1-, yh, )

k1gl,

-----Fig. 4.8 Bode plots of the turbine/governor

(Option 1).

81

channel without

PSS

50

m
....

..................

CA
-1^0
10 -2

10

................

01

10

10

10

FREQUENCY

.......

-90

LU

....

CO
...........

(L -180

............

-270

10

16,10

10-1

10-2

FREQUENCY

Fig. 4.9 Bode plots of the exciter channel without

frequency
the
of

Comparison
open-loop

channel

turbine/governor
lines)

show

frequency
This,
4.3,

them

together

turbogenerator

with
the

different

significantly

the discussion
original

see

that

with

of the

over

the

(0 to 1 rad/sec).

system to consist

systems,

(2.1),

by equation

be

to

not

of the turbine/governor

(dashed4.8
in
Fig.
k1g,
subsystem
l(s)

open-loop

interest
range of

justifies

feedback

C1, given

response

PSS (Optionl)

immediately
of

approach

prior

considering

of two single-input
4.2.

Section

82

The

to Section
the

single-output

corresponding

step

responses

of speed

and terminal

voltage

in
(Figs.
are shown

4.10,

4.11) and are quite satisfactory.

1.2

.............................

0.8

..............

................

....................................

...

. ..........

................

..............

0.6

0.4

0.2 .........

...... ................

.............................................................

123
Time (secs)

Fig.

4.10

Step response

of the system without

(Option 1)

83

PSS : speed output

1.2

0.8

:g-0.6
E

0.4

0.2

3
Time (secs)

Fig. 4.11

Step response of the system without PSS: terminal

(Option
1)
output.
if set point

Thus,
and

terminal

then

requirements,

rad/sec

suffices,

was the option

C2)

(channel

low

the relatively

are

channel

the

taken with

smaller

before

machines

the introduction

control

and smaller

Cj)

control

only

C2 bandwidth

first
This
PSS.
need of a

without

reactance

synchronous

(channel
the
shaft speed
of

regulation

voltage

voltage

3
of

option
per unit

in
PSS
the
the
of

[701.
1960's
early
Control

option

1 does however

highlighted

by the terminal

disturbance

in Fig. 4.12.

have a major

disadvantage

10%
to
voltage
voltage output response

84

1.1

08

1.06

1.04

1.02

0.981023456

Time (sec. )

Fig. 4.12 Terminal

10%
to
voltage output response
voltage
disturbance (Option 1).

is corroborated

This

function

1/(l+C2)

by the Bode diagram

of the excitation

channel

in
4.13.
Fig.
shown

'77

...........

-5

-10

-15

of the sensitivity

..........

..........

.........................

..................

..........

.................

-20

.........

................

-25

.........

..............................

-30'
10

I. I. 1 .1.

I;

.................

---Z.

- I- .;.;

-:-; ;-

................

10-1

0
10
Frequency rad/sec.

1
10

10

Fig. 4.13

Bode plot of the sensitivity

(Option

85

1).

element

I+C2

without

PSS.

This disadvantage
disturbance

poor sensitivity

4.5-

rejection

of the closed-loop

properties

0
to
range

rad/sec.
Control

option

2: Excitation1governor

The second control

bandwidth

strong voltage
10 rad/sec.

disturbance

However,

7
near
rad/sec,

prevents

option

is compatible

the

with

use of such

discussed under Control

features

: the

high

presence

of the switch-back

Control

option

lightly
the
of

C2

1, this

to provide

frequency
the
over

high-bandwidth
a

0
to
range

damped LHP zero

response

4.3,

Fig.
of

excitation

channel

as

1.
how the additional

these two seemingly

bandwidth

channel

requirement

in the frequency

Let us now explore

alleviate

Unlike

rejection

option

Fig. 4.2 might

control

a control

the presence
shown

PSS
with

is to seek an excitation

10 rad/sec.

of roughly

specification

pair

1 lies in the relatively

option

channel over the latter part of the frequency

excitation
10

Control
of

for

frequency

in
PSS
the
use of

irreconcilable

disturbance

rejection

response characteristic

control
and the
round

7 rad/sec. From Fig. 4.2, it is observed that the PSS acts as a postcompensator

on the

speed output,

86

so that

taking

Fig.

3.11

or

(4.2)

equation

into

the system with

account,

PSS assumes the post-

form

compensated

G"(s) = P(s)G(s)

Fl 0 Fgll

=[P

1][921

9121

(4.4)

9221

F
911
[921

1Fg

11
19"21

912

+ P911

922+P9121

912
9"221

where,

in

element

denotes
function
the
transferp(s)

the

p(s) =k ' 1+sT,

and
kl(s)

control

denotes

ks =9.5

denotes

and the excitation

In equation

function

LHP

is observed

filter

washout

(Tw=1.4)

to

responses

and

(T3=0.145,

The governor

the stabiliser

gain.

system control

k2(s) are as defined

(4.4), the PSS transfer function

zeros,

922(s)

frequency

[221

(4.2) and (4.3).

912(s) subsystem transfer

damped

by
PSS
the
given
of

denotes a phase advance compensator

in
before
equations
as

benign

cross-coupling

I+ sT4]

+ sT3)/(l+sT4)]

T4=0.03)

the

(4.5)

sTw/(l+sTw)

where

matrix

P(s),

sT,

[(l

post- compensator

function

be

amended

to

subsystem

922+P912-

transfer

Comparing

in
4.14,
Fig.
to
those
of 922+P912
of 922(s),

that the effect

inverted
the
of

87

the

lightly
which contains no

the electrical

so causing

p(s) multiplies

notch characteristic

the
it

of

the PSS, p(s),

dynamics,

in equation

(4.5),

multiplied

is

dominate

the

912(s),

characteristic

of

to

922(s) at 7 rad/sec.

is
approach
effective

for different

by the benign system

problematic

It should

be noted

loading conditions,

because the dynamics of the turbogenerator

switch-back
that

this

4.15
Fig.
see

itself are exploited.

jM -20

-40

.............................

..........

...........

'
-60 2
10

1
10

0
10

10

FREQUENCY

...............

-90
LU
CL
-180

-270'-- 2
10

..........

............

...............

..............

1
10

0
10

10-1

FREQUENCY

Fig.
Working
instead
facilitates
excitation

created

4.14 Bode plot of

with

of the original
in

a most

channel.

which

922+P912 and 922(S)

the amended
electrical
ingenious
Essentially,

is the weighted

subsystem
way
an

PSS
thereby
the
922(s),

the use of a high-bandwidth

artificial

system

sum of the terminal

88

922+P912

subsystem

electrical

output

voltage

is

and the

shaft speed output.

is essentially
to

that

artificial

of

At low and high frequency,

the terminal
the

LHP

voltage

pole-zero

as desired;

pairs

of

the artificial

output

at frequencies

922(s) at 7 rad/sec,

close
the

output is dominated by the speed output.

...........

...........

m
c3-20

...................................

-40

--60'
10

10

1
10

0
10
FREQUENCY

-90

.......

....

............................

..............

-180

'
-270
2
10

10-1

0
10

1
10

FREQUENCY

Fig.

4.15a

Bode plots of the electrical

subsystem
different
loading conditions

89

for
three
922

T.:.;

................

............

.............

.....................

..............

m
a
(9-20

.............

-40

..........

10-2

10-1

10

10 1

FREQUENCY
0
..............

..............

-30
LU -60
co

............

...........
.....................

-90

..............

.........

-120

..............

-150L-- 2
10

0
10

10-1

.......
.............
1
10

FREQUENCY

for
4.15b
Bode
Fig.
plots of the electrical subsystem g,, + pg,,,
different loading conditions.
frequency

The resulting

C, and excitation

channel

4.16 where

channel

C, and C2 without

of the terminal
quite
voltage
output

satisfactory

PSS are shown

C2 with

channel

they may be compared

of the turbine/governor

responses

to those of the lower

PSS in Fig. 4.9-10.

with

little

detrimental

effect

by
the
creation
caused
response
step
near

frequency

The step responses

be
4.17
in
Fig.
to
and shaft speed are seen

voltage

7 rad/sec

in Fig.

in the frequency

90

on the

of the artificial

response.

terminal
speed

50

Ln

%.............

.... ...

............

. ... .

................

-50

L-

-100

10-2

0
10

10-1

10

10

FREOUENCY

LU
CO

-180

a- -360

.......................

r:
--540

Z---*,- .:.; ...........

10-2

10-1

::

-, ..-.:

;::.

.. -, .:." .:..............

1 op
FREQUENCY

--, .:..:.,.

10

10

Fig. 4.16

Bode plots of the turbine/governor

(
C1
channel
) with PSS (Option 2).
and the exciter channel C2 (- ------1

....................

43
CL
E
< 0.5

OL
0

.........

0.5

..........

.........

...................

..........

1.5

2.5
3
Time (secs)

3.5

4.5

1.5

2.5
3
Time (secs)

3.5

4.5

1.5

CD

CL

0.5

0
0

0.5

Fig. 4.17 Step response of the system with PSS , (a) speed output, (b)
terminal voltage output. (Option 2).
91

It is observed in Fig. 4.17 that the step response of the

excitation

channel

is not due to lack

4.16,

Fig.

observed
residual
output
observed

the

C2 with

PSS is still

of damping

phase

to be above

margin

of

600; rather,

response

10%

the Bode diagram

the

open-loop

channel

the oscillation

disturbance,

in

function

1/(I+C2)

in

C2 is

is due to the

Fig.

to that of Fig. 4.12 without

of the sensitivity

since

system

The
terminal
system poles.

voltage

to be much superior

oscillatory.

in the closed-loop

effects of the open-loop

to

somewhat

This

voltage
4.18,

is

PSS as is

4.19.
Fig.
of

1.1

1.08

1.06

CL

1.04

1.02

0.98'
0

Fig. 4.18 Terminal

Time (sec)

10%
disturbance
to
output
response
voltage
voltage
(Option 2)

92

-5

92-10

2-15

-20

-25

L-_qn
10 -2

10-1

0
10
FREQUENCY rad/sec.

