Whole PDF

THE DYNAMIC ANALYSIS AND CONTROL OF A
SELF-EXCITED INDUCTION GENERATOR DRIVEN

BY A WIND TURBINE
by
Dawit Seyoum
A thesis submitted to
The University of New South Wales for the Degree of
Doctor of Philosophy
School of Electrical Engineering and Telecommunications
March, 2003
ii
CERTIFICATE OF ORIGINALITY
I hereby declare that this submission is my own work and that, to the best of my
knowledge and belief, it contains no material previously published or written by another
person nor material which to a substantial extent has been accepted for the award of any
other degree or diploma of a university or other institute of higher learning, except
where due acknowledgment is made in the text.
_______________________
Dawit Seyoum
iii
ACKNOWLEDGEMENTS
First, thanks be to God who gave me the physical and spiritual health to pursue my
Ph.D. study.
I would like to thank my supervisors Associate Professor M.F. Rahman and Associate
Professor Colin Grantham for their guidance and financial assistance throughout this
study.
Special acknowledgement is due to Mr. Doug McKinnon for proof reading the thesis
and for sharing ideas. I thank Daniel Indyk from Energy Australia for his assistance to
visit a wind power site. Thank you to the laboratory staff for their logistical support.
Thanks also go to my colleagues in the Energy Systems Research Group for their
suggestions, Mr. Baburaj Karanayil, Mr. Chathura Mudannayake, Dr. Enamul Haque,
Mr. Lixin Tang, Mr. Phuc Huu To and Mr. Phop Chancharoensook.
I thank my late father who encouraged me to go to school when I was a little boy and
my mother who raised me and helped me to go to school as a single mother.
Last, but foremost, thanks go to my family. To my wife Abeba, thank you for your
patience, understanding, encouragement and help, especially when I was spending most
of the time doing research. And thanks to my little daughter, Lwam, for your patience in
enjoying the little time that I had to spend with you.
iv
ABSTRACT
This thesis covers the analysis, dynamic modelling and control of an isolated self-
excited induction generator (SEIG) driven by a variable speed wind turbine. The
voltage build up process of an isolated induction generator excited by AC capacitors
starts from charge in the capacitors or from a remnant magnetic field in the core. A
similar voltage build up is obtained when the isolated induction generator is excited
using an inverter/rectifier system with a single DC capacitor on the DC link of the
converter. In this type of excitation the voltage build up starts from a small DC voltage
in the DC link and is implemented using vector control.
The dynamic voltage, current, power and frequency developed by the induction
generator have been analysed, simulated and verified experimentally for the loaded and
unloaded conditions while the speed was varied or kept constant. Results which are
inaccessible in the experimental setup have been predicted using the simulation
algorithm.
To model the self excited induction generator accurate values of the parameters of the
induction machine are required. A detailed analysis for the parameter determination of
induction machines using a fast data acquisition technique and a DSP system has been
investigated. A novel analysis and model of a self-excited induction generator that takes
iron loss into account is presented in a simplified and understandable way.
The use of the variation in magnetising inductance with voltage leads to an accurate
prediction of whether or not self-excitation will occur in a SEIG for various capacitance
values and speeds in both the loaded and unloaded cases. The characteristics of
magnetising inductance, L
m
, with respect to the rms induced stator voltage or
magnetising current determines the regions of stable operation as well as the minimum
generated voltage without loss of self-excitation.
In the SEIG, the frequency of the generated voltage depends on the speed of the prime
mover as well as the condition of the load. With the speed of the prime mover of an
isolated SEIG constant, an increased load causes the magnitude of the generated voltage
v
and frequency to decrease. This is due to a drop in the speed of the rotating magnetic
field. When the speed of the prime mover drops with load then the decrease in voltage
and frequency will be greater than for the case where the speed is held constant.
Dynamic simulation studies shows that increasing the capacitance value can
compensate for the voltage drop due to loading, but the drop in frequency can be
compensated only by increasing the speed of the rotor.
In vector control of the SEIG, the reference flux linkage varies according to the
variation in rotor speed. The problems associated with the estimation of stator flux
linkage using integration are investigated and an improved estimation of flux linkage is
developed that compensates for the integration error.
Analysis of the three-axes to two-axes transformation and its application in the
measurement of rms current, rms voltage, active power and power factor from data
obtained in only one set of measurements taken at a single instant of time is discussed.
It is also shown that from measurements taken at two consecutive instants in time the
frequency of the three-phase AC power supply can be evaluated. The three-axes to two-
axes transformation tool simplifies the calculation of the electrical quantities.
vi
CONTENTS
ACKNOWLEDGEMENTS....................................................................................... ...iii
ABSTRACT ........................................................................................................... ...iv
CONTENTS ........................................................................................................... ...vi
LIST OF FIGURES ................................................................................................... ..xii
LIST OF TABLES ..................................................................................................... .xix
LIST OF SYMBOLS ................................................................................................. ..xx
1 INTRODUCTION............................................................................................... 1
1.1 General ........................................................................................................ 1
1.2 Thesis outline............................................................................................... 4
1.3 Literature review ......................................................................................... 8
1.3.1 Self-excited induction generator........................................................ 8
1.3.2 Capacitance and rotor speed for self-excitation ................................ ..11
1.3.3 Representation of magnetising inductance........................................ ..11
1.3.4 Control of generated voltage and frequency...................................... ..13
1.3.5 Wind powered generators.................................................................. ..13
1.3.6 Cross saturation ................................................................................. ..15
1.4 References ................................................................................................... ..16
2 WIND POWER.................................................................................................... ..21
2.1 Source of wind............................................................................................. ..21
2.2 Wind Turbine............................................................................................... ..22
2.2.1 Vertical axis wind turbine.................................................................. ..22
vii
2.2.2 Horizontal axis wind turbine ............................................................. ..23
2.3 Power extracted from wind.......................................................................... ..24
2.4 Torque developed by a wind turbine........................................................... ..31
2.5 Tip-Speed Ratio........................................................................................... ..35
2.6 Power control in wind turbines.................................................................... ..36
2.6.1 Pitch control....................................................................................... ..38
2.6.2 Yaw control ....................................................................................... ..38
2.6.3 Stall control........................................................................................ ..39
2.7 Wind powered electric generation............................................................... ..40
2.8 Economics of wind powered electric generation......................................... ..41
2.9 Summary...................................................................................................... ..42
2.10 References ................................................................................................... ..43
3 THREE AXES TO TWO AXES TRANSFORMATION AND ITS
APPLICATION................................................................................................... ..44
3.1 Introduction ................................................................................................. ..44
3.2 General change of variables in transformation............................................ ..45
3.2.1 Transformation into a stationary reference frame ............................. ..46
3.2.2 Transformation into a rotating reference frame................................. ..51
3.3 Voltage measurement .................................................................................. ..53
3.4 Current measurement................................................................................... ..55
3.5 Power measurement..................................................................................... ..58
3.6 Power factor measurement .......................................................................... ..60
3.7 Frequency measurement .............................................................................. ..61
3.8 Measurement in a balanced non sinusoidal three phase system.................. ..63
3.9 Summary...................................................................................................... ..64
3.10 References ................................................................................................... ..64
4 INDUCTION MACHINE MODELING........................................................... ..66
4.1 Introduction ................................................................................................. ..66
4.2 Conventional induction machine mode ....................................................... ..67
4.3 D-Q axes induction machine model ............................................................ ..70
viii
4.4 Simulation of induction machine................................................................. ..74
4.5 D-Q axes induction machine model in rotating reference frame ................ ..86
4.6 Development of D-Q axes induction machine model with R
m
.................... ..87
4.7 Summary...................................................................................................... ..93
4.8 References ................................................................................................... ..93
5 DATA ACQUISITION AND DIGITAL SIGNAL PROCESSING................ ..95
5.1 Introduction ................................................................................................. ..95
5.2 DS1102 DSP board...................................................................................... ..96
5.3 Data acquisition........................................................................................... ..98
5.3.1 Voltage and Current measurement .................................................... ..98
5.3.1.1 Anti-aliasing filter................................................................. ..99
5.3.1.2 Voltage measurement ........................................................... 101
5.3.1.3 Current measurement............................................................ 102
5.4 Speed and angle measurement..................................................................... 103
5.4.1 Angle measurement ........................................................................... 105
5.4.2 Speed measurement ........................................................................... 107
5.5 Digital signal processing ............................................................................. 108
5.5.1 Digital filter ....................................................................................... 108
5.5.1.1 Infinite Impulse Response (IIR) filter .................................. 109
5.5.1.2 Finite Impulse Response (FIR) filter .................................... 110
5.5.1.3 Comparison of IIR and FIR filters........................................ 111
5.5.2 Digital filter design from analog filter............................................... 111
5.5.3 Implementation of a digital filter by approximating analog
filter circuits....................................................................................... 112
5.6 Summary...................................................................................................... 113
5.7 References ................................................................................................... 114
6 PARAMETER DETERMINATION FOR AN INDUCTION MACHINE.... 115
6.1 Introduction ................................................................................................. 115
6.2 Open-circuit and short-circuit test ............................................................... 117
6.2.1 Open-circuit test ................................................................................ 117
ix
6.2.2 Short-circuit test ................................................................................ 118
6.2.3 Induction machine with constant rotor parameters............................ 119
6.2.4 Induction machine with variable rotor parameters............................ 120
6.2.5 Results for DSP based parameter determination .............................. 125
6.3 Sensitivity study on variable rotor parameters ............................................ 137
6.3.1 The effect of combining measurement errors ................................... 138
6.3.1.1 Percentage errors ................................................................. 138
6.3.1.2 Combining errors.................................................................. 139
6.3.2 Induction machine parameters for analysis of measurement error..... 139
6.3.3 Statistical tools.................................................................................... 140
6.3.4 Simulation of parameter determination with measurement error....... 142
6.4 Summary...................................................................................................... 146
6.5 References ................................................................................................... 147
7 EXCITATION OF THREE PHASE INDUCTION GENERATOR USING
THREE AC CAPACITORS............................................................................... 149
7.1 Introduction ................................................................................................. 149
7.2 Model of self-excited induction generator .................................................. 151
7.3 Analysis of self-excitation process.............................................................. 153
7.3.1 RLC circuit characteristics ................................................................ 154
7.3.2 Conditions for self-excitation in induction generator........................ 156
7.3.2.1 Using matrix partition........................................................... 158
7.3.2.2 Direct matrix inversion......................................................... 162
7.4 Characteristics of magnetising inductance in induction machine ............... 164
7.5 Minimum speed and capacitance for self-excitation................................... 166
7.6 Magnetising inductance and its effect on stability of generated voltage .... 170
7.7 Onset of self-excitation when the SEIG is loaded....................................... 173
7.8 Simulation of self-excited induction generator ........................................... 175
7.8.1 The modelling of self-excitation process........................................... 175
7.8.1.1 Determination of initial conditions....................................... 175
7.8.1.2 The dynamic representation of self-excitation at no load .... 176
7.8.2 The dynamic representation of a loaded SEIG.................................. 186
7.9 Characteristics of wind turbine and its effect on generator output.............. 194
x
7.10 Effect of rotor parameters variation on self-excitation................................ 199
7.11 Summary...................................................................................................... 205
7.12 References ................................................................................................... 207
8 MODELLNG OF AN ISOLATED SELF-EXCITED INDUCTION
GENERATOR TAKING IRON LOSS INTO ACCOUNT............................. 208
8.1 Introduction ................................................................................................. 208
8.2 SEIG dynamic model including R
m
............................................................. 209
8.3 Characteristics of L
m
and R
m
........................................................................ 210
8.4 Analysis of SEIG including R
m
................................................................... 211
8.5 Simulation of dynamic self-excitation including R
m
................................... 213
8.5.1 Simulation of dynamic self-excitation at no load.............................. 213
8.5.2 Dynamics of SEIG during loading .................................................... 216
8.6 Summary...................................................................................................... 220
8.7 References ................................................................................................... 221
9 INVERTER/RECTIFIER EXCITATION OF A THREE-PHASE
INDUCTION GENERATOR............................................................................. 222
9.1 Introduction ................................................................................................. 222
9.2 Vector control .............................................................................................. 224
9.2.1 Rotor flux oriented vector control ..................................................... 225
9.2.1.1 Direct (feedback) flux oriented vector control ..................... 227
9.2.1.2 Indirect (feed forward) flux oriented vector control............. 231
9.2.2 Rotor flux oriented control with voltage as the controlled variable.. 232
9.2.3 Stator flux oriented vector control..................................................... 234
9.3 System description....................................................................................... 239
9.4 Establishment of reference flux linkage...................................................... 241
9.5 Details for the implementation of vector control ........................................ 243
9.5.1 Implementation of direct rotor flux oriented vector control.............. 244
9.5.2 Implementation of indirect rotor flux oriented vector control........... 245
9.5.3 Implementation of rotor flux oriented vector control with voltage as a
control variable.................................................................................. 246
xi
9.5.4 Implementation of stator flux oriented vector control....................... 247
9.6 Results ......................................................................................................... 248
9.7 Summary...................................................................................................... 254
9.8 References ................................................................................................... 256
10 FLUX LINKAGE ESTIMATION AND COMPENSATION IN
INDUCTION MACHINES ................................................................................ 258
10.1 Introduction ................................................................................................. 258
10.2 Theory of Integrator .................................................................................... 259
10.3 Numerical integrator.................................................................................... 263
10.4 Proposed integration offset adjustment ....................................................... 263
10.4.1 Strategy I - without input offset minimization ................................ 264
10.4.2 Strategy II - with input offset minimization .................................... 265
10.5 Stator flux linkage estimation with the proposed method........................... 265
10.6 Summary...................................................................................................... 267
10.7 References ................................................................................................... 268
11 CONCLUSIONS AND SUGGESTION FOR FUTURE WORK.................... 269
11.1 Conclusions ................................................................................................. 269
11.2 Suggestions for future work ........................................................................ 277
APPENDICES
A DETERMINATION OF INERTIA AND FRICTION COEFIENT OF
THE INDUCTION GENERATOR SYSTEM........................................ 278
B MEASUREMENT AND CONTROL SYTEMS HARDWARE............ 283
C DETAILS IN INDUCTION MACHINE MODELLING ...................... 289
C.1 Introduction ........................................................................................ 289
C.2 Relationship of parameters in steady state model and d-q model of
induction machines ............................................................................. 289
C.3 Expanded equations for induction machine modelling including R
m
.. 292
D LIST OF PUBLICATIONS...................................................................... 296
xii
LIST OF FIGURES
Fig. 1.1 Kooragang wind turbine generator, Newcastle, NSW, Australia ...............2
Fig. 1.2 Wind farm around San Francisco, California, USA (Photo 2002) .............3
Fig. 2.1 Vertical axis wind turbine...........................................................................23
Fig. 2.2 Horizontal axis wind turbine (a) upwind machine (b) downwind machine24
Fig. 2.3 Detail of a wind turbine driven power generation system..........................25
Fig. 2.4 Change of wind speed and wind pressure around the wind turbine ............ ..27
Fig. 2.5 Power coefficient versus V
2
/V
1
.................................................................... . 30
Fig. 2.6 Wind turbine output power to shaft speed characteristic curve................... . 31
Fig. 2.7 Air flow around cross section of a blade of a wind turbine......................... . 32
Fig. 2.8 Air flow around cross section of a blade during stall condition .................. . 32
Fig. 2.9 Wind turbine output torque to shaft speed characteristic curve................... . 33
Fig. 2.10 Detail of a twisted rotor blade...................................................................... . 34
Fig. 2.11 Cross section of a twisted rotor blade from tip to base................................ . 35
Fig. 2.12 Typical power coefficient versus tip speed ratio ......................................... . 36
Fig. 2.13 Histogram and Weibull function for the probability of a given wind
speed (data measured in 1m/s intervals) ...................................................... ..37
Fig. 2.14 Wind turbine control regions ....................................................................... ..38
Fig. 2.15 Power coefficient verses tip speed ratio under yaw control ........................ ..39
Fig. 2.16 Growth of wind energy capacity worldwide................................................ ..41
Fig. 2.17 Trend in the cost of electricity generated from wind energy....................... ..42
Fig. 3.1 Three-axes and two-axes in the stationary reference frame......................... ..46
Fig. 3.2 Three-axes and two-axes in the stationary reference frame with d-axis
and a-axis aligned ........................................................................................ ..49
Fig. 3.3 Steps of the abc to rotating dq axes transformation..................................... ..52
Fig. 3.4 Voltage vector and its component in dq axes .............................................. ..54
Fig. 3.5 Current vector and its component in stationary dq axes .............................. ..58
Fig. 3.6 Voltage and current vectors with their components in the stationary
dq-axes ......................................................................................................... ..59
xiii
Fig. 4.1 Stator side of the per-phase equivalent circuit of a three-phase
induction machine........................................................................................ ..67
Fig. 4.2a Rotor side of the per-phase equivalent circuit of a three-phase
induction machine....................................................................................... ..68
Fig. 4.2b Rotor side of the induction machine with adjustment ................................. ..68
Fig. 4.3 Per-phase equivalent circuit of three-phase induction machine neglecting
core loss........................................................................................................ ..69
Fig. 4.4 Per-phase equivalent circuit of three-phase induction machine including
core loss........................................................................................................ ..69
Fig. 4.5 D-Q representation of induction machine.................................................... ..71
Fig. 4.6 Detailed d-q representation of induction machine in stationary reference
frame (a) d-axis circuit (b) q-axis circuit ..................................................... ..72
Fig. 4.7 Experimental setup to find the characteristics of induction machine
in the motoring and generating regions ....................................................... ..75
Fig. 4.8 Variation of stator phase current for constant supply voltage and frequency
(a) Current and voltage when the rotor speed is varied from standstill to
twice the synchronous speed (b) detail of motoring region (c) detail around
the synchronous speed (d) detail in the generating region........................... ..77
Fig. 4.9 Relationship between phase voltage vector and phase current vector (a) in
the motoring region (b) between motoring and generating (at synchronous
speed) (c) in the generating region............................................................... ..78
Fig. 4.10 Induction machine torque, power and efficiency characteristics (a) torque
(b) electrical power (c) mechanical power (P
m
=e
m
T
e
) (d) efficiency ......... ..79
Fig. 4.11 Space vector angles measured with respect to the stator voltage space
vector angle for (a) stator current I
s
(b) stator flux linkage
s
(c) rotor current I
r
(d) magnetising current I
m
.............................................. ..80
Fig. 4.12 Magnitude of space vector for (a) stator voltage (b) stator current I
s
(c) stator flux linkage
s
(d) rotor current I
r
(e) magnetising current I
m
...... ..81
Fig. 4.13 Space vector diagram for stator voltage, stator current, rotor current,
magnetising current and stator flux linkage (a) during motoring mode
(b) during generating mode.......................................................................... ..82
Fig. 4.14 Stator current in the d
e
-q
e
axes of the excitation reference frame
xiv
(a) q
e
-axis current (b) d
e
-axis current........................................................... ..84
Fig. 4.15 Stator voltage in the d
e
-q
e
axes of the excitation reference frame (a) d
e
-axis
voltage (b) q
e
-axis voltage ........................................................................... ..84
Fig. 4.16 Magnetising current in the d
e
-q
e
(a) d
e
-axis magnetising current (b) q
e
-axis magnetising current.................. ..85
Fig. 4.17 Rotor current in the d
e
-q
e
(a) d
e
-axis rotor current (b) q
e
-axis rotor current ......................................... ..85
Fig. 4.18 Rotor current in different reference frames (a) rotor current in a rotating
reference frame that is rotating at the rotor speed (b) rotor current in the
stator (stationary) reference frame ............................................................... ..86
Fig. 4.19 D-Q representation of induction machine in the excitation (e
e
) reference
frame (a) d-axis circuit (b) q-axis circuit ..................................................... ..87
Fig. 4.20 D-Q model of induction machine including core loss represented by R
m
(a) d-axis (b) q-axis...................................................................................... ..90
Fig. 5.1 Block diagram for data acquisition and signal processing........................... ..95
Fig. 5.2 Hardware and software system configuration.............................................. ..96
Fig. 5.3 Block Diagram of the DS1102..................................................................... ..97
Fig. 5.4 Voltage measurement system (a) voltage sensor (b) signal conditioning
for the sensed voltage................................................................................... 101
Fig. 5.5 Current measurement system (a) current transducer (b) signal
conditioning for the sensed current in terms of voltage signal .................... 103
Fig. 5.6 Output signals of and incremental angle encoder ........................................ 104
Fig. 5.7 Block diagram of an incremental encoder interface .................................... 105
Fig. 5.8 Block diagram for FIR filter ........................................................................ 110
Fig. 5.9 Simple first order analog low pass filter ...................................................... 112
Fig. 6.1 The per-phase equivalent circuit with shunt magnetising branch
impedance represented in parallel................................................................ 115
Fig. 6.2 Per-phase equivalent circuit with shunt magnetising branch impedance
represented in series form............................................................................ 116
Fig. 6.3 Per-phase equivalent circuit of three-phase induction machine under
no load test ................................................................................................... 117
Fig. 6.4 Per-phase equivalent circuit at standstill (short-circuit test)........................ 118
Fig. 6.5 Current displacement with rotor speed a) zero speed b) intermediate
xv
speed c) close to synchronous speed............................................................ 121
Fig. 6.6 Rotor parameter variations with slip for deep bar induction machine......... 123
Fig. 6.7 Per-phase equivalent circuit with variable rotor parameters ....................... 123
Fig. 6.8 Monitoring system for parameter determination ......................................... 126
Fig. 6.9 Three-phase induction motor input quantities as a function of time
(a) measured line voltage (b) measured line current (c) measured input
power............................................................................................................ 128
Fig. 6.10 Three-phase induction motor input quantities as a function of speed
(a) measured line voltage (b) measured line current (c) measured input
power............................................................................................................ 130
Fig. 6.11 Variation of rotor parameters for machine single-cage rotor....................... 131
Fig. 6.12 Variation of rotor parameters with slip and supply line to line voltage ...... 132
Fig. 6.13 Effect of temperature on rotor parameters ................................................... 134
Fig. 6.14 Variation of (a) magnetizing reactance with voltage at 95
o
C (b) iron loss
resistance with voltage at 95
o
C (c) magnetizing reactance with temperature
and voltage (d) iron loss resistance with temperature and voltage .............. 137
Fig. 6.15 Values of rotor resistance, R
r,
and rotor leakage reactance, X
lr ...............................
140
Fig. 6.16 Measurement error with a normal distribution ............................................ 141
Fig. 6.17 Data generated for simulation of measurement error................................... 143
Fig. 6.18 Error in rotor parameters due to 0.5% error in voltage current and/or
power............................................................................................................ 144
Fig. 6.19 Error in rotor parameters due to 1% error in voltage current and/or
power............................................................................................................ 144
Fig. 6.20Error in rotor parameters due to 1.5% error in voltage current and/or
power............................................................................................................ 145
Fig. 6.21 Simulated shaft torque for variable and constant rotor parameters .............. 146
Fig. 7.1 SEIG with a capacitor excitation system driven by a wind turbine............. 150
Fig. 7.2 D-Q representation of self-excited induction generator............................... 151
Fig. 7.3 Detailed d-q model of SEIG in stationary reference frame (a) q-axis circuit
(b) d-axis circuit ........................................................................................... 152
Fig. 7.4 RLC circuit................................................................................................... 154
Fig. 7.5 Current in series RLC circuit (a) for R = 1.2O and (b) for R = -1.2O......... 156G
xvi
Fig. 7.6 Variation of magnetising inductance with phase voltage at rated frequency
165
Fig. 7.7 Flow chart to determine the minimum speed and minimum capacitance
for SEIG at no load ...................................................................................... 167
Fig. 7.8 Values of minimum capacitance and rotor speed for self-excitation at
no load.......................................................................................................... 169
Fig. 7.9 Error in capacitance when calculated using the approximate method......... 170
Fig. 7.10 Measured unsuccessful self-excitation at C=60F (a) generated phase
voltage (b) speed.......................................................................................... 171
Fig. 7.11 Measured self-excitation at C = 60F and lower speed (a) generated
phase voltage (b)speed................................................................................. 172
Fig. 7.12 Measured self-excitation at C = 60F with speed and generated voltage
close to rated values (a) generated phase voltage (b)speed ......................... 173G
Fig. 7.13 Required capacitance and speed for self-excitation with load, R
L
............... 174
Fig. 7.14 Relationship between capacitance value, rotor speed and generated
voltage at no load......................................................................................... 178
Fig. 7.15 Variation of magnetising inductance with phase voltage at different
frequencies ................................................................................................... 179
Fig. 7.16 Variation of magnetising inductance with magnetising current .................. 180
Fig. 7.17 DC motor speed regulator ............................................................................ 181
Fig. 7.18 Measured self-excitation at C = 60F and with regulated speed
(a) generated phase voltage (b) speed (c) stator current .............................. 182G
Fig. 7.19 Simulated self-excitation at C = 60F and with regulated speed
(a) generated phase voltage (b) speed (c) stator current .............................. 183
Fig. 7.20 Simulated self-excitation at C = 60F and with regulated speed
(a) magnetising inductance (b) rms magnetising current (c) peak stator
flux-linkage .................................................................................................. 184G
Fig. 7.21 Three dimensional d-axis flux-linkage and q-axis flux-linkage as a
function of time during self-excitation process ........................................... 185
Fig. 7.22 Self-excitation process initiated by a charged capacitor of 60F and
rotor speed of 1480rpm (a) experimental result (b) simulated result........... 186G
Fig. 7.23 d-q model of a loaded SEIG in a stationary reference frame (a) q-axis
xvii
circuit (b) d-axis circuit............................................................................... 187
Fig. 7.24 Relationship between rotor speed and synchronous speed in a SEIG......... 189
Fig. 7.25 Experimental loading of SEIG after the voltage has developed to its steady
state value (a) phase voltage (b) speed (c) frequency (d) rms phase voltage
(e) generated power (f) rms stator current ................................................... 190
Fig. 7.26 Simulated loading of SEIG after the voltage has developed to its steady
state value (a) phase voltage (b) speed (c) frequency (d) rms phase voltage
(e) generated power (f) rms stator current ................................................... 191
Fig. 7.27 Simulated loading of SEIG (a) rms stator current (b) rms capacitor current
(c) rms load current ...................................................................................... 192
Fig. 7.28 Simulated loading of SEIG (a) Lm (b) peak flux-linkage (c) rms
magnetising current...................................................................................... 192
Fig. 7.29 Measured variation of generated voltage with load for a 60F capacitance 193G
Fig. 7.30 Measured variation of generated frequency with load for a 60F
capacitance................................................................................................... 193G
Fig. 7.31 Wind turbine output torque as a function of rotor speed ............................. 195
Fig. 7.32 Simulated results for wind turbine with variable rotor speed (a) load
resistance (b) capacitance (c) rotor speed (d) phase voltage (e) frequency
as a function of time..................................................................................... 197
Fig. 7.33 Simulated results for wind turbine with variable rotor speed (a) rms stator
current (b) rms capacitor current (c) rms load current (d) electromagnetic
torque (e) output power as a function of time............................................. 198
Fig. 7.34 Input to the hypothetical SEIG (a) capacitance, F (b) load resistance, OG
(c) speed, rpm............................................................................................... 200
Fig. 7.35 Comparison of constant and variable rotor parameters performance in
SEIG (a) rms phase voltage (b) rms stator current (c) rms capacitor current
(d) rms load current (e) rms magnetising current (f) magnetising
inductance .................................................................................................... 203
Fig. 7.36 Comparison of constant and variable rotor parameters performance in
SEIG (a) generated frequency (b) slip (c) electromagnetic torque
(d) electrical generated output power (e) mechanical input power
(f) efficiency................................................................................................. 204
xviii
Fig. 8.1 No load D-Q model of a SEIG including core loss represented by R
m
(a) d-axis (b) q-axis........................................................................................ 210
Fig. 8.2 Values of capacitance and speed for self-excitation with and without R
m
at no load........................................................................................................ 213
Fig. 8.3 No load RMS phase voltage during self-excitation with and without R
m
..... 216
Fig. 8.4 Variation of connected capacitor and resistor................................................ 218
Fig. 8.5 The dynamic rms generated voltage with variation of load and capacitance 218
Fig. 8.6 Dynamic currents in the load, capacitor and stator with variation in load
and capacitance .............................................................................................. 219
Fig. 8.7 The dynamic output power with variation in load and capacitance .............. 219
Fig. 8.8 The dynamic electromagnetic torque with variation in load and capacitance220
Fig. 9.1 Electrical and mechanical connections ........................................................ 224
Fig. 9.2 Vector diagram for rotor flux oriented vector control ................................. 226
Fig. 9.3 Vector diagram for stator flux oriented vector control ................................ 235
Fig. 9.4 System description....................................................................................... 240
Fig. 9.5 Relationship between generator rotor speed and flux linkage..................... 242
Fig. 9.6 Implementation of direct rotor flux oriented vector control with current
controlled PWM VSI ................................................................................................... 244
Fig. 9.7 Implementation of indirect rotor flux oriented vector control with current. 246
Fig. 9.8 Implementation of direct rotor flux oriented vector control with stator
voltage as a control variable ........................................................................ 247
Fig. 9.9 Implementation of stator flux oriented vector control with current
controlled PWM VSI ................................................................................... 248
Fig. 9.10 Generated DC voltage for different capacitance value................................ 250
Fig. 9.11 Rotor speed and angular frequency of the generated voltage for different
capacitance value ......................................................................................... 250
Fig. 9.12 Flux linkage at different rotor speeds of the induction generator
for 1000F ................................................................................................... 251
Fig. 9.13 Generated line to line voltage at the terminals of the induction generator .. 251
Fig. 9.14 Loading of the induction generator (a) R
L
(b) rotor speed (c) V
DC
(d) flux
linkage (e)
e
ds
i (f)
e
qs
i (g) I
dc
(h) Output power (i) Slip (j) Electromagnetic
torque ........................................................................................................... 254
xix
Fig. 10.1 Offset error equal to A
m
as a result of the integration initial condition........ 261
Fig. 10.2 No integrator error ....................................................................................... 261
Fig. 10.3 Error produced due to measurement offset .................................................. 262
Fig. 10.4 Error produced due to measurement offset and integration initial condition262
Fig. 10.5 Numerical integrator representation ............................................................ 263
Fig. 10.6 Proposed offset adjustment in a numerical integrator.................................. 265
Fig. 10.7 Proposed integrator with input offset adjustment ........................................ 265
Fig. 10.8 Detail for integration error compensation ................................................... 266
Fig. 10.9 Stator flux linkage estimation using the proposed method.......................... 267
Fig. A.1 Electromagnetic torque versus motor speed at steady state......................... 280
Fig. A.2 Variation of speed with time (a) DC motor field supply on
(b) DC motor field supply off ...................................................................... 280
Fig. B.1 Interconnection of hardware system............................................................ 283
Fig. B.2 DSPACE DS1102 DSP controller board ..................................................... 283
Fig. B.3 Multiplexer board control to dSPACE DS1102 DSP card connection........ 283
Fig. B.4 DAC output for DC motor speed control ..................................................... 284
Fig. B.5 Dead time Generator board and DS1102 DSP card connection .................. 284
Fig. B.6 Incremental encoder DS1102 DSP card connection.................................... 284
Fig. B.7 Four isolated 15V Power supply for optocoupler circuit............................. 285
Fig. B.8 Optocoupler to Mitsubishi PM50RVA120 IPM.......................................... 286
Fig. B.9 8 to 4 multiplexer with Sample and Hold .................................................... 287
Fig. B.10 Cross over protection board (dead time generator)...................................... 288
Fig. D.1 Student award............................................................................................... 300
LIST OF TABLES
Table 2.1 Rough Categories of Wind Generator Sizes ............................................... 39
xx
LIST OF SYMBOLS
Generally symbols are defined locally. The list of principal symbols is given below
V
1
- Upwind velocity, m/s
V
2
- Downwind velocity, m/s
V
T
- wind velocity at the wind turbine, m/s
- density of air, Kg/m
3
m - mass of air, Kg
V - velocity of air, m/s
F - force applied on rotor blades, N
P
T
- power extracted by the wind turbine, Watt
A - area swept by the blades of the wind turbine, m
2
e
T
- angular velocity of the wind turbine, rad/s
V
tn
- tangential speed of the blades at the tips
T
T
- torque produced by the wind turbine, Nm
V
w
- the undisturbed wind speed in the site, m/s
V
e
- the maximum fraction of the undisturbed wind that can be absorbed by the rotor
blade for maximum capture of wind power, V
e
= 2/3*V
w
, m/s
V
a
- is the wind created due to rotation of the wind turbine and increases with radius (V
a
is perpendicular to V
e
and V
w
), m/s
V
res
- the resultant incident wind speed due to V
a
and V
e
, m/s
r - total radius the rotor blade respectively, m
r
1
, r
2
and r
3
- radiuses at points 1, 2 and 3 of the rotor blade respectively, m
TSR - Tip-Speed Ratio (dimensionless ratio of tip linear speed of blades to V
w
)
Pr
- Steady state wind pressure, which is equal to atmospheric air pressure, N/m
2
Pr
- wind pressure just after the wind turbine, N/m

2
Pr
+
- wind pressure just before the wind turbine, N/m
2
m - Mass flow rate of air per unit time, K
g
/s
Q
- Volume flow rate of air per unit time, m

3
/s
C
p
- Dimensionless power coefficient
f
a
s
, f
b
s
, and f
c
s
a b c axes instantaneous quantities in stationary reference frame
xxi
f
q
s
, f
d
s
, and f
o
s
dq axes instantaneous quantities in stationary reference frame
f
q
e
, and f
d
e
dq axes DC quantities in excitation reference frame
v
a
, v
b
and v
c
phase voltages in three axes system (stationary reference frame), V
i
a
, i
b
and i
c
phase currents in three axes system (stationary reference frame), A
v
q
s
, and v
d
s
phase voltages in two axes system (stationary reference frame), V
i
q
s
, and i
d
s
phase currents in two axes system (stationary reference frame), A
i
q
e
, and i
d
e
phase currents in two axes system (excitation reference frame), A
d
s
-q
s
stationary dq axes
d
e
-q
e
dq axes in rotating reference frame (rotating at excitation frequency)
v
ds
d-axis stator voltage, V
v
qs
q-axis stator voltage, V
v
dr
d-axis rotor voltage, V
v
qr
q-axis rotor voltage, V
i
ds
d-axis stator current, A
i
qs
q-axis stator current, A
i
dr
d-axis rotor current, A
i
qr
q-axis rotor current, A
i
md
d-axis magnetising current, A
i
mq
q-axis magnetising current, A
ds
d-axis stator flux linkage, web-turn
qs
q-axis stator flux linkage, web-turn
dr
d-axis rotor flux linkage, web-turn
qr
q-axis rotor flux linkage, web-turn
dm
d-axis air gap flux linkage, web-turn
qm
q-axis air gap flux linkage, web-turn
V
m
peak phase voltage, V
I
m
peak phase current, A
V
rms
rms phase voltage, V
I
rms
rms phase current, A
V
dq
phase voltage space vector, V
I
dq
phase current space vector, A
T
s
sampling time(period), seconds
xxii
u angle between the two axes and three axes, rad
| phase shift between current and voltage
e angular speed of the space vector, speed of the general reference frame, rad/s
e
e
angular speed of the excitation reference frame, synchronous speed, rad/s
e
r
electrical rotor angular speed, rad/sec
e
m
mechanical rotor (shaft) angular speed(e
m
= e
r
/P
p
), rad/sec
f
e
excitation frequency, Hz
s the slip of the rotor with respect to the stator magnetic field
P
p
number of pole pairs of the induction machine
N
e
synchronous speed in revolutions per minute (rpm)
V
s
rms stator voltage, V
I
s
rms stator current, A
I
r
rms rotor current, A
R
s
stator winding resistance, O
R
r
rotor winding resistance, O
R
m
equivalent resistance representing iron loss or core loss, O
L
ls
stator leakage inductance, H
L
lr
rotor leakage inductance, H
L
m
magnetising inductance, H
L
s
stator leakage inductance (L
ls
) + magnetising inductance (L
m
) , H
L
r
rotor leakage inductance (L
lr
m
), H
p d/dt, the differential operator
E
s
rms induced emf in the stator winding due to the rotating magnetic field that
links the stator and rotor windings, V
E
r
rms induced voltage in the rotor when the rotor is stationary, V
se
e
rotor current angular frequency
T
e
electromagnetic torque, Nm
T
m
mechanical torque
m
JJG
air gap flux linkage
r
I
JG
rotor current space vector
D friction coefficient, Nm/rad/sec
J inertia, Kg-m
2
xxiii
A
incr
incremental count of the position counter, incremental steps
|
incr
incremental position, radians
e
res
speed measurement resolution, rad/s
u
res
angle measurement resolution, rad
V
O
the measured open-circuit phase voltage, V
I
O
the measured open-circuit phase current, A
P
O
the measured open-circuit three-phase power, W
V
sh
the measured short-circuit input phase voltage, V
I
sh
the measured short-circuit input phase current, A
P
sh
the measured short-circuit three-phase input power, W
Superscript
* commanded variables
Abbreviations
SEIG Self-Excited Induction Generator
emf Electromotive force
PWM Pulse Width Modulation
IGBT Insulated Gate Bipolar Transistor
RMS root mean square
DSP Digital Signal Processor (Processing)
ADC Analog to digital converter
IIR Infinite impulse response
FIR Finite impulse response
PI Proportional and integral (PI controller)
VSI Voltage source inverter
IPM Intelligent power module
VAR Volt ampere reactive
1
CHAPTER 1
INTRODUCTION
1.1 General
Today, most of the electricity generated comes from fossil fuels (coal, oil, and natural
gas). These fossil fuels have finite reserves and will run out in the future. The negative
effect of these fossil fuels is that they produce pollutant gases when they are burned in
the process to generate electricity. Fossil fuels are a non-renewable energy source.
However, renewable energy resources (solar, wind, hydro, biomass, geothermal and
ocean) are constantly replaced, hence will not run out, and are usually less polluting [1].
Due to an increase in greenhouse gas emissions more attention is being given to
renewable energy. As wind is a renewable energy it is a clean and abundant resource
that can produce electricity with virtually no pollutant gas emission. Induction
generators are widely used for wind powered electric generation, especially in remote
and isolated areas, because they do not need an external power supply to produce the
excitation magnetic field. Furthermore, induction generators have more advantages such
as cost, reduced maintenance, rugged and simple construction, brushless rotor (squirrel
cage) and so on.
In the literature, starting in the 1930s, it is well known that a three-phase induction
machine can be made to work as a self-excited induction generator (SEIG) [2, 3]. In an
isolated application a three-phase induction generator operates in the self-excited mode
by connecting three AC capacitors to the stator terminals [2-4] or using a converter and
a single DC link capacitor [5]. The dynamic performance of an isolated induction
generator excited by three AC capacitors or a single DC capacitor with a converter is
discussed in detail in this work.
CHAPTER 1 INTRODUCTION
2
Induction machines are more robust and cheaper than other electrical machines for the
same rating. They need less maintenance when manufactured with a squirrel cage rotor.
Depending on the condition of operation the induction machine can be used as a motor
or generator. Induction machines are available in single-phase or three-phase
constructions. In this work the modelling and analysis given is only for the three-phase
induction machine and the induction machine is operated as a generator. The definition
of slip in this study is the usual one and is the same for the induction generator and
induction motor.
In a grid connected induction generator driven by a wind turbine the magnetic field is
produced by excitation current drawn from the grid. In different countries there are
many induction generators with high power ratings that use wind power as their prime
mover. These export electric power to the grid. The Kooragang wind turbine generator,
shown in Fig. 1.1, which is owned and operated by Energy Australia, in Newcastle,
NSW, Australia, is connected to the grid and has rated power of 600KW and the turbine
is a Vestas V44-600KW machine [6].
Fig. 1.1 Kooragang wind turbine generator, Newcastle, NSW, Australia (Photo 2002)
3
For this generating system the angular speed of the wind turbine rotor measured on the
wind turbine side is 28rpm. A gear box steps up the shaft speed and on the generator
side the angular speed of the generator rotor is approximately 1500rpm [6].
Multiple wind turbine generators can be installed at a given site to form a wind farm.
Fig. 1.2 shows part of a wind farm around San Francisco, California, USA.
Fig. 1.2 Wind farm around San Francisco, California, USA (Photo 2002)
The output voltage and frequency of an isolated induction generator vary depending on
the speed of the rotor and the load connected to the generator. This is due to a drop in
the speed of the rotating magnetic field [7]. The wind turbine can be designed to operate
4
at constant speed or variable speed. When the speed of the prime mover of the isolated
induction generator drops with load, then the decrease in voltage and frequency will be
greater than for the case where the speed is held constant. The AC voltage can be
compensated by varying the exciting AC capacitors or using a controlled inverter and a
DC capacitor. However the frequency can be compensated only if there is a change in
the rotor speed. Because the frequency of the three-phase isolated induction generator
varies with loading its application should be for the supply of equipment insensitive to
frequency deviations, such as heaters, water pumps, lighting, battery charging etc.
For applications that require constant voltage and frequency the rectified DC voltage of
the isolated induction generator should be controlled to remain at a given reference
value. Then the constant DC voltage can be converted to constant AC voltage and
frequency using an output inverter. In this way a control mechanism is implemented to
regulate the output voltage and frequency from an induction generator.
1.2 Thesis outline
There are eleven chapters and four appendices in this thesis. The thesis presents the
modelling of the dynamic characteristics of an isolated self-excited induction generator
driven by a wind turbine. To have a good understanding of the prime mover an
overview of the characteristics of wind turbines is presented. Analysis of an induction
generator is discussed using modelling and the theory of induction machines.
In Section 1.3 of this chapter the literature related to isolated induction generators and
wind turbines is reviewed. This involves clarifying the strengths and limitations of the
previous works and highlighting the advantages of the research covered in the thesis.
In Chapter 2 a detailed explanation about wind as a power source and the mechanism of
conversion of wind power to mechanical power is presented. The variation of output
power and output torque with rotor angular speed and wind speed is discussed. The
economics and growth of wind powered electric generation is given and the projection
for the future is also discussed.
5
The three-axes to two-axes transformation presented in Chapter 3 is applicable for any
balanced three-phase system. In electrical machines analysis a three-axes to two-axes
transformation is applied to produce simpler expressions that provide more insight into
the interaction of the different parameters. The D-Q model for dynamic analysis is
obtained using this transformation. It is shown that the three-axes to two-axes
transformation simplifies the calculation of dynamic rms current, rms voltage, active
power and power factor in a three-phase system and more specifically for this
application, the three-phase induction machine. Traditional methods of measuring these
quantities are unable to obtain peak values of current and voltage in less than one
quarter of a cycle. However using the three-axes to two-axes transformation in the
manner described in Chapter 3, it is possible to evaluate the rms or peak magnitudes of
three-phase AC currents and voltages from one set of measurements taken at a single
instant of time. Furthermore from measurements taken at two consecutive instants in
time the frequency of the three-phase AC power supply can be evaluated.
In Chapter 4 the modelling of an induction machine using the conventional or steady
state model and the D-Q or dynamic model are explained. The voltage, current and flux
linkage in the rotating reference frame and their phase relationships in the motoring
region and generating region are presented. Chapter 4 gives the fundamentals of
induction machine modelling and characteristics as a preparation of the modelling and
analysis of an isolated induction generator. The induction machine model in D-Q axes
has been improved to include the equivalent iron loss resistance, R
m
. This improved
model is presented in a simple and understandable way. Using this model the dynamic
current, torque and power can be calculated more accurately.
In Chapter 5 the data acquisition system and signal processing are discussed. The
measurement of voltages, currents, rotor angle and angular speed with their appropriate
sensors is explained. The detail of the digital signal processing (DSP) card and
transducer board used in the experimental setup is given. The sensors for current and
voltage are Hall-Effect devices. Rotor speed and angle measurements are taken using an
optical incremental encoder. The resolution of angle and speed for a given encoder is
derived. Anti-aliasing filters are introduced in the analog signals of the sensor outputs to
prevent the high frequencies appearing as a low frequency when the analog signal is
6
digitised in the A/D converter. The advantage of digital signal processing is discussed
and different types of filter design are presented which are used in the simulation and
experimental procedures.
Machine modelling requires knowledge of the parameters of the machine. Whether the
three-phase induction machine is modelled using the conventional per-phase equivalent
circuit or the D-Q method the parameters of the machine are required. Chapter 6
discusses a rapid way of determining the parameters that is fast enough to determine the
parameters at rated voltage of the induction machine without damaging it due to
overheating. The error in the values of induction motor parameters arising from
measurement error in voltage, current and power have been presented. Rotor parameter
variations in squirrel cage induction machines and the cause of this variation is
examined. The variation of induction machine parameters with temperature is also
presented.
Chapter 7 deals with the modelling, analysis and dynamic performance of an isolated
three-phase induction generator excited by three AC capacitors connected at the stator
terminals. The mathematical model of a self-excited induction generator including the
representation of the remnant magnetic flux in the iron core and the initial charge in the
capacitor is given. The initiation and process of self-excitation is presented, starting
from a simple RLC circuit as an analogy to a complete dynamic representation of a self-
excited induction generator, i.e. the complete representation includes both steady state
and transient conditions. The variation of magnetising inductance of the induction
machine is important in the voltage build up and stabilisation of the generated voltage.
It is shown that the characteristics of magnetising inductance with respect to the rms
induced stator voltage or magnetising current determines the regions of stable operation
as well as the minimum generated voltage without loss of self-excitation. The variation
of the generated voltage and frequency for a self excited induction generator driven by a
wind turbine at constant and variable speeds has been investigated. Using simulation
algorithms more results which are not accessible in an experimental setup have been
predicted.
7
In Chapter 8 the modelling of an isolated self-excited induction generator taking iron
loss into account is discussed. Iron loss or core loss is represented in the induction
machine model using R
m
, a resistance value which has the same power loss as the total
iron loss in the induction machine. The method presented here is a novel analysis and
modelling for the dynamics of the self-excited induction generator driven by a variable
speed prime mover taking iron loss into account. It is noted that this method is easily
understood, having drawn on many familiar concepts and using the standard
terminology and nomenclature of D-Q unified machine theory. This improved model
takes into consideration the variations of R
m
with air gap voltage and, as in Chapter 7,
the variation of magnetising inductance. This model is then coupled to the
characteristics of a variable speed prime mover and the analysis of this system is
produced and discussed.
In Chapter 9 the voltage build up process and terminal voltage control in an isolated
wind powered induction generator using an inverter/rectifier excitation with a single
capacitor on the DC link is discussed. A vector control technique is developed to
control the excitation and the active power producing currents independently. That is,
the current control scheme causes the currents to act in the same way as in a DC
generator where the field current and the armature current are decoupled. When the
speed of the prime mover is varied the flux linkage in the induction generator is made to
vary inversely proportional to the rotor speed so that the generated voltage will remain
constant. Since the torque produced by a wind turbine drops at high turbine rotor speed
the induction generator will run at high generator rotor speed when loaded with a small
load and the rotor speeds decrease with an increase in load. As the turbine rotor shaft
and the generator rotor shaft are connected via a gear box, both rotor speeds will
increase and decrease proportionally at constant gear ratio. The flux linkage of the
induction generator is controlled by controlling the d-axis current in the synchronously
rotating reference frame. Two vector control strategies: rotor flux oriented vector
control and stator flux oriented vector control are presented. It is shown that the
estimation of rotor flux linkage is more dependent on the induction machine parameters
whereas estimation of stator flux linkage is dependent only on the stator resistance.
8
Chapter 10 investigates the problems and the solutions in the estimation of stator flux
linkage using integration of the voltage behind the stator resistance. This voltage is
calculated from the measured voltages and currents. Accurate flux estimation is very
crucial in the control of induction motor drives and induction generators using vector
control. The method of flux linkage estimation proposed in this chapter is new and
effective. It eliminates the error produced by the measurement offset error and
integrator output error due to initial integration in a continuous time integrator or
numerical/discrete time integrator. It is shown that if the integration ramp output due to
the existence of measurement offset error is large then subtracting the output of a low
pass filter of the signal from the signal to be integrated minimizes the offset. A signal
with small input offset will have a small increment of ramp that will appear at the
output of the integrator. As the time increases the ramp keeps on increasing and
eventually the distortion in flux will be unacceptable. However, if the ramp is
eliminated every cycle, the flux distortion due to the offset correction at the output is
insignificant.
In Chapter 11 conclusions and suggestions for future work are given.
1.3 Literature review
In this section previous work carried out in the area of self-excited induction generators
that are driven by variable speed prime movers and in particular by wind turbines are
reviewed. If there is a controller to regulate the output voltage and frequency, then an
isolated induction generator can be driven by a variable speed prime mover. However,
for loads which are insensitive to frequency, then the controller needs only to regulate
the generated voltage.
1.3.1 Self-excited induction generator
The early work on three-phase SEIGs excited by three capacitors was mainly
experimental analysis [2, 3]. The main methods of representing a SEIG are the steady
state model and the dynamic model. The steady state analysis of SEIG is based on the
steady state per-phase equivalent circuit of an induction machine with the slip and
angular frequency expressed in terms of per unit frequency and per unit angular speed.
The steady state analysis includes the loop-impedance method [8-13] and the nodal
9
admittance method [14-15]. The loop-impedance method is based on setting the total
impedance of the SEIG, i.e. including the exciting capacitance, equal to zero and then to
find the steady state operating voltage and frequency using an iteration process. In the
nodal admittance method the real and imaginary parts of the overall admittance of the
SEIG are equated to zero. The equations are formulated based on the steady state
conditions of the SEIG.
The main draw back of using the per-phase steady state equivalent circuit model is that
it cannot be used to solve transient dynamics because the model was derived from the
steady state conditions of the induction machine.
The dynamic model of a SEIG is based on the D-Q axes equivalent circuit or unified
machine theory. For analysis the induction machine in three axes is transformed to two
axes, D and Q, and all the analysis is done in the D-Q axes model. The results are then
transformed back to the actual three axes representation. In the D-Q axes if the time
varying terms are ignored the equations represent only the steady state conditions. The
SEIG represented in D-Q axes and analysed under steady state conditions are reported
in [16-17]. In [18-21] the dynamic equations for the representation of SEIG conditions
are given. In these papers the initial conditions that take into account the initial charge
in the exciting capacitors and the remnant magnetic flux linkage in the iron core are not
given and in some of the papers the complete dynamic equations are not presented.
The D-Q axes model of SEIG given in [20] reported that the dynamic generated voltage
varies with the applied load, but there are no results that show what happens to the
dynamic speed of the rotor when the generator is loaded. Hence it cannot be proven
whether the variation in voltage is exaggerated due to a change in speed or not. To
investigate this, the characteristic of the dynamic voltage is simulated and measured
keeping the speed at a constant value by applying a speed regulator to a DC motor
which is used as a prime mover for the SEIG. For the constant speed drive test a PI
(proportional and integrator) speed controller and an inner loop PI current controller is
used. The dynamic frequency of the generated voltage, during loading conditions, is
calculated from measured voltages or from measured voltages and currents. A three-
axes to two-axes transformation is used in the calculation of the dynamic frequency
10
value. Here the transformation is used to simplify the calculation. The measured and
simulated dynamic currents, active power and electromagnetic torque generated by the
SEIG are also given in this paper.
The normal connection of a SEIG is that the three exciting capacitors are connected
across the stator terminals and there is no electrical connection between the stator and
rotor windings. However, in the literature a SEIG with electrical connection between
rotor and stator windings is reported [22]. This paper deals with the steady state
performance of a SEIG realised by a series connection of stator and rotor windings of a
slip-ring type induction machine and solved using D-Q analysis. In this type of
connection it has been claimed that it has the advantage of operating at a frequency
independent of load conditions for a fixed rotor speed, however the angular frequency
of the output voltage is equal to half of the rotor electrical angular speed, which means
the prime mover should rotate at twice the normal speed to generate voltage with
standard frequency. There is also concern regarding the current carrying capability of
the rotor and stator windings because both of them are carrying the same current.
Whether any wound rotor induction machine can be used in this way or not is not
specified.
Shridhar et al reported that if a single valued capacitor bank is connected, i.e. without
voltage regulator, a SEIG can safely supply an induction motor rated up to 50% of its
own rating and with a voltage regulator that maintains the rated terminal voltage the
SEIG can safely feed an induction motor rated up to 75% of its own rating [23]. In this
case the SEIG can sustain the starting transients of the induction motor without losing
self-excitation.
Since a SEIG operates in the saturation region, it has been shown that to saturate the
core, the width of the stator yoke is reduced so that the volume and the weight of the
induction generator will be less than the corresponding induction motor [24]. The
voltage drop for a constant capacitor induction motor used as a generator was 30%
while the voltage drop of the corresponding designed induction generator was 6% [24].
11
A three-phase SEIG can be used as a single-phase generator with excitation capacitors
connected in C-2C mode where capacitors C and 2C connected across two phases
respectively and nil across the third phase [25].The steady state performance of an
isolated SEIG when a single capacitor is connected across one phase or between two
lines supplying one or two loads is presented in [26]. However in these applications the
capacity of the three-phase induction generator cannot be fully used.
1.3.2 Capacitance and rotor speed for self-excitation
The minimum and maximum values of capacitance required for self-excitation of a
three-phase induction generator have been analysed previously using a current model
[9, 11, 20]. Calculation of the minimum capacitance required for self excitation using a
flux model has also been reported [27].
In the calculation of capacitance required for self-excitation, economically and
technically, it is not advisable to choose the maximum value of capacitance. This is due
to the fact that for the same voltage rating the higher capacitance value will cost more.
In addition, if the higher capacitance value is chosen then there is a possibility that the
current flowing in the capacitor might exceed the rated current of the stator due to the
fact that the capacitive reactance reduces as the capacitance value increases.
It has been shown that a de-excited induction generator can re-excite even if the load is
already connected to it [30], but the relationship between the value of the load,
capacitance and speed has not been given. In this thesis the relationship between speed,
capacitance and load is given so that the characteristics of the induction generator for
self-excitation with a load can be established. This relationship is also important to find
the region where the induction generator can continue to operate without loss of self-
excitation.
Wind speed can change from the minimum set point to the maximum set point
randomly and the SEIG can be started at any point within the range of speed. It is
essential to find the minimum and maximum speed required for self-excitation, when
the generator is loaded. In this thesis the author has developed the analysis and
12
calculation of the minimum and maximum speeds for self-excitation to occur and for a
particular value of capacitance.
1.3.3 Representation of magnetising inductance
In the SEIG the variation of magnetising inductance is the main factor in the dynamics
of voltage build up and stabilisation. Several papers have reported on the representation
of the variation of magnetising inductance (L
m
) or magnetising reactance (X
m
) during
voltage build up.
One of the ways of representation is X
m
as a function of V
g
/f (V/Hz to relate to flux) [8-
9, 11-13, 15, 21], where V
g
is the voltage across X
m
and f is the frequency of excitation,
or L
m
as a function of V
g
[14, 26] for a known frequency of operation. In these papers it
has been shown that the value of X
m
, as the value of V
g
/f or V
g
increases from zero,
starts at a given unsaturated value, remains constant at the unsaturated value for low
values of air-gap voltage or ratio of air gap voltage to frequency, and then starts to
decrease up to its rated value, which is a saturated value. In fact, in [9] the measured
values show the actual variation of magnetising reactance. This is the magnetizing
reactance as the air gap voltage increases from zero. It starts at a given value, increases
until it reaches its maximum value and then starts to decrease down to its rated value,
which is a saturated value. However, in the analysis of the SEIG the magnetising
reactance for values of air gap voltage close to zero were ignored. Since X
m
is
dependent on frequency it is not good for transient dynamic analysis, rather L
m
should
be used.
The other representation is X
m
as a function of magnetising current [20, 28] or L
m
as a
function of magnetising current [16, 29, 30]. In these papers it has been illustrated that
the magnetising inductance or magnetising reactance starts at a maximum unsaturated
value and then decreases when the iron core saturates, however in [16] the authors have
indicated that the value of magnetising inductance starts at a given unsaturated value,
increases and then finally decreases as the magnetising current increases from zero.