1
10

Fig. 4.19 Bode plot of the sensitivity element of the voltage

(Option
2).
PSS.
channel with
4.6-

(4.4) and (4.5) take the form of a non-

The PSS of equations

diagonal

system

Leithead

O'Reilly
and

individual
any

(4.4)

system

tran sfer- function

element

equation

uncertainty

AgO/g-U (i, j=1,2)

compensated

function
transfersystem

(4.4) is related to the relative

of

compared

93

elements

G'(S)=P(s)G(s)

uncertainty

for

to the

element gij(s) of

Specifically,

the

shown

in

g-i'j(s) of equation

of the transfer- function

(4.1).

is

(uncertainty)
error

by system post-compensation

system

it

general,

[36] that the relative

(uncertainty)
error

original

In

post- compensation.

is increased

relative
the

PSS
system with

Robustness of turbogenerator

the
of

the

relative
post-

of equation

Ag, /gij of the elements of

the

un-compensated

(4.1) as follows

equation
lgli
gil

[(l - Aj

-1y

tran sfer- function

system

gll

matrix

G(s)

of

[36]
(A,

_, y

) JA921

(4.6)

921

A912

-[(BI

1-Y

9'12

A921
-1-7

422

921

+(l-B

922

(A2
-'Y
+

F(I-A2)
L

l-,

9"22

912

(4.7)

922

[(B

9/21

(1
+ - Bj

2)

(4.8)

911

(4.9)

912

where
912921

4512j521

A2

cr/ er
t511e522

(4.10)

9119/22

911912

B,

B2

=1
911912

(4.11)

_9219122
9121922

Thus, the price paid for the use of

that relative
with
factor

that of the original

Since

is

in the post- compensated system, compared

uncertainty

1/(l-y).

PSS (post- compensation)

system without

in the frequency

PSS, is multiplied
interest
range of

by the
(0 to 10

(1,0),
is
the
),
the
the
point
of
y(s)
nowhere
near
plot
polar
rad/sec.
factor

1/(I-y)

is not

does not significantly

very

large

reduce

and the PSS (post- compensation)

robustness

94

to system uncertainty.

4.7-

Control

Option

input-output

swapped

interest,

the problematic

only

with

One
is to

particular

swap

switch-back
7
at
response

channel bandwidth,

is to use non-diagonal

control).

precompensation

control

frequency
subsystem

high excitation

yet retain

diagonal

to avoid

option,

of the electrical

characteristic

theoretical

Excitation1governor
:

pairings

A third

rad/sec

the

least
at
of
(non-

precompensation
form

simple

assignment

of

system

of

system
inputs

to

outputs so that the output speed Aco is paired with the field voltage
input

AEfd and the output terminal

mechanical
assignment

torque
of

input

system

voltage

ATm. The effect

inputs

AVt is paired with the

of this swapping

is to pre-compensate

to outputs

in
3.5
Fig.
system
such that the amended pre-compensated

of the
the

is
system

by
given

F911
[921

9121FO

G* (s) = G(s)P(s) =
The governor
kl =

control

9221[l

11
01

Fg129111
=[922
9211

kl(s) of Fig. 3.11 is given by

30

(4.13)

S(S+10)

The excitation

(4.12)

control

k2(s) of Fig. 3.11

95

is given by

2+5.779s
100(s+ 3.686)(s+ 0288)(s
+ 23.892)
k2 =2+0.955s
+ 1.365)
s6 + 63.9584)(s+ 0.47)(s
the high-frequency

Just as before
function

structure

(s)=I/y(s)
one, y*
it

Also,

electrical

Y(S)

is observed
subsystem

in

be used (20 rad/sec)

channel

of the multivariable

(4.1)

was

less

than

frequency

lightly

the problem

of

the

damped LHP zeros is

advantages of this third contro

for the excitation

2 with

Option

channel

the actual outputs

and secondly

Control

that

element.

also robust to plant uncertainty.

by

(4.12)

equation

that a high bandwidth

unlike

equation

922(S) possessing

As a consequence,

controlled

limit

for the amended system (4.12) is greater than one.

removed to the off-diagonal

firstly

of

(4.14)

4.22.
Fig.
4.20
in
Fig.
to
plots

96

are

C2 can now

directly
are

the PSS. Performance

The good performance

responses,

option

step responses

is

is confirmed

and sensitivity

50

20
'0
..........

-50

-1001-- 2
10

...

...........

0
10

10-1

10

10

FREQUENCY

-180
CO

:Z

0--360

-540

..........

.........

..................

10-2

..............

...............

10-1

.............

10

. ... .......

...........

1
10

10

FREQUENCY

Fig.

(.
C1
4.20 Bode plots of the turbine/governor
channel
the exciter channel C2 ( ----+
with swapped assignment
(Option
3).
4.12
inputs
to
system
outputs of equation

and
of

1.5

01
Z

...........

...............................

.........

........................

CL

E
< 0.5

OL
0

.............................
10

23456
Time (secs)

1.5

V.1
CL

0.5

0023456789

10
Time (secs)

(a)
PSS
Step
the
4.21
system with
output,
speed
Fig.
response of
,
terminal voltage output. (Option 3).

97

(b)

1 05 ................................
.

I ................................................................

I
3
'rime (Sec)

10

.........

...............

-10

-20
.....

a -30
Ei
-40

I. .1..............

............

. ......

...

..........

.....................

.................

....................................

............................................

-50

Z. Z.

. ... .....

-60
L-70 2
10

10-1

a
10
Frequency rad/soc.

1
10

10

(b)
Fig.

4.22

(a)

disturbance,
without

Terminal
(b)
and

PSS. (Option

output

voltage

response

to

Bode plot of the sensitivity

3).

98

10% voltage
1
element I+C2

There are, however, two practical

Control

Option

I+

subsystem hl(s) of

(4.15)

k1912

is unstable.

This causes a system integrity

loop fail,

exciter

is that the governor

for the amended system (4.12) given by

equation(2.5)
hi (s)

3. The first

disadvantages to the use of

consequence

the whole

the

of

system will

for,

problem;

This is a direct

go unstable.

multivariable

structure

the

should

function,

(S)
7*

associated with the amended system (4.12), having a high-frequency

limit

greater

[68].

The second practical

output

pairing

electrical

subsystem

satisfactory

Option

of the swapping

near 7 rad/sec.

the

4.1

of

inputof

to high frequency

have
above
options evaluated
in the methods by which
frequency

characteristic

On balance,

most

they
of the

the use of the PSS

practical

option

with

performance.

4.1 below

multivariable

switch-back

2 remains

Result
see

of the governor

disadvantages
and

the awkward

Control

Table

be avoided,

disadvantage

is the subjection

advantages

deal with

in

one and cannot

In summary, all three control

signals.
their

than

summarises

turbogenerator

the control
control

99

options

system

available

for the

Control

Advantages

Suitable

Disadvantages

option
Low

machine
1-more

1- low stability

small
machines

band-

economical

"no

margins

width

fast exciters or

2-highly

Exciter

PSS are required"

exciter channel

per unit

channel

2-direct

3- some loss of

synchronous

on outputs

performance

reactance

1- no loss of

1- increase in

all types of

(PSS+fast
cost

machines

exciter)

high
with

channel available"

2- no direct

damping

2- no integrity

control on the

"low
problem

second output

performance

(effectively

Power

control

System

performance

Stabiliser

bandwidth

(PSS)

exciter

on

controller
governor

"high

channel

sensitive

it is

speed+voltage)
3- some stability

OK"

if
the
problems

3-increased

two channels are

stability

margin

strongly

100

coupled

with smaller

Control

Advantages

Disadvantages

Suitable

option

machine

swapping

1- no loss of

assignme

performance

"high

bandwidth

1- an integrity

all machines

due
to
problem

and

the instability

nt of

exciter

system

OK99

h,
subsystem

f ast

input-

2- actual outputs

2-needs a very

governor

output

are directly

fast governor

pairs

controlled

Table
Final

remark

provides
used

Control

4.1

multivariable

and

exciter

analysis

justifies

[15]

[14],

system.
treating

In

loop

purpose of rejecting

of this

section

is treated

as a SISO

disturbances.
voltage

101

being

despite
other

words,

the turbogenerator

system where the governor

as a pseudo-SISO
the

the turbogenerator

control

control

(cross-coupled)
analysis

to

extremely

loop
loop
design
by
the
practice
of

justification

turbogenerator

multivariable

avialable

The
multivariable
:

formal

for

options

of

loop is first
system

for

a
the

system
closed

the prime

ChaDter

5:

Analysis

Some
Recent Developments
of

Turbogenerator
5.1-

stability

problems

conditions,

reactance

transmission

excitation

damping

of

oscillatory

is
considered
which

response

gain

lower
and

Engineers

oscillatory

complexity

voltage

of

a very

Engineers

[7 1].

In the

look
to
used

system

on the

analysis

of

imPortant
the
at

its
;
effective

gain and time

have a larger effective

[72].
response systems,

than normal

that the effect

damping

it is known

influence

strong

higha

to have an effect on machine damping.

constant

are convinced

oscillatory

and

these conditions,

systems generally

excitation
time

reactance,

oscillations,

of the excitation

High

Under

have
can

machine/system

characteristics

constant

synchronous

system.

Power

that

in
high
load
the
occur
presence of a

high

systems

stability,

it was illustrated

chapters,

generally

operating

unit

Control

Introduction
In previous

that

in

depends

of these characteristics

upon

is
be
PSS
to
applied
a

whether

or

to supplement

not
the

on

added
normal

controil.