Although this representation depicts the actual variation of magnetising inductance, the
significance of this characteristic has not been presented.
13
The reason for this variation in magnetising reactance and the effect on self-excitation is
discussed in this thesis. As the magnetizing reactance is dependent on frequency,
magnetizing inductance is used in the analysis and its effect on the initiation of self-
excitation and stabilisation is discussed in detail and confirmed experimentally.
1.3.4 Control of generated voltage and frequency
The main problem in using a SEIG is the control of the generated voltage because the
voltage amplitude and frequency drops with loading as well as with a decrease in the
generator rotor speed [7]. The magnitude and frequency of the output voltage of a stand
alone induction generator driven by a variable speed rotor can be controlled by
employing the rotor excitation of a wound-rotor induction machine [31]. In a similar
way it can be controlled by varying the rotor resistance of a self-excited slip-ring
induction generator [32]. However a self-excited slip-ring induction generator will
require more maintenance than a squirrel cage rotor due to the slip-rings and brush gear.
The rms value of the generated voltage, irrespective of its frequency, can be controlled
using variable capacitance values [33], or a fixed capacitor thyristor controlled reactor
static VAR compensator [34], or continuously controlled shunt capacitors using
antiparallel IGBT switches across the fixed excitation capacitor [35].
It has been shown that copper loss decreases in the stator and increases in the rotor in
the generating mode when compared to the motoring mode [36]. In a SEIG, a squirrel
cage rotor is preferable to a wound rotor because the squirrel cage rotor has a higher
thermal withstand capability and requires less maintenance. Due to the higher thermal
withstand capability of the squirrel cage rotor, a higher copper loss in the rotor is
acceptable.
1.3.5 Wind powered generators
For a fixed speed wind turbine system that can be connected to the grid, maintaining a
constant frequency is not a problem, irrespective of whether an induction or
synchronous generator is used. Such systems typically employ induction machines
connected directly to the grid. In grid connected systems there are two generating
schemes for variable speed wind turbine systems [37-43]. The first scheme employs
14
machine control using power electronics feeding the rotor circuit (wound rotor
induction machine) or a second winding in the stator of an induction machine (squirrel
cage rotor or wound rotor) to adjust the frequency and generated voltage when the
generator rotor speed is varied. The second scheme applies to single stator winding fed
induction generators which produce a constant DC output voltage that is then inverted
to have an output of constant rms voltage and frequency. The generation of constant DC
voltage is implemented using scalar or vector control [44-45] or using a DC-DC
converter to produce constant DC voltage from the variable rectified DC voltage [46].
In a variable speed wind turbine system the mechanical stresses caused in the structural
elements by gusts and varying wind speed are diminished by letting the rotor follow the
wind. Also when the rotor speed is allowed to vary with the wind the turbine can be
operated at peak efficiency. However, the necessary power electronics can be
expensive.
Brushless doubly-fed induction machines have two stator windings of different pole
number [39-42]. Although the system has reduced size and cost of the power
electronics, the induction machine is expensive because it is specially made. A double
output induction generator is a wound rotor induction machine with the control power
electronics connected on the rotor circuit [43, 45]. In this arrangement the induction
generator gives more than its rated power without being overheated. The power
generation can be realised for a wide range of wind speed. They have a rotor inverter
and front end converter while the stator is linked directly to the grid.
The methods discussed above can also be used to control the output voltage from a
stand alone induction generator. In the literature it is reported that a stand alone
induction generator excited by a single DC capacitor and inverter/rectifier system can
be used instead of the AC capacitor excited system. If a constant DC voltage is achieved
then a load side inverter is used to produce a constant rms voltage and frequency. For
this application an inverter/rectifier can be shunt connected so that it carries only the
exciting current [47-49] or a converter can be connected in series so that it carries the
full current [50-51], i.e. the exciting and load current. In both cases the initiation of
voltage build up is the same. However in these papers the details of the control
mechanism and the generation of reference currents are not given. The minimum DC
15
capacitance required for the initiation of voltage build up has been discussed [50].
When the converter carries only the exciting current an additional rectifier is required to
produce the DC voltage that supplies the load.
Artificial Intelligence is the branch of science that concentrates on making computers or
computer-based technology to function like humans. Advanced intelligent control of a
variable speed wind generation system has been reported in the literature [52-57].
Artificial intelligence techniques include fuzzy logic, neural network, and genetic
algorithm, etc [56-57]. The evolving adaptive and elastic versions of fuzzy logic
control in combination with the artificial neural network algorithms promise to
revolutionize the applicability of fuzzy logic control in reference trajectory tracking,
state estimation and parameter adaptation of control strategies [52]. It has been shown
that fuzzy control algorithms are universal, give fast convergence, are parameter
insensitive, and accept noisy and inaccurate signals [57]
It has been reported that artificial intelligent has been used extensively to optimize
efficiency and enhance performance in a grid connected variable speed wind generation
system [52-57]. A fuzzy controller tracks the generator speed with the wind velocity to
extract the maximum power. For a grid connected variable speed wind generation
system any power generated by the induction generator is absorbed by the grid. The
draw back of using Artificial Intelligence control is, at the current price of
microprocessors, it is expensive to implement in induction generators with small power
rating. Of course the general drawback of Artificial Intelligence is the inability to
untangle the complicated web of human intelligence.
1.3.6 Cross saturation
The idea of cross saturation is to have an interest focused on the so called cross
coupling effect by which two windings with their magnetic axes in space quadrature
exhibit a specific magnetic interaction due to saturation of the main flux paths. With the
concept of cross saturation the flux produced by the current flowing in the d-axis
magnetising inductance will have effect on the q-axis circuit. With the consideration of
cross saturation some authors have argued that the model describes physically sound
phenomenon in the smooth air gap machine that are known from the theory of salient
16
pole machines [58-60]. However others have argued that cross coupling in smooth air
gap machines is not possible and that the existence of cross coupling terms and the
inequality of mutual inductances along the d-axis and q-axis is purely a consequence of
the mathematical derivation and it has been shown that the concept of cross saturation
has no physical significance; it is an erroneous conclusion only, obtained from
otherwise correct mathematical formulation [61].
The motivation for transforming the actual three-axes model into a fictitious two-axes
or D-Q axes model is to avoid cross coupling between the D and Q phases and simplify
the analysis of induction machines. Therefore, since the contribution of cross saturation
in the area of induction machine analysis has not been resolved and since the intention
of the D-Q analysis is to decouple the axes, cross saturation will not be considered in
this thesis. Any variation due to saturation will be represented by the variation of
magnetising inductance along the q-axis and the d-axis. In this way, the basic essence of
transforming the three-axes model to the D-Q axes model is maintained.
1.4 References
[1] M. R. Patel, Wind and Solar Power Systems, CRC Press LLC, Boca Raton, 1999.
[2] E. D. Basset and F.M. Potter, Capacitive excitation of induction generators, Trans. of the Amer.
Inst. Electr, Eng., Vol. 54, No. 5, May 1935, pp. 540-545.
[3] C. F. Wagner , Self-excitation of induction Motors, Trans. of the Amer. Inst. Electr, Eng., Vol.
58, Feb. 1939, pp. 47-51.
[4] J. M. Elder, J.T. Boys and J.L. Woodward, Self-excited induction machine as a small low-cost
generator, IEE Proc. C, Vol.131, No. 2, March 1984, pp. 33-41.
[5] D. W. Novotny, D. J. Gritter and G. H. Studtmann, Self-excitation in inverter driven induction
machines, IEEE Transactions on Power Apparatus and Systems, Vol. PAS-96, no.4, July/August
1977, pp. 1117-1125.
[6] Energy Australia Kooragang wind turbine generator fact sheet, 2001.
[7] D. Seyoum, C. Grantham and F. Rahman, "The dynamic characteristics of an isolated self-excited
induction generator driven by a wind turbine", Proceedings IEEE- IAS 2002 Annual Meeting
Pittsburgh, USA, October 13-18, 2002, pp 731-738.
[8] S. S. Murthy, O. P Malik and A. K. Tandon, Analysis of self-excited induction generators, IEE
Proc., Vol. 129, No. 6, Nov. 1982, pp. 260-265.
[9] N. H. Malik, and A. H. Al-Bahrani, A.H., Influence of the terminal capacitor on the performance
characteristics of a self-excited induction generator , IEE Proc C., Vol. 137, No. 2, March 1990,
pp. 168-173.
17
[10] A. K. Al Jabri and A. I. Alolah, "Capacitance requirements for isolated self-excited induction
generators", IEE Proc., B, Vol. 137, No. 3, 1990, pp.154-159.
[11] A. K. Al Jabri and A.I. Alolah, Limits on the performance of the three-phase self excited induction
generators IEEE Trans. on Energy Conversion, Vol. 5, No. 2, June 1990, pp. 350-356.
[12] A.K. Tandon S. S. Murthy and G. J. Berg, Steady state analysis of capacitor self-excited induction
generators, IEEE Transaction on Power Apparatus and Systems, Vol. 103, No.3, March 1984, pp.
612-618.
[13] L. Shridhar, B. Singh and C. S. Jha, A step towards improvements in the charactersitics of self-
excited induction generator, IEEE Transactions. on Energy Conversion, vol. 8, No .1, March 1993,
pp. 40-46.
[14] T. F. Chan, Capacitance requirements of self-excited induction generators, IEEE Trans. on EC,
Vol.8, No.2, June 1993, pp. 304-311.
[15] S. Rajakaruna and R. Bonert, A technique for the steady-state analysis of a self-excited induction
generator with variable speed, IEEE Transactions on Energy Conversion, Vol. 8, No. 4, December
1993, pp. 757-761.
[16] M. Elder, J. T. Boys and J. L. Woodward, Self-excited induction machine as a small low-cost
generator, IEE Proc. C, Vol. 131, No. 2, March 1984, pp. 33-41.
[17] Y. Uctug and M. Demirekler, Modelling, analysis and control of a wind-turbine driven self-excited
induction generator, IEE Proceedings, Vol. 135, Pt. C, No. 4, July 1988, pp. 268 -275.
[18] C. Grantham, D. Sutanto and B. Mismail, Steady-state and transient analysis of self-excited
induction generators, IEE Proc. B, Vol. 136, No. 2, pp. 61-68, March 1989.
[19] M. H. Salama, and P.G. Holmes, Transient and steady-state load performance of stand-alone self-
excited induction generator, IEE Proc. -Electr. Power Appl., Vol. 143, No. 1, pp. 50-58, January
1996.
[20] L. Wang and L. Ching-Huei, A novel analysis on the performance of an isolated self-excited
induction generator, IEEE Trans. on Energy Conversion, Vol. 12, No. 2, pp. 109-117, June 1997.
[21] L. Wang and Jian-Yi Su, Effect of long-shunt and short-shunt connection on voltage variations of
a self-excited induction generator, IEEE Trans. on Energy Conversion, vol. 12, No .4, pp. 368-
374, December 1997.
[22] A. S. Mostafa, A. L. Mohamadein, E. M. Rashad, Analysis of series-connected wound-rotor self-
excited induction generator, IEE Proceedings-B, Vol. 140, No. 5, September 1993, pp. 329-336.
[23] L. Shridhar, B. Singh, C. S. Jha and B. P. Singh, Analysis of self excited induction generator
feeding induction motor, IEEE Transactions on Energy Conversion, vol. 9, No .2, June 1994, pp.
390-396.
[24] J. Faiz, A. A Dadgari, S. Horning and A. Keyhani, Design of a three-phase self-excited induction
generator, IEEE Transactions on Energy Conversion, Vol. 10, No. 3, September 1995, pp. 516-
523.
[25] J. L. Bhattacharya and J. L. Woodward, Excitation balancing of a self-excited induction generator
for maximum power output, IEE Proceedings, Vol. 135, Pt. C, No. 2, March 1988, pp. 88-97.
18
[26] Y.H.A. Rahim, Excitation of isolated three-phase induction generator by a single capacitor, IEE
Proceedings-B, Vol. 140, No. 1, January 1993, pp. 44-50.
[27] R. J. Harrington and F. M. M. Bassiouny, New approach to determine the critical capacitance for
self-excited induction generators, IEEE Transactions on Energy Conversion, Vol. 13, No. 3,
September 1998, pp. 244-249.
[28] C. H. Lee and L. Wang, A novel analysis of parallel operated self-excited induction generators,
IEEE Transactions on Energy Conversion, Vol. 13, No. 2, June 1998, pp. 117-123.
[29] Li Wang and Jian-Yi Su, Dynamic performances of an isolated self-excited induction generator
under various loading conditions, IEEE Transactions on Energy Conversion, Vol. 14, No. 1,
March 1999, pp. 93-100.
[30] L. Shridhar, B. Singh and S. S. Jha, Transient performance of the self regulated short shunt self
excited induction generator, IEEE Transactions on Energy Conversion, Vol. 10 No. 2, June 1995,
pp. 261-267.
[31] Y. Kawabata Y. Morine, T. Oka, E. C. Ejiogu and T. Kawabata, New stand-alone power
generating system using wound-rotor induction machine, IEEE-Power Electronics and Drive
Systems Conference, Vol. 1, October 2001, pp. 335-341.
[32] T. F. Chan, K. Nigim and L. L. Lai, Voltage and frequency control of self-excited slip ring
induction generators, IEEE IEMDC Conference, 2001, pp. 410-414.
[33] P. G. Casielles, L. Zarauza and J. Sanz, Analysis and design of wind turbine driven self-excited
induction generator, IEEE-IAS Conference, 1988, pp. 116-123.
[34] E. S. Abdin and W. Xu, Control design and dynamic performance analysis of a wind turbine-
induction generator unit, IEEE Transactions on Energy Conversion, Vol. 15, No. 1, March 2000,
pp. 91-96.
[35] M. A, Al-Saffar E.-C. Nho and T. A. Lipo, Controlled shunt capacitor self-excited induction
generator, IEEE - Industry Applications Conference, 12-15 October 1998, pp. 1486-1490.
[36] S. S.; Murthy, C. S. Jha and P. S. N. Rao, Analysis of grid connected induction generators driven
by hydro/wind turbines under realistic system constraints, IEEE Transactions on Energy
Conversion, Vol. 5, No. 1, March 1990, pp. 1-7.
[37] I. Schiemenz and M. Stiebler, Control of a permanent magnet synchronous generator used in a
variable speed wind energy system, IEEE- IEMDC 2001 Conference, 2001, pp. 872-877.
[38] J. Jayadev, Harnessing the wind, IEEE Spectrum, Vol. 32, No. 11, Nov. 1995, pp. 78-83.
[39] C. S Brune, R. Spee and A. K. Wallace, Experimental evaluation of a variable-speed, doubly-fed
wind-power generation system, IEEE Transactions on Industry Applications, Vol. 30 No. 3,
May/June 1994, pp. 648-655.
[40] R. Spee, S. Bhowmik and J. H. R. Enslin, Adaptive control strategies for variable-speed doubly-
fed wind power generation systems, IEEE-Industry Applications Society Conference, 2-6 Oct.
1994, pp. 545-552.
19
[41] S. Bhowmik, R. Spee and J. H. R Enslin, Performance optimization for doubly fed wind power
generation systems, IEEE Transactions on Industry Applications, Vol. 35, No. 4, July/August
1999, pp. 949-958.
[42] M. Y. Uctug, I. Eskandarzadeh and H. Ince, Modelling and output power optimisation of a wind
turbine driven double output induction generator, IEE Proc. Electr. Power Appl., Vol. 141, No. 2,
March 1994, pp. 33-38.
[43] L. Zhang and C. Watthanasarn A matrix converter excited doubly-fed induction machine as a wind
power generator, IEE - Power Electronics and Variable Speed Drives Conference, 21-23
September 1998, pp. 532 -537.
[44] A. Miller, E. Muljadi and D. S Zinger, IEEE Transactions on Energy Conversion, Vol. 12, No. 2,
June 1997, pp. 181-186.
[45] R. S Pena, R. J. Cardenas, G. M. Asher and J. C. Clare, Vector controlled induction machines for
stand-alone wind energy applications, IEEE - Industry Applications Conference, 2000, pp. 1409-
1415.
[46] S. Jiao, G. Hunter, V. Ramsden and D. Patterson, Control System Design for a 20KW wind
Turbine Generator with a Boost Converter and Battery Bank Load, IEEE -Power Electronics
Specialists Conference, Vancouver, 2001, pp. 2203-2206.
[47] S. R. Silva, R. O. C Lyra, PWM converter for excitation of induction generators, Fifth European
Power Electronics Conference, 1993, pp 174-178.
[48] M. S. Miranda, R. O. C. Lyra, and S. R. Silva, An Alternative Isolated Wind Erlecric Pumping
System Using Induction Machines, IEEE Transactions on Energy Conversion, Vol. 14, No. 4, Dec.
1999, pp 1611-1616.
[49] L. A. C. Lopes and R. G. Almeida, Operation aspects of an isolated wind driven induction
generator regulated by a shunt voltage source inverter, IEEE - Industry Applications Conference,
Oct 2000, pp. 2277-2282.
[50] S. N. Bhadra, K. V. Ratnam and A. Manjunath, Study of voltage build up in a self-excited,
variable speed induction generator/static inverter system with DC side capacitor, International
Conference on Power Electronics, Drives and Energy Systems for Industrial Growth, 1996, 8-11
Jane 1996, pp. 964 -970.
[51] R. Cardenas, R. Pena, G. Asher and J. Clare, Control strategies for enhanced power smoothing in
wind energy systems using a flywheel driven by a vector-controlled induction machine, IEEE
Transactions on Industrial Electronics, Vol. 48, No. 3, June 2001, pp. 625-635.
[52] H. M. Mashaly; A. M. Sharaf, A. A. El-Sattar and M. M. Mansour, Implementation of a fuzzy
logic controller for wind energy induction generator DC link scheme Proc. of the Fuzzy Systems
IEEE Conference on Computational Intelligence, 26-29 June 1994, pp. 978 -982.
[53] R. M. Hilloowala and A. M. Sharaf, A rule-based fuzzy logic controller for a PWM inverter in a
stand alone wind energy conversion scheme, Proc. of the IEEE- IAS 1993 Annual Meeting
Pittsburgh, 2-8 Oct. 1993, pp. 2066 -2073.
20
[54] R. M. Hilloowala and A. M. Sharaf, A rule-based fuzzy logic controller for a PWM inverter in a
stand alone wind energy conversion scheme, IEEE Transactions on Industry Applications, Vol. 32,
No. 1, January/February 1996, pp. 57 65.
[55] H. M. Mashaly; A. M. Sharaf, A. A. El-Sattar and M. M. Mansour, A fuzzy logic controller for
wind energy utilization, Proc. of the IEEE Conference on Control Applications, 24-26 Aug. 1994,
pp. 221 -226.
[56] M. G. Simoes, B. K. Bose and R. J. Spiegel, Fuzzy logic based intelligent control of a variable
speed cage machine wind generation system, IEEE Transactions on Power Electronics, Vol. 12,
No. 1, Jan. 1997, pp. 87 -95.
[57] M. G. Simoes, B. K. Bose and R. J. Spiegel, Design and performance evaluation of a fuzzy-logic-
based variable-speed wind generation system, IEEE Transactions on Industry Applications, Vol.
33, No. 4, July-Aug. 1997, pp. 956 965.
[58] J. E. Brown, K. P. Kovacs, P. Vas, A method of including the effects of main flux path saturation
in the generalised equations of AC machines, IEEE Transactions on Power Apparatus and
Systems, Vol. PAS-102, No. 1, January 1983, pp. 96-106.
[59] K. E. Hallenius, P. Vas, J. E. Brown, The analysis of a saturated self-excited asynchronous
generator, IEEE Transactions on Energy Conversion, Vol. 6, No. 2, June 1991, pp. 336-345.
[60] E. Levi, Impact of cross saturation on accuracy of saturated induction machine models, IEEE
Transactions on Energy Conversion, Vol. 12, No. 3, September 1997, pp. 211-216.
[61] K. P. Kovacs, On the theory of cylindrical rotor AC machines including main flux saturation,
IEEE Transactions on Power Apparatus and Systems, Vol. PAS-103, No. 4, April 1984, pp. 754-
761.
21
CHAPTER 2
WIND POWER
2.1 Source of wind
Wind is a result of the movement of atmospheric air. Wind comes from the fact that the
regions around the equator, at 0 latitude, are heated more by the sun than the polar
region. The hot air from the tropical regions rises and moves in the upper atmosphere
toward the poles, while cool surface winds from the poles replace the warmer tropical
air. These winds are also affected by the earths rotation about its own axis and the sun.
The moving colder air from the poles tends to twist toward the west because of its own
inertia and the warm air from the equator tends to shift toward the east because of this
inertia. The result is a large counterclockwise circulation of air streams about low-
pressure regions in the northern hemisphere and clockwise circulation in the southern
hemisphere [1]. The seasonal changes in strength and direction of these winds result
from the inclination of the earths axis of rotation at an angle of 23.5
o
to the axis of
rotation about the sun, causing variations of heat radiating to different areas of the
planet.
Local winds are also created by the variation in temperature between the sea and land.
During the daytime, the sun heats landmasses more quickly than the sea. The warmed
air rises and creates a low pressure at ground level, which attracts the cool air from the
sea. This is called a sea breeze. At night the wind blows in the opposite direction, since
water cools at a lower rate than land. The land breeze at night generally has lower wind
speeds, because the temperature difference between land and sea is smaller at night.
Similar breezes are generated in valleys and on mountains as warmer air rises along the
heated slopes. At night the cooler air descends into the valleys. Although global winds,
due to temperature variation between the poles and the equator, are important in
CHAPTER 2 WIND POWER
22
determining the main winds in a given area, local winds have also influence on the
larger scale wind system.
Meteorologists estimate that about 1% of the incoming solar radiation in converted to
wind energy. Since the solar energy received by the earth in just ten days has an energy
content equal to the worlds entire fossil fuel reserves (coal, oil and gas), this means that
the wind resource is extremely large. As of 1990 estimation, one percent of the daily
wind energy input, i.e. 0.01% of the incoming solar energy, is equivalent to the world
daily energy consumption [2]. It is encouraging to know that the global wind resource is
so large and that it can be used to generate more electrical energy than what is currently
being used.
2.2 Wind Turbine
A wind turbine is a turbine driven by wind. Modern wind turbines are technological
advances of the traditional windmills which were used for centuries in the history of
mankind in applications like water pumps, crushing seeds to extract oil, grinding grains,
etc. In contrast to the windmills of the past, modern wind turbines used for generating
electricity have relatively fast running rotors [1].
In principle there are two different types of wind turbines: those which depend mainly
on aerodynamic lift and those which use mainly aerodynamic drag. High speed wind
turbines rely on lift forces to move the blades, and the linear speed of the blades is
usually several times faster than the wind speed. However with wind turbines which use
aerodynamic drag the linear speed can not exceed the wind speed as a result they are
low speed wind turbines. In general wind turbines are divided by structure into
horizontal axis and vertical axis.
2.2.1 Vertical axis wind turbine
The axis of rotation for this type of turbine is vertical. It is the oldest reported wind
turbine. The modern vertical axis wind turbine design was devised in 1920s by a French
electrical engineer G.J.M. Darrieus [2]. It is normally built with two or three blades. A
typical vertical axis wind turbine is shown in Fig. 2.1. Note that the C-shaped rotor
blade is formally called a 'troposkien'.
23
Guy wires
C-shaped rotor
Tower support
Generator and gearbox
Fig. 2.1 Vertical axis wind turbine
The primary aerodynamic advantage of the vertical axis Darrieus machine is that the
turbine can receive the wind from any direction without the need of a yaw mechanism
to continuously orient the blades toward the wind direction. The other advantage is that
its vertical drive shaft simplifies the installation of gearbox and electrical generator on
the ground, making the structure much simpler. On the disadvantage side, it normally
requires guy wires attached to the top for support. This could limit its applications,
particularly for offshore sites. Wind speeds are very low close to ground level, so
although it might save the need for a tower, the wind speed will be very low on the
lower part of the rotor. Overall, the vertical axis machine has not been widely used
because its output power can not be easily controlled in high winds simply by changing
the pitch. Also Darrieus wind turbines are not self-starting, however straight-bladed
vertical axis wind turbines with variable-pitch blades are able to overcome this problem
[3].
2.2.2 Horizontal axis wind turbine
Horizontal axis wind turbines are those machines in which the axis of rotation is parallel
to the direction of the wind. At present most wind turbines are of the horizontal axis
type. Depending on the position of the blades wind turbines are classified into upwind
machines and down wind machines as shown in Fig. 2.2. Most of the horizontal axis
wind turbines are of the upwind machine type. In this study only the upwind machine
design is considered.
24
Wind
direction
Wind
direction
Fig. 2.2 Horizontal axis wind turbine (a) upwind machine (b) downwind machine
Wind turbines for electric generation application are in general of three blades, two
blades or a single blade. The single blade wind turbine consists of one blade and a
counterweight. The three blades wind turbine has 5% more energy capture than the two
blades and in turn the two blades has 10% more energy capture than the single blade.
[2]. These figures are valid for a given set of turbine parameters and might not be
universally applicable.
The three blade wind turbine has greater dynamic stability in free yaw than two blades,
minimising the vibrations associated with normal operation, resulting in longer life of
all components [4].
2.3 Power extracted from wind
Air has a mass. As wind is the movement of air, wind has a kinetic energy. To convert
this kinetic energy of the wind to electrical energy, in a wind energy conversion system,
the wind turbine captures the kinetic energy of the wind and drives the rotor of an
electrical generator.
The kinetic energy (KE) in wind is given by
2
1
KE mV
2
= (2.1)
where m- is the mass of air, in Kg
V-is the speed of air, in m/s
25
Generator
Controller
Anemometer
Nacelle
Wind vane
Gear box
High speed
shaft
Rotor brake
Pitch
Rotor hub
Yaw drive
Foundation
Tower
Wind direction
Yaw motor
Rotor blade
Wind direction
Wind direction
Fig. 2.3 Detail of a wind turbine driven power generation system
26
The power in wind is calculated as the flux of kinetic energy per unit area in a given
time, and can be written as
( )
2 2
d KE
1 dm 1
P V mV
dt 2 dt 2
= = = (2.2)
where m is the mass flow rate of air per second, in kg/s, and it can be expresses in
terms of the density of air ( in kg/m
3
) and air volume flow rate per second ( Q
in m
3
/s)
as given below
m Q AV = =
(2.3)
where A-is the area swept by the blades of the wind turbine, in m
2
.
Substituting equation (2.3) in (2.2), we get
3
1
P AV
2
= (2.4)
This is the total wind power entering the wind turbine. Remember that for this to be true
V must be the wind velocity at the rotor, which is lower than the undisturbed or free
stream velocity. This calculation of power developed from a wind turbine is an idealised
one-dimensional analysis where the flow velocity is assumed to be uniform across the
rotor blades, the air is incompressible and there is no turbulence where flow is inviscid
(having zero viscosity).
The volume of air entering the wind turbine should be equal to the volume of air leaving
the wind turbine because there is no storage of air in the wind turbine. As a result
volume flow rate per second, Q
, remains constant, which means the product AV

remains constant. Hence when the wind leaves the wind turbine, its speed decreases and
expands to cover more area [5]. This is illustrated in Fig. 2.4.
Fig. 2.4 shows the idealised case where the speed of wind continues to flow at a value
of V
2
downstream of the rotor. In reality the slow air in the wake 'diffuses' into the
surrounding air through turbulence, so that further down stream the velocity of air will
be equal to the undisturbed up stream wind speed because of the gain of energy from
the surrounding wind.
27
V
1
A
1
V
T
V
2
A
T
A
2
Wind speed
Distance in the direction of wind
V
2
V
1
V
T
Wind turbine
Pr
Wind pressure
Distance in the direction of wind
P
rT
Pr
P
rT
+
_
Fig. 2.4 Change of wind speed and wind pressure around the wind turbine
As shown in Fig. 2.4 Pr
is the wind at atmospheric pressure. The turbine first causes

the approaching wind to slow down gradually, which results in a rise in wind pressure.
Applying Bernoullis equation the wind has highest pressure, Pr
+
, just before the wind
turbine and the wind has lowest pressure (lower than atmospheric pressure), Pr
, just
after the wind turbine. As the wind proceeds down stream, the pressure climbs back to
atmospheric value, causing a further slowing down of the wind speed.. The pressures
immediately upwind and downwind of the rotor are related to the far upwind and
downwind velocities V
1
and V
2
by applying Bernoulli's equation separately upwind and
downwind. Using momentum theory the downwind force on the rotor is equal to the
pressure drop across it times the rotor blade area [2].
The force F on the rotor blades can be given by the rate of change of momentum,
28
( )
1 2
F m V V = (2.5)
Using equation (2.2), the power extracted by the wind turbine P
T
is the difference
between the upstream wind power, at A
1
, and the downstream wind power, at A
2
given
by ,
( )
1 2
2 2
T
1
P m V V
2
= (2.6a)
( )( )
T 1 2 1 2
1
P m V V V V
2
= + (2.6b)
This power is calculated assuming that all the power lost by the wind has been extracted
by the wind turbine and none has been lost through turbulence.
The force F on the rotor blades multiplied by the wind speed at the rotor blades, V
T
produces power given by
T T
P FV = (2.7)
Substituting equation (2.5) in (2.7) gives
( )
T 1 2 T
P m V V V = (2.8)
Equating equations (2.6b) and (2.8) gives
1 2
T
V V
V
2
+
= (2.9)
Therefore the wind speed at the rotor blades, V
T
is the average of the undisturbed up
stream wind speed, V
1
, and the down stream wind speed, V
2
.
Using equation (2.3) the mass flow rate of air through the rotating blades of the wind
turbine is
T T
m A V = (2.10)
Substituting equation (2.9) in (2.10) the mass flow rate of air at the wind turbine is
given by
1 2
T
V V
m A
2
+
= (2.11)
Substituting equation (2.11) in (2.6a) gives the power absorbed by the wind turbine,
which is the mechanical power at the shaft of the wind turbine, as
29
( )
2 2 1 2
T T 1 2
V V 1
P A V V
2 2
+ | |
=
|
\ .
(2.12)
This power is calculated assuming that all the power lost by the wind has been extracted
by the wind turbine and none has been lost through turbulence. If all the power in the
wind were extracted, the wind speed V
2
would be zero and the air could not leave the
wind turbine. However, if there is no wind leaving the wind turbine the power extracted
is zero because air has to exit the wind turbine in order to make the rotor blades rotate.
Rearranging the equation (2.12) to express the mechanical power developed in the wind
turbine in terms of the upstream wind speed at A
1
, shown in Fig. 2.2, gives
2
2 2
1 1
3
T T 1
V V
1 1
V V
1
P A V
2 2
| | | |
| | | |
| | +
| |
|
|
\ . \ .
\ .
= |
|
|
|
\ .
(2.13)
From equation (2.4) the total wind power P
1
at area A
1
is
3
1 T 1
1
P A V
2
= (2.14)
Then the ratio of wind power extracted by the wind turbine to the total wind power at
area A
1
is the dimensionless power coefficient C
p
, where
2
2 2
1 1
p
V V
1 1
V V
C
2
| |
| | | |
| +
| |
|
\ . \ .
\ .
= (2.15)
Substituting equation (2.15) into equation (2.14) the wind power extracted by the wind
turbine can be written as
3
T T 1 p
1
P A V C
2
= (2.16)
or
2 3
T T 1 p
1
P D V C
2
t = (2.17)
where D
T
is the sweep diameter of the wind turbine.
30
In equation (2.17) it is clearly shown that the power output of a given wind turbine
depends on the square of the rotor blade diameter and the cube of the wind speed. For a
given turbine if the wind speed is doubled the output power will be multiplied by 8.
The maximum value of power coefficient C
p
gives the maximum power absorbed by the
wind turbine. To simplify the calculation of C
p
, the substitution of x=V
2
/V
1
in Equation
(2.15) is made.
The maximum C
p
is then obtained from
p
dC
0
dx
=
which gives a solution of x = 1/3 or x = -1 and in terms of V
1
and V
2
, V
1
= 3V
2
or
V
1
= -V
2
. The solution is V
1
=3V
2
because V
1
=-V
2
shows equal and opposite winds
coming from upstream and down stream towards the wind turbine, which is not
realistic.
0 0.2 0.4 0.6 0.8 1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
V
2
/V
1
Power coefficient C
p
Fig. 2.5 Power coefficient versus V
2
/V
1
Substituting V
1
=3V
2
in equation (2.11), gives the maximum value of C
p
as
p
16
C
27
= (2.18)
31
Here maximum C
p
is about 0.59. This is called the Betz limit [2]. In practical designs,
the maximum achievable C
p
is below 0.5 for high speed, two blade wind turbines, and
between 0.2 and 0.4 for slow speed turbines with more blades [6].
From equations (2.14) and (2.16) the power extracted by the wind turbine is given by
T 1 p
P PC = (2.19)
Then the theoretical maximum extracted power by the wind turbine is given by
T 1
16
P P
27
= (2.20)
This shows that the maximum theoretical efficiency of a wind turbine is about 59%.
For the same wind speed the output power of the wind turbine varies with the shaft
speed. Fig. 2.6 shows a typical set of wind turbine output power versus shaft speed
characteristics for a 7.5kW machine at fixed pitch angle.
Fig. 2.6 Wind turbine output power to shaft speed characteristic curve [7]
2.4 Torque developed by a wind turbine
The torque in a wind turbine is produced due to the force created as a result of pressure
difference on the two sides of each blade of the wind turbine. From fluid mechanics it is
known that the pressure in fast moving air is less than in stationary or slow moving air.
This principle helps to produce force in an aeroplane or in a wind turbine. To explain
the detail of the force created due to the wind, a cross sectional area of the blade of a
32
wind turbine is given in Fig. 2.7. The air travelling from A to B follows two paths. The
shape of the upper surface (path 1) results in higher velocity than the lower surface
(path 2). This will create a low pressure on path 1 side of the blade. Hence force F, at
90
o
to the air flow, will be produced and pushes the blade upwards. This force F
multiplied by the radial distance from the hub at which the force is created gives the
torque.
A
B
1
2
F
Fig. 2.7 Air flow around cross section of a blade of a wind turbine
In reality the angle of incidence of the incoming wind can be different from the one
given in Fig. 2.7 but the principle remains the same.
If the angle between the incoming wind and blade increases for some small angle, the
force produced increases. However, if the angle increases above a given value then the
air flow on path 1 stops sticking to the surface of the blade. Instead the air whirls around
in an irregular path and creates turbulence as shown in Fig. 2.8. Then the force that was
pulling upward on the low pressure side of the blade disappears. This phenomenon is
known as stall [8].
A
B
1
2
Fig. 2.8 Air flow around cross section of a blade during stall condition
On a wind turbine rotor the blades are at some angle to the plane of rotation. At low
shaft speeds, the angle of incidence on a blade element at some radius from the hub is
large, the blades are stalled and only a small amount of driving force will be created. As
a result small torque will be produced at low shaft speed. As the shaft speed increases
33
the velocity of the wind hitting the blade element increases, because of the additional
component of wind due to the blade's rotational speed. In addition, the angle of
incidence decreases. If this angle is below the blade's stall angle, lift increases and drag
decreases, resulting in higher torque. As the shaft speed increases further, the angle of
incidence on the blade element decreases towards zero as the free wind speed becomes
insignificant relative to the blade's own velocity. Since lift generated by a blade is
proportional to the angle of incidence below stall, the torque reduces towards zero at
very high shaft speeds. This variation in torque produced by the wind turbine is shown
in the typical wind turbine torque versus shaft speed characteristics for a 7.5kW
machine given in Fig. 2.9 [9]. In Fig. 2.9, at zero shaft speed, the wind turbine produces
a small starting torque otherwise it will not self start.
Fig. 2.9 Wind turbine output torque to shaft speed characteristic curve
For a horizontal axis wind turbine, operating at fixed pitch angle, the torque developed
by the wind turbine, T
T
, can be expressed as
T
T
T
P
T
e
= (2.21)
where e
T
- angular velocity of the wind turbine, rad/s.
The wind which hits the rotor blades of a wind turbine will not come from the direction
in which the wind is blowing at the site, i.e. from the front of the turbine. This is
because the rotor blades themselves are moving. As the rotor blade rotates it will see
34
different wind speed along its length from its base to its tip. The wind will be coming
from a much steeper angle (more than from the general wind direction at the site) as you
move towards the base of the blade. Therefore, the rotor blade has to be twisted, so as to
achieve an optimal angle of incidence throughout the length of the blade and follow the
change in direction of the resultant wind as shown in Fig. 2.10. Otherwise, as discussed
above, if the blade is hit by wind at an angle of incidence which is too steep, the rotor
blade will stop producing the turning force, causing the blade to stall [2].
r
1
r
2
r
3
Plane of rotation
Plane of rotation
Plane of rotation
Blade
Cross sections
V
a1
= e
T
r
1
V
e
= (2/3)V
w
V
res1
V
e
=(2/3)V
w
V
e
=(2/3)V
w
V
res2
V
res3
V
a2
= e
T
r
2
V
a3
= e
T
r
3
1
3
2
Wind direction viewed from
blade cross sections
Direction of rotation at this
instant of time
Fig. 2.10 Detail of a twisted rotor blade
Fig. 2.10 shows the detail of the twisting of the rotor blade at different radius from the
center of rotation where
V
w
- undisturbed wind speed in the site
V
e
- maximum fraction of the undisturbed wind that can be absorbed by the rotor
blade for maximum capture of wind power, V
e
= (2/3)V
w
V
a
- is the wind created due to rotation of the wind turbine and increases with
radius, V
a1
= e
T
r
1
(V
a
at blade radius of r
1
is V
a1
). V
a
is perpendicular to V
e
and V
w
V
res
- the resultant wind speed of V
a
and V
e
(V
res
at blade radius of r
1
is V
res1
)
r
1
, r
2
and r
3
- radiuses at points 1, 2 and 3 of the rotor blade respectively
The twisting of the rotor blade when viewed from its tip is given in Fig. 2.11 [10].
35
Fig. 2.11 Cross section of a twisted rotor blade from tip to base
2.5 Tip-Speed Ratio
A tip speed ratio TSR is simply the rate at which the ends of the blades of the wind
turbine turn (tangential speed) in comparison to how fast the wind is blowing. The tip
speed ratio TSR is expressed as:
tn T
w w
V r
TSR
V V
e
= =
(2.22)
Where V
tn
- tangential speed of the blades at the tips
e
T
- angular velocity of the wind turbine
r - radius of the wind turbine
V
w
- undisturbed wind speed in the site.
The tip speed ratio dictates the operating condition of a turbine as it takes into account
the wind created by the rotation of the rotor blades. A typical power coefficient C
p
versus tip speed ratio TSR is given in Fig. 2.12 [11]. The tip speed ratio shows
tangential speed at which the rotor blade is rotating compared with the undisturbed
wind speed.
As the wind speed changes, the tip speed ratio and the power coefficient will vary. The
power coefficient characteristic has single maximum at a specific value of tip speed
ratio. Therefore if the wind turbine is operating at constant speed then the power
coefficient will be maximum only at one wind speed.
36
Power Coefficient C
p
Tip speed ratio TSR
Fig. 2.12 Typical power coefficient versus tip speed ratio [11]
Usually, wind turbines are designed to start running at wind speeds somewhere around
4 to 5 m/s. This is called the cut in wind speed. The wind turbine will be programmed to
stop at high wind speeds of 25 m/s, in order to avoid damaging the turbine. The stop
wind speed is called the cut out wind speed [6].
2.6 Power control in wind turbines
The output power of a wind turbine is a function of the wind speed. The determination
of the range of wind speed at which the wind turbine is required to operate depends on
the probability of wind speed obtained from wind statistics for the site where the wind
turbine is to be located. A typical histogram of the wind speed is shown in Fig.2.13
[12]. This histogram is derived from long term wind data covering several years. The
histogram indicates the probability, or the fraction of time, where the wind speed is
within the interval given by the width of the columns. The sum of the height of the
columns is 1 or 100%, since the probability that the wind will be blowing at some wind
speed including zero must be 100%. Fig. 2.13 shows the probability of wind speed
being in a 1m/s interval centred on a given value. For example the probability of the
wind speed being between 4.5 and 5.5 m/s is 0.104 or (0.1048760) = 910 hours per
year. When the width of the columns becomes smaller, the histogram becomes a
continuous function called a probability density function, which can be fitted to a
Weibull distribution or function as shown in Fig. 2.13.
37
Pr obabi l i t y (%)
Fig. 2.13 Histogram and Weibull function for the probability of a given wind speed
(data measured in 1m/s intervals)
In a wind power system typically the wind turbine starts operating (cut in speed) when
the wind speed exceeds 4-5m/s, and is shut off at speeds exceeding 25 to 30m/s. In
between, it can operate in the optimum constant C
p
region, the speed-limited region or
the power limited region as shown in Fig. 2.14 [6]. This design choice was made in
order to limit the strength and therefore the weight and cost of the components of the
wind turbine. Over the year some energy will be lost because of this operating decision.
However, considering the typical wind speed distribution of Fig. 2.13, the number of
hours per year is quite small when the wind speed exceeds 15m/s.
As discussed in Section 2.3 the power absorbed by the wind turbine is proportional to
the cube of the wind speed. Hence there should be a way of limiting the peak absorbed
power. Wind turbines are therefore generally designed so that they yield maximum
output at wind speeds around 15 metres per second. In case of stronger winds it is
necessary to waste part of the excess energy of the wind in order to avoid damaging the
wind turbine. All wind turbines are therefore designed with some sort of power control
to protect the machine. There are different ways of doing this safely on modern wind
turbines.
38
Fig. 2.14 Wind turbine control regions
2.6.1 Pitch control
In pitch controlled wind turbines the power sensor senses the output power of the
turbine. When the output power goes above the maximum rating of the machine, the
output power sensor sends a signal to the blade pitch mechanism which immediately
pitches (turns) the rotor blades slightly out of the wind. Conversely, the blades are
turned back into the wind whenever the wind speed drops again. On a pitch controlled
wind turbine, in order to keep the rotor blades at the optimum angle and maximise
output for all wind speeds, the pitch controller will generally pitch the blades by a small
angle every time the wind changes. The pitch mechanism is usually operated using
hydraulics [5].
2.6.2 Yaw control
Yaw control is a mechanism of yawing or tilting the plane of rotation out of the wind
direction when the wind speed exceeds the design limit. In this way the effective flow
cross section of the rotor is reduced and the flow incident on each blade considerably
modified. The effect of yawing on the power coefficient is given in Fig. 2.15 [5].
Fig. 2.15 shows a drastic drop in the power coefficient resulting from turning wind
turbines out of the wind, with consequent blade stall.
39
Power coefficient C
p
Tip speed ratio
Fig. 2.15 Power coefficient verses tip speed ratio under yaw control
2.6.3 Stall control
Under normal operating conditions, a stalled rotor blade is unacceptable. This is
because the power absorbed by the wind turbine will decrease, even to the point where
no power is absorbed. However, during high wind speeds, the stall condition can be
used to protect the wind turbine. The stall characteristic can be designed in to the rotor
blades so that when a certain wind speed is exceeded, the power absorbed will fall to
zero, hence protecting the equipment from exceeding its mechanical and electrical
ratings. In stall controlled wind turbines the angle of the rotor blades is fixed. The cross
sectional area of the rotor blade has been aerodynamically designed to ensure that the
moment the wind speed becomes too high, it creates turbulence on the side of the rotor
blade which is not facing the wind, similar to that shown in Fig. 2.8. The stall prevents
the creation of a tangential force which pulls the rotor blade to rotate. The rotor blade
has been designed with a slight twist along its length, from its base to the tip, which
helps to ensure that the wind turbine stalls gradually, rather than abruptly, when the
wind speed reaches its critical value.
The main advantage of stall control is that it avoids moving parts in the rotor blade
itself, and a complex control system. However, stall control represents a very complex
aerodynamic design problem, and related design challenges in the structural dynamics
of the whole wind turbine, e.g. to avoid stall-induced vibrations. Around two thirds of
the wind turbines currently being installed in the world are stall controlled machines
[13].
40
2.7 Wind powered electric generation
In general electrical generators for wind powered application include induction
machines and synchronous machines. The generation of electricity using wind energy
systems has found application for grid connected and stand alone systems. The capacity
of wind powered electric generation has been growing. The category of electricity
produced from wind power with respect to size is given in Table 2.1.
Type Rotor Size Electricity Produced
Micro 0.5 1.25 m 20 300 W
Mini 1.25 2.75 m 300 850W
Household 2.75 7 m 0.85 10KW
Industrial 7 30 m 10 100KW
Utility 30 90 m 0.1 4 MW
Table 2.1 Rough Categories of Wind Generator Sizes [14]
Wind projects are relatively easy to site and expand, have low environmental impacts
(including no carbon emissions) and are highly desirable to buyers of green power
[15].
Wind power has emerged as the world's fastest growing electricity generating
technology, growing by more than 40 percent annually since 1993. Fig. 2.16 shows the
worldwide growth of wind energy capacity [17]. Total world wide wind power capacity
is now estimated at more than 24,000MW [16].
Harnessing the power of the wind has a rich tradition that is enjoying resurgence due to
recent political, economic and technological developments. The strategic plan in USA is
to provide 5 percent of the nation's electricity from wind power by the year 2020. This
plan equates to 80,000 MW of new USA wind power capacity (compared to today's less
than 3,000 MW) [18, 19].
41
Fig. 2.16 Growth of wind energy capacity worldwide
One major trend driving the development of wind power is concern over global climate
change. In Europe, countries have established specific target dates for reducing carbon
dioxide (CO
2
), a pollutant which is linked to rising global temperatures. Wind power,
which is projected to reach 40,000 MW on the European continent by 2010, has become
the technology of choice to reduce CO
2
emissions in Europe and much of the rest of the
world [18].
2.8 Economics of wind powered electric generation
The increasingly attractive economics of wind energy is perhaps the most important
factor in the expansion of wind powered electric generation. The costs of wind power
have declined almost 90% since the early 1980s [19]. Between 1990 and 2000, the
average cost to produce electricity from wind turbines has decreased from around ten
cents to less than five cents per kWh in regions with very favourable wind resources.
Some projects are selling power under long-term contracts at 3.5 cents per kWh [20].
The cost depends upon the particular wind turbine, the nature of the local wind
resource, and economies of scale associated with the size of the wind farm. Prices are
projected to drop another 20 to 40 percent over the next ten years. The goal of the USA
Department of Energy (DOE) program is to get the cost down to 2.5 cents/kWh (at 6.7
m/s wind sites). However, wind turbines, particularly off-grid models, are already cost
effective throughout much of the developing world where the cost of dirty diesel-fired
42
power is often as high as 30 cents/kWh. The dramatic cost reductions are a result of
increased size and an increase in efficiency when converting kinetic wind energy into
electricity. The graphical decrease in the cost of electricity generated from wind energy
is given in Fig. 2.17 [21].
Fig.2.17 Trend in the cost of electricity generated from wind energy
2.9 Summary
The general definition of wind and the source of wind have been presented in this
chapter. The analysis of power absorbed by a wind turbine is based on the horizontal
axis wind turbine. The mechanism of production of force from wind that causes the
rotor blades to rotate in a plane perpendicular to the general wind direction at the site
has been discussed in detail. The importance of having twisted rotor blades along the
length from the base to the tip is given. The variation of the torque produced by the
wind turbine with respect to the rotor angular speed has been presented.
Power absorbed by a wind turbine is proportional to the cube of the wind speed. Wind
turbines are designed to yield maximum output power at a given wind speed. In case of
stronger winds it is necessary to waste part of the excess energy of the wind in order to
avoid damaging the wind turbine. Different ways of power control to protect the
43
machine have been presented. The economics and growth of wind powered electric
generation has been discussed and the projection for the future has been presented.
2.10 References
[1] N. P. Cheremisnoff, Fundamentals of Wind Energy, Ann Arbor Science Publishers, Michigan,
1978.
[2] L.L, Freris, Wind Conversion Systems, Prentice Hall International, London, 1990.
[3] Nick Pawsey, Private conversation with Nick Pawsey, School of Mechanical and Manufacturing
Engineering, The University of New South Wales, Sydney, Australia, 2002.
[4] WASTWIND Turbines, Australia (http://www.westwind.com.au/tubines.htm).
[5] S. Heier, Grid Integration of Wind Energy Conversion Systems, John Wiley & Sons Ltd,
Chichester, 1998, pp. 23.
[6] M. R. Patel, Wind and Solar Power Systems, CRC Press LLC, Boca Raton, 1999, pp.37.
[7] Z. Zhang, C. Watthanasarn and W. Shepherd, Application of a Matrix Converter for the Power
Control of a Variable-Speed Wind-Turbine Driving a Doubly-Fed Induction Generator, in Proc.
1997 IEEE IECON97 Conference, pp. 906-911.
[8] Danish wind Industry Association (http://www.windpower.dk/tour/wtrb/rotor.htm)
[9] Z. Zhang and C. Watthanasarn, A Matrix Converter Excited Doubly-Fed Induction Machine as a
Wind Power Generator, IEE- Power Electronics and Variable Speed Drives conference, 21-23
September 1998, pp. 532-537.
[10] http://www.tuat.ac.jp/~akilab/renewables/wind.html.
[11] R. Spee, S. Bhowmik and J. HR Enslin, Adaptive Control Strategies for Variable-Speed Doubly-
Fed Wind Power Generation System, IEEE IAS Annual Meeting Conf., 1994, pp. 545-552.
[12] J. F. Walker and N. Jenkins, Wind Energy Technology, John Wiley & Sons, New York, Inc,
1997.
[13] http://www.windpower.org/tour/wtrb/powerreg.htm.
[14] http://www.aocwind.net/wbasics.htm.
[15] M. Begovic, A. Pregelj, A. Rohatgi and C. Honsberg, Green power: status and perspectives,
Proceedings of the IEEE, Vol. 89 Issue: 12, Dec. 2001, pp. 1734 -1743.
[16] IEEE The Institute Vol.26, No.5, May 2002, pp. 1-8.
[17] National Renewable Energy Laboratory, USA Department of Energy Wind energy Information
report, May 2001.
[18] http://www.ceert.org/ip/wind.html.
[19] R.; Swisher, C.R.; De Azua and J. Clendenin, Strong winds on the horizon: wind power comes of
age, Proceedings of the IEEE, Vol. 89 Issue: 12, Dec. 2001, pp. 1757 -1764.
[20] http://www.eren.doe.gov/consumerinfo/refbriefs/ad2.html.
[21] National Renewable Energy Laboratory, USA Department of Energy Wind energy Information
report, 1997.
44
CHAPTER 3
THREE-AXES TO TWO-AXES TRANSFORMATION AND
ITS APPLICATION
3.1 Introduction
Mathematical transformations are tools which make complex systems simple to analyse
and solutions easy to find. For instance, the differential equation is transformed into its
corresponding Laplace transform representation. The Laplace equation is then simply
analyzed using algebra. Once the solution is obtained in algebraic form the inverse
Laplace transform is applied to yield the time domain solution.
In electrical machines analysis a three-axes to two-axes transformation is applied to
produce simpler expressions that provide more insight into the interaction of the
different parameters.
The different transformations studied in the past are [1]:
-In the late 1920s, R.H.Park formulated a change of variables, which, in effect, replaced
the variables (Voltages, currents, and flux linkages) associated with the stator windings
of a synchronous machine with variables associated with fictitious windings rotating
with the rotor. He transformed, or referred, the stator variables to a frame of reference
fixed in the rotor.
-In the late 1930s, H.C Stanley employed a change of variables in the analysis of
induction machines. He transformed the rotor variables to a frame of reference fixed in
the stator.
-G. Kron introduced a change of variables which eliminated the time-varying
inductances of a symmetrical induction machine by transforming both the stator
CHAPTER 3 THREE-AXIS TO TWO-AXIS TRANSFORMATION AND ITS APPLICATION
45
variables and the rotor variables to a reference frame rotating in synchronism with the
rotating magnetic field. This reference frame is commonly referred to as the
synchronously rotating reference frame.
-D. S. Brereton et al, employed a change of variables which also eliminated the time-
varying inductances of a symmetrical induction machine by transforming the stator
variables to a reference frame fixed in the rotor. This is essentially Parks
transformation applied to induction machines.
Park, Stanley, Kron and Brereton et al. developed changes of variables each of which
appeared to be uniquely suited for a particular application. Consequently, each
transformation was derived and treated separately in literature until it was noted in
1965[2] that all known transformations used in induction machine analysis are
contained in one general transformation which eliminates all time-varying inductances
by referring the stator and rotor variables to a frame of reference which may rotate at
any angular velocity or remain stationary. All known real reference transformations
may then be obtained by simply assigning the appropriate speed of rotation to this so-
called arbitrary reference frame.
3.2 General change of variables in transformation
Mathematical transformations are often used to decouple variables. Transformations
help to facilitate the solution of difficult equations with timevarying coefficients, or to
refer all variables to a common reference frame [3].
The three axes are representing the real three phase supply system. However, the two
axes are fictitious axes representing two fictitious phases perpendicular, displaced by
90
o
, to each other. The transformation of three-axes to two-axes can be done in such a
way that the two-axes are in a stationary reference frame, or in rotating reference frame.
The transformation actually achieves a change of variable, creating the new reference
frame. Transformation into a rotating reference frame is more general and can include
the transformation to a stationary reference frame. Transformation to a stationary
reference frame is one particular condition of transformation to a rotating reference
46
frame. If the speed of rotation of the reference frame is zero it becomes a stationary
reference frame.
If the reference frame is rotating at the same angular speed as the excitation frequency,
when the variables are transformed into this rotating reference frame, they will appear
as a constant value instead of time-varying values.
3.2.1 Transformation into a stationary reference frame
Here the assumption taken is that the three-axes and the two-axes are in a stationary
reference frame. It can be rephrased as a transformation between abc and stationary dq0
axes.
To visualize the transformation from three-axes to two-axes, the trigonometric
relationship between three-axes and two-axes is given below.
u

f
b
s
e
f
q
s
f
a
s
f
d
s
f
c
s
Fig. 3.1 Three-axes and two-axes in the stationary reference frame
In the above diagram, Fig. 3.1, f can represent voltage, current, flux linkage, or electric
charge. The subscript s indicates the variables, parameters, and transformation
47
associated with stationary circuits. The angular displacement u shows the displacement
of the two-axes, dq-axes, from the three-axes, abc-axes.
f
q
s
and f
d
s
variables are directed along paths orthogonal to each other. f
a
s
, f
b
s
, and f
c
s
may be considered as variables directed along stationary paths each displaced by 120
electrical degrees.
A change of variables which formulates a transformation of the three-phase variables of
stationary circuit elements to an arbitrary reference frame can be expressed as [1]:
2 2
cos cos cos -
3 3
2
2 2
sin sin sin
3
3 3
1 1 1

2 2 2
s s
q a
s s
b d
s s
c o
f f
f f
f f
t t
u u u
t t
u u u
( | | | |
+
| |
(
\ . \ .
( (
(
( (
( | | | |
=
+ ( ( | |
(
\ . \ .
( (
(

(
(

(3.1)
This transformation is a special case of the classical symmetrical components
transformation [3]. In Equation (3.1) f
o
s
is a variable that takes care of the unbalance in
the variables of the three-axes system and is the same as the zero-sequence component
in three phase system. It is important to note that the zero-sequence variables are not
associated with the arbitrary reference frame. Instead, the zero-sequence variables are
related arithmetically to the abc variables, independent of u.
It is essential not to confuse f
a
s
, f
b
s
, and f
c
s
with phasors, because phasors are used in
steady state expressions. Instead f
a
s
, f
b
s
, and f
c
s
are instantaneous quantities, which may
be any function of time. Portraying the transformation as shown in Fig. 3.1 is
particularly convenient when applying it to ac machines where the direction of f
a
s
, f
b
s
,
and f
c
s
may also be thought of as the direction of the magnetic axes of the stator
windings. They can also represent space vectors or the axes of distribution of the phase
windings. The direction of f
q
s
and f
d
s
can be considered as the direction of the magnetic
axes of the new windings created by the change of variables.
48
The inverse of Equation (3.1), which can be derived directly from the relationship given
in Fig. 3.1, is
cos sin 1
2 2
cos sin 1
3 3
2 2
cos sin 1
3 3
s s
q a
s s
b d
s s
c o
f f
f f
f f
u u
t t
u u
t t
u u
(
(
( (
| | | |
(
( (
| |
=
( \ . \ . ( (
(
( (
| | | |
(
+ +
| |
(
\ . \ .
(3.2)
The change of variables may be applied to variables of any waveform and time
sequence. However, the transformation given above is particularly appropriate for an
abc sequence. That is, as the magnetic field rotates at the speed e in the direction
shown in Fig. 3.1, it will induce an abc phase sequence in the abc axes, i.e. voltage will
be induced first in axis-a then b then in c. This shows voltage in phase a leads the
voltage in phase b and so on. And for the dq-axes, voltage will be induced first in the d-
axis then in the q-axis. That is the voltage in the d-axis will lead the voltage in the q-
axis.
In Fig. 3.1, if the q-axis is aligned with the a-axis, i.e. u = 0, Equation (3.1) will be
written as:
1 1
1
2 2
2
3 3
0
3
2 2
1 1 1

2 2 2
s s
q a
s s
b d
s s
c o
f f
f f
f f
(

(
( (
(
( (
(
=
( (
(
( (
(

(
(

(3.3)
and Equation (3.2) will be simplified to:
1 0 1
1 3
1
2 2
1 3
1
2 2
s s
q a
s s
b d
s s
c o
f f
f f
f f
(
(
( (
(
( (

= (
( (
(
( (
(
(

(3.4)
49
In Equation (3.3) and (3.4) the magnitude of the phase quantities, voltages and currents,
in the three (abc) axes and two (dq) axes remain the same. This transformation is based
on the assumption that the number of turns of the windings in each phase of the three
axes and the two axes are the same. Here the advantage is the peak values of phase
voltages and phase currents before and after transformation remain the same. The detail
of transformation for voltages and currents is given in Sections 3.3 and 3.4 respectively.
However, for the power, to have the same magnitude in the three axes and two axes,
there should be a multiplying factor 3/2 in the two axes power calculation as given in
Section 3.5.
Another way of transforming the three axes to two axes, with the magnitude of per
phase voltages and per phase currents remaining unchanged in both axes systems, is to
align the d-axis along the a-axis, u = 0 in Fig. 3.2. Fig. 3.2 is similar to Fig. 3.1 except
that the d-q axes are rotated by 90
o
while the abc axes remain the same.
u
f
b
s
e
f
a
s
f
d
s
f
c
s
f
q
s
Fig. 3.2 Three-axes and two-axes in the stationary reference frame with d-axis and
a-axis aligned
50
From Fig. 3.2, for u = 0 and taking the zero sequence current into account, the
transformation equation can be written as [4, 5]
1 1
1
2 2
2
3 3
0
3
2 2
1 1 1

2 2 2
s s
d a
s s
q b
s s
c o
f f
f f
f f
(

(
( (
(
( (
(
=
( (
(
( (
(

(
(

(3.3a)
and the inverse transform will be
1 0 1
1 3
1
2 2
1 3
1
2 2
s s
d a
s s
q b
s s
c o
f f
f f
f f
(
(
( (
(
( (
= (
( (
(
( (
(

(

(3.4b)
Whether the q-axis or d-axis is aligned along the a-axis is a matter of choice. The
difference is only a rotational translation of the d-q axes by 90
o
. In this thesis the option
of aligning the q-axis along the a-axis has been used.
There is another way of transformation from three-axes into two-axes where the
magnitude of the power remains the same but the magnitude of phase voltages and
currents in the two axes are higher than that of the three-axes [6, 7, 8]. In these
references the d-q axes are given as o-| axes. Replacing the o-| with d-q and making
some rearrangement, it can be written as
1 1
1
2 2
2
3 3
0
3 2 2
1 1 1

2 2 2
s s
q a
s s
b d
s s
c o
f f
f f
f f
(

(
( (
(
( (
(
=
( (
(
( (
(

(
(

(3.5)
Here the q-axis is aligned along the a-axis but their magnitude is different. The inverse
transformation will be
51
1
1 0
2
1 3 1 2
2 2 3 2
1 3 1

2 2 2
s s
q a
s s
b d
s s
c o
f f
f f
f f
(
(
( ( (
( ( (
=
( ( (
( ( (
(
(
(

(3.6)
In this type of transformation the transformation matrix from three-axes to two-axes is
the transpose of the inverse transformation matrix and it can be shown that
(f
a
s
)
2
+(f
b
s
)
2
+(f
c
s
)
2
= (f
q
s
)
2
+(f
d
s
)
2
+(f
o
s
)
2
. The phase voltages and currents in the two-axes
are 3 2 times the quantities in the three-axes. As a result when the power is calculated
in the two-axes the 1.5 multiplying factor is not required.
In this thesis the transformation that gives constant magnitude of voltages and currents
in the two-axes and three-axes is used. In addition the q-axis of the two axes is aligned
along the a-axis of the three axes.
3.2.2 Transformation into a rotating reference frame
The rotating reference frame can have any speed of rotation depending on the choice of
the user. Selecting the excitation frequency as the speed of the rotating reference frame
gives the advantage that the transformed variables, which had instantaneous values,
appear as constant (DC) values. In other words, an observer moving along at that same
speed will see the space vector as a constant spatial distribution, unlike the time-varying
values in the stationary abc axes.
In the previous section the transformation from abc axes to a stationary qd axes is given.
Here the stationary qd axes will be transformed into a rotating qd reference frame,
which is rotating at e
e
, excitation frequency.
To see through the eyes of the so-called moving observer is equivalent, mathematically
to resolving whatever variables that we want to see onto a rotating reference frame. This
reference frame moves at the same speed as the observer. Since we are dealing with
two-dimensional variables, the rotating reference can be any two independent basic
52
space vectors, which for convenience another pair of orthogonal qd axes will be used
[3]. The zero-sequence component remains the same as before. Fig. 3.3 shows the abc
to rotating dq transformation in two steps, ie, first transforming to stationary dq axes
and then to rotating dq axes.
f
b
s
e
at u = 0
f
a
s
f
q
s
f
d
s
f
c
s
f
d
s
f
d
e
f
q
e
f
q
s
(a) (b)
u
u =e t
Fig. 3.3 Steps of the abc to rotating dq axes transformation (a) abc to stationary dq axes
b) stationary d
s
-q
s
to rotating d
e
-q
e
axes
The equation for the abc to stationary dq-axes transformation is given in Equation (3.7).
Using geometry, it can be shown that the relation between the stationary d
s
-q
s
axes and
rotating d
e
-q
e
axes is expressed as:
cos sin
sin cos
e s
q q
e s
d d
f f
f f
u u
u u
( (
(
=
( (
(
( (

(3.7)
The angle,u, is the angle between the q-axis of the rotating and stationary d
s
-q
s
axes. u
is a function of the angular speed, e(t), of the rotating d
e
-q
e
axes and the initial value,
that is
0
( ) ( ) (0)
t
t t dt u e u = +
)
(3.8)
If the angular speed of the rotating reference frame is the same as the excitation
frequency then the transformed variables in the rotating reference frame will appear
constant (DC).
53
3.3 Voltage measurement
For a balanced, pure sinusoidal three phase supply the sum of the three phase voltages is
zero, as a result the zero-sequence voltage will be zero. The voltage transformation for
this type of source can be shown to be
1 1
1
2 2 2
3
3 3
0
2 2
a s
q
b
s
d
c
v
v
v
v
v
(
(
(
(
(
( =
(
(
(
(