But as was demonstrated

in Chapter 4, there exist a complex pair

lightly
of

damped zeros in the exciter/voltage

subsystem

922 of the

synchronous

102

machine)

loop (the electrical

of the turbogenerator

system,

which

the system
creating

at the mechanical
"switch-back"

increasing

characteristic

lightly
are

as they

The presence
be

cannot

pair of pole of
(4-10

mode of the machine

which

high synchronous

real power,

these zeros

they

close to the complex

damped

gets

limits

worse

It is undesirable
(very

the

)
rad/sec.
with

of a synchronous

to cancel

imaginary
to
the
near

in
the closed
zeros

of these complex
cancelled)

reactance

lines.

and weaker transmission

machine

axis).

dangerously
are

achievable

loop (since
closed

loop

loop
bandwidth
4).
(see
4.3
in
Chapter
If
Fig.
the
closed
;

bandwidth

loop is greater than the natural

of the exciter/voltage

frequency

of

these zeros, then

(a) the closed

loop system time response will

characteristics;

the

system

will

have

a very

Fig. 5.2.

function;

sensitivity

(b)

5.1,

Fig.

have a very oscillatory

1.8
1.6 .... ...............................
1.4

...........

1.2 ...

....

...

.....

.....

..................

...................
..............

............

............
..............

.............

1
E 0.8
0.6
0.4

..............................................

.......................

.......................................................

0.2
n
-

10

15
Time (sece)

zu

zo

Fig. 5.1 Excitation

ju

channel step response

103

poor

20

10

0
co
-10

-20
-30
-40'
10

10-1

0
10
Frequency rad/sec.

1
10

2
10

Fig. 5.2 Excitation

Three

options

machine performance
is to have a closed
frequency

channel

were presented

in Chapter

by considering

these lightly

loop

bandwidth

of the mechanical

part of the machine

the assignment

is to "bury"
dynamics

the effects

changes

subsystemg22

function

transfer

is
to
one

another

matrix

to the offby swapping

last
the
and
one

of the machine;

by
these
crossfeed
of
zeros

the previously

was to, get a signal

which

damped zeros; one

mode of the system,

to outputs

the

of the natural

PSS.
by
the
a
a way of
system
of
Unlike

main

inputs
of

4 for enhancing

much less than the natural

from
the electrical
these
get rid of
zeros
diagonal

sensitivity

in
phase with
that are

purpose

believed
which

objective

should

speed changes;

is
PSS
the
presented
of

104

force

designed
PSS,
of a
terminal

voltage

here, in this study,

as to overcome

effects

the

of the

lightly

machine

damped

the three options

discussed,

in
one used
practice
maintain

good

the plant

zeros using

the one involving

natural

in

performance

wide

Of

the PSS is the only

The PSS is comparatively

today.

dynamics.

cheap and can

range

of

operating

conditions.
5.2- Recent Studies involving
Remarkable

appropriate
techniques

the PSS
have been devoted

efforts

PSS; many

methods

[73], pole placement

such

as root

to the design

locus,

[74], adaptive control

[76].. etc have been used. One of the very recent

of

eigenvalue

[75], and
studies

Hoo

of such

is the :

methods

Augmented

PSS

In 1995 Saidy and Hughes [77], proposed an augmented control

scheme to enhance the performance
(CPSS) under weak tie-line

stabiliser

of an additional

design

of

this

loop in parallel

additional
excitation

conditions.

control
system

and

additional

loop seeks to co-ordinate

loop

105

system

stabiliser.

The

is based on the inherent

dynamic

characteristics.

the contributions

performance.

power

It is consist simply

with the conventional

generator

damping
CPSS
to
and the

of conventional

The benefits

The

of the d-axis
of the scheme

were presented
frequency

as contributing

is to rid
from

positive

range and contributing

But in Chapter

positive

4, it was established

the system

the presence

natural mechanical

damping power in the lower

power.

that, what the PSS really

"switch-back"
of a

of a lightly-damped

synchronising

characteristic

which

does
arises

pair of zero very near to the

mode of the turbogenerator.

To see the effects of

this scheme on a power system model taken from [77], in the light of
Chapter
treated

4's
with

clarification

PSS
the
of
objective;

the model

is

first

PSS, of the form

a conventional

(1 + ST1Xl + STO

STw
CPSS= Ks
;F(1+ST2Xl+ST4)(l+S

(5.2)

w)

where
Ks = 10, Tw = 10, Tj = T3 = 03, T2 = T4 = 0.0595
and then an augmented

PSS is used in parallel

with

the CPSS. The

form
has
PSS
the
augmented
(I + sTi)

Kc

sT,
APSS = Ks
(i
(i + STf) (i + sTdo) + STW)
the APSS uses the natural

dynamics

system, these are :

Tf
Te
Kf
*
Ke
+
Ti =
+

106

(5.3)

CPSS
the
and the exciter
of

Ks =-

Kc =

K4d
a*K3
O)o
-TSO)*- Ke

(c*Tdo

where

Ke = 198, Te = 0.03, Kf = 0.01, Tf = 0371, are characteristics

excitation

of the

system, and

K3, K4d, Tdog T&

a, and c,

are

characteristics

of

the

synchronous

machine.
The effect
illustrated

mainly

synchronous
Form

APSS
the
to the system with
of
added
by analysing

the electrical

subsystem

Saidy

Hughes
and

the 5-th

[77],

order

linearised

line is :

314.6
539.954s
35.903S2
0.512S3
+
+
+
gl , --.
2

922

2_9.49s 124.715
--0.168s
3_ 3347s 2- 12382s 49506
--0.043s
4+0.064S3
0-001S
+ 0956S2 + 4.948s+ 36518

den

2+ 605212s 170.06
4+15.84S3
5+0976s
0.014s
+
+73.78s

921

922 of the

machine.

connected to a strong transmission

912

the CPSS was

The same model for the weak tie-line

107

is
:
case

model

1.6992S3
90.01S2
+
=
+ 855.624s+ 314.6

911

107.743
-0.2542 -12.091s 2_
12.411s
93124s
112.064
--0168S3 _
4+
2+
0.006s
0316S3 + 3.06s 10.758s+ 80.816
4
0.046S5+2.446s +24.74S3 + 70.225S2+ 574.62s+ 128.28

912
921
922
den

The nominal

important

range

plant

(1-10)

of

rad/sec;

has a low synchronous

system

line.

transmission
behaved"

diagonal

922, Fig.

feedback

The plant
5.4;

control

has a fairly

good characteristic

is expected

because the

and is connected

to a strong

this

reactance

has a small y(s), Fig. 5.3, with

and so the addition

element)

will

in the

"wella

(as
PSS
the
of
a non-

increase
its sensitivity
not

(see Section 4.6 in the last Chapter).

-0.05
12
.

..............................................

-0.15

-0.2
-0.05

Fig. 5.3 Nyquist

(Wo

U.1
Real A)ds

U.I Z)

U.id

for
y(s)
strong transmission
plot of

108

line model

......................................

-20

'a
0
2-40

...........

............

L--60 2
10

10-1

0
10
FREQUENCY

101

-30
Co
x
CL-60

-90
10 -2

10

10-1

10

FREOUENCY

line model

5.4 Bode plot of 922 for strong transmission

Fig.

For the system connected to the weak transmission

y(s)

of

the

characteristics
frequency

plant

increases,

developed

Fig.

5.5,

in the electrical

5.6.
Fig.
(4-6)
rad/sec,
range

109

and

subsystem

line, the

"switch-back"
in
the
922(s)

-0.2

-0.4

-0.6
A

x
d)-0.8
0
E

-1.2

-1.4

-0.8

-0.6

-0.4

Fig. 5.5 Nyquist

-0.2

0
0.2
Real Axis

0.4

0.6

0.8

for
the weak transmission
plot of Y(S)

line model

......................

co -20
'a
0
2-40

L-6o
10 -2

..................

1
10

0
10
FREQUENCY

10-1

w
-90

............

.....................

m
CL

............

-180

10-1

10 -2

10 u

10 1

FREQUENCY

Fig.

5.6 Bode plot of 922 for the weak transmission

The Bode plot

with

conventional

of the electrical

subsystem

line model

for
the model
922(s)

PSS and augmented PSS is shown in Fig. 5.7 (for

110

the

nominal

model),

Fig.
and

5.8

.......

............

(for

the weak

transmission

line

model).
0

.................

m -20

...............

2-40

............

.....................

'..

:.. -, .

I
'"''
......
.... ...................................
L-6o10-2

10-1

0
10
FREQUENCY

.*,

1
10

0
w

-30

...

.........................................

...

.....................

..................

Z:

12--60 ...........
-90

..........

..................
................

............

10

10-1

10-2

10

FREQUENCY

(,

-_
Fig.

) no PSS, (5.7 Bode

plot

--

-)

CPSS, (--

---.

-)

APSS

for
+
strong
of 922 and 922 P912

ill

tie-line

model

m -20

-40

...............

...

....

......

..........................................

L-6o
10-2

....

........

0
10
FREQUENCY

10-1

............

1
10

..............

CO

-90
CL

........

...........................

.......

.............

-180
10 -2

10-1

10

10

FREQUENCY

(S,
PS
no
Fig.

---)

CPS S, (-----

--

) APSS

5.8 Bode plot of 922 and 922 + P912 for weak tie-line

model

From Fig. 5.8, it is clear that the augmented PSS has an slightly
larger gain and phase advance than that provided
PSS; and the step response of the terminal
for the strong transmission
transmission

line illustrate

line model,

by the conventional

5.9,
in
Fig.
voltage shown
for
5.11
Fig.
the weak
and

improvement
some

in the APSS case over

5.12
Fig.
5.10
Fig.
the
CPSS
effect
show
and
the
one.

input

in the rotor speed output,

of the Exciter

form which also, it appears that the

better
disturbance
has
(APSS)
PSS
a
augmented

112

rejection.

1.2

0.8

,30.6
0-

0.4

0.2

01
0123456

111111
Time (Sec.)

plant
Fig.

5.9 Step response

0.06.1

CPSS
plant with

of terminal

voltage

of strong tie-line

1111

0.04 ........

0.02 .......

.......

....

0 ..... .......

-0.02

with

APSS,

....