(

(

(3.9)
Assume that the three phase supply voltages are given by
cos
a m e
v V t e = (3.10)
cos( 2 3)
b m e
v V t e t = (3.11)
cos( 2 3)
c m e
v V t e t = + (3.12)
Applying Equation (3.9) to the three phase voltages given above
( )
2
3
2 2
s
b c
q
a
v v
v
v
=

(3.13)
( )
2 3
[ ]
3
2
s
d
b c
v
v v
=
+
(3.14)
For a balanced system since
a b c
v v v = then (3.13) becomes
cos
s
q a m e
v v V t e = = (3.15)
Simplifying Equation (3.14)
( )
1
3
s
d c b
v v v =
( )
1
cos( 2 3) cos( 2 3)
3
m e m e
V t V t e t e t = +
Applying Eulers identity,
( ) ( ) ( ) ( ) 2 / 3 2 / 3 2 / 3 2 / 3
( ) ( )
3
2 2
e e e e
j j j j w t w t w t w t
m
V e e e e
t t t t + +
| |
+ +
=
|
\ .
(3.16)
Simplifying Equation (3.16)
sin
s
d m e
v V t e = (3.17)
For a balanced three phase system the dq voltages,
s
d
v and
s
q
v , are orthogonal and they
have the same peak values as the abc phase voltages. The explanation given in Section
54
3.2.1 indicates that the voltage in the d-axis leads the voltage in the q-axis. This is
confirmed by the results given in Equations (3.15) and (3.17).
Based on the discussion given above the dq component voltages can be written as
( ) cos sin
s s
dq q d m e e
V v jv V t j t e e = = +
e
t
dq m
V V e
e
= (3.18)
So
dq
V , the resultant voltage vector in the stationary reference frame, rotates counter
clockwise at a speed of
e
e , as shown in Fig. 3.4, from an initial position at t=0, which
is co-phase with the a-phase axis. At t=0 v
q
s
= V
dq
and v
d
s
= 0.
V
dq
v
q
s
e
e
t
v
d
s
q- axis
d- axis
Fig. 3.4 Voltage vector and its component in dq axes
For a direct on line measurement v
d
s
and v
q
s
will be calculated from the measured
instantaneous values of
a
v ,
b
v and
c
v . For a balanced three-phase supply the magnitude
of the peak phase voltage can be calculated as:
2 2 2 2 2
[(sin ) (cos ) ]
s s
dq d q m e e
V v v V t t e e = + = +
dq m
V V = (3.19)
Then the rms voltage will be
2
dq
rms
V
V = (3.20)
From Equation (3.20) it is shown that the rms voltage can be calculated from the
measured instantaneous values of
a
v and
b
v .
c
v is not included because for a balanced
system
c a b
v v v = . The rms value is readily available by taking only one set of
measurements, i.e.
a
v and
b
v , at one instant of time [9].
55
If the measured voltages are line-to-line values, like in a delta connected induction
machine, then two line-to-line voltages will be measured. The third line-to-line voltage
ca
v can be calculated from the other two measured values,
ab
v and
bc
v , as it is assumed
the supply is a balanced three-phase system.
If the instantaneous measured line to line voltages are
ab
v and
bc
v , then for a balanced
system
ca ab bc
v v v = (3.21a)
and based on the fact that
a
v +
b
v +
c
v =0 (3.21b)
ab b a
v v v = (3.21c)
bc b c
v v v = (3.21a)
ca c a
v v v = (3.21a)
the phase quantities can be calculated from line quantities using the above equations to
give
( )
1
3
a ab ca
v v v = (3.22)
( )
1
3
b bc ab
v v v = (3.23)
( )
1
3
c ca bc
v v v = (3.24)
3.4 Current measurement
When the general transformation given in Equation (3.3) is applied to three phase
currents the equation becomes
1 1
1
2 2
2
3 3
0
3
2 2
1 1 1

2 2 2
s
q
a
s
b
d
s
c
o
i
i
i
i
i
i
(

(
(
(
(
(
(
(
=
(
(
(
(
(
(

(
(

(3.25)
56
If the three-phase currents are balanced sinusoids which are 120 electrical degrees apart,
the sum of the three currents flowing in the phases of the system is zero. Therefore the
zero-sequence current in the system, provided there is a path for the zero-sequence
current, will be zero and Equation (3.25) is simplified to
1 1
1
2 2 2
3
3 3
0
2 2
a s
q
b
s
d
c
i
i
i
i
i
(
(
(
(
(
( =
(
(
(
(

(

(

(3.26)
Then,
( )
2
3
2 2
s
b c
q
a
i i
i
i
=

| | ( )
2 3
3
2
s
d
b c
i
i i
=
+
Since i
a
+i
b
+i
c
=0, the above equations can be reduced to
s
q a
i i = (3.27)
( )
1
3
s
d c b
i i i = (3.28)
Corresponding to the three phase voltages given in equations (3.10) to (3.12), the
currents flowing in the system may be described as
cos( )
a m e
i I t e | = (3.29)
cos( 2 3 )
b m e
i I t e t | = (3.30)
cos( 2 3 )
c m e
i I t e t | = + (3.31)
Where | is the phase shift between the voltage and the current for a lagging power
factor. Substituting the instantaneous values of the currents in the abc axes into the dq
axes
cos( )
s
q m e
i I t e | = (3.32)
| |
1
cos( 2 3 ) cos( 2 3 )
3
s
d m e m e
i I t I t e t | e t | = +
( ) ( ) ( ) ( ) 2 / 3 2 / 3 2 / 3 2 / 3
( ) ( )
3
2 2
e e e e
j w t j w t j w t j w t
s
m
d
I e e e e
i
t | t | t | t | + +
| |
+ +
=
|
\ .
(3.33)
Simplifying the above equation results in,
57
sin( )
s
d m e
i I t e | = (3.34)
To compare the phase of
s
d
i with the phase of
s
q
i ,
s
d
i can be expressed as
cos( 2 )
s
d m e
i I t e t | = + . This shows that the current in the d-axis leads that in the q-
axis by 2 t . The peak value of
s
q
i and
s
d
i is the same as the peak value of
a
i ,
b
i and
c
i .
The resultant current in the stationary dq axes,
dq
I , can be written in terms of the
components in the d-axis and the q-axis as,
( ) cos( ) sin( )
s s
dq q d m e e
I i ji I t j t e | e | = = +
( )
e
j t
dq m
I I e
e |
= (3.35)
Equation (3.35) highlights the fact that
dq
I is a vector with a magnitude of I
m
and rotates
at the excitation angular frequency
e
e . The components of
dq
I along the d-axis and q-
axis vary with time. It can also be stated that the magnitudes of the components along
the d-axis and the q-axis are instantaneous values, similar to the three-phase currents in
the abc axes. The magnitude of I
dq
can be calculated as
2 2 s s
dq d q
I i i = +
2 2 2
[(sin( )) (cos( ))
dq m e e
I I t t e | e | = +
dq m
I I = (3.36)
The currents in the stationary dq-axes can be shown as in Fig. 3.5.
q-axis
i
d
s
d-axis
i
q
s
I
dq
e
e
t-|
Fig. 3.5 Current vector and its component in stationary dq axes
58
Since the magnitude of I
dq
is equal to I
m
and is the same as the peak magnitude of phase
current in the abc-axes, the rms current can be evaluated from the instantaneous values
in the dq-axes. Therefore using Equation (3.36)
2
dq
rms
I
I = (3.37)
If the system is a balanced three-phase system, then only two phase currents (i
a
and i
b
)
are required to be measured, the third one (i
c
) can be derived from the assumption that
the three-phase currents are 120 electrical degrees apart and instantaneously add to zero.
Taking one sample of instantaneous values of currents flowing in any two phases of a
three-phase system, the rms and the peak currents of the three-phase system can be
obtained instantaneously.
3.5 Power measurement
The transformation from three axes to two axes is done based on the concept that the
peak values of the phases in three axes as well as two axes are the same. The total
power in the system under consideration should remain the same regardless of the
choice of reference frame and choice of axes.
Since the voltages and currents in the three axes have the same peak values to the
currents and voltages in the two axes, the power in the two-axes system should be
multiplied by a factor 3/2 so that the transformation will keep the value of total power
the same.
Fig. 3.6 shows the voltage and current vectors with their components in the stationary
dq-axes. Once the components of the currents and voltages are calculated in the d and q
axes then power is evaluated as
3
( )
2
s s s s
d d q q
P i v i v = + (3.38a)
59
If the currents and voltages are substituted in Equation (3.38a) with the expressions
given in the voltage and current measurement sections then the classical power
expression becomes
3
cos
2
m m
P I V | = (3.38b)
q-axis
d-axis
V
dq
v
d
s
I
dq
e
e
t-|
i
d
s
i
q
s
v
q
s
|
e
e
t
Fig. 3.6 Voltage and current vectors with their components in the stationary dq-axes
The two axes can be visualized as a machine having two windings. The power in the
two axes system is related to that of the three axes system by the factor 3/2. Therefore,
with the currents and voltages in the dq-axes, calculated from the instantaneous values
of two-phase currents and two-phase voltages, the value of the active power (average
power) can be computed instantaneously.
This is quite a remarkable result. Traditionally to calculate the average power, measured
values are averaged over one cycle. With the method described above, the average
power can be calculated from a single set of instantaneous voltages and currents
measured at a single instant in time.
60
3.6 Power factor measurement
Any circuit with effective resistive and reactance will have a phase shift, |, between the
voltage vector and the current vector. To transform the voltages and currents into a
rotating dq axes reference frame Equation (3.7) is applied in the following way
cos sin
sin cos
e s
q q
e s
d d
v v
v v
u u
u u
( (
(
=
( (
(
( (

(3.39)
cos sin
sin cos
e s
q q
e s
d d
i i
i i
u u
u u
( (
(
=
( (
(
( (

(3.40)
If it is assumed that the reference frame is rotating at the excitation frequency, e
e
, then
u =e
e
t and for voltages
cos sin
e s s
q q d
v v v u u = (3.41)
sin cos
e s s
d q d
v v v u u = + (3.42)
With u =e
e
t and substituting the expressions cos
s
q dq e
v V t e = and sin
s
d dq e
v V t e = into
Equations (3.41) and (3.42), then simplifying gives
e
q dq
v V =
0
e
d
v =
Hence with the given arrangement and analysis of the rotating reference frame, the q-
axis voltage in the rotating reference frame,
e
q
v
,
and the rotating voltage space vector,
dq
V , are in the same axis and they are equal. That is why the d-axis component of the
voltage vector is always zero in the rotating reference frame rotating at the excitation
frequency e
e
. The AC voltage in the stationary reference frame appears as if it is DC or
a constant value in the rotating reference frame.
For the currents, if it is assumed that the reference frame is rotating at the excitation
frequency, e
e
, then Equation (3.40) will become
cos sin
e s s
q q e d e
i i t i t e e = (3.43)
sin cos
e s s
d q e d e
i i t i t e e = + (3.44)
Substituting cos( )
s
q dq e
i I t e | = and sin( )
s
d dq e
i I t e | = in Equation (3.43)
61
cos( ) cos [ sin( )]sin
e
q dq e e dq e e
i I t t I t t e | e e | e =
[cos cos sin sin ]cos [sin cos sin cos ]sin
e
q dq e e e dq e e e
i I t t t I t t t e | e | e e | | e e = + +
2 2
[cos cos sin sin cos sin cos sin cos sin ]
e
q dq e e e e e e
i I t t t t t t e | e | e e | | e e = + +
cos
e
q dq
i I | = (3.45)
Applying the same analysis to Equation (3.44) gives
sin
e
d dq
i I | = (3.46)
From Equation (3.45) it can be seen that the component of the current in the q-axis
contains the term cos| which is the power factor of the system
The expressions cos
s
q dq e
v V t e = and sin
s
d dq e
v V t e = lead to
cos
s
q
e
dq
v
t
V
e = (3.47)
sin
s
d
e
dq
v
t
V
e
= (3.48)
Substituting Equations (3.47) and (3.48) in Equation (3.43)
e
q
i can be rewritten as,
s
s
e s s q
d
q q d
dq dq
v
v
i i i
V V
= + (3.49)
and from Equation (3.45)
cos
e
q
dq
i
I
| = (3.50)
Applying the above equation, the power factor of the system can be calculated
instantaneously from data measured at any single instant of time.
3.7 Frequency measurement
For a balanced three-phase system it is enough to have measured the instantaneous
values of voltages and currents from two phases to determine all the required values
including frequency. If the measured instantaneous voltages and currents are given by
cos
a m e
v V t e =
cos( 2 3)
b m e
v V t e t =
cos( )
a m e
i I t e | =
cos( 2 3 )
b m e
i I t e t | =
62
The equations given above have four known values
a
v ,
b
v ,
a
i and
b
i , which are
measured at a single instant of time, and four unknowns
m
V ,
e
t e ,
m
I and| . Having four
equations and four unknowns there is a solution for all the unknowns. The three axes to
two axes transformation helps to simplify the mathematics in attaining this objective.
The information for the excitation frequency e
e
is concealed in the e
e
t term.
Differentiating e
e
t with respect to time give e
e
from which the frequency can be
obtained as f
e
=e
e
/2t.
Differentiation can be implemented by having two measurements at a small interval of
time and experimental differentiation is implemented as the difference between two
consecutive measurements divided by the sampling time.
Applying Equation (3.47) at two instants of time with a small interval of time between
the two measurements
1
1
cos
s
q
e
dq
v
t
V
e = ===>
1
1
1
cos
s
q
e
dq
v
t
V
e
| |
=
|
\ .
(3.51)
2
2
cos
s
q
e
dq
v
t
V
e = ===>
1
2
2
cos
s
q
e
dq
v
t
V
e
| |
=
|
\ .
(3.52)
Then
2 1
2 1
( )
( )
e e
e
t t
t t
e e
e
(3.53)
Here t
2
-t
1
can be replaced by a sampling time T
s
of the measurement and T
s
should be
less than one fourth of the period of the frequency to be measured. For power frequency
it is more than adequate to have a sampling time less than one millisecond.
With a sampling time of T
s
,
1 1
2 1
cos cos
s s
q q
dq dq
e
s
v v
V V
T
e

| | | |
| |
\ . \ .
= (3.54)
63
3.8 Measurement in a balanced non sinusoidal three phase system
It has been explained above that for a sinusoidal (free of harmonics) balanced three
phase system the electrical quantities can be measured at a single instant of time from
only one set of measurements. However, most systems contain harmonics.
Fourier analysis states that any periodic waveform consists of a summation of
sinusoidal waveforms that are integral multiples of the fundamental. These integral
multiples of the fundamental are harmonics. A signal with harmonics will have a
distorted sinusoidal waveform [10]. To identify and isolate each harmonic the original
signal can be filtered using a digital filter. The digital filter technique will be discussed
in Chapter 5. The harmonic filters are easily instituted in software. These digital filters
are applied to the data produced by a Digital Signal Processor (DSP). This can be done
on-line in real-time using filters that are generally supplied with most DSP cards or off-
line using filters generated in some mathematical analysis software, such as the
MATLAB software package.
Using different digital filters all harmonics with magnitude greater than a given value
can be separated. Then each three phase harmonic will be a pure sinusoidal waveform.
To transform the three axes to two axes Equations (3.9) and (3.26) are applied for
voltages and currents respectively. In the two axes the rms quantities of each harmonic
can be calculated individually.
The rms quantity for each harmonic of h
th
order is expressed as
h _ dq
h _ rms
f
f
2
= (3.55)
where
h _ rms
f -rms value of the h
th
harmonic
h _ dq
f -the peak value
f - represents either voltage or current
The total rms value of a distorted sinusoidal waveform is calculated from each harmonic
and is given by [11]
( )
2 2 2 2
rms 1_ rms 2 _ rms 3 _ rms h _ rms
f f f f ... f = + + + + (3.55)
64
And the total power measured from the distorted waveform is[12]
1 2 3 h
P P P P ... P = + + + + (3.56)
where P
h
- power due to h
th
harmonic.
The space vector of the fundamental signal of the distorted waveform rotates at angular
speed e
e
. Consequently the angular frequency of the space vector of the h
th
harmonic
will be he
e
.
3.9 Summary
The three-axes to two-axes transformation presented in this chapter is applicable for any
balanced three-phase system. It has been discussed that the three-axes to two-axes
transformation simplifies the calculation of rms current, rms voltage, active power and
power factor in a three-phase system. Only one set of measurements taken at a single
instant of time is required when using the method described to obtain rms current, rms
voltage, active power and power factor. And from measurements taken at two
consecutive instants in time the frequency of the three-phase AC power supply can be
evaluated. Existing electrical measuring methods, such as the Fast Fourier Transform,
require many samples from a significant period of the measured waveforms cycle to be
processed by elaborate computation techniques in order to evaluate rms or peak
magnitudes of AC currents and voltages. These traditional methods are unable to obtain
peak values in less than one quarter of a cycle. However using the three-axes to two-
axes transformation it is possible to evaluate the rms or peak magnitudes of three-phase
AC currents and voltages from one set of measurements taken at a single instant of
time.
3.10 References
[1] P.C. Krause, O. Wasynczuk and S. D Sudhoff, Analysis of Electric Machinery, IEEE press, New
York, 1995.
[2] P.C. Krause and C. H. Thomas, Simulation of Symmetrical Induction Machinery, IEEE Trans.
Power Apparatus and Systems, Vol. 84, November 1965, pp.1038-1053.
[3] Chee-Mun Ong, Dynamic Simulation of Electric Machinery using Matlab/Simulink, Prentice-
Hall Inc., Upper Saddle River, New Jersey, 1998.
65
[4] Peter Vas, Sensorless Vector and Direct Torque Control, Oxford University Press, New York,
1998.
[5] Andrzej M. Trzynadlowski, The field Orientation Principle in Control of Induction Motors,
Kluwer Academic Publisher, Boston, 1994.
[6] H. Akagi, Y. Kanazawa and A Nabae, Generalized Theory of the Instantaneous Reactive Power in
Three-Phase Circuits, IPEC83-Int. Power Electronics Conf., Tokyo, Japan, 1983, pp. 1375-1386.
[7] H. Akagi, Y. Kanazawa and A Nabae, Instantaneous Reactive Power Compensator Comprising
Switching Devices without Energy Storage Components, IEEE Trans. Industry Application,
vol.20, May/June 1984, pp. 625-630.
[8] J. Afonso, C. Couto and J. Martins, Active Filters with Control Based on the p-q Theory, IEEE
Industrial Electronics Society Newsletter, September 2000, pp. 5-11.
[9] C. Grantham, D. Seyoum and H. Tabatabaei-Yazdi, Very fast and accurate Electrical
Measurements, AUPEC/EECON99 Conference Proceedings, 26-29 September 1999, Darwin,
Australia, pp 99-103.
[10] D. McKinnon, D. Seyoum and C. Grantham, "Rapid Determination of Fundamental and Harmonic
RMS Quantities in a 3-Phase System, Proc. AUPEC 2001, Perth, 23 - 26 September. 2001, pp. 73 -
78.
[11] Muhammad H. Rashid, Power Electronics Circuits, Devices and Applications, Prentice-Hall Inc,
New Jersey, 1993.
[12] L. Peretto, R. Sasdelli and G. Serra, Measurement of harmonic losses in transformers supplying
nonsinusoidal load currents, IEEE Transactions on Instrumentation and Measurement, Vol. 49, No.
2, April 2000, pp. 315 319.
66
CHAPTER 4
INDUCTION MACHINE MODELLING
4.1 Introduction
The main aspect which distinguishes the induction machine from other types of electric
machines is that the secondary currents are created solely by induction, as in a
transformer, instead of being supplied by a DC exciter or other external power source,
through slip rings or a commutator, as in synchronous and DC machines. Depending on
the condition of operation, the induction machine can be used as a motor or generator.
Induction machines are available in single-phase or three-phase winding configurations.
In this thesis the modelling and investigation is given only for the three-phase induction
machine.
When the stator is excited from a balanced three-phase supply, the three phases together
create a constant magnitude, synchronously revolving mmf or field in the air gap with a
crest value 3/2 times the peak value of the alternating field due to one phase alone [1].
This field rotates around the air-gap at synchronous speed N
e
, which can be calculated
as
60
e
e
p
f
N
P
= (4.1)
Where f
e
- excitation frequency in cycles per second (Hz)
P
p
- number of pole pairs
N
e
- synchronous speed in revolutions per minute (rpm)
N
e
is also expressed as the rotational speed of the stator magnetic field, or mmf.
The slip of a motor, s, which is defined as the slip of the rotor with respect to the stator
magnetic field, can be given as
CHAPTER 4 INDUCTION MACHINE MODELLING
67
e r
e
N N
s
N
= (4.2a)
where N
r
- the rotational speed of the rotor in rpm.
If the speeds are expressed in radians per second the slip is given by
e r
e
s
e e
e
= (4.2b)
where e
e
- synchronous speed in radians per second (rad/sec)
e
r
- rotor speed in rad/sec.
4.2 Conventional induction machine model
The relative speed between the synchronous speed and the rotor speed is expressed in
its equivalent electrical speed as e
e
-e
r
or se
e
, where the electrical rotor speed is the
product of the mechanical speed and the number of pole pairs.
Rotation of the rotor changes the relationships between stator and rotor emfs. However,
it does not directly change the inductance and resistance parameters. The angular
frequency of the induced current in the rotor is se
e
and the induced voltage in the rotor
will be sE
r
, where E
r
is the induced voltage in the rotor when the rotor is stationary.
This is based on the assumption that the induction machine is only supplied from the
stator terminals.
Assuming that the winding is distributed sinusoidally in angular space around the stator
to produce a sinusoidally distributed magnetic field [2] and the rotor winding is similar
in form to the stator winding, then the per-phase equivalent circuit of the stator side of
the three-phase induction machine can be represented as follows
V
s
je
e
L
ls
R
s
E
s
I
s
Fig. 4.1 Stator side of the per-phase equivalent circuit of a three-phase induction
machine
where V
s
- stator voltage, V
68
I
s
- stator current, A
R
s
- stator winding resistance, O
L
ls
- stator leakage inductance, H
E
s
- induced emf in the stator winding due to the rotating magnetic field that
links the stator and rotor windings, V
e
e
- stator current angular frequency, rad/sec
For constant stator flux the voltage induced in the rotor depends solely on the slip,
which is the relative speed between the stator flux rotating at synchronous speed and
rotor speed. Maximum induced voltage occurs in the rotor when the rotor is stationary.
Without any external input on the rotor side, the rotor circuit is given by
sE
r
jse
e
L
lr
R
r
I
r
Fig. 4.2a Rotor side of the per-phase equivalent circuit of a three-phase induction
machine
where sE
r
- induced voltage in the rotor, V
I
r
- rotor current, A
R
r
- rotor winding resistance, O
L
lr
- rotor leakage inductance, H
se
e
- rotor current angular frequency, rad/sec.
If all the terms in the rotor side are divided by the slip, s, a modified circuit is obtained
as shown in Fig. 4.2b.
E
r
je
e
L
lr
R
r
/s
I
r
Fig. 4.2b Rotor side of the induction machine with adjustment
69
Using the appropriate voltage transformation ratio between the stator and rotor, the
rotor voltage, E
r
, referred to the stator is then equal to E
s
, in Fig. 4.1. The stator and
rotor circuits are linked because of the mutual inductance L
m
. When all circuit
parameters are referred to the stator, the stator and rotor circuits can be combined to
give the circuit shown in Fig. 4.3.
V
s
R
s
I
s
R
r
/s
I
r
je
e
L
m
je
e
L
ls
je
e
L
lr
Fig. 4.3 Per-phase equivalent circuit of three-phase induction machine
neglecting core loss
In Fig. 4.3 the core loss, which is due to hysteresis and eddy current losses, is neglected.
It can be compensated by deducting the core loss from the internal mechanical power at
the same time as the friction and windage losses are subtracted [3]. The no load current
in three-phase induction machines consists of the iron loss or core loss component and
the magnetizing component. From the iron loss current component and from the applied
voltage the equivalent resistance for the excitation loss can easily be calculated. There
is also some core loss in the rotor. Under operating conditions, however, the rotor
frequency is so low that it may reasonably be assumed that all core losses occur in the
stator only [4].
The core loss can be accounted for by a resistance R
m
in the equivalent circuit of the
induction machine [18]. R
m
is dependent on the flux in the core and frequency of
excitation. For constant flux and frequency R
m
remains unchanged. As R
m
is
independent of load current it is connected in parallel with the magnetising inductance
L
m
. The equivalent circuit including R
m
is shown in Fig. 4.4.
V
s
je
e
L
ls
R
s
I
s
je
e
L
lr
R
r
/s
I
r
je
e
L
m
R
m
I
m
I
c
70
Fig. 4.4 Per-phase equivalent circuit of three-phase induction machine
including core loss
4.3 D-Q axes induction machine model
Using the D-Q representation, the induction machine can be modelled as shown in
Fig.4.5. This representation is a general model based on the assumption that the supply
voltage can be applied to both the stator and/or rotor terminals. In squirrel cage
induction machines voltage is supplied only to the stator terminals. In general power
can be supplied to the induction machine (induction motor) or power can be extracted
from the induction machine (induction generator). It all depends on the precise
operation of the induction machine. If electrical power is applied to the stator of the
induction machine then the machine will convert electrical power to mechanical power.
As a result the rotor will start to rotate and the machine is operating as a motor. On the
other hand, if mechanical power is applied to the rotor of the induction machine then the
machine will convert mechanical power to electrical power. In this case the machine is
operating as an induction generator. When the induction machine operating as a
generator is connected to the grid or supplying an isolated load, driven by an external
prime mover, then the rotor should be driven above synchronous speed.
When the machine is operated as a motor, power flows from the stator to the rotor,
crossing the air gap. However, in the generating mode of operation, power flows from
the rotor to the stator. Only these two modes of operation are dealt with in this
investigation. The braking region, where the rotor rotates opposite to the direction of the
rotating magnetic field, is not dealt with here.
The conventional model and the d-q (or D-Q) axes model are the same for steady state
analysis. The advantage of the d-q axes model is that it is powerful for analysing the
transient and steady state conditions, giving the complete solution of any dynamics.
The general equations for the d-q representation of an induction machine, in the
stationary stator reference frame, are given as [5]:
71
0 0
0 0
- -
qs qs
s s m
ds s s m ds
qr m r m r r r r qr
r m m r r r r
dr dr
v i
R pL pL
v R pL pL i
v pL L R pL L i
L pL L R pL
v i
e e
e e
( ( + (
( (
(
+
( (
(
=
( (
(
+
( (
(
+
( ( (

(4.3)
where R
s
- stator winding resistance, O
R
r
- rotor winding resistance, O
L
m
- magnetising inductance, H
L
s
- stator leakage inductance (L
ls
m
), H
L
r
- rotor leakage inductance (L
lr
m
), H
e
r
- electrical rotor angular speed in rad/sec and p=d/dt, the differential operator.
v
dr
v
ds
v
qr
v
qs
i
qs
i
qr
i
dr
i
ds
Q-axis
D-axis
e
r
Fig. 4.5 D-Q representation of induction machine
Equation (4.3) can be written in a first order differential equation form as given in the
following matrix equation
72
2
r m qs qs
r s m r m r m r r
2
r m ds ds
m r s s m r r m r
qr qr m s m s s r m s r s r r
m m dr dr s r m m s s r r s r
L 0 L 0 pi v
L R L L R L L
0 L 0 L pi v
L L R L L L R 1 1
pi v L L L 0 L 0 L R L L L R L L
0 L 0 L pi v L L L R L L L R
e e
e e
e e
e e
( ( ( (
( ( (
( ( (
= +
( ( (

( ( (
( ( (

qs
ds
qr
dr
i
i
i
i
(
( (
( (
( (
( (
( (

(4.4)
where
2
s r m
L L L L = .
Using the matrix shown in Equation (4.3), the d-q representation given in Fig. 4.5 can
be redrawn in detail, in a stationary stator reference frame, with separate direct and
quadrature circuits as shown in Fig. 4.6.
L
ls
R
s
i
ds
L
lr
i
dr
L
m
i
md
R
r
dr
ds
v
ds
v
dr
-e
r
qr
+ -
(a)
L
ls
R
s
i
qs
L
lr
i
qr
L
m
i
mq
R
r
qr
qs
v
qs
v
qr
e
r
dr
+ -
(b)
Fig. 4.6 Detailed d-q representation of induction machine in stationary reference
frame (a) d-axis circuit (b) q-axis circuit
From the stator side (for simplicity the superscript s which indicates stationary
reference frame is not included with the currents, voltages and fluxlinkages)
ds s ds m dr
L i L i = + (4.5)
qs s qs m qr
L i L i = + (4.6)
ds
ds s ds
d
v R i
dt
= + (4.7)
qs
qs s qs
d
v R i
dt
= + (4.8)
73
From the rotor side
dr m ds r dr
L i L i = + (4.9)
qr m qs r qr
L i L i = + (4.10)
dr
dr r dr r qr
d
v R i
dt
e = + + (4.11)
qr
qr r qr r dr
d
v R i
dt
e = + (4.12)
For the air gap flux linkage
dm m md m ds m dr
L i L i L i = = + (4.13)
qm m mq m qs m qr
L i L i L i = = + (4.14)
The stator electrical input power to the induction machine during motoring operation or
the stator electrical output power in generating mode is given by
( )
3
2
e ds ds qs qs
P i v i v = + (4.15)
The electromagnetic torque T
e
generated by the induction machine is given by [6]
3
2
e p m r
T P I =
JJG JG
(4.16)
where
m
JJG
- air gap flux linkage
r
I
JG
- rotor current space vector
P
p
- number of pole pairs of the induction machine.
Solving the cross product in Equation (4.15) gives
( )
3
2
e p m qs dr ds qr
T P L i i i i = (4.17)
The mechanical equation in the motoring region is
m
e m m
d
T J D T
dt
e
e = + + (4.18)
and in the generating region it is given as
m
m e
d
T J D T
dt
e
e = + + (4.19)
74
where T
m
- mechanical torque in the shaft, Nm
Te - electromagnetic torque, Nm
e
m
- mechanical shaft speed (e
m
= e
r
/P
p
), rad/sec
D - friction coefficient, Nm/rad/sec
J - inertia, Kg-m
2
.
The mechanical power generated during motoring or the mechanical power required to
drive the induction generator is given by
m e m
P T e = (4.20)
4.4 Simulation of induction machine
The simulation of the induction machine shows its characteristics in the motoring and
generating regions without the need to use a hardware experimental setup, which is time
consuming, space demanding, noisy and expensive. Also, at low rotor speeds, away
from synchronous speed, the induction machine will draw high current compared to its
rated value. If the data acquisition takes a long time the induction machine may be
damaged due to excessive temperature rise. The simulation is useful for finding the
characteristics of the induction machine from standstill to twice the synchronous speed
or for any rotor speed. In order to find the characteristics of a real induction machine in
the motoring and generating region, an experimental setup such as shown in Fig. 4.7 is
used. With a fast data acquisition system the conventional meters can be replaced with
sensors so that all the data is measured in a personal computer, quickly and accurately,
using a Digital Signal Processing (DSP) interface.
The induction machine used in this investigation is a three-phase wound rotor induction
motor with specification: 415V, 7.8A, 3.6kW, 50Hz, and 4 poles. The simulation of the
induction machine is done using the d-q equivalent circuit because it provides the
complete solution, comprising of both the transient and the steady state. The parameters
of the induction machine obtained using the standstill test and the no-load test data at
rated values of voltage, current and frequency are L
ls
= L
lr
= 12mH, L
m
= 181mH,
R
s
= 1.6O, R
r
= 2.75O.
75
Three-phase
supply
Ammeter and
Voltmeter
d
Ammeter, Voltmeter and
Watmeter
Speed meter
Induction
machine
DC machine
DC Supply
Fig. 4.7 Experimental setup to find the characteristics of induction machine in the
motoring and generating regions
The simulation is carried out based on the fact that when the induction machine is
operating as a motor it drives the DC machine so that the DC machine is operating as a
generator. When the induction machine is operating as a generator it is driven by the
DC machine, operating as a DC motor. Since the area of interest is in the characteristics
of the induction machine the investigation is done only for the induction machine. The
DC machine is used as a prime mover and as a load for the induction machine. In the
simulation, to avoid the transients in the motoring region, the induction machine is
started by rotating in the reverse direction. However, the region of operation
investigated here is between zero speed and twice the synchronous speed. A constant
full voltage is applied and the induction machine is allowed to accelerate at constant
acceleration. The system takes 10 seconds to accelerate from standstill (zero speed) to
twice the synchronous speed. In this case the synchronous speed is 1500rpm (4-pole,
50Hz machine).
Using Equations (4.4) and (4.16) a model is developed in Matlab/Simulink. The
simulation model has two inputs namely stator voltage and rotor speed. As the rotor
terminals of the induction machine are shorted there is no electrical input from the rotor
side. With the help of the developed model the necessary characteristics of the induction
machine are obtained.
76
Whether the induction machine is operating as a motor or a generator, it needs an
exciting current (reactive current) to magnetise the core and produce the required
amount of magnetomotive force (mmf). This mmf will produce an equivalent amount of
magnetising flux linkage determined by the air gap electromotive force (emf), which is
dependent on the stator terminal voltage and the voltage drop in the stator resistance
and stator leakage inductance.
77
0 1 2 3 4 5 6 7 8 9 10
-500
-400
-300
-200
-100
0
100
200
300
400
500
Phase voltage(V) and 10*Phase current(A) in stationary reference frame
10*Current
Voltage
time(sec)
A
B C
2.62 2.63 2.64
-500
0
500
5 5.01 5.02
-500
0
500
6.64 6.65 6.66
-500
0
500
Voltage
10*Current
Detail of A
Detail of B
Detail of C
10*Current
10*Current
Voltage
Voltage
(a)
(b) (c)
(d)
Fig. 4.8 Variation of stator phase current for constant supply voltage and frequency
(a) Current and voltage when the rotor speed is varied from standstill to twice the
synchronous speed (b) detail of motoring region (c) detail around the synchronous
speed (d) detail in the generating region
From Fig. 4.8 it can be seen that the phase current lags the voltage by an angle less than
90
o
in the motoring region. Close to synchronous speed the current lags by about 90
o
78
and in the generating region the lag angle is greater than 90
o
. The time varying phase
current shown in Fig. 4.8 can be i
qs
or i
ds
and for the voltage it can be v
qs
or v
ds
. The
currents i
qs
and i
ds
correspond to the voltages v
qs
and v
ds
respectively. The angle between
the phase current and the phase voltage remains the same in the q-axis and the d-axis as
well as in the three axis representation a, b and c.
V
I
u
(a)
V
I
u
(b)
V
I
u
(c)
Fig. 4.9 Relationship between phase voltage vector and phase current vector
(a) in the motoring region (b) between motoring and generating (at synchronous speed)
(c) in the generating region
In Fig. 4.9, V is the stator voltage and I is the stator current. In the two axis
representation V can be V
ds
or V
qs
and in three axis representation V is the phase voltage
and can be V
a
, V
b
or V
c
and I represents phase current, i.e. I
ds,
I
qs,
I
a
, I
b
or I
c
.
The stator electrical input/output power, the mechanical output/input power, the
electromagnetic torque and efficiency are given in Fig. 4.10. The electrical input power
and the mechanical output power are considered as positive during motoring operation.
Hence during generating operation the electrical output power and the mechanical input
power are negative because the flow of power is reversed. Positive slip is motoring and
negative slip is generating.
79
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
25
50
75
100
E
f
f
i
c
i
e
n
c
y

(
%
)
Slip
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-100
0
100
T
o
r
q
u
e

(
N
m
)
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-10
0
10
20
P
e

(
K
W
)
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-20
-10
0
10
P
m

(
K
W
)
(a)
(b)
(c)
(d)
Fig. 4.10 Induction machine torque, power and efficiency characteristics
(a) torque (b) electrical power (c) mechanical power (P
m
=e
m
T
e
) (d) efficiency
The efficiency calculation is only taking into account the losses in the induction
machine and is related to the electrical power, P
e
, and mechanical power, P
m
, in the
equations following. It does not include the friction and windage loss in the system nor
does it include stray losses. The efficiency in the motoring region is calculated by
100
m
e
P
Efficiency
P
=
and for the generating region
100
e
m
P
Efficiency
P
=
80
The instantaneous space vector angles and magnitudes of the stator voltage V
s
, stator
current I
s
, magnetising current I
m
, rotor current I
r
, and stator flux linkage
s
are
calculated using
2 2
f d q
f f u = + and
1 q
d
f
f tan
f
| |
=
|
\ .
Where f can be V
s
, I
s
, I
m
, I
r
, or
s.
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-120
-90
-60
-30
0
Slip
Magnetising current space vector angle
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-150
-120
-90
-60
-30
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-90
-60
-30
0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
30
60
90
120
150
180
Rotor current space vector angle
Stator flux likage space vector angle
Stator current space vector angle
u
s
(deg)
u
I s
(deg)
u
I r
(deg)
u
I m
(deg)
(a)
(d)
(b)
(c)
Fig. 4.11 Space vector angles measured with respect to the stator voltage space vector
angle for (a) stator current I
s
(b) stator flux linkage
s
(c) rotor current I
r
(d) magnetising current I
m
81
To have a comparison between the different space vectors, all the space vector angles
are calculated with respect to the stator voltage space vector angle. Fig. 4.11 shows the
space vector angle for stator current, stator flux linkage, rotor current and magnetising
current.
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
20
40
60
I
r

(
A
)
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
2
4
6
I
m

(
A
)
Slip
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
200
340
V
s

(
V
)
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
20
40
60
I
s

(
A
)
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
0.5
1
s

(
w
e
b
-
T
u
r
n
)
(a)
(b)
(c)
(d)
(e)
Fig. 4.12 Magnitude of space vector for (a) stator voltage (b) stator current I
s
(c) stator flux linkage
s
(d) rotor current I
r
(e) magnetising current I
m
The magnitude of stator voltage, stator current, stator flux linkage, rotor current and
magnetising current are shown in Fig. 4.12. Using the angles given in Fig. 4.11 and the
82
magnitudes given in Fig. 4.12, the space vectors for stator voltage, stator current, stator
flux linkage, rotor current and magnetising current are shown in Fig. 4.13.
q
s
d
s
V
s
I
s
I
r
d
e
q
e
s
I
m
e
e
u
e
(a)
u
Ir
u
Im
u
s
u
Is
q
s
d
s
V
s
I
s
I
r
d
e
q
e
s
I
m
e
e
u
e
(b)
u
Ir
u
Im
u
s
u
Is
Fig. 4.13 Space vector diagram for stator voltage, stator current, rotor current,
magnetising current and stator flux linkage (a) during motoring mode
(b) during generating mode
From Fig. 4.10d maximum efficiency occurs close to synchronous speed and efficiency
is zero at synchronous speed (s = 0) and at standstill when the rotor is stationary (s = 1).
83
In Fig. 4.11 it can be observed that compared to the angle of the stator voltage space
vector, the stator current space vector always lags. Because of the stator winding
resistance, the maximum magnitude of angle for the stator flux linkage is close to 90
o
lag but not exactly 90
o
. Close to synchronous speed the rotor current is more resistive,
and there is a reversal of current when it goes from the generating mode to the motoring
mode. The magnetising space vector, which is the sum of the stator current space vector
and the rotor current space vector, has an angle of 90
o
lag close to synchronous speed.
As can be seen in Fig. 4.12, the magnitude and frequency of the terminal voltage space
vector are kept constant during the motoring and generating modes. There is a minimum
stator current space vector at synchronous speed. The magnitude of the stator flux
linkage space vector shows that there is a small increase as the machine goes from
motoring to generating. The variation in stator flux linkage is a reflection of the
variation in the emf of the induction machine. There is no closed-loop control to control
the exciting current and the stator flux linkage. As will be discussed in the vector
control of induction machines in Chapter 8 the magnitude of the terminal voltage space
vector can be controlled by varying the stator flux linkage. At synchronous speed the
rotor current is zero and the magnetising current approaches its maximum value.
Fig. 4.13 shows the diagram of all the magnitudes and angles of the space vectors
discussed above. The space vectors shown in Fig. 4.13 are helpful in analysing and
designing stator oriented vector control of the induction machine in the generating and
motoring regions.
To transform the signals into the rotating reference frame the d-axis of the rotating
reference frame (d
e
) is aligned with the stator flux linkage. Hence at u
e
= 0 the d-axis
component of the stator flux linkage in the stationary reference frame is equal to the
total stator flux linkage. At a given operating slip the space vectors are rotating at the
same speed e
e
. Therefore, all the components of the space vectors appear as DC values
in the rotating reference frame.
84
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-40
-20
0
20
Iqs in exictation reference frame (A)
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
20
40
60
Ids in exictation reference frame (A)
Slip
(a)
(b)
Fig. 4.14 Stator current in the d
e
-q
e
axes of the excitation reference frame (a) q
e
-axis
current (b) d
e
-axis current
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
50
100
Vds in the exitaion reference frame (V)
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
100
200
300
slip
Vqs in the exitaion reference frame (V)
(a)
(b)
Fig. 4.15 Stator voltage in the d
e
-q
e
(a) d
e
-axis voltage (b) q
e
-axis voltage
85
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0
2
4
6
Imd in the exitaion reference frame (A)
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-2
-1
0
1
2
slip
Imq in the exitaion reference frame (A)
(a)
(b)
Fig. 4.16 Magnetising current in the d
e
-q
e
(a) d
e
-axis magnetising current (b) q
e
-axis magnetising current
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-40
-30
-20
-10
0
Idr in the exitaion reference frame (A)
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-20
0
20
40
slip
Iqr in the exitaion reference frame (A)
(a)
(b)
Fig. 4.17 Rotor current in the d
e
-q
e
(a) d
e
-axis rotor current (b) q
e
-axis rotor current
86
As shown in Figures 4.14 4.17, all time varying signals transformed to the excitation
reference frame look like DC quantities. In the excitation reference frame it is easier to
see the components of currents that produce the electromagnetic torque.
Transforming the rotor current to a reference frame fixed on the rotor gives an
alternating current with a frequency equal to the slip frequency e
e
-e
r
or se
e
. Fig. 4.18
shows the rotor current in a reference frame fixed to the rotor and in a reference frame
fixed to the stator, which is a stationary reference frame. The transformation angle to
the rotor reference frame is calculated using the rotor speed as
r r r
t o
dt u e u
=
= +
)
(4.21)
The magnitude of the rotor current given is when it is referred to the stator side.
0 1 2 3 4 5 6 7 8 9 10
-50
-25
0
25
50
rotor current in the rotor reference frame (A)
0 1 2 3 4 5 6 7 8 9 10
-50
-25
0
25
50
time (sec)
(a)
(b)
rotor current in the stationary reference frame (A)
Fig. 4.18 Rotor current in different reference frames (a) rotor current in a rotating
reference frame that is rotating at the rotor speed (b) rotor current in the stator
(stationary) reference frame
4.5 D-Q axes induction machine model in rotating reference frame
The transformation of currents and voltages to a rotating reference frame gives a
characteristic from a different perspective. The speed of the rotating reference frame can
87
have any value. If the reference frame is rotating exactly at the excitation frequency
then the difference between the speed of the rotating reference frame, e
e
, and the rotor
speed, e
r
, gives the slip frequency e
sl
.
Assuming the induction machine is only supplied from the stator side the equivalent
circuit in the excitation reference frame of the d and q axes is shown in Fig. 4.19.
(e
e
-e
r
)
qr
e
(a)
R
s
i
dr
e
+
L
ls
L
lr
R
r
i
ds
e
L
m
dr
e
ds
e
V
ds
e
-e
e
qs
e
+
-(e
e
-e
r
)
dr
e
(b)
R
s
i
qr
e
+
L
ls
L
lr
R
r
i
qs
e
L
m
qr
e
qs
e
V
qs
e
e
e
ds
e
Fig. 4.19 D-Q representation of induction machine in the excitation (e

e
) reference frame
(a) d-axis circuit (b) q-axis circuit
Unlike the stationary reference frame, in the excitation or synchronous reference frame
the reference frame is rotating at the same speed as the excitation frequency or the
synchronous speed. Since the voltages and currents have the excitation frequency they
will appear as DC values.
4.6 Development of D-Q axes induction machine model with R
m
The conventional way of three-phase induction machine steady state modelling is to use
the per-phase equivalent circuit, including R
m
, as given in Fig. 4.4. The equivalent
resistance for core loss or iron loss is incorporated into the circuit by adding R
m
in
88
parallel with the magnetizing reactance, X
m
. To simplify the analysis of the three-phase
induction machine, R
m
is often neglected from the per-phase equivalent circuit [7].
To date the three-phase induction machine d-q axes model has been represented
neglecting R
m
[8, 9, 19]. Neglecting R
m
will definitely simplify the analysis of the
induction motor, but it will introduce some error in the results that are obtained from the
d-q axes model. The error, which is introduced by neglecting R
m
, will have more effect
especially if the application of the analysis is to compute efficiency or analyse losses in
the machine [10, 11].
There is an interest in modelling the iron loss in the d-q model of an induction machine.
In the literature different ways of representing iron loss have been presented that
attempt to treat iron loss in a unified way [12, 13, 14, 15, 16]. These representations do
not give the relationship between the conventional model and the d-q model of a three-
phase induction machine. The d-q model of a three-phase induction machine including
iron loss needs to be presented in an easily understandable way.
In this analysis a novel d-q model of an induction machine for analysing the effects of
iron loss under steady-state and transient conditions is provided. In collaboration with
D. McKinnon, a fellow research student, the d-q model including iron loss is presented
in a simple, understandable way [17]. It should be noted that Mr McKinnon contributed
more to the inclusion of iron loss into the traditional d-q model. The author, Mr
McKinnon and C. Grantham exchanged ideas frequently in this area of the
investigation.
Using Fig. 4.3, the conventional steady state per-phase equivalent circuit model of a
three-phase induction machine with R
m
neglected and all rotor quantities referred to the
primary/stator side, the following substitution is made (e
e
-e
r
)/ e
e
= s and p=je
e
, and
with power being supplied only on the stator side, the voltages and currents can be
related as:
0
s s m s s
m r m r r r r r
R +L p L p i v
=
L p-j L R +L p-j L i
( ( (
( ( (

(4.22)
89
Where
L
s
= L
ls
+L
m
, L
r
= L
lr
+L
m
, p=d/dt and e
r
is the rotor speed in electrical rad/sec.
From the conventional per-phase equivalent circuit of a three-phase induction machine
given in Fig. 4.4 with R
m
included in the model, the relation between the input stator
voltage, and the stator and rotor currents can be derived as:
0
s sNEW N s s
N r N r rNEW r rNEW r
R +L p L p i v
=
L p-j L R +L p-j L i
( ( (
( ( (

(4.23)
Where
m m
N
m m
R L
L =
R +L p
sNEW ls N
L =L +L
rNEW lr N
L =L +L
From Equation (4.3), neglecting R
m
and with v
dr
= v
qr
= 0, the following matrix equation
can be obtained.
0 0
0 0
0
0
qs
qs s s m
s s m ds
ds
m r m r r r r qr
r m m r r r r
dr
i
v R +L p L p
R +L p L p i
v
=
L p - L R +L p - L i
L L p L R +L p
i
(
( (
(
( (
(
( (
(
( (
(
( (
( ( (

(4.24)
To represent R
m
in the d-q model, Equation (4.24) needs to be modified. Using the
comparison between Equation (4.22) and (4.24), Equation (4.23) gives the matrix
Equation in the d-q model, i.e. including R
m
, as:
0 0
0 0
0
0
qs
qs s sNEW N
s sNEW N ds
ds
N r m r rNEW r rNEW qr
r N N r rNEW r rNEW
dr
i
v R +L p L p
R +L p L p i
v
=
L L p L R +L p
i
(
( (
(
( (
(
( (
(
( (
(
( (
( ( (

(4.25)
90
This is a new form of matrix expression, which takes into consideration the effect of R
m
in the model. Equation (4.25) is the matrix form for the relationship between voltages
and currents and it can be used for dynamic analysis of induction machines with R
m
included.
With only a stator supplied induction machine, i.e. v
dr
= v
qr
= 0, the d-q model,
including R
m
, is derived from Equation (4.25) and is given in Fig. 4.20. The d-q model
of the induction machine, including R
m
, given in Fig. 4.20 is the same as the model
given in Fig. 4.6 except that R
m
is now added in parallel with L
m
.
L
ls
R
s
i
ds
L
lr
i
dr
L
m
i
md
R
r
dr
ds
v
ds
-e
r
qr
+ -
R
m
(a)
L
ls
R
s
i
qs
L
lr
i
qr
L
m
i
mq
R
r
qr
qs
v
qs
e
r
dr
+ -
R
m
(b)
Fig. 4.20 D-Q model of induction machine in the stationary reference frame including
core loss represented by R
m
(a) d-axis (b) q-axis
From Fig. 4.20 it is clear that
qs qr mq
i i i + = and
ds dr md
i i i + = . There is current flowing in
R
m
, which is the representation of the core loss.
Expanding and rearranging Equation (4.25) gives
P
2
I = A
o
pV + A
1
V + B
o
pI + B
1
I (4.26)
91
Where
qs
ds
qr
dr
i
i
I
i
i
(
(
(
=
(
(
(

,
qs
ds
v
V
v
(
=
(

,
ls
o
ls
1
0
L
A
1
0
L
(
(
(
=
(
(

,
m
m ls
1
m
m ls
R
0
L L
A
R
0
L L
(
(
(
=
(
(

s m m m
ls m ls ls
s m m m
ls m ls ls
o
m m m r
r
lr lr m lr
m m m r
r
lr lr m lr
R R R R
0 0
L L L L
R R R R
0 0
L L L L
B
R R R R
0
L L L L
R R R R
0
L L L L
e
e
( | |
+ +
( |
\ . (
(
| |
(
+ +
|
(
\ .
=
(
| |
(
+ +
|
(
\ .
(
(
| |
+ + (
|
(
\ .
m s
m ls
m s
m ls
1
r m m r m m
r
lr m lr lr m
r m m m m r
r
lr lr m m lr
r
R R
0 0 0
L L
R R
0 0 0
L L
B
R R R R R
0
L L L L L
R R R R R
0
L L L L L
e
e
e
e
(
(
(
(
(
(
=
(
| |
+ (
|
( \ .
(
| |
(
+
|
(
( \ .

From the magnetising inductance branch circuit of Fig. 4.20 the q-axis magnetising
current is
( )
m
mq qs qr
m m
R
i i i
R L p
= +
+
(4.27a)
and in integral form it can be written as
( )
m
mq qs qr mq
m
R
i i i i dt
L
= +
)
(4.27b)
For the d-axis magnetising current
( )
m
md ds dr
m m
R
i i i
R L p
= +
+
(4.28a)
and in integral form it is written as
92
( )
m
md ds dr md
m
R
i i i i dt
L
= +
)
(4.28b)
The q-axis air gap flux linkage is
( )
m m
mq m mq qs qr
m m
R L
L i i i
R L p
= = +
+
(4.29)
and the d-axis air gap flux linkage is
( )
m m
md m md ds dr
m m
R L
L i i i
R L p
= = +
+
(4.30)
From the stator side the q-axis stator flux linkage is given as
( )
m m
qs ls qs m mq ls qs qs qr
m m
R L
L i L i L i i i
R L p
= + = + +
+
(4.31)
and the d-axis stator flux linkage is given as
( )
m m
ds ls ds m md ls ds ds dr
m m
R L
L i L i L i i i
R L p
= + = + +
+
(4.32)
From the rotor side the q-axis stator flux linkage is given as
( )
m m
qr lr qr m mq lr qr qs qr
m m
R L
L i L i L i i i
R L p
= + = + +
+
(4.33)
and the d-axis stator flux linkage is given as
( )
m m
dr lr dr m md lr dr ds dr
m m
R L
L i L i L i i i
R L p
= + = + +
+
(4.34)
Based on the definition of electromagnetic torque [6]
3
2
e p m r
T P I =
JJG JG
(4.35)
where
m
JJG
and
r
I
JG
are the space vectors for the air gap flux linkage and for the rotor
current respectively.
Substituting
m
JJG
and
r
I
JG
in Equation (4.35) and rearranging using vector manipulation
gives:
( )
3
2
m m
e p qs dr ds qr
m m
R L
T P i i i i
R L p
=
+
(4.36)
93
and in integral form
3
2
m
e p m qs dr ds qr e
m
R
T P R i i i i T dt
L
| |
=
|
\ .
)
(4.37)
4.7 Summary
The modelling of an induction machine using the conventional or steady state model
and the d-q or dynamic model are explained in detail. The voltage, current and flux
linkage in the rotating reference and their phase relationships in the motoring region and
generating region are presented. For the same stator terminal voltage the magnitude of
the electromagnetic torque in the generating region is higher than the electromagnetic
torque in the motoring region. The reason for the difference in electromagnetic torques
is that during the motoring region all the electrical losses in the induction machine are
supplied by an external electrical power source and the electromagnetic torque is the
output of the system. However, in the generating region the electromagnetic torque is
equivalent to the external mechanical input torque and all the electrical power losses in
the induction machine are indirectly supplied by the external mechanical power source
and the terminal voltage is the output of the system. Hence to overcome all the internal
power losses in the induction machine and have the same terminal voltage as in the
motoring region the electromagnetic torque in the generating region is higher than the
motoring region.
D. McKinnon and C. Grantham, in collaboration with the author, has improved the
induction machine model in D-Q axes including the equivalent iron loss resistance, R
m
,
is presented in a simple and understandable way. Using this model the dynamic current,
torque and power can be calculated.
4.8 References
[1] P. L. Alger, The nature of induction machines, Gordon and Breach Inc., New York, 1965.
[2] G. R. Slemon, Electric Machines and Drives, Addison-Wesley Publishing Company, New York,
1992.
[3] A. E. Fitzgerald and JR. Charles Kingsley, Electric Machinery, McGraw-Hill Book Company,
New York, 1961.
[4] J. Rosenblatt, and M. H. Friedman, Direct and Alternating Current machinery, CBS Publishers
and Distributors, New Delhi, 2000.
94
[5] N. N. Hancock, Matrix Analysis of Electrical Machinery, Pergamon Press, New York, 1974.
[6] B. K. Bose, Modern Power Electronics and AC Drives, Printice-Hall, New Jersey, 2002.
[7] B. K. Bose, Power Electronics and AC Drives, Printice-Hall, New Jersey, 1968.
[8] Chee-Mun Ong, Dynamic Simulation of Electric Machinery using Matlab/Simulink Printice-hall,
New Jersey, 1998.
[9] C. Grantham, D. Seyoum, H.Tabatabaei-Yazdi, Very Fast and Accurate Electrical
Measurements, AUPEC 99, The Northern Territory University, Darwin, Australia, 26
th
-29
th
Sept.
1999, pp. 99 - 103.
[10] H. Auinger, Efficiency of electric motors under practical conditions, IEE Power Engineering
Journal, June 2001, pp. 163 - 167.
[11] C. Grantham and H. Tabatabaei-Yazdi, Rapid Parameter Determination for use in the Control
High Performance Induction Motor Drives, IEEE 1999 International Conference on Power
Electronics and Drive Systems, PEDS99, July 1999, Hong Kong, pp. 267 - 272 .
[12] E. Levi, Impact of iron loss on behavior of vector controlled induction machines, IEEE Trans.
On Industry Applications, Vol. 31, No. 6, November/December 1995, pp. 1287 - 1296.
[13] E. Levi, M. Sokola, A. Boglietti and M. Pastorelli, Iron loss in rotor-flux-oriented induction
machines: identification, assessment of detuning, and compensation, IEEE Transactions on Power
Electronics, Vol. 11 No. 5, Sept. 1996, pp. 698 - 709.
[14] G.O. Garcia, J.A. Santisteban and S.D. Brignone, Iron losses influence on a field-oriented
controller, In Proc. IECON '94, Vol. 1, 1994, pp. 633 - 638.
[15] Sung-Don Wee, Myoung-Ho Shin and Dong-Seok Hyun, Stator-flux-oriented control of induction
motor considering iron loss, IEEE Trans. On Industrial Electronics, Vol. 48, No. 3, June 2001, pp.
602 - 608.
[16] J.W. Choi, D. W. Chung, and S. K. Sul, Implementation of Field Oriented Induction Machine
Considering Iron Losses, IEEE- APEC '96. Conference Proceedings, 1996, pp. 375-379.
[17] D. McKinnon, D. Seyoum and C. Grantham, "Novel Dynamic Model for a Three-Phase Induction
Motor With Iron Loss and Variable Rotor Parameter Considerations", Proc. AUPEC 2002,
Melbourne, 29 Sept. - 2 Oct. 2002. ISBN 0-7326-2206-9.
[18] M.G. Say, Alternating current machines Pitman, London, 1983, pp. 260-262.
[19] P. C., Krause, O. Wasynczukand and S. D. Sudhoff, Analysis of Electric Machinery, IEEE press,
New York, 1995.
95
CHAPTER 5
DATA ACQUISITION AND DIGITAL SIGNAL
PROCESSING
5.1 Introduction
The remarkable increase in the speed and power of digital computers and special
purpose hardware over recent years has ensured the continued growth of interest in data
acquisition and Digital Signal Processing (DSP) for monitoring and control of electrical
machines. A signal is any variable that contains information from a transducer or from a
controller that can be manipulated for a given application. Digital Signal Processing, as
the term suggests, is the processing of signals by digital means.
Each measurement system in the experimental set up involves sensors, signal
conditioning, transmission, sampling, manipulation and interpretation of the resultant
data. The output of the measurement process is presented to be displayed or processed
for the control of physical parameters. The over all interconnection of a data acquisition
and signal processing system is shown in Fig. 5.1.
Parameter to
be measured
sensor
A/D
converter
Analog
signal
Digital
signal
D/A
converter
Digital
signal
Analog
signal
Digital
signal
processor
Fig. 5.1 Block diagram for data acquisition and signal processing
Filtering reduces noise errors in the signal. This can be done with either an analog or
digital filter. For most applications a low-pass filter is used. This allows the lower
frequency components through but attenuates the higher frequencies. The cut-off
frequency must be compatible with the frequencies present in the actual signal (as
opposed to possible contamination by noise) and the sampling rate used for the A/D
CHAPTER 5 DATA ACQUISITION and DIGITAL SIGNAL PROCESSING
96
conversion. A high pass filter is used to remove DC component or signals with low
frequency.
Fast data acquisition and Digital Signal Processing systems are used to acquire
thousands of samples of a measured signal in a second. This was impossible to
implement a few years ago. Data acquisition deals with the way a signal is sensed and
conditioned, whilst Digital Signal Processing deals with the manipulation of a signal in
digital form.
b
i
b
i
q
v
ab
Sensor
board
a
c
i
a
v
bc
DS1102 DSP
board
Hardware
Software
3 axes to 2 axes
transformation
i
d
v
q
v
d
Induction
machine
Anti-aliasing filter
Speed measurement
Fig. 5.2 Hardware and software system configuration
5.2 DS1102 DSP board
The fast data acquisition and control system in the experimental set up is based on a
DS1102 controller board produced by dSPACE GmbH. The DS1102 board is built
around a Texas Instruments TMS320C31 floating point Digital Signal Processor (DSP)
running at 40MHz clock rate and the slave-DSP is a Texas Instruments TMS320P14
DSP running at 25MHz clock rate. The processor can be accessed from the host
computer through graphically oriented tools for on-line visualization (TRACE) and on-
line parameter tuning (COCKPIT). Control desk combines the cockpit and trace options
together in the same window. The most significant advantage in using the DS1102 is the
97
relative ease of use, freeing the designer from the burden of manually programming the
DSP. The operation of the DS1102 DSP board requires the Texas Instrument C
compiler, assembler, linker and loader. The LD31 utility program performs the actual
loading of the TMS320C31 DSP object code modules and optionally the TMS320P14
object code modules. The C environment of the DS1102 controller board contains a
library with predefined functions, header files and other application files.
The TMS320P14 can be used for numerical calculation through loading the program
into its memory. The TMS320C31 communicates with the slave TMS320P14 through a
32 bit communication port and programs can be executed in both processors
concurrently by two separate timers. The block diagram of the DS1102 is shown in Fig.
5.3 [1].
128K32
static RAM
zero wait states
Host
Interface
16-bit ADC 1
TMS 320P14
Digital I/O
s
e
r
i
a
l

i
n
t
e
r
f
a
c
e
J
T
A
G
c
o
n
n
e
c
t
o
r
TMS320C31
JTAG
Interface
Incr. encoder 1
16-bit ADC 2
12-bit ADC 3
12-bit ADC 4
Incr. encoder 2
Noise filter
Noise filter
12-bit DAC 1
12-bit DAC 2
12-bit DAC 3
12-bit DAC 4
A
n
a
l
o
g
/
D
i
g
i
t
a
l