..............

....................................................

................

....................

....................................................

.......

-0.04 . ... .............................................

-0.06 ..............

...............................

.................

........

-0.08 02345

(
APSS,
--plant with

Fig. 5.10 Effect

-)

of speed on terminal

113

CPSS
plant with
voltage strong tie-line

case

1.2

0.8

O.6
E
0.4

0.2

ON
023456

IaIIII
Time (Sec.)

plant with APSS,

plant with CPSS

Fig. 5.11 Step response of terminal

voltage of strong tie-line

0.08
0.06
0.04
0.02
0
E 0.02
<
-0.04-0.06-0.08.

i.
-n

2356
Time (Sec.)

APSS,
(
plant with
Fig. 5.12 Effect

CPSS
plant with

of speed on terminal

114

voltage weak tie-line

case

5.3-

Recent Method not involving

Some studies

design

PSS

did investigate

controllers

methods not involving

improve

which

the

performance

Here, one of these methods will

turbogenerator.

the PSS to
of

be analysed using

the ICAD technique.

The method

by Rosenbrock

proposed
degree

is based on the Inverse Nyquist

interaction
of

measurable
methods
decoupling.

Once

loops

control

stability

overall

Nyquist

distinct

in

the

what

measurable

this

separately,

(and the stability

decoupling

distinct

SISO
to

referred

as

decoupled,

well

found
be
can
loops.

the

the

and the
from

Rosenbrock

the
[29]

is
applied to the
procedure which

function

diagonally
as a

is

and

and

so that

one at a time

margins)

the

of

transfer

reduction

has been fairly

system

diagrams

and speed)

the

voltage

a turbogenerator)

of

voltage

designed
be
can

of the plant

is known

case

applicable;

devised a mathematical
inverse

the

(field

signals

technique

is to reduce

approach

the input

(terminal

outputs
become

The design

between

torque

mechanical

[29].

Array

matrix,

dominant

has
to
correspond
outputs

the number

of

and which

systems;

to the number of input signals

in order to have a square transfer matrix.

115

results

in

I-

---------

Fig. 5.13 an INA turbogenerator

control problem

Fig. 5.13 above shows an INA turbogenerator

turbogenerator

system
the

exciter

and

matrix,

having

off-diagonal

turbine/governor

input
an

the synchronous
systems

in

U (mechanical

vector

[G(s)]
elements of
SISO

well-known

regulators.
creates

which

However,

new plant transfer

methods

with

function

designing

of

field
and

torque

voltage).

of matrices

inputs
new

choosing

R and new outputs

off-diagonal

(5.4)
of the pre-compensator

[L(s)],

[Q(s)]
a

terms are much smaller

[G(s)].
to
compared

Z and a

Q(s) such that

the elements

and the post- compensator

the use

[L(s)]
and

[Q(s)] = [L(s)] [G(s)] [K(s)]

By carefully

The

feedback

stable
[K(s)]

the

function

are not zero, thus prohibiting

the addition

a new system

machine,

a transfer

(speed
Y
and an output vector
and terminal

voltage),

of

include

G(s)
the
system, where

The interaction

116

[K(s)]

found
be
of which the
can

than the diagonal

terms when

between [R] as inputs and [Z] as

is

outputs

therefore

greatly

reduced

and

dominant.

This

reduction

is

diagonally

[Q(s)]

is

also

said

referred

to

be

to

as

decoupling.
In applying

is always

feedback
A

matrix

decoupler

ICAD

[78],

design

obtained
designed

the

on

[80]
et. el
will
is a 3rd

Concordia
and

basis

of

[K(s)]

or

is to be used in the

the

INA

the

using

The machine

from DeMello

it was

[79].

developed

technique.

problem,

to either

gain values

by
Bollinger
system

turbogenerator
the

if higher

necessary
[F(s)],

control

lead
of
compensation

that the addition

noticed
[L(s)]

the INA to the turbogenerator

[11],

following

method

for

be analysed using

order

linear

model

decoupler
the
and
loading

conditions

was
and

data
:
machine
(i) loading

conditions

l pu.
P=1.0 pu; Q=0.5 pu; and X,
_=O.
data
:
machine
Xd

=1.6 pu;

M=3 pu;
From
is

Xq=

55 pu;
.

Xd=0.32

D=1.0 pu; and T60=5 pu.

Appendix

calculated

using

(3), the machine

the

machine

transfer

data

above.

117

function

and the

matrix

loading

[G(s)]

conditions

The

decoupler

(a pre-compensator)

in a transfer

function

matrix

form is :
0.412(0.ls+ 1)
(1235s+l)

P(S)

3.48(s+ 77.6)

(5.5)

(0.1s+ 1)(S2+ 0.33s+ 226)

in a block diagram,

the plant G(s)

decoupler P(s) is shown
the
and
,

in Fig. 5.14 below

AT-A

A r.-%

MM

Fig.

5.14 a Turbogenerator

system with

decoupler
a

From the machine data, the plant would not be expected to

have

a y(s)

conditions

function

high,
are

near

the effect

very strong transmission

After

applying

(1,0)

the

point;

and

although

high
this
of
real power

loading

is offset

by a

line (tie line reactance Xe=O. 1 pu)

the new plant Q(s) is

the pre-compensator,

].
[1
[921
911 912
Q(s) = G(s) P(S) =
D2
922

DI]
1

118

q12]
=[qll
q2l

q22

(5.6)

The y(s) function

of the original

pre-compensated
that

the

the

cross-coupling

0.8 ...................

......................................

0.6 ...................

.................

0.2 ................
0 ............

in

effects

it
making
more diagonally

0.4 .........................

has

plant

a very

means that, the pre-compensater

which

of the

are shown in Fig. 5.15 ; where it is observed

plant

pre-compensated

function,

plant and the 'y"(s) function

the

original

much

smaller

has indeed

nominal

y(s)

reduced

plant

model

dominant.

......................................
.........................................

.........

.......

..................

... ...........

....................................

..................

..........................

..................................

1: .................

..................

............

-0.2 ...........

.......

-0.4 .............

.......................

............................................

-0.6 ..................

................................

-0.8-..., ..............

...................

-1

-1.5

............
.

......

........................

...................

-0.5

nominal

Nevertheless,

......................

........................
0

0.5

RealAxis

plant ,( -------Fig. 5.15 Nyquist

the Bode plot

) compensated
plot of y(s)

plant

of the y(s) of the uncompensated

frequency
in
lack
5.16
the
Fig.
of robustness
shows some
plant
(12-18)

rad/sec,

where the switch-back

5.17.

119

characteristic

occurs,

range
Fig.

0
,o-20
0
-40
-60
-80
10

100
FREQUENCY

10-1

10

...........

-90
w
CO)
1:
cL -180

-270

.........

....

....

... .....

. ...

.......

...

...........

...........

............

. .............
.........
..........................
.

10 -2

10-1

10

............

10

FREQUENCY

Fig. 5.16 Bode plot of y(s) for nominal plant

is this pre-compensation

How

system stabiliser
answer

in

lies

synchronous

(PSS) effect
analysing

machine

before

compares to the power

effect

discussed earlier

the

electrical

in Chapter 4 ?. The

subsystem

922

and after the application

of

the

of the pre-

compensator.
The Bode

plot

(of
the original
of 922

is
it
observed
where

plant)

that the switch-back

is shown

in Fig.

characteristic

is not that

has
because
This
the
a relatively
machine
is expected
severe.
synchronous
(see

is
transmission
to
connected
a
strong
and
reactance

Section

4.3,

Chapter

4).

The Bode

120

plot

of

5.17,

small
line

the electrical

subsystem
together

of the pre-compensated
922 of the original

with

plant

q22

is

shown

in Fig.

plant.

0
-20 .......

........

.... ............

-40 .......
..........

-60

.......

......

............

........ ........

........

..........

.......

...........
........
...........................

0
10

-2

lo-

10

10

FREQUENCY

0
-30 ...............

......................

..................

UJ -60 .............
::::::::::::.:..... ......................
co
IL

..................

-90 ....................
-120 ............

.............................

....
:mI--,

-150,1

................

-,
lo-

10

111---.

10

id,
FREQUENCY

5.17 Bode plot of 922 for the nopminal

Fig.

plant

0
..........

-20

.........

...

im
-40

.................

.............

-60

...............
......................................................

L-80
10

id,
FREMENCY

10-1

::::..

-30 ............

-- .:. .;

---:............

..........

Lu -60
Go

..........
.... .......

-90 ............

......................

-120
-9

10

16-1
) 922,

Fig.

10,

(-

FREWENCY

----'

10

+
P912
922
-4

5.18 Bode plot of 922 and 922+P912

121

5.189

It is clear that the switch-back

has been "buried"
much

sought

without

deteriorating

bandwidth

direct

has a bandwidth

That

rotor

speed

closed

used to

the

control

and this loop has a

IEEE-1s
the
type
of

of

is used to control

the

because
of the presence of a
and
loop

the exciter/voltage

is used here for the sole purpose of disturbance

is,

the

input
system

the synchronism

loop",

characteristic

the exciter/voltage

while

).
rad/sec.
excitation

An exciter

(DGR) in the exciter,

(the idswitch-back"

10 rad/sec,

loop

5
rad/sec.
under

The decoupler
rejection

is

[Ke(l+sTf)/(l+sTf)(l+sTe)+sKeKfI

gain reduction

to be used in the
the

[1/41+sTgXl+sTt)],

less than 0.8 rad/sec.

loop
"exciter/voltage
second

the very

system.

turbine/governor

loop of the form

torque/speed

thus permitting

of 50 pu or more

conventional

very small,

albeit

gain

of the turbogenerator

form

high

loop

performance

the

by the pre-compensator,

after

exciter/voltage

characteristic,

loop has a bandwidth

decoupler

on the rotor

frequency
occurs at

reduces

the

effect

above

less than

of

the

speed output thus maintaining

the
the
of
response
the
step
power system; and
of

output

as affected

by

the

excitation

system

input

is

transmission
for
to
5.19
a
strong
the
Fig.
connected
in
system
shown
line with and without

the decoupler,

122

for
5.20
in
Fig.
the
system
and

connected

to

weak

transmission

line

with

and

without

decoupler.
It is obvious

that, the decoupler

did indeed reduces the disturbance.

0.02.

-0.01

3
Time (sec. )

) un-compensated
Fig. 5.19 Effect

(plant ,
----

compensated
plant
_

"strong
tie-line
of speed on volatge

case"

0.04
A
0.03---

.........

...............

.................................................

..............

0.02

0 0.01

-0.01

(sec.
)
-nme

) un-compensated
Fig. 5.20 Effect

plant ,(----

-)

compensated plant

"weak
tie-line
on
volatge
speed
of

123

case"

the

5.4- Non-diagonal

control

As mentioned

[36],

in Chapter

the advantages of non-diagonal

and there is no improvement

diagonal

feedback
of

example

Section

of

plant

(pre-compensation),
illustrate

in structure

inputs
5.3

in

and

plant

4,

non-

by swapping

attained

Nevertheless,

outputs.

Chapter

from

in using

Control

the
the

a decoupler

option

also

in
(postPSS
the
great advantages
compensation).
using

structure

specified
specifications

however,

conditions

strongly- coupled
pu

reducing

the

of

model

(high

reactance

the interconnection

damper
windings
number of
function.
the y(s)

real power)

possible;

e. g.

line

And with

in the synchronous
the y(s) function

then the system exhibits

neither

124

be

should

between the channels),

if

to a

and/or
used

(so,

increase
also an

machine will

nowhere

excessive

high
a

(leading

is required

machine

control

function
a small y(s)

structure,
always

its

and

then a strong transmission

system),

synchronous

is

due to the well-

are mainly

turbogenerator

its
By
:
well-defined

a turbogenerator

operating

point,

that

in

are not great,

or robustness

benefit
some

shows

These advantages,

small

to

by examples

feedback control

beyond

control

assignment

for

2 and illustrated

reduce

(1,0)
the
near

phase sensitivity

nor

excessive

diagonal

structural

therefore

sensitivity,

On the other hand, in turbogenerator

exists

between

characteristic

requirement

non-

a considerable

only

in

bandwidth

(the

gov/speed

the

latter

the exciter/voltage
frequency

loop

loop,

hence eliminating

from
this channel.
even

125

is

not

separation
and

the

high

use a Direct

Gain

requiring

to ensure

is well below the natural

mode of the turbogenerator

occur),

specification,

bandwidth)

channel

channels
with

control

Some power system utilities

gain crossover

of the mechanical
back"

two

control.
(DGR)

reduction
channel

the
loop),

exciter/voltage

performance

(high

control

for both channels,

requirement

additional

increase
the overall sensitivity.
not

element will

high -performance

the

that

the

frequency

(where the "switchthe high -performance

Chapter

Conclusions
:

Introduction
Excitation/governor

control with power system stabilisers (PSS) has been

immensely successful for the control of turbogenerators

infinite
to
connected an
bus or as part of a multi-machine power system.This research provides a thorough
multivariable analysis of a single turbogeneratorsystem connectedto an infinite
bus using the control engineering analysis framework known as Individual Channel
Analysis and Design (ICAD). This emphasis on the multivariable nature of the

turbogenerator system, as compared with conventional power system analysis,

interesting
interpretation
(linearised)
the
provides a very
new
of
smaH-signal
dynamical characteristics of a single machine connected to an infm*ite bus.
Moreover, complete frequency-domain insight into the dynamic effects of machine

is
by
their explicit parameterisationin the
and system parameters made possible
is
2-input
2-output
multivariable system.
multivariable structure of what a

6.2- The Primary Result

The main results concern the basic control function of the PSS and what
is
it
First,
it
that
the
to
shown
since
out
achieve.
sets
control specification

frequency responseof the relatively slow govemor/turbine channel rolls off above
is
dynamic
turbogenerator
the
system
1 rad/sec, the
overaH
performance of
in
by
the
the
typical
determined
that
922(S)
of
electrical
subsystem
essentially

126

exciter frequency range of interest of 1 rad/sec to 10 rad/sec, [see Fig. 4.3].

Focusing therefore on the electrical subsystem
it
is
922(S) 9
observed that its

frequency responsepossessesthe extremely awkward switch-back characteristicof

(lightly
damped left half-plane pole pair) closely followed by an
a resonance

inverted resonance(lightly dampedleft half-plane zero pair) round 7 rad/sec.

So, taking the above electrical subsystem frequency-response characteristics
into account, what are the control specification options with or without the

?
PSS
If
additional use of
set point regulation of the tenninal voltage and shaft
low
the
then
speed are
only control requirements,
excitation channel
a relatively
bandwidth of about 3 rad/sec suffices, weR short of the awkward frequency
characteristic

at

7rad/sec.

PSS

is

not

required.

However,

the

sensitivity/disturbance rejection properties of the closed-loop excitation channel

first
is
This
be
the
option and was the option exercisedwith
relatively poor.
would
before
the
reactance
synchronous
smaRer
per
and
unit
smaffer machines
introduction of the PSS in the early 1960's [ 11].
A second control option is to seek an excitation channel bandwidth of up to 10

is
to
strong
voltage
Provide
control
requirement
a
this
with
compatible
rad/sec;
damped
LBP
lightly
But
frequency
the
this
zero
range.
disturbance rejection over
frequency
in
the
the
prevents
7
Of
922(s),
response
rad/sec, shown
pair round

PSS
This
discussed
directly
the
is
where
above.
as
option
the
second
attainment of

127

comes into play in a most ingenious way. 71bePSS transfer function multiplies the
benign 912(s) subsystem transfer-function
(no
lightly damped LIHP zeros),
element
so causing

e electrical subsystem 922(S)to be amended to 922+P912-In this way,

the inverted notch characteristic of the PSS removesor dominatesthe problematic

switch-back characteristic of the LBP pole-zero pairs round 7 rad/sec.In effect, an
is
artificial system output created which is the weighted sum of the terminal voltage
output and the speed output. At low and high frequency, the artificial output is
essentially terminal voltage; at frequencies round that of the LBP pole-zero pairs (7

is
dominated
is
longer
by
There
the
the
rad/sec),
artificial output
no
speedoutput.
first
bandwidth,
there
the
any restriction on excitation channel
as
option,
was with
needed to provide

the required dynamical

perfomiance

Of 922+P912 over the

frequency range 0 to 10 rad/sec; the price paid in tenns of the change of transient

is
7
the
tenninal
voltage round rad/sec small.
responseof
The PSS is shown to take the form of a non-diagonal (cross-coupled)system
for
The
the
terminal
the
paid
price
speed
outputs.
and
voltage
post-compensatoron
increase
in
leads
is
to
that
PSS post-compensator
a general
post-compensation
increase
in
it
is
PSS,
In
that
the
the
established
of
case
uncertainty.
model
system

)
(0
10
interest
frequency
in
to
the
rad/sec.
of
range
system model uncertainty
.
is
loading
not
excessive.
conditions,
typical
under

128

This control interpretation as to the function

of the PSS and the control

specification it sets out to achieve in the face of a problematic machine dynamical

characteristic is strikingly

different from that afforded by conventional

synchronising and damping torque analysis [24]. Interestingly, it is however

supported by one other multivariable analysis carried out on the turbogenerator

[281,
frequency-response
the
system
using
methodsof Rosenbrockand MacFarlane
and pole-zero analysis. A similar mutlivariable control analysis to that carried out
here could also be conducted into the use of stabilisers based on acceleration or
electrical power feedback signals insteadof the speedfeedbackbasedPSS[ 15].
A third and final control option consideredis the use of excitation/governor
control

only

with

swapped

input-output

system pairings

(system

pre-

While
has
this
third
the advantages that a high bandwidth
compensation).
option

for the excitation channel can be used and the actual outputs are directly controlled
from
PSS,
does
disadvantages:
lack
two
the
the
option
suffer
practical
unlike with
in
face
failure
loop
input
integrity
highthe
the
of
actuator
and
of
of system
frequency signals to the governor. Consequently, the use of the conventional PSS

remains the preferred option.

6.3- Some Other Results
Other results of the multivariable analysis of the turbogeneratorsystem are as
fonows

129

(a) It confirms that the system

suffers an increasinglack of stability robustnessto
system uncertainty for increasing levels of real power P, in particular at heavy
loading conditions in excessof P=0.9.
(b) It confirms that the system suffers an increasing lack
of stability robustness to
system uncertainty for increased values of synchronous reactance. For values of

in
synchronousreactance the usual range 1.6 pu to 2.6 pu, this is shown not to be a
problem.
(c) For systems with a given synchronous reactance at a given operating condition,

it is shown that the multivariable structuresof the 3-rd order and 6-th order smallsignal models are essentially the same, coffoborating the widely held view that a 3-

is
for
design.
rd order model adequate control
(d) The multivariable structure of the turbogeneratorsystemis encapsulatedby the
is
function
This
a scalar complex-frequency
y(s).
so-called multivariable structure
function,

infonnation
whose polar plot provides graphical

robustness to

system uncertainty,

in

on the stability

this case, the conventional

use of

for
3-rd
PSS.
Furthennore,
turbogenerator
a
order
control
with
excitation/govemor

by
function
is
all
explicitly
parameterised
y(s)
the
structure
multivariable
model,
dynamics.
describing
the
machine
the key system parameters

6.4- Future Work

ICAD,
framework
provides a complete
The multivariable analysis
chosen,
for
been
has
PSS
so successful
insight into why excitation/govemor control with

130

the control of turbogenerators connected to infinite bus. ffie

multivariable analysis
justifies treating the turbogenerator
system as a pseudo-SISO system where the
govemor loop is first closed and the exciter loop is treated as a SISO system for the

prime purpose of rejecting voltage disturbances.While the analysishas focused on

the use of the PSS to remove local turbogeneratormode oscillation, due to an
awkward frequency responsecharacteristicround 1.2 Hz, it is readily extended to
the use of the PSS to remove the slower inter-area mode oscillations in the
frequency range 0.1 to 0.7 Hz [14] and the common low-frequency swing in the

0.02
0.08
[16].
Hz
Likewise,
filters
to
the
to remove rotor
range
use of notch
torsional oscillations [13] and/or the use of additional compensationelements are
framework
in
the
a transparent and physically
all readily accommodatedwithin
itself
lends
Furthermore,
to an examination of the
the
approach
meaningful way.
for
PSS
different
the
the
tuning/gain
parameters of
scheduling strategies
effect of
in
frequency-domain
the
the
excitation channel
characteristics of
small-signal
on

the neighbourhood of different systemoperatingpoints.