I
/
O

c
o
n
n
e
c
t
o
r
PC/AT Expansion Bus
26
DS1102 DSP-board
PC
Fig. 5.3 Block Diagram of the DS1102
From Fig. 5.3 the signal groups used in the experimental set up are:
- Four ADC (analog to digital conversion) inputs: The two ADC inputs are of 16
bit with A/D conversion time of 10s and two ADC inputs are of 12 bit with
98
A/D conversion time of 3s. All ADCs have single ended bipolar inputs with
10V input span. 10V appears as 1V in the software environment.
- Four DAC outputs: All DAC outputs are of 12 bit with programmable output
voltage range. The DACs have single ended voltage outputs with 10V span.
- Two channels of incremental encoder interfaces.
- Digital I/O subsystem: This is based on the TMS320P14 DSP-microcontroller. It
contains 16 pins bit selectable parallel I/O port, six PWM circuits and other lines
which are not used in the experimental setup.
5.3 Data acquisition
Taking measurements using conventional meters is very slow and is able to show only
the steady state values. For a fast data acquisition system fast sensors are required to
track the change in the measured quantities. The voltage, current and speed of the
induction machine are measured using appropriate sensors to be processed in the Digital
Signal Processing part for machine analysis and control. The temperature of the
induction machine is sensed and displayed on a dedicated display system as a
conventional thermometer.
5.3.1 Voltage and Current measurement
In a balanced three-phase system with the measurement of only two phase voltages and
two line currents it is possible to calculate the rms voltage, rms current, three-phase
power, power factor and frequency of the three-phase system at an instant of time. This
was illustrated in Chapter 3. To implement a fast current and voltage measurement, fast
responding (high bandwidth) and high accuracy sensors are required.
The sensors used for voltage and current measurement are based on the Hall-Effect
principle and are able to measure voltage and current signals in the frequency range of
DC up to 100 KHz. The principle of measurement for the Hall-Effect voltage
measurement and current measurement are the same. Hall-Effect produces Hall-Voltage
based on a magnetic flux created by current. If a conducting material is placed in a
magnetic field perpendicular to the direction of current flow, a Hall-Voltage is
99
developed across the material in a direction perpendicular to both the initial current
direction and the magnetic field.
The sensors for current and voltage are isolated from the circuit being measured and use
a separate power supply to avoid any effect on the voltages and currents being
measured. To avoid ground loops between the sensor boards and the DSP card separate
return lines are used for the connected sensors and the signal line is shielded to avoid
any external interference. The voltage and current sensors and their associated
amplification circuits are located on one PCB. This is referred to as the Sensor Board.
The DS1102 has only 4 analog ADC inputs, however to measure 6 analog inputs (2 AC
voltages, 2 AC currents, 1 DC voltage and 1 DC current) an 8 to 4 multiplexer is used
and the multiplexer circuit is controlled by 2 lines from the 16 I/O pins. The function
ds1102_ad(channel) returns the ADC input value of the converter specified by the
parameter channel which must be set to 1, 2, 3 or 4. The ADC data is read subsequently
and scaled to its floating-point value in the range -1.0 +1.0. Since the ADC input
value is a 16 or 12-bit signed integer left aligned within a 32-bit data word, the factor
2
-31
is required for scaling. The conversion of the ADCs must be started by a preceding
call to the function ds1102_ad_start( ).
5.3.1.1 Anti-aliasing filter
According to the Sampling Theorem [2], any signal can be accurately reconstructed
from values sampled at uniform intervals as long as it is sampled at a rate that is at least
twice the highest frequency present in the signal. Failure to satisfy this requirement will
result in aliasing of higher-frequency components, meaning that these components will
appear to have frequencies lower than their true values.
One way of avoiding the problem of aliasing is to apply a low-pass filter to the signal
prior to the sampling stage to remove any frequency components above the "folding" or
Nyquist frequency (half the sampling frequency) [3]. The anti-aliasing filters are
introduced at the end of the analog signal conditioning and just before the analog to
digital converter. These anti-aliasing filters are implemented using conventional analog
100
circuitry with cut off frequency below half of the minimum sampling frequency
(maximum sampling time) used in the DS1102 application programs.
The anti-aliasing filter also helps to have smooth voltage and current signals in the
PWM controlled inverter source. Each analog signal from voltage and current sources
must have an anti-aliasing filter.
5.3.1.2 Voltage measurement
For voltage measurement the sensor circuit is designed in such a way that it will be able
to measure the full range of the applied line to line voltage at the terminals of the
induction machine and the maximum DC voltage available in the DC link of the
inverter. The peak voltage to be measured is 590V (4151.414). LV 25-P, produced by
LEM components, is the transducer used for the electronic measurement of DC and AC
voltages. The detail of the voltage measurement system is shown in Fig. 5.4.
When the voltage required for measurement is connected across terminals a-b, a
proportional current flows through resistor R
V
(47K). This current produces a magnetic
flux, which in turn is proportional to the voltage. The magnetic flux is constantly
controlled at zero by a compensating current flowing through the secondary coil using
the Hall-Effect device and associated electronic circuit. The amount of secondary
(compensating) current required to hold the zero flux is a measure of the primary
current flowing multiplied by the ratio of the secondary winding. The secondary current
multiplied by R
M
is the output voltage of the sensor. The signal conditioning part of the
circuit is calibrated by adjusting the potentiometer P
2
in such a way that the maximum
voltage to be measured will give an output voltage signal of peak value 10V. To
protect the ADC input of the DS1102 board from over voltage, two zener diodes
(BZX55-C9V1), with breakdown voltage of 9.1V, are connected back to back. The
zener voltage of 9.1V is chosen so that when the back to back diode combination is
conducting, the voltage drop will not exceed 10V (9.1V + forward biased voltage drop).
The output of the voltage measurement circuits are connected to ADC3 and ADC4 of
the DS1102 board. ADC3 and ADC4 contain 12 bit successive approximation analog to
101
digital converters with integrated sample and hold units. Each converter achieves a
conversion time of 3s.
+15V
R
M
100R
I
M
-15V
+
-
M
LV 25-P
Turns ratio 1:2.5
+
-
Voltage to be
measured
R
V
47K
+
-
Sensed voltage
a
b
(a)
R
1
5K6
From voltage
sensor
R
2
5K6
R
3 10K
P
1 2K
R
4 10K
R
5 270R
+15V
-15V
+
_
9
10
8
11
4
-15V
+15V
+
_
12
13
14
R
6
9K1
P
2
20K
LT1058
LT1058
R
7
2K2
To DS1102
DSP board
Jumper
C
1
0.01F
9.1VZ
9.1VZ
(b)
Fig. 5.4 Voltage measurement system (a) voltage sensor (b) signal conditioning for the
sensed voltage
The connection of resistor R
5
to ground in the signal conditioning part of the voltage
measurement system reduces the effect of offset error to a negligible value as compared
to the measured signal of interest. The voltage sensor given here is good for measuring
voltages above 10V. For input voltages below 10V there is more error because the
accuracy is given as absolute error relative to the nominal current. The voltage sensor
can go up to its rated insulation capability by increasing the value of the resistance R
V
.
102
The combined resistance of R
6
and potentiometer P
2
in parallel with capacitor C
1
provides the anti-aliasing filter with a cut off frequency f
c
= 1/(2tRC).
5.3.1.3 Current measurement
The current transducer used in the current sensor circuit is an LTA 50P/SP1 produced
by LEM Components. The current transducer LTA 50P/SP1, similar to the voltage
transducer, employs Hall-Effect to measure DC, AC and impulse signals. The LTA
50P/SP1 is connected in its open-loop mode to give an output voltage proportional to
the current under measurement. This arrangement gives a bandwidth from DC up to
25KHz. The principle of operation of the LTA 50P/SP1 is that the magnetic flux created
by the primary current (current to be measured) is connected in a magnetic circuit and
measured using the Hall-Effect device. The output voltage from the Hall-Effect device
and associated electronic circuit provide an exact representation of the primary current.
The output voltage is 100mV/Amp, i.e. 10A primary current gives 1V output voltage.
The detail of the current measurement system is shown in Fig. 5.5.
In Fig 5.5(a) V
out
is the output voltage signal of the current transducer and this has to be
signal conditioned to be fed to the DS1102 board. The signal conditioning part of the
circuit is calibrated by adjusting the potentiometer P
2
in such a way that the maximum
current to be measured will give an output voltage signal of peak value 10V.
The output of the current measurement circuits are connected to ADC1 and ADC2 of
the DS1102 board. ADC1 and ADC2 contain 16 bit successive approximation analog to
digital converters with integrated sample and hold units. Each converter achieves a
conversion time of 10s.
The purpose of the resistor R
5
, capacitor C
1
, zener diode and other resistors is similar to
that of the voltage measurement set up described in the previous Section.
103
-15V
+15V
1
2
4
5
6
Current to be measured
(primary current)
Vout
(a)
R
1
5K6
From current
sensor
R
2
5K6
R
3 10K
P
1 2K
R
4 10K
R
5 270R
+15V
-15V
+
_
9
10
8
11
4
-15V
+15V
+
_
12
13
14
R
6
9K1
P
2
20K
LT1058
LT1058
R
7
2K2
To DS1102
DSP board
Jumper
C
1
0.01F
9.1VZ
9.1VZ
(b)
Fig. 5.5 Current measurement system (a) current transducer (b) signal conditioning for
the sensed current in terms of voltage signal
5.4 Speed and angle measurement
Speed and angular position can be measured using different types of sensors. An optical
incremental encoder is the most frequently used. Incremental measurement means
measurement by counting. That is, the output signals of incremental rotary encoders are
supplied to an electronic counter in which the measured value is derived by counting
individual increments. The information for speed and angle calculation is mainly
determined by counting the encoder pulses during a given sampling interval.
104
In the experimental set up, the position measuring system is based on a ROD 426
incremental rotary encoder produced by Heidenhain GmbH with 5,000 lines (or
increments) per revolution [4]. The ROD 426 operates on the principle of photo
electrically scanning very fine gratings forming an incremental track. Output signals are
incremental with TTL square wave pulse trains U
a1
, U
a2
, and their inverted pulse trains
a1
U and
a2
U produced by four-field scanning. A reference signal of a single square
wave pulse U
a0
is produced every revolution and there is a
a0
U , which is the inverted
pulse of U
a0
. U
a2
lags U
a1
by 90
o
(1/4 of a cycle) with clockwise rotation (viewed from
shaft side) as shown in Fig. 5.6. One signal cycle of 360
o
electrical corresponds to the
angle of rotation for one pitch of the radial grating (i.e. one line and one space).
U
a2
U
a0
U
a1
90
o
elec.
90
o
elec.
phase shift
a edge separation
Signal period
360
o
elec.
t
d t
d
Fig. 5.6 Output signals of and incremental angle encoder
The output U
a2
and U
a1
pulses produce four states represented by 2 bits 00, 01, 11 and
10 for each line space of the encoder, as shown in Fig. 5.6. Hence the minimum
resolution of the increment encoder is one fourth of the grating pitch of the radial
grating. The value of the edge separation, a, is greater than or equal to 0.45s at the
maximum scanning frequency 300 KHz. The lag time of pulse U
a0
to signals U
a1
and
U
a2
is less than 50ns. In this application with a sampling time around 200s, equivalent
105
to a scanning frequency of about 5KHz, the requirements for edge separation is satisfied
and the effect of lag time is negligible.
5.4.1 Angle measurement
The 5,000 line count per revolution of the ROD 426 encoder with the four states per
line space gives 20,000 measuring steps (counts) per revolution. Then the value of the
minimum angle measurement step is 360
o
/20,000 = 0.018
o
. Hence the angle resolution
is 0.018
o
or as a fraction of a revolution is 510
-5
revolution.
The DS1102 DSP card contains two incremental sensor interfaces (Channel 1 and
Channel 2) to support optical incremental sensors commonly used in position control.
Each interface contains line receivers for the input signals, a digital noise pulse filter
eliminating spikes on the phase lines, a quadrature decoder which converts the sensors
phase information to count-up and count-down pulses, a 24-bit counter which holds the
current position of the sensor and a 24-bit output latch [1]. Using a C program the
current value of the position counter is read from the output latch. Fig. 5.7 shows a
block diagram for an incremental sensor interface.
Line
receiver
Noise
filter
24-bit
position
counter
Quadrature
decoder
Line
receiver
24-bit
output
latch
U
a1
U
a2
Phi0 1
Phi90 1
index U
a0
25MHz RESET STROBE
Fig. 5.7 Block diagram of an incremental encoder interface
In the experimental set up Channel 1 incremental encoder position counter register
address 024000H of the DS1102 is used. The incremental position can be determined
from the 24 bit position counter that counts up when U
a1
is the leading sequence and
106
counts down when U
a2
is the leading sequence. When digitized, both edges of U
a1
and
U
a2
are counted, thus the incremental position |
incr
in radians is given by
incr incr o
2
A
4 N
t
| | = +
(5.1)
where A
incr
- incremental count of the position counter in incremental steps
N - the line count of the encoder (here it is 5000 lines)
|
o
- the initial position
One incremental step is equivalent to a 90 electrical phase shift of the signals U
a1
and
U
a2
.
To express the incremental position between 0 and 2t, when the angle exceeds 2t it
resets to 0. To have this option the angle measurement can be implemented using
( )
_ _ incr incr incr new incr old
2
A A
4 N
t
| | = +
(5.2)
where A
incr_new
and A
incr_old
are the new and old incremental count of the position counter
respectively. At start |
incr
is equal to |
o
.
A function read_inc(1) is derived from the original dSPACE function ds1102_inc from
the DS1102 software environment to read Channel 1 of the 24 bit position counter
register and return the output as a 32 bit left aligned data word. The left alignment is
done by shifting the 24 bits by 8 bits towards the left or it is multiplied by 2
8
= 256. In
the DS1102 software environment the angle measurement is implemented as
( )
_ _ incr incr incr new incr old
2
A A
4 N 256
t
| | = +

(5.3)
where A
incr_new
and A
incr_old
are the new and old values of the function
read_inc(Channel1) incremental count of the position counter respectively. The 256 is
included to cancel the multiplication factor of 256 introduced due to the conversion
from 24 bit to the left aligned 32 bit.
107
5.4.2 Speed measurement
The speed information is determined by counting the number of pulses, which represent
an increment in angular position, during a specified sampling period. The 5000 line
encoder gives 20000 measuring steps per revolution. In general the speed resolution is
expressed as
res
res
s
T
u
e = (5.4)
where u
res
- incremental angular position resolution equal to 510
-5
of a revolution
= 3.14210
-4
radians
T
s
- sampling period (seconds)
Hence the speed resolution expressed in revolutions per minute (rpm) is
.
5
res
s s
5 10 0 003
60
T T
e
= = rpm (5.6)
With a sampling time of 200sec (T
s
= 200s) the speed resolution will be 15rpm. This
means that the speed measurement is expressed as a discrete value in 15rpm steps. The
maximum error, irrespective of the speed, will be 15rpm. At high speed the relative
error in speed measurement is small however at low speed the relative error is large. If
the sampling time is 500s then the speed resolution is 6rpm. For a sampling time of
1ms the speed resolution is 3rpm.
The accuracy of speed measurement is dependent on the sampling time. The sampling
time has to be large enough so that the accuracy of the speed measurement will be
improved. However the sampling time should not be increased to an extent where it will
not be able to follow the change in speed quickly and affect a control system dependent
on rotor speed.
A function ds1102_inc(1) updates the Channel 1 incremental encoder counter output
register and returns the position counter value. The 24-bit position counter value is
scaled to a floating point value in the range -1.0 to 1.0 by the factor 2
-31
because the data
word is a 24 bit signed integer left aligned within the 32 bit data word [5]. The speed is
calculated as
108
_ _ new old
count count
speed k
Ts
= (5.7)
where k is the multiplication factor, k = 25165.824. count
_new
and count
_old
are the new
and old value of the function ds1102_inc(Channel1) respectively. The value of count
_new
and count
_old
is between -1.0 and 1.0 floating point value.
The multiplying factor k is derived from the way the counted value is stored in the
register and returned from the function ds1102_inc(Channel1) to the software
environment. Based on the encoder used 1rpm = 333.333counts/sec. The 333.333 is
stored in the 24-bit signed integer left aligned position counter register. When the 24-bit
is converted to a left aligned 32-bit word, it has effectively been multiplied by 256 or 2
8
.
Finally the function ds1102_inc(Channel1) returns the counted value in 32-bit form
divided by 2
31
to have a scaling in the range -1.0 +1.0 floating point value so that
1rpm will be returned from the function as 3.97410
-5
/sec. To express the speed
measurement in rpm the incremental counted value per second should be multiplied by
a factor k calculated as
.
31
60 2
k 25165 824
20000 256
= = (5.8)
5.5 Digital Signal Processing
The currents and voltages used in the three axes to two axes transformation should be a
sinusoidal waveform. Due to the resolution in the incremental encoder the speed is
filtered. The filters used to smooth the currents and voltages are designed using digital
signal processing tools. The filters used are first order and second order low pass filters.
A high pass filter is also used to remove offset from the signal. For band pass filtering
the combination of the low pass filter and high pass filter is used.
5.5.1 Digital filter
Digital filters are applied to sampled data in the discrete time domain. There are two
types of digital filters: infinite impulse response (IIR) filters which are recursive and
finite impulse response (FIR) filters which are non recursive. For the recursive type
there is at least one feed back path, however for the non recursive type there is no feed
back path. Either type of filter, in its basic form, can be represented by its impulse
109
response sequence, h(k) (where k = 0, 1, 2, ). The input and output signals to the filter
are related by the convolution sum, which is given in Equation (5.9) for the IIR and
Equation (5.10) for the FIR filter [2, 6].
( ) ( ) ( )
k 0
y n h k x n k
=
=
_
(5.9)
( ) ( ) ( )
M 1
k 0
y n h k x n k
=
=
_
(5.10)
From Equations (5.9) and Equations (5.10) it is clear that for the IIR filter the impulse
response is of infinite duration, whereas for the FIR it has finite duration.
5.5.1.1 Infinite Impulse Response (IIR) filter
An IIR filter can not be realized using Equation (5.9) because of the theoretical infinite
length of its impulse response. Instead, the IIR filtering equation is expressed in a
recursive form given by
( ) ( ) ( ) ( ) ( )
M L
k k
k 0 k 0 k 1
y n h k x n k a x n k b y n k
= = =
= =
_ _ _
(5.11)
where x(n) is the input signal, y(n) is the output signal and the constants a
0
, a
1
, a
2
a
M
,
b
1
, b
2
b
L
are filter coefficients.
The current output sample y(n), is a function of past outputs as well as present and past
input samples. That is, an IIR filter is some sort of feedback system. The transfer
function for the IIR filter consists of poles and zeroes and the generalized form is:
( )
( )
( )
M
k
k
k 0
L
k
k
k 1
a z
Y z
H z
X z
1 b z
=
= =
| |
+
|
\ .
_
_
(5.12)
IIR filters are implemented in a recursive fashion. An important part of the IIR filter
design process is to find suitable values for the coefficients a
k
and b
k
such that some
aspect of the filter characteristic behaves in a desired manner. The transfer function of
the IIR filter, H(z), given in Equation (5.10) can be factorised as
( )( ) ( )
( )
( )( ) ( )
1 2 N
1 2 L
K z z z z z z
H Z
z p z p z p

=

"
"
(5.13)
110
where z
1
, z
2
, , z
N
are the zeros of H(z), and p
1
, p
2
, , p
L
are the poles of H(z).
Instability can occur if any of the magnitudes of the roots of the poles are greater than 1
(outside the unit circle in the z plane).
5.5.1.2 Finite Impulse Response (FIR) filter
Finite Impulse Response (FIR) Filters are generally non recursive digital filters and they
have a finite duration. From Equation (5.10) the current output sample y(n), is a
function of present and past input samples. This means that the FIR filter appears as a
moving average filter where the n
th
output is a weighted average of the most recent M
inputs. The block diagram realization for FIR filters is relatively straight forward. For a
FIR polynomial, the realization is simply a combination of multipliers and delay
elements. A generalized version in block diagram format is presented in Fig. 5.8.
z
-1
z
-1
h(0)
z
-1
y(n)
_
x(n)
h(1) h(2) h(M-1) h(M-2)
Fig 5.8 Block diagram for FIR filter
Using Equation (5.10) the transfer function for FIR filters is expressed as
( )
( ) ( )
( )
M 1
k
k 0
Y z
H z h k z
X z
=
= =
_
(5.14)
Upon expansion the equation yields M-1 poles at the origin, meaning that FIR filters are
unconditionally stable as the poles cannot ever lie outside the unit circle in the z-plane
[7].
( )
( )
( ) ( )
( )
M 1
M 1 k
M 1
k 0
Y z 1
H z h k z
X z z
=
= =
_
(5.15)
In Equation (5.11) when b
k
= 0 the filter is of non recursive type and becomes an FIR
filter where h(k) = a
k
, (k= 0,1, ).
111
5.5.1.3 Comparison of IIR and FIR filters
The comparison of IIR and FIR filters is summarised as follows
- FIR filters are realised non recursively and are always stable. The stability of IIR
filters cannot always be guaranteed. Round off noise and quantisation errors are
much less severe in FIR filters than in IIR filters.
- FIR filters can have linear phase response however IIR filters have non linear
phase response, especially at the band edges.
- IIR filters require fewer coefficients than FIR filters for sharp cut off filters.
Fewer coefficients mean less processing time and less storage requirement.
- In IIR filters there is no direct relationship between complexity and the length of
the impulse response, which is infinite by definition. Filters with high selectivity
can be realised with relatively low complexity. However, in FIR filters
complexity is proportional to the length of the impulse response.
- Analog filters can be readily transferred into equivalent IIR digital filters
meeting similar specifications, but FIR filters have no analog counter part.
- IIR filters can be designed using design formulae. FIR filter design procedures
are normally iterative procedures and design equations do not exist.
5.5.2 Digital filter design from analog filter
In the literature there are many ways of converting analog filters to digital filters [2, 6,
8]. The Matlab software provides a function, c2dm, to change a transfer function from
continuous time to discrete time. One of the ways of converting an analog filter (s-
domain) to a digital filter (z-domain) is to use a binomial Bilinear Transformation.
A bilinear transformation is implemented by substituting
1
1
2 1 z 2 z 1
s
T 1 z T z 1

= =
+ +
(5.16)
in the analog filter in the s-domain. With the bilinear transformation, the entire left half
s-plane maps to the interior of the unit circle in the z-plane. Hence all stable analog
filters will result in stable digital ones. Also, the bilinear transformation maps the entire
imaginary axis in the s-plane onto the unit circle in the z-plane.
112
It can be shown that the digital frequencies, (e
D
), are related to frequencies in the s-
plane, (e
A
), by a mapping of the frequency in the s-plane to the frequency in the z-
plane, that is,
tan
1 A
D
T 2
T 2
e
e
| |
=
|
\ .
(5.17)
where T is the sampling time.
Correction for this frequency scale wrapping may be accomplished by redesigning (pre-
warping) the critical frequencies of the desired transfer function G(s) before applying
the bilinear z-transform. If the digital filter cut-off frequency is e
k
then the equivalent
analog filter cut-off frequency (e
ke
) is found using the relation (pre-warping) given by
tan
1 k
ke
T 2
T 2
e
e
| |
=
|
\ .
(5.18)
5.5.3 Implementation of a digital filter by approximating analog filter
circuits
Continuous time filters can be converted to discrete time filters by approximating the
derivative in a differential equation representing the input output relationship of a
circuit. For example the derivative
dv
dt
can be approximated by using a backward
difference
( ) ( ) v n v n 1
t

+
or
( ) ( )
s
v n v n 1
T

,where T
s
is the sampling time. The
sampling rate is the time between two samples and its inverse is the sampling
frequency.
A simple analog RC low pass filter is given in Fig. 5.9.
R
C x
y
Fig. 5.9 Simple first order analog low pass filter
113
In Fig. 5.9 y is the filtered output and x is the unfiltered input. The differential equation
relating the input and output is
dy
RC y x
dt
+ = (5.19)
The Laplace transform, which is the analog transfer function in the frequency domain, is
( )
( )
c
c
Y s
X s s
e
e
=
+
(5.20)
where e
c
=
1
RC
is the cut-off frequency.
Using the backward Euler method the differential equation given in Equation (5.19) can
be approximated by a finite difference equation given by
( ) ( )
( ) ( )
y n y n 1
RC y n x n
t