The multivariable analysis framework itself is not confined to two-input two-

interest
it
be
in
future
Also
[681.
the
to
analyse
of
would
work,
output systems
for
integrated
systems.
power
multi-machine
control
schemes
of
performance

have
is
desire
deserve
to
the
a governor channel
Another area which
someattention
fast
due
I
technology
to
the
higher
of
than
growing
ever
bandwidth
rad/sec,
with

govenung.

131

Appendix

I KIMEBARK

1: Bibliography

EDWARD W. 'Power system stability Vol. 1,11,111' John Wiley

,
.

and sons Inc., N. Y. 1956.

2 ALDRED, A. S. and SHACKSHAFT, G.: 'A frequency-responsemethod for the
predetermination of synchronous machine stability' Proc. IEE, Vol-105A, 1959,
2
pp
3 CONCORDIA, C. and BROWN, P.G. : 'Effects of trends in large steam turbine
driven generator parameters on power system stability' IEEE Trans PAS-89
1970, pp 2211
4

CONCORDIA,

C.

Steady-state stability of synchronous machine as affected

by voltage regulator characteristics'AIEE., 1944 Vol-63, pp 215-220.

5 SCHLEEF, F. R. and WHITE, J.H. :' Damping of northwest-southeast tie-line

PAS-85,1963,
1239-1247.
IEEE
Trans.,
pp
oscillations: an analoguestudy'
6 BERGEN, A. R. 'Power system analysis' Prentice-Hall Inc. 1986

7 KIMBARK

EDWARD W, : 'Improvement of power system stability by

,

PAS-88,
5,1969,
773-778
IEEE
No.
Network'
in
trans,
the
pp
changes
1&2'
'Two-Reaction
H.
trans.
R.
theory
PARK,
pt
machine
of
synchronous
:
AIEE

1929, Vol-48, pp 716-730.

132

DANDENO, P. L., KARAS, A. M., McCLYMONT,

K. R., and WATSON, W.:

'Effect of high speed rectifier excitation systems on generator

stability limits' IEEE

trans., 1968, PAS-87, pp 190-201.

10

SCHEIF, F. R., HUNKINS, H. D., MARTIN, G.E., and HAFFAN, E.: 'Excitation

improve
line
to
control
power
stability' ibid., 1968, PAS-87, pp 1426-1434.
11 DEMIELLO, F. P., and CONCORDIA, C. : 'Concepts of synchronous machine
by
as effected
excitation control' ibid., 1969, PAS-88, pp 316-328.
12 Ll WANG 'A

damping
comparative study of
schemes on damping generator

IEEE
Trans.,
PWRS-8,
No. 2.1993, pp 613-619.
oscillations'
13 KUNDUR, P., LEE, D. C., and ZER*MLDIN, H.M. : 'Power system stabiliser
for thermal units: Analytical techniques and on-site validations' IEEE trans., 1981,

PAS-100, pp 81-95.
14 KUNDUR, P., KLEIN, M., ROGERS, G.J., and ZYV; NO, M. S.: 'Application

IEEE
for
enhancementof overall system stability'
of power system stabilisers
trans., 1989, PWRS-4, pp 614-626
15 LARSEN, E.V., SWANN, D.A.: 'Applying power system stabilisers pt. I, H
3017-3046.
PAS-100,1981,
Trans,
IEEE
HI'
pp
and
16 GRONDIN,
CHAWAGNE,

R.,
R.,:

KAMWA,
'An

I.,

SOULEERES, L.,

for
design
PSS
to
approach

POTVIN,

J.,

and

transient stability

low-frequency'
damping
the
common
of
improvement through supplementary
954-963.
PWRS-8,
1993
pp
IEEE trans.,

133

17 GIBBARD, M. J. 'Robust design of fixed-parameterPSS

over a wide range of
operating conditions' IEEE trans., 1991 PWRS-6, pp 794
,
18 CHEN, S. and MALIK,

O.P. : TSS design using g synthesis' IEEE trans., EC-

10, No. 1,1995, pp 175.

19 El-MATWALLY,

K. A.

; HANCOCK,

G. C ; and MALIK,

O.P.

'Implementation of a fuzzy logic PSS using a micro-controller' ibid, EC-11, No. 1.

1996, pp 91-117.
20 YOUNG-MOON

PARK and Y. LEE 'A

fuzzy
PSS
self-organising
using

ARMA model' ibid, EC- 11, No.2,1996, pp 442-447.

21 ANDERSON, P., and FOUAD 'Power system control and stability' (Iowa state
)
Ames.
1977,
Press,
university

22 YAO, Y. N.,: Electric Power System Dynamics' (Academic Press, 1983, New
York).
23

GUPTA, D.P., NARAHARI, N. G., BOYD, I., and HOGG, B. : An adaptive

damping
torque
the
of
a
negative
cancels
which
stabiliser
power system
109-117.
Vol-132.
1985,
IEE,
pp
generator'Proc.
synchronous
24

KUNDUR,

P.,: 'Power system stability and control' (McGraw-Hill

1994,

NewYork)
25

by
'Irnprovement
of turbogeneratortransient perfonnance
HUGHES, F.M. :

233-240.
Vol-120,
1973,
IEE
pp
means'Proc.
control
9

134

26

CHOW, J.H., HARRIS, L. P., OTHMAN, M. A.,

SANCHEZ-GASCA,
J.:
and

'Robust control design of power system

stabilisers using multivariable frequency
domain technique' Proc. of the 29th conf. on Decision and Control,

1993

Honolulu, HI, pp 2067-2073.

27 CHEN, S.; and MALIK, O.P. : 'H.

PSS
design'
Procoptimisation-based

IEE, Pt.C, Vol-142, No. 2,1995, pp 179.

28

HAMDAN,

A. M., and HUGHES, F. M. : 'Analysis and design of power

system stabiliser' International Journal of Control., 1977, Vol-26, pp 769-782.

29 ROSENBROCK,

H. H. : 'Design of multivariable control system using the

inverse Nyquist Array' Proc-IEE, Vol- 116, No. 11,1969, pp 1929-1936.

30 MacFARLANE, A. G.J. :' Multivariable control system design techniques: a
117,1972,
Vol1039-1047.
Proc-IEE,
tour'
pp
guided
31 HANSON, C. W. ; GOODWIN, C.J.; and DANDENO, L. P. :'

Influence of

intersystem
in
oscillations'
excitation and speed control parameters stabilizing
IEEE Trans, PAS-87,1968, pp 1306-1313
32 SCHIER, R.M.; and BLYTHE, A. L. : 'Field tests of dynamic stability using a
PAS-87,1968,
Trans,
IEEE
stabilising signal and computer program varifications'
315-322.
pp
33 OREILLY,

J., and LEITHEAD,

W. E. : 'Multivariable control by Individual

1-46.
Vol-54,
Control,
Journal
1991,
International
pp
of
Channel Design'

135

34 LEITUEAD, WE and OREELLY, J: 'Perfonnance issues in

the individual
channel design of 2-input 2-output systems : pt 1 structural issues' ibid, 1991 Vol.

54, pp 47-82.
35 ROBERTSON, S., LEITHEAD, W.E., and O'REIILLY, J. 'Graphicafly based
multivariable frequency domain package' Proc. EURACO Workshop on Recent
Results in Robust and Adaptive Control, September1995,Florence,pp 11-14.
36 LEITHEAD,

W. E., and OREELLY, j. : 'Performance issues in ICD of 2-input

2-output system pt-3 :Non-diagonal control and related issues' International

Journal of Control, 1992, Vol-55, pp 265-312.

37 LICEAGA-CASTRO,

J.; VERDE, C. ; O'REILLY, J., and LEITHEAD, W. E:

' Heliocopter flight control by ICD'

38 LICEAGA-CASTRO,

design'
system
control

Proc-IEE, pt. D, Vol-142,1995, pp 58-72

E.; and VANDERMOLEN,

G. : 'A submarine depth

International Journal of Control, No. 1 1995, Vol-61, pp

279-308.
39 DOHERTY, R.; and NICKLE, C. : 'Synchronous machine I& H' MEE Trans.,

Vol-45,1926, pp 912-942.
40 SHACKSHAFr, G.; and HENSER, P. :' Model of generator saturation for
Vol-126,
8,1979,
759-763.
Proc-IEE,
No.
in
pp
use power system studies'
41

in
'Current
force
power system
task
usage and suggested practices
:

1,
1,1986,
ECNo.
Trans,
IEEE
for
pp
synchronous
machines'
simulation
stability

77-93

136

42 CONCORDIA,

C.: ' Synchronous machines: theory and performance' John

Wiley and sons Inc. N. Y. 1951

43 BUSBY, E.L.; IH[UIRLY,J.; KEAY, F. W.;
and RACZKOWSKI,

C. : 'Dynamic

stability improvement at Monticello station-Analytical study and field tests' IEEE

trans., 1977 PAS-98, pp 889-897.

,
44 DANDENO, P. ; SCHULTZ, R.P. :'

Effects of synchronous machine

modelling in large-scale system studies' IEEE Trans., PAS-92,1973,

pp

574-582.
45 AHSON,

S.I. : 'Multivariable

control of turboaltemator'

Ph. D. Thesis,

University of Sheffeild, UK, 1975.

46 SCHULTZ, R.P. : 'Synchronous machine modelling' in IEEE Publication
Adequacy and Philosophy of Modelling SystemDynamic Performance, 1972
47

DAVISON,

E. J., and RAO, N. S. : "Ibe optimal output feedback control of a

2123-2133.
1971,
PAS-90,
trans.,
pp
synchronousmachineIEEE
48 F[EFFERON, W. G., and PMILEPS,R.A. : 'Effect of a modem voltage regulator
PAS-71,
1952,
AIEE
large
trans.,
turbine
generator'
on under-excited operation of
692-697.
pp
VolProc-IEE,
'General-purpose
G.
turboalternator
SHSCKSHAFT,
49
model'
110,

No. 4,1963, pp 703-713

137

50 AHSON, S.I., and NICHOLSON, H.

'Improvement
:
of turbo-altemator
response using the Inverse Nyquist Array methods' International Journal of
Control, 1976, Vol-23, pp 657-672.
51 LAUGHTON,

M. A. 'Matrix

analysis of dynamic stability in synchronous

multimachine systems' Proc-IEE, Vol-113,1966, pp 325-336.

52 AGGOUNE, M. E., BOUDJEMA, F. and BENSENOUCI : 'Design of variable
,
structure voltage regulator using pole assignement technique' IEEE trans., AC-

39, No. 10,1994, pp 2106.

53 SAIDY, M., HUGBES9 F. 'Block diagram transfer function model of generator

including damper winding' Proc-IEE, Vol-141, No.6,1994, pp 599-608

54 BAYNE, J., KUNDUR, P., WATSON, W. 'Static exciter control to improve
transient stability' IEEE Trans. PAS-94, No.4,1975, pp 1141-1146
55

Committee Report 'Excitation system models for power system stability

PAS-100,1981,
Trans.
IEEE
pp494-509.
studies'
56 WOODWARD,

J. 'Hydraulic-turbine

transfer function for use in governing

424-426.
3
115,
No.
1968,
VolProc-IEE
pp
studies'
57

Committee Report 'Dynamic models for steam and Hydro turbines in

1904-1915.
PAS-92,1973,
Trans.
IEEE
pp
power system studies'
58 COOWAR,

F., MAGDY,

M., and GRAINGER, W.: 'Decoupling negative

Gain
Reduction
Dynamic
in
through
measures'
damping signals
a power system
1279-1284.
3,1992,
No
PWRS-7,
pp
IEEE-Trans,

138

59 AHSON, S.I., and NICHOLSON,

H. :'Inverse Nyquist Array design with

minimum sensitivity'Proc. IEE 1977,123, pp 457-461.

,
60 SALEH, A., and EI-SHARBIM, M. K. 'Optimal design of an overall controller
of saturatedsynchronousmachine' IEEE Trans. PAS-102,1983, pp 1651-1657
61 EI-SHARBINI,

M. K., and MEHTA, D. 'Dynamic system stability pt I' IEEE

Trans. PAS-92,1973, pp 1538-1546

62 HAMDAN,

A. M. A. : 'Vector frequency response control applied to turbo-

Ph.
D.
Thesis,
1977,
UK
University
Machester,
UMIST,
alternator control'
of

63 HARELY, R.G.,

LAHOUD, M. A., and SECKER, A. :'

Optimal and

System
Electric
turbogenerator'
research,
models'
power
multivariable control of a
Vol-10,1986, pp 35.
,eA

64

AHSON, S.I., and HOGG, B. W. : 'Application of multivariable frequency

Control,
Journal
International
turbogenerator'
of
response methods to control of
1979, Vol-30, pp 533-548.
65 HUGHES, F. and HAMDAN. A. M.

'Design of turboalternator excitation

Proc-IEE,
frequency-response
methods'
mutfivariable
using
controllers

V01-

123, No. 9 1976, pp 901-905

66

FENWICK,

in
'Review
F.
W.
trends
WRIGHT,
excitation
D. R., and
of
:

413-420.
123,
Vol1976
ibid.
development'
pp
future
,
systems and possible
,

139

67 VOURNAS, C.D., and PAPADIAS, B.C. 'Excitation

in
:
control schemes the
Hellenic

Interconnected

system for

the improvement of

dynamic
system

perfonnance' CIGRE paper 39-50,1984, Paris, France

68

LEITHEAD,

W. E., and OREELLY, J. : 'm-Input m-Output feedback control

by 1CD pt- 1: structural issues'International Journal of Control, 1992 Vol-56, pp

,
1347-1397.
69 EI-SHARBINI,

M. K. and El-SERAIFI

A. M. : 'Analysis
,

performance of a staurated machine and analog simulation'

dynamic
of

IEEE Trans. PAS-

101, No. 7,1982, pp 1899-1906.

R. T., and KIMBARK,

70 BYERLY,

Systems' IEEE Press Book,

W. :' Stability of large Electric Power

N. Y. 1974.
J.W., KEAY,

F., SOUTH

W. and
,

71 MILLMAN,

T., SKOOGLUND,

RACZKOWSKI,

C. 'A high initial response brushless excitation system' IEEE

Trans. PAS-90,197 1, pp2089-2094.

72

HURLY,

J.D., and BALDWIN,

M. S. 'High-response excitation systems on

11,1982,
No.
PAS-101,
Trans.
IEEE
turbine generators: a stability assessement'

4211-4221.
pp
'Power
K.
M.
EI-SHARBM,
M.
system control
and
73 SAKAR, M., SOLIMAN,
Computer
to
conditions'
operating
sensitivity
eignvalue
mininlurn
with
139-146.
13,1987,
No.
Vol-13,
pp
Electrical Engineering,

140

74 CHOW, J.H. and SANCHEZ-GASCA, J.J.: 'Pole-placementdesigns Power

of
Systems Stabilisers' IEEE trans., PWRS-4. No. 1,1989,
271-278.
pp
75 WANG, H.F., SWIFT, F.J., HOGG, B.W., CHEN, H.
TANG,
G.:
'Ruleand
based variable-gain power system stabiliser' Proc. IEE,
1995,
Vol-142,
Nol,
pt-C,
29-32.
pp.
76 KLIEN, M.; LE, LX;

ROGERS, G.; FARROKHPAY,

S.; BALU, N.: ' H.

damping controller design in large power systems' IEEE trans., PWRS-10,

No.

1,1995 pp 158-166.
77 SAIDY, M., HUGHES, F. 'Performance improvement of a conventional PSS'
Electrical Power & Energy Systems,Vol-17,1995, pp 313-323.
78 LIMBEER

D., and HARLEY,

.

R. G. : 'Synchronous machine stability using

compsite govemor and volatge regulator models' Electric power System research,

Vol- 11977, pp 97.

79 PAPADOPOULOS, D. and PARASKEVOPOULOS, N. ' Design of a
Vol-132,1985,
Proc-IEE,
for
pp
an unregulated synchronousmachine'
controller
277-280.
80 VVFNYAN Gu, and BOLLINGER,

K. E. 'A

decoupler design method for

EC-4,1989,
54-61.
Trans.
infinite-bus
pp
system'IEEE
machine
synchronous

141

Appendix 2: Nomenclature
Voltages, currents andfluxes

ed: direct-axis annature voltage

eq

: quadrature-axisannaturevoltage

infinite
bus
eo :
voltage
field
: generator
voltage

fd

Eqo
I
fd

id:

iq:

xyfd

direct-axis
flux
linkage
to
: voltage proportional

field
current
: generator
direct-axis annature current
quadrature-axisarmaturecurrent
flux
linkage
field
:

Vt : generator terminal voltage

14fd

linkage
flux
direct-axis
:

Tq

linkage
flux
: quadrature-axis
: real power

: reactivepower
Reactances, resis nces
xd

direct-axis synchronousreactance

142

xq:

quadrature-axis synchronous reactance

direct-axis transient reactance

machine stator resistance
resistive part of per phaseparameter
B:
Xe

6:

reactive part of per phaseparameter

: machine stator reactance
rotor angle
angular position of direct-axis with respectto stator

(o : speed
wo : synchronousspeed
field
T",
open circuit time constant
do :
U,dz : effective field time constantunder load
The subscript zero denotes an initial operating condition. A quantities denotes
initial
operating point.
small changesabout an
Mechanical symbols
Tm : mechanical torque
inertia constant (6 s)
damping constant (3 pu)
Exciter symbols

k: exciter 91

143

Te

: exciter time constant

Vref : voltage input to the exciter

Governorl turbine
governor gain
Tg : governor time constant

Tt

: turbine time constant

Ug: speed input to the governor

144

Appendix3:

Derivation of the smaU-signal3-rd order synchronous

machine model of

Fig. 3.4 (equations (3.11) to (3.13))

Ile general equations describing the performanceof machine with no damper

infinite
to
and connected
an

windings

bus system, shown in Fig. Al,

are

[81,[221,[481:
(Al)

'yd - 'Yq

(A2)

eq " Vq + Wdo
sinb

ed 'i0+e
xeld - xe q0

(M)

+
+
eocos6
xe'do
elq

eX:
q

Wd " lfd - xd'd

(A5)

Tq = _x q'q

(A6)

T-mWfd

Efd

=1

Vt 2=

i. fd

ed

(x,

+TL

Vd'q

Vq'd

-x-,

(A7)

D(6 - 1)

)i,

(A8)
(A9)

*fd

(AlO)

+eq

is
to
the
rewritten
are
initial
equations
assignedand
A particular
operating point
forth
in
their
A
about
so
and
voltages,
currents,
changes
small
the
of
effect
consider
differential
linear
following
is
The
equations:
the
of
set
result
this
point.
values at
[*d
A&e d=

[AVqO 4

IA&lVq

(All)

145

[*q]
AWd

Aed "

lxeAd I-

Aeq =

[XeAQ

AVd

4"Wq

(M

AI

fd

[V&IA6]

(A12)

XeAiq +

+ Xelid -

(e.

cOs6o)Ab -

(eo

sinbo)A6+

[iqoxeA]

(A 13)

kdoxe4

(A14)

XdAd

(A15)

-" -XqAq

(A16)

b)AF)
=

sl.

AlVd
q-i

ATfd ` ''Ifd -

(xd

qo

+ 14fqo Ad

+ 'do ATq

(A17)

Iid
- xa),

(A18)

AEfd = AIM +T 6o'61*fd

AVt

eqo

d-o
= -!VtD

(A19)

Ae dV+

Ae q

(A20)

to

Infinite bus
Synch

X---0.4pu, R=G=B=O
Fig. Al Single machine connectedto an infinte bus [22]

146

The eight terms in brackets [equations (A 11) to (A 14)] are neglected;this is

justified, because the quantities neglected have negligible effect [48]. Next, all
variables except AVt , A6, AxVfd , ATm , and AEfd are eliminated leaving three
equations in three unknowns. AEfd and ATm are independent variables. The three
final equations in AVt, A6, and AiVfd are :

e.
sin,
ATfd
i,
+ Ab= AT. -
X, +X
_X,

1
'Vfd

+x

de

= -.

l+T

AE

_.
fd

+Xd'

Eqe cosb

6
ee sin

+X

X/ +X

-00+

x dde.-x

qde

do o0

X /d
1-

Ab

(A21)

(A22)

A
sin,

Xe + Xd

-Xe

eqo

eqo

xq

Xe

edo
xa
60
Ab
sin
eo cosbo --Vto
Vto
+
xe
xa

Avt =
AlVfd +x+x
Vto xa + xe
eq

(A23)

X, '

de+x T"
do
Xd + Xe

Tdz
where

(A24)

define:
if
and we
K, =

K2 '-:

K3

Eeqo

5(ee
cos
o0+
+X

x cqd

do

6
oosin

1_.

(A25)

X.
X"+

60
eo sin
xa +

(A26)

Xe

X/ +x
xd

(A27)

+X

60
eo sin (xd
K4 "'
x1d)
Xe
+
xa

(A28)

147

K5 = ___jq
Xe + xq

K6

ed
-0
Vto eoc( 3S60-

eq.

xe

Vto

/
Xd + Xe

xa

eqo

X + xe Vto

60
eo sin

(A29)

(A30)

Tben, the 3-rd order small-signal model, consistentwith equations(3.11) to (3.13)

Fig.
3.4,
is as fbHows :
or
(M

b)A6

AT
=
K3

AVfd =-.
1+ K3Tao

K2ATfd-

(AEfd

K, A6

(A31)

K4A8)
-

(A32)

AVt = K6A4ffd + K5A6

(A33)

The 3-rd order small-signal model of equations (A3 1) to (A33) for a single
synchronousmachine connectedto an infinite bus is defmed by the parametersKj
to K6 of equations (A25) to (A30). These parameters in turn are determined by the

initial currents, voltages and torque angle as follows

.
The initial currents, voltages and torque angle in steady-state are usually found
from a load flow study for a multi-machine system. For a single-machine infinitebus system, they can be calculated as follows [221.

Given real and reactive power P and Q and the machine terminal voltage Vto we
have :
)

P+jQ=('d

)*(ed+ jeq
+ji,,

(A34)

From equation (57) above,

'd ed

(A35)

+'qeq

148

and

Q=i

deq

(A36)

-iqed

Also we have
ed

(A37)

Xqlq

Eqo
eq =
- xdid
Vt 2=

ed

(A38)

(A39)

+eq

From thesefive independentequations,ed, eq, id, iq, andEqo canbe calculated.

Thus :

Pvt

(A40)

ed p2 + Q+
V(V
eq

vt2)2
xq

2)
t2

(A41)

-ed

ed

(A42)

Xq

(p

(Q i,
+ ed)

-'qeq)
ld

=
ed

or

(A43)

eq

(A44)

Eqo = eq + Xd'id

be
(rotor
60
(infinite
bus
to
angle) remain
The initial values of eo
voltage) and
[22]
define
Al,
Fig.
From
we
determined.
C, =1 +'RG - XB

(A45)

XG-RB

(A46)

149

whereX = xe and R=G=B

Vod=

Cled -C2eq-

Rd

V, = C, e, + C,eq-

Xid-

Alsowehave
=0.

+ Xq

(A47)

Rq

(A48)

Then
VVod

2+V

eo=

(A49)

oq

Vod

60

-1

= tan

(A50)

Voq

From (A40) to (A50) with Vt=1.0 pu, X---0.4 pu and R--G=B=O and a given P and

the machine parametersK, to K, are determinedusing equations(A25) to (A30).

150

AwDendix

4:

Derivation of 3-rd order modeltransfer-functionmatrix of Fig.

3.11 or equations (3.17) to (3.21)fiom equations(3.11) to (3-13)

:
From Fig. 3.4 -A(o = SA8

(A5 1)

Substituting (A5 1) into (3.11) and (3.12)

(Ms + D)A(o ATM K2 ATfd
=
-

Kl(oo

K3

A'ffd =AEfd 1+ sK3Tao

Aco

(A52)

s
Alo

(A53)

Substituting (A53) into (A52)

K2K3

K2K3K4

(Ms + D)Aco AT
=
-- 1+sKJ

AEfd

+
)

Ts
s(l+sK 3

Ao)-

KI(po

AO)

(A54)

becomes
which
(Ms+ D)
-(I

K2K3K4(Oo
s(l + sK3rdo)

Kjo)o
+

s1+

A(o = ATm -

K2K3_

AEfd
sK3rFdo

K2K3s
+ sK3rdo)s
ATm AEfd
den
den

(A55)

(A56)

where
(1
den = (Ms + D)(1 + sK3rdo)s + K, + sK3rFdo)O)o- K2 K3K4(Qo

(A57)

(3.18)
(3.17)
Therefore, equations
and
are obtained, namely
gl 1(s) =
(S)
912 =

(1

(A58)

+ sK3ydo)s

(A59)

-K2K3S

151

Now, substituting (A5 1) and (A53) into equation (3-13) :K3K 6_,
Avt =
1+sK 3T

K3K4K6(1)o
E
)
fd_ 41+sK 3Ts

Aw+

Aw

(A60)

or
K3K6

K3K4K6(1)0

Avt =
-AEfd
1+sK3TL

KjI+sKJ)(1)o 41+sK 3TL)

AO)

(A61)

Substituting (A56) into (A61)

y"5)
(KZ,
K6(N+D)s+ K, -KrY,
2
-I<
lAvt
den

wo
Agd

(1
+ sY,-P) - K, Y, Y,
AT.
den

(A 62)

Therefore :92,

(S)=K3[K6(MS+

(S)

9
21

D)s+

(KK6-K2K,

)(y). ]

)346]

-[Kjl+sK

"

*' -KKK

3Tdo

(A63)
(A64)

wo

3-rd
transfer-function
the
this
representationof
model
order
completes
and
Fig. 3.11 or equations (3.17) to (3.21).

152

Avvendix

The 3rd order model :a2l -

K,

a22

b2l

a23

32

a3l =-

,3L

TL

a33

KT

K6
C33 =

K5
C21 =

TL

K4

K2

The 6th order model :a 21 =-

a3l =-

a34

a4l

K,

a22 = a23=m ,

R
o).
Lfd

fd m,

R
co. Id
Lld

Llq

Llq

(ooR 2q
a61 =

L"
m,

1Lfd

L2q

R
co fd
0
Lfd

R
(oo ld

+ M3 L#
a45 =ads

1-

aqs

Llq

a53 '

Llq

+ n4Lffq

Lld

n2 L,, '
aqs,

a63

L2q

R
(a fd

L'
M4

a36 = _o
Lfd

m5Lds

Lfd

woRld
ads,

a54

a46

Lld

(ooRlq
Llq

. n3Laqs

L//
R
(oo lq
ags
n5 Lff _.
aqs L2q
Llq

o)0R2q
n2L9

153

aqs

L-`
ads

R2q
o).
Laqs,
@'
n,

a 26=-

+ M2 L4'
ads

La
n14 "d

o)oRld_

a56 --

, a25--

2 ads

Lld

Rlq
e).
L'
n,

Lfd

a35

a43

ads

-M

Lfd

Lfd

Lld

Rlq
(9.
a55 =

a33

acis

Lfd

Rlq
(o.
a5l =-

L'
R
e)0 fd
ads
M3 L'9 ads

woRld,
a44 =-

L"

a 24-

K23

K22

K21

K2

Kd

a64 --

L2q

n3L* aqs

LO
m5
ads

o). R2q
a65

L2q

Llqs
, n4L: qs Llq

b2l

b32

R2q
co,,,
a66 --

L2q

+ n5L//
L2q
aqs

Rfd
(o,,,
"2 Lad,,

I
edo(Llnl
e
(L//
(Lids
qo(Llml+n,
C21
+ MI
I
+R
-R
a)) eto
eto
aqs a))
ed.

C23

e to

Lln2+M2(L/

ds-R
a)-

[e
do Lln3+M3(L//
C2A "',,
eto

ed,, (Lln4+M4(L#
C25 "" -

-e

C26

to

ed. (Lln5+M5(Lo
-

eto

-qo (LIM2
-

Lfd

!! kS
ads -R a)-

ads-R a

-R.
ads

e to
eqo

Llds

e q.

Lqs

))+

eto

+ n2

(LIM3

eto

,Llq

e to

L2q

t
GLASGOW,
'Q'LMIBRARY-,,
UNIVERSftt

154

+R
a'/qs a

(L//
+ n3 aq,s+R

LIM4 - n4

Lo
!
))+
qo
, qs LIM5
.
-

(L

- n5

(Llqs

(Ul

+R

a))]

a)

+R
aqs
a)