+ =
+
(5.21)
substituting e
c
=
1
RC
and T
s
= At and rearranging leads to
( ) ( ) ( ) ( )
c
c s
1
y n y n 1 Tx n
1 T
e
e
= +
+
(5.22)
Here y(n) is the filtered value and x(n) is the unfiltered value.
A second order low pass filter and high pass filter can be derived in a similar way as
given in the first order low pass filter.
5.6 Summary
The data acquisition system which is used for the measurement of voltages, currents,
angle and speed with their sensors is explained. The outputs of the voltage and current
measurements are calibrated so that the voltage signals of the sensor outputs will not
exceed the ADC input rating of the DS1102 DSP board. The sensors for current and
voltage are Hall-Effect devices. Anti-aliasing filters are introduced in the analog signals
of the sensor outputs to prevent the high frequencies appearing as a low frequency when
the analog signal is digitised in the A/D converter.
114
Speed and angle measurements are taken using an optical incremental encoder. The
resolution of angle and speed for a given encoder is derived. For a given encoder the
resolution of the angle measurement is constant; however the resolution of speed
measurement is dependent on the sampling period used. High sampling period (low
sampling frequency) gives small resolution in speed (discrete steps) producing less error
in the measurement of the speed.
The advantage of digital signal processing is discussed. Different types of filter design
are presented. Digital filters are used in the simulation and experimental results
presented in the following chapters.
5.7 References
[1] DSPACE, DS1102 Users Guide - Document Version 2.0, Paderborn, Germany.
[2] Alan V. Oppenheim and Ronald W. Schafer, Discrete time Signal Processing, Prentice-Hall Inc.,
Upper Saddler River, N.J., 1999.
[3] Paul A. Lynn and Wolfgang Fuerst, Introductory Digital and Signal Processing with Computer
Applications, John Wiley and Sons, New York, 1989.
[4] Heidenhain, Rotary Encoders Catalogue 1999.
[5] DSPACE, DSP-CITeco DS1102 Software Environment - Document Version 2.0, Paderborn,
Germany.
[6] Emmanuel C. Ifeachor and Barrie W. Jervis, Digital Signal Processing a Practical Approach,
Addison-Wesley Publishers Ltd., Wokingham, England, 1993.
[7] John G. Proakis and Dimitris G. Manolakis, Digital Signal Processing Principles, Algorithms, and
Applications, Prentice-Hall Inc., Upper Saddler River, N.J., 1996.
[8] Charles L. Philips and Troy H. Nagle, Digital Control System Analysis and Design, Prentice-Hall,
Englewood Cliffs, N.J., 1995.
115
CHAPTER 6
PARAMETER DETERMINATION FOR AN INDUCTION
MACHINE
6.1 Introduction
Machine modelling requires knowledge of the parameters of the machine. Whether the
three-phase induction machine is modelled using the conventional equivalent circuit or
dq method, the parameters of the machine are required. To have an accurate model of
the machine, which represents all the characteristics of the physical machine, the
parameters need to be determined accurately. An in depth analysis and simulation of an
induction machine can be carried out only with accurate parameters that represent the
actual machine. Consequently to accurately model a three-phase induction machine,
accurate parameter values which represent the actual operating conditions being
modelled should be known.
There are different ways to determine the parameters of an induction machine modelled
by the conventional or steady state method. In this work the parameters are obtained by
taking measurements of input voltage, current and power over a wide speed range [1, 2].
For a three-phase induction machine with variable rotor parameters or constant rotor
parameters, the determination of parameters is dependent on phase voltage, phase
current, phase power and rotor speed.
The parameter determination method is based on the well known equivalent circuit
shown in Fig. 6.1. In this equivalent circuit the arrow through X
lr
and R
r
/s indicates that
these two parameters may be treated as variables for the case where rotor parameter
variations are taken into account
CHAPTER 6 PARAMETER DETERMINATION FOR AN INDUCTION MACHINE
116
jX
m R
m
jX
ls
R
s
jX
lr
R
r
/s
Fig. 6.1 The per-phase equivalent circuit with shunt magnetising branch impedance
represented in parallel
To convert the shunt magnetising branch impedance from parallel to series form
2
2 2
m m
M
m m
R X
R
R X
=
+
(6.1)
2
2 2
m m
M
m m
R X
X
R X
=
+
(6.2)
Where R
M
and X
M
are the resistive and the reactive equivalent components, respectively,
of the shunt magnetising branch represented in series form.
And to revert, to the parallel branch parameters:
2 2
M M
m
M
R X
R
R
+
= (6.3)
2 2
M M
m
M
R X
X
X
+
= (6.4)
The modified form of the per-phase equivalent circuit is given in Fig. 6.2. Here the
shunt magnetising branch elements are connected in series.
jX
ls
R
s
jX
lr
R
r
/s
jX
M
R
M
Fig. 6.2 Per-phase equivalent circuit with shunt magnetising branch impedance
represented in series form
117
6.2 Open-circuit and short-circuit test
Open-circuit and short-circuit tests are conducted to find the parameters of the induction
machine. At each test all the parameter values were taken into consideration. Even
when two parameters having a large ratio in their values were compared, the smaller
value was not neglected. It should be noted that the rotor leakage reactance is referred
to the stator frequency and from the usual assumption X
ls
= X
lr
.
R
s
is obtained from a DC measurement of stator resistance taking some consideration
for skin effect. Alternatively a Ware test [3] can be used, where R
s
is measured by
removing the rotor and supplying the stator with AC voltage at 50Hz .The difference
between the Ware test and the simple DC test can be about 5%. The details of the tests
are given below.
6.2.1 Open-circuit test
The open-circuit test is conducted by supplying rated voltage to the stator while driving
the induction motor at its synchronous speed using an external prime mover. When the
motor runs at synchronous speed the slip, s, will be zero and as a result the current
flowing in the rotor becomes zero. Then for the open-circuit test, the conventional
equivalent circuit model can be reduced to the one shown in Fig. 6.3.
X
ls
R
s
X
M
R
M
V
O
I
O
Fig. 6.3 Per-phase equivalent circuit of three-phase induction machine under no load
test
With V
O
- the measured open-circuit phase voltage
I
O
- the measured open-circuit phase current
P
O
- the measured open-circuit three-phase power
So at slip s = 0:
118
Total input resistance under open-circuit condition
2
3
=
O
O
O
P
R
I
(6.5)
Total input impedance under open-circuit condition
O
O
O
V
Z
I
= (6.6)
Total input reactance under open-circuit condition
2 2
O O O
X Z R = (6.7)
Then
M O s
R R R = (6.8)
M O ls
X X X = (6.9)
6.2.2 Short-circuit test
The short-circuit test (or locked rotor or standstill test) is conducted by blocking the
motor using a locking mechanism or using another prime mover to hold the induction
motor at zero speed. At standstill, rated current is supplied to the stator. When the speed
of the rotor is zero, the slip will be unity. At this slip, the resistive value on the rotor
side will be R
r
, which is the referred rotor winding resistance. Fig. 6.4 shows the per-
phase equivalent circuit for the short-circuit or standstill test condition.
jX
ls
R
s
jX
lr
R
r
jX
M
R
M
I
sh
V
sh
Fig. 6.4 Per-phase equivalent circuit at standstill (short-circuit test)
At slip s = 1
Total input resistance under short-circuit condition
2
3
sh
sh
sh
P
R
I
= (6.10)
Total input impedance under short-circuit condition
sh
sh sh sh
sh
V
Z R jX
I
= = + (6.11)
Total input reactance under short-circuit condition
2 2
sh sh sh
X Z R = (6.12)
119
Where V
sh
- the measured short-circuit input phase voltage
I
sh
- the measured short-circuit input phase current
P
sh
- the measured short-circuit three-phase input power
6.2.3 Induction machine with constant rotor parameters
For a wound rotor induction machine the rotor winding arrangement is similar to that of
the stator winding. The rotor parameters for the wound rotor induction machine do not
change with speed or slip.
From Equation (6.11) and Fig. 6.4 the exact expression for the short-circuit test is given
as
( )( )
( ) ( )
M M r lr
sh sh s ls
r M lr M
R jX R jX
R jX R jX
R R j X X
+ +
+ = + +
+ + +
(6.13)
Using the assumption that X
ls
= X
lr
and substituting the expressions for R
M
and X
M
,
Equations (6.8) and (6.9) respectively, into Equation (6.13) gives
( ( )( )
( )
+ +
+ = + +
+ +
O s O ls r lr
sh sh s lr
r O s O
R R j X X R jX
R jX R jX
R R R jX
(6.14)
Multiplying both sides of Equation (6.14) by ( )
r O s O
R R R jX + + and then simplify by
equating the real and imaginary parts on both sides of the equation, gives an expression
for the rotor winding resistance and the rotor leakage reactance.
From the real part, the expression for
r
R is
2 2
2
sh O sh s sh O s O s ls O ls
r
O sh
R R R R X X R R R X X X
R
R R
+ +
=
(6.15)
and from the imaginary part the expression for
lr
X is
2( )
+ +
=
sh O sh r sh O sh s s O O r
lr
O s
R X X R X R X R R X X R
X
R R
(6.16)
Simplifying Equation (6.16), gives
2
1 1 1
0 + + =
lr lr
a X b X c (6.17)
Where
1 O sh
a X X =
1 1 0 0
2( )
O
b X a b c = +
120
2 2
1 0 0 1 sh O O sh
c c X d X X X a = +
0 O sh
b R R =
0 O s
c R R =
0 sh s
d R R =
Solving Equation (6.17) gives the solution for
lr
X or
ls
X as follows
2
1 1 1 1
1
4
2

=
lr
b b a c
X
a
(6.18a)
Here there are two mathematical solutions for the rotor leakage reactance. To determine
the realistic solution the terms can be partially substituted to obtain
( )( )
( )
2
1 1 1
1
4
2 2

= +
O sh O s
lr O
O sh
R R R R b a c
X X
X X a
(6.18b)
In Equation (6.18b), if the + sign is chosen the rotor leakage reactance, X
lr
, will be more
than the total input reactance during the Open-circuit test, X
O
. This is not a realistic
solution. Hence the solution must be
2
1 1 1 1
1
4
2

=
lr
b b a c
X
a
(6.18c)
Minimising Equation (6.15) using the coefficients given in Equation (6.17) gives
2
0 0
0
2 +
=
sh O lr O lr
r
d c X X X X X
R
b
(6.19)
Hence R
r
is calculated by substituting the value of X
lr
in Equation (6.19).
6.2.4 Induction machine with variable rotor parameters
The rotor parameters of induction machines with a deep bar rotor or double-cage
squirrel-cage rotor vary due to rotor current displacement effect, or skin effect [4]. The
ideal machine for motoring application will have varying rotor resistance, large at
standstill and decreasing as the speed rises. The referred rotor leakage reactance will
increase as the speed rises. The explanation for the variation in rotor parameters is
illustrated in Fig. 6.5.
121
(a) (b) (c)
Stator
Air gap
Rotor leakage
magnetic flux
Rotor bar
Rotor
Mutual magnetic flux
Current
Fig. 6.5 Current displacement with rotor speed a) zero speed b) intermediate speed
c) close to synchronous speed
The variation of rotor resistance and rotor leakage reactance can be explained as
follows. At zero rotor speed, slip equal to one, there is maximum relative speed between
the rotor and the rotating magnetic flux set by the stator current. As a result, at zero
rotor speed the frequency of the induced rotor current is the same as the stator excitation
frequency. The leakage flux produced by the rotor current will pass through the iron
below the bottom of the rotor bar, as shown in Fig. 6.5, because of high permeability of
the iron core. Now imagine the bar to consist of an infinite number of layers of
differential depth. The layers at the lower part of the bar will be linked by all rotor
leakage flux produced by the rotor current flowing in all layers of the bar. However the
layers at the upper part of the bar will be linked only by the flux produced due to the
current flowing in these layers. As inductance is flux linkage per unit current, the
leakage inductance at the bottom of the rotor bar is greater than that of the upper part of
the rotor bar. With alternating current flowing in the bars this leakage inductance effect
produces leakage reactance, which is higher at the bottom of the bar than at the top of
the bar. This is significant when the frequency of the rotor current is the same as the
frequency of the stator excitation current, i.e. at slip s = 1. The leakage reactance
decreases along the bar from the bottom to the top. Hence at slip equal to one current
will be displaced upwards and almost all the current will flow in the upper part of the
bar, as shown in Fig. 6.5a. This phenomenon is basically the same as skin and proximity
effect in any system of conductors with alternating current in them. Since the upper part
122
is close to the air gap the possibility of the flux produced by the rotor current crossing
the air gap and linking the stator core to become part of the mutual flux increases. As
the flux produced by the rotor current crossing the air gap to link the stator increases the
leakage flux in the rotor will decrease. A decrease in rotor leakage flux per unit current
decreases the rotor leakage inductance and thereby the rotor leakage reactance when
referred to the stator side.
However the resistance will increase because the effective area utilized by the current
decreases. Hence, referred to the stator, at high slip, the rotor resistance is high but the
rotor leakage inductance is at its lowest value. The frequency of the rotor current
decreases with an increase in rotor speed (a decrease in slip) because of the decrease in
relative motion between the rotating magnetic field and the rotor bars. For this reason at
a speed half way between standstill and synchronous speed, about s = 0.5, the cross-
sectional area of the bar used by the current has increased, as shown in Fig. 6.5b. At a
rotor speed close to synchronous speed, s = 0, the rotor current frequency is low and
there is almost no rotor current displacement, as shown in Fig. 6.5c. Close to
synchronous speed the magnitude of the induced current is low and this small current
flows over almost all of the cross-sectional area of the rotor bar. The spread of the
current over almost the whole cross-sectional area of the bar increases the effective
cross-sectional area and consequently the resistance to the flow of current decreases. As
a result the rotor resistance will have a lower value at higher speeds. In this condition
the rotor flux linking only the rotor increases and the rotor leakage inductance also
increases. A typical example of rotor parameter variation for an induction motor with
double-cage rotor is given in Fig. 6.6 [5].
123
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
1
2
3
4
5
6
7
8
9
Slip
Rotor resistance
Rotor leakage reactance
R
o
t
o
r

r
e
s
i
s
t
a
n
c
e

a
n
d

r
e
a
c
t
a
n
c
e

(
o
h
m
)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
1
2
3
4
5
6
7
8
9
Slip
Rotor resistance
Rotor leakage reactance
R
o
t
o
r

r
e
s
i
s
t
a
n
c
e

a
n
d

r
e
a
c
t
a
n
c
e

(
o
h
m
)
Fig. 6.6 Rotor parameter variations with slip for deep bar induction machine
For an induction machine with variable rotor parameters, once the open-circuit test and
short-circuit tests are carried out the rotor parameters at slip equal to one can be
calculated in the same way as in the induction machine with constant rotor parameters.
The variation of rotor parameters at different slip can be obtained by measuring the
stator phase voltage (V
S(S)
), phase current (I
S(S)
) and the three-phase power (P
S(S)
)
supplied to the induction machine at a given slip. At slip equal to one X
lr(1)
= X
ls
. Fig.
6.7 shows the variable rotor parameters and associated variation in input quantities
which are functions of slip.
jX
ls
R
s
jX
lr(S)
( ) r S
R
s jX
M
R
M
I
S( S)
V
S( S)
Fig.6.7 Per-phase equivalent circuit with variable rotor parameters
124
X
lr(S)
and R
r(S)
are the rotor leakage reactance and rotor resistance respectively, referred
to the stator. They vary as functions of slip (rotor frequency) or rotor speed. Viewed
from the stator input side
Total input impedance at any slip, s,
( )
( )
( )
=
S S
S S
S S
V
Z
I
(6.20)
Total input resistance at any slip, s,
( )
( ) 2
( )
3
=
S S
S S
S S
P
R
I
(6.21)
Total input reactance at any slip, s,
2 2
( ) ( ) ( )
=
S S s S s S
X Z R (6.22)
Then using Fig. 6.7
( )
( )
( )
( )
( ) ( )
( )
( )
| |
+
+
|
\ .
+ = + +
| |
+ +
+
|
\ .
r S
M M
lr S
S S S S s ls
r S
lr S M
M
R
R jX
jX
s
R jX R X
R
X X j
R
s
(6.23)
Multiplying both sides of Equation (6.23) by
( )
( )
( )
r S
lr S M
M
R
X X j
R
s
| |
+ +
+
|
\ .
and then
equating the real parts leads to
( ) ( ) ( )
( )
( ) ( ) ( ) ( ) ( )
(
r S
S S s M ls S S M s S S M S S ls lr S M
R
R R R X X X R R X X X X R
s
+ + + =
(6.24)
and equating the imaginary parts gives
( ) ( ) ( ) ( )
( )
( ) ( ) ( ) ( ) ( )
r S
S S ls M S S s M ls S S s S S lr S M M
R
X X X R R R X X R R X R X
s
+ = +
(6.25)
Equations (6.24) and (6.25) can be written as
( ) ( ) r S lr S
AR BX E + = (6.26)
( ) ( ) r S lr S
CR DX F + = (6.27)
Solving the above two equations simultaneously, the rotor resistance and the rotor
leakage reactance at a given slip can be calculated from:
( ) r S
BF ED
R
BC AD
(6.28)
125
( ) lr S
CE AF
X
BC AD
(6.29)
Where
( ) S S s M
R R R
A
s

=
( ) ls S S M
B X X X = +
( ) S S ls M
X X X
C
s

=
( ) S S s M
B R R R = +
( ) ( )
( ) ( ) s S S S S ls M M
R R X X E R X =
( ) ( )
( ) ( ) ls S S s S S M M
X X R R F R X =
Brown & Grantham[6] devised a method whereby the parameters are determined using
a variable speed test, where a speed control method, such as Ward-Leonard, is used to
vary the speed, and the values of voltage, current and power are measured at each value
of speed. However, the test time can be quite long for this method so the temperature
can change during the test, hence changing the resistance values. Also, because of the
period of time spent at each speed to record the data, it is not possible to determine the
parameters under rated conditions. The motor temperature would become excessive at
high slips due to the increased current (up to ten times the full-load current at standstill)
causing overheating and damage to the induction machine.
To avoid the temperature change during the test, the data can be collected rapidly using
a DSP data acquisition system. The DSP data acquisition can be done so quickly that
the required data can be collected as the induction machine runs up to speed without
any variation in temperature and it allows the standstill test to be implemented at full
voltage.
6.2.5 Results for DSP based parameter determination
Variation of rotor parameters exists in squirrel-cage induction machines. Even in
induction machines with a single-cage rotor the variation in rotor parameters is
significant [1, 2]. The monitoring system used for the parameter determination is as
shown in Fig. 6.8. The monitoring system for this part of the experimental setup is based
126
on a 603 PowerPC floating-point processor running at 250MHz with a TMS320F240
slave DSP. The testing and measurements described below were carried out in
collaboration with D. McKinnon [2].
Temperature display
2
3
2
3
Digital Real Time Filters
Digital Signal Processing System
Digital
Filters
Digital
Filters
Analogue Anti-Aliasing Filters
Induction
Motor
A
B
C
V I
R R R
I
n
c
r
e
m
e
n
t
a
l

E
n
c
o
d
e
r
Thermocouple
Temperature display
2
3
2
3
Digital
Filters
Digital
Filters
Induction
Motor
A
B
C
V I
R R R
I
n
c
r
e
m
e
n
t
a
l

E
n
c
o
d
e
r
2
3
2
3
Digital
Filters
Digital
Filters
Induction
Motor
A
B
C
V I
R R R
I
n
c
r
e
m
e
n
t
a
l

E
n
c
o
d
e
r
Thermocouple
Fig. 6.8 Monitoring system for parameter determination
The parameter determination algorithm discussed in the previous section together with
the above monitoring system were used for the parameter evaluation of a 3-phase,
7.5kW, 4-pole, 13.8A, 415V, 50Hz delta connected induction motor with a single-cage
rotor and fully enclosed 215T frame. The speed was monitored by a very accurate
incremental encoder with 10000 lines/revolution, together with a four-fold pulse
multiplication mechanism i.e. 40000 pulses/revolution and a 24-bit counter at the
incremental encoder input.
In any induction machine parameter determination technique the most important data to
be measured are the ones at zero and synchronous speed, which for on-line systems, are
the most difficult values to sample. The difficulty at zero speed is that when a machine
127
is switched on with full voltage supply, there exists a transient current, the peak value of
which can be substantially higher than the steady state current. Furthermore, as soon as
voltage is applied to the machine, the rotor will start to rotate. Consequently a separate
test is normally needed to measure the necessary data at zero speed.
Both of the above problems can be overcome by simply rotating the motor in the
reverse direction, reversing the phase sequence of the supply and then start sampling as
soon as the machine reaches zero speed. It has also been established that a convenient
method of reducing the severity of the speed reversal transients is to ramp up the
voltage in the negative speed region such that the voltage is at rated value as the
machine approaches zero speed.
Initially 50% of the rated voltage was applied while the induction machine was rotating
in the reverse direction. The motor decelerated in the reverse direction to zero and then
accelerated in the forward direction. All the transients are settled when the machine
crosses zero speed. The measured voltage, current and power as a function of time is
shown in Fig. 6.9 and Fig. 6.10 shows the input quantities as a function of speed.
1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5
0
20
40
60
80
100
120
140
160
180
200
220
Time (Sec)
L
i
n
e

V
o
l
t
a
g
e

(
V
)
(a)
128
1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5
0
5
10
15
20
25
30
Time (Sec)
L
i
n
e

C
u
r
r
e
n
t

(
A
)
(b)
1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5
-1000
-500
0
500
1000
1500
2000
2500
3000
3500
4000
Time (Sec)
T
o
t
a
l

I
n
p
u
t

P
o
w
e
r

(
W
)
(c)
Fig. 6.9 Three-phase induction motor input quantities as a function of time (a) measured
line voltage (b) measured line current (c) measured input power
129
0 300 600 900 1200 1500
0
20
40
60
80
100
120
140
160
180
200
Speed (rpm)
L
i
n
e

V
o
l
t
a
g
e

(
V
)
(a)
0 300 600 900 1200 1500
0
5
10
15
20
25
30
Speed (rpm)
L
i
n
e

C
u
r
r
e
n
t

(
A
)
(b)
130
0 300 600 900 1200 1500
-1000
-500
0
500
1000
1500
2000
2500
3000
3500
4000
Speed (rpm)
T
o
t
a
l

I
n
p
u
t

P
o
w
e
r

(
W
)
(c)
Fig. 6.10 Three-phase induction motor input quantities as a function of speed
(a) measured line voltage (b) measured line current (c) measured input power
The negative power close to synchronous speed clearly indicates that the speed actually
passes through synchronous speed and consequently synchronous speed data can be
evaluated from this test to calculate X
m
and R
m
. The value of power measured when the
rotor speed crosses the synchronous speed is still in the transient condition. It is difficult
to calculate the steady state parameters from this transient condition. However, the
measured power when the rotor speed crosses the synchronous speed for the last time or
when it just reaches the synchronous speed, gives parameters close to the actual values.
Even though the values are not exact, this method of parameter determination does not
suffer from several problems identified for the existing method. That is, there is no need
to carry out a separate open-circuit and short-circuit tests, there is no need for a variable
speed drive and the problem associated with temperature rise and voltage levels is
eliminated due to the very much shorter period of time required for the test. Fig. 6.11
shows the variation of X
lr
and R
r
with rotor frequency for the test machine with single-
cage rotor. The test was carried out at room temperature and the value of stator winding
resistance measured was 1.8O. The small dip in power Fig. 6.10(c) is related to the 5th
131
harmonic torque [7] and it is suspected also that the natural resonance frequency of the
mechanical system has contribution to this slight dip.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
1
2
3
4
5
6
7
8
Slip
Rotor Leakage Reactance
Rotor Resistance
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
1
2
3
4
5
6
7
8
Slip
Rotor Resistance
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
1
2
3
4
5
6
7
8
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
1
2
3
4
5
6
7
8
Slip
Rotor Resistance
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
Fig. 6.11 Variation of rotor parameters for machine single-cage rotor
In Fig. 6.11 it is important to notice that even for the single-cage type of rotor there is a
variation of rotor parameters due to rotor current displacement effect i.e. the skin and/or
proximity effect. The variation in rotor parameters cannot be identified by the
conventional open-circuit and short-circuit tests.
At rated voltage, the free-running motor suffers under severe transient conditions and
the steady-state model can no longer be applied. Therefore, in order to capitalize on the
benefits of the fast data acquisition system and the parameter determination method, an
inertial load was coupled to the shaft of the machine to provide damping of the transient
conditions. Whilst this treatment diminishes the transient effect, it has the disadvantage
of preventing the machine from accelerating up through synchronous speed, thus
inhibiting the calculation of the effective iron loss resistance, R
m
, and the magnetization
reactance, X
m
. Therefore, a synchronous machine was coupled to the induction motors
shaft, providing both the inertial load and the means for driving the induction motor at
synchronous speed. When the induction motor neared synchronous speed the
synchronous motor was activated to draw the machine up to exactly synchronous speed.
With this method the parameters at synchronous speed are obtained very accurately,
132
because there is no transient condition and the synchronous motor is locked at precisely
synchronous speed.
Fig. 6.12 shows the results of this test method for various supply voltages at a constant
motor temperature of 40
o
C. The run up to speed tests were conducted over an
appropriate time period for the applied voltage, that is, a long duration (30 seconds) for
low voltage and a short duration (5 seconds) for high voltage.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
Slip
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
lowest Voltage
Rotor Resistance
200V
L-L
250V
L-L
300V
L-L
350V
L-L
highest Voltage
lowest Voltage
highest Voltage
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
Slip
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
lowest Voltage
Rotor Resistance
200V
L-L
250V
L-L
300V
L-L
350V
L-L
highest Voltage
lowest Voltage
highest Voltage
Fig. 6.12 Variation of rotor parameters with slip and supply line to line voltage
The set of results in Fig. 6.12 shows that both the rotor leakage reactance and rotor
resistance decrease with increasing supply voltage at each slip. This can be attributed to
saturation of the iron core. Due to saturation, for the same slip or rotor current
frequency, if the voltage is increased the rotor leakage inductance (rotor leakage flux
per unit rotor current) will decrease. The decrease in rotor leakage reactance with
increased voltage is as expected due to saturation of the leakage paths. The rotor
leakage inductance decreases and there will be less shifting of the rotor current towards
the upper part of the rotor bars due to current displacement effect. When current flows
in an increased effective cross-sectional area, the resistance of the current path drops
and therefore rotor resistance decreases.
133
The induction machine can operate at different temperature values depending on the
loading condition and surrounding air temperature. The temperature of the induction
machine was varied to see the effect of temperature on the rotor parameters. The data
acquisition system and the parameter determination algorithm were used to identify the
effect of temperature variation on induction machine parameters. Because the test is so
rapid the temperature remains almost the same even as the speed of the induction
machine varies. It is not possible to achieve the result with the conventional way of
manual data recording.
Fig. 6.13 shows the variation of X
lr
and R
r
with slip at 50% of the rated voltage. The test
was carried out under the same operating conditions, but at two different temperatures:
40 and 75
o
C. A thermocouple is embedded in the stator winding for temperature
measurement. When the temperature of the stator is changed the temperature of the
rotor will also change. In this experiment the aim is to see the variation of rotor
parameters with temperature. The rotor temperature will follow the stator temperature
variation. If necessary the rotor temperature can be measured by inserting a
thermocouple in the rotor when the induction machine stops and compare with the
stator temperature. This will give the gradient of temperature between the stator core
and rotor core and be able to determine the exact temperature of the rotor while the
temperature sensor is in the stator.
From Fig. 6.13 it can be seen that, as expected, a rise in temperature causes a rise in
resistance at each slip or rotor current frequency. This rise in resistance results in a
lower induced current in the rotor and therefore causes a small rise in the leakage
inductance because the rotor leakage flux linkage per unit of rotor current slightly
increases. That is, since at each rotor current frequency the rotor flux linkage is almost
constant due to saturation for a particular supply voltage, the leakage inductance must
increase as current decreases. However, it is clear that temperature rise affects the rotor
resistance significantly more than the rotor reactance.
134
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
40
o
C
75
o
C
Slip
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
Rotor Resistance
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
40
o
C
75
o
C
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
40
o
C
75
o
C
Slip
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
Rotor Resistance
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
40
o
C
75
o
C
Slip
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
Rotor Resistance
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
40
o
C
75
o
C
0 0.1 0.2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
40
o
C
75
o
C
Slip
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
Rotor Resistance
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
40
o
C
75
o
C
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
6
7
8
40
o
C
75
o
C
Slip
R
o
t
o
r

R
e
s
i
s
t
a
n
c
e

a
n
d

R
e
a
c
t
a
n
c
e

(
O
)
Rotor Resistance
Fig. 6.13 Effect of temperature on rotor parameters
Fig. 6.14 shows the variation for magnetising reactance X
m
and iron loss resistance R
m
with supply voltage. These results were derived from measurements made as the
induction machine was driven at synchronous speed (open-circuit test) by a
synchronous machine and the supply voltage varied accordingly. Two techniques were
used to confirm the synchronous speed test data. In the first technique discrete data is
stored for 30 seconds for each temperature at each fixed voltage level. This method is
referred to as the discrete method. The average of each quantity during this period is
then used to evaluate X
m
and R
m
. The second technique varies the supply voltage
continuously during the test, from 50V to rated voltage, for each temperature. This
method is referred to as the continuous method. The duration of this test is 60 seconds
for each temperature.
Fig. 6.14(a) and Fig. 6.14(b) show the comparison of the results from the test methods
carried out at synchronous speed. The result shows that there is no difference between
discretely varying and continuously varying the supply voltage. The results were taken
135
when the motor temperature was 95
o
C. A low pass filter is applied during the
continuous method to aid in identifying R
m
. It is clear that the two test methods are in
excellent agreement.
Fig. 6.14(c) and Fig. 6.14(d) present the effects that variations in supply voltage and
temperature have on the effective magnetizing reactance, and iron loss resistance
respectively. Only the results from the discrete voltage method have been shown for
clarity in these figures.
As shown in Fig 6.14 the magnetizing reactance rises to a peak value and then, due to
saturation, decreases steadily as the voltage rises. The iron loss resistance curve exhibits
the predicted linear rise with voltage up to close to the rated supply voltage. It is noted
that the magnetizing reactance rises slightly with temperature, whereas the iron loss
resistance appears to have no consistent relationship with temperature.
0 50 100 150 200 250 300 350 400 450
0
20
40
60
80
100
120
140
160
180
200
Line voltage (V)
M
a
g
n
e
t
i
z
i
n
g

r
e
a
c
t
a
n
c
e

-
X
m
(
O
)
Discrete voltages
Continuous voltages
0 50 100 150 200 250 300 350 400 450
0
20
40
60
80
100
120
140
160
180
200
Line voltage (V)
M
a
g
n
e
t
i
z
i
n
g

r
e
a
c
t
a
n
c
e

-
X
m
(
O
)
Discrete voltages
Continuous voltages
(a)
136
0 50 100 150 200 250 300 350 400 450
0
1000
2000
3000
4000
5000
6000
Discrete voltages
Continuous voltages
Line voltage (V)
I
r
o
n

l
o
s
s

r
e
s
i
s
t
a
n
c
e

R
m
(
O
)
0 50 100 150 200 250 300 350 400 450
0
1000
2000
3000
4000
5000
6000
Discrete voltages
Continuous voltages
0 50 100 150 200 250 300 350 400 450
0
1000
2000
3000
4000
5000
6000
Discrete voltages
Continuous voltages
Line voltage (V)
I
r
o
n

l
o
s
s

r
e
s
i
s
t
a
n
c
e

R
m
(
O
)
(b)
0 50 100 150 200 250 300 350 400 450
0
20
40
60
80
100
120
140
160
180
200
Line voltage
X
m
55
o
C
X
m
40
o
C
X
m
65
o
C
X
m
75
o
C
X
m
95
o
C
M
a
g
n
e
t
i
z
i
n
g

r
e
a
c
t
a
n
c
e

-
X
m
(
O
)
(c)
137
0 50 100 150 200 250 300 350 400 450
0
1000
2000
3000
4000
5000
6000
Line Voltage (V)
I
r
o
n

l
o
s
s

r
e
s
i
s
t
a
n
c
e

R
m
(
O
)
X
m
55
o
C
X
m
40
o
C
X
m
65
o
C
X
m
75
o
C
X
m
95
o
C
0 50 100 150 200 250 300 350 400 450
0
1000
2000
3000
4000
5000
6000
Line Voltage (V)
I
r
o
n

l
o
s
s

r
e
s
i
s
t
a
n
c
e

R
m
(
O
)
X
m
55
o
C
X
m
40
o
C
X
m
65
o
C
X
m
75
o
C
X
m
95
o
C
X
m
55
o
C
X
m
40
o
C
X
m
65
o
C
X
m
75
o
C
X
m
95
o
C
(d)
Fig. 6.14 Variation of (a) magnetizing reactance with voltage at 95
o
C (b) iron loss
resistance with voltage at 95
o
C (c) magnetizing reactance with temperature and voltage
(d) iron loss resistance with temperature and voltage
6.3 Sensitivity study on variable rotor parameters
To effectively model a three-phase induction motor the parameter values should be
accurately known. These parameters are obtained from the measured values of voltage,
current and power as discussed in Section 6.2. Any practical meter will have
discrepancy between the actual value and the measured value. The accuracy of
measurement influences the accuracy of the calculated parameters. Consequently any
measurement error in obtaining the values of current, voltage and power would be
directly reflected as an error in the calculated parameters. The extent of the error
reflected as a deviation from the actual values of parameters was studied [8].
To achieve this analysis a hypothetical model of a three-phase induction motor was
considered and a MATLAB program was used. Rotor parameter variations due to skin
or proximity effect are taken into account. MATLAB functions can generate any data
with a given standard deviation and percentage error of the measured sample values
138
randomly, which follow the normal curve in statistics. All the randomly generated data
are above and below the true value within a predetermined range.
6.3.1 The effect of combining measurement errors
The parameters of a three-phase induction machine can be determined using the most
well known method that uses no load and locked rotor tests or the fast DSP based
parameter determination discussed previously. The parameters can be calculated taking
into consideration the effect of temperature, speed, or any factor that affects the values
of the parameters and can be achieved in different ways. All the procedures used in
calculating the parameters could be correct, but that does not show anything related to
the measurement errors in the conventional meters and sensors used to measure voltage,
current and power. There is a need to calculate the magnitude of error in the parameters
for a given percentage of error, or range of error, in the measurement devices, i.e. a
sensitivity study on the effect of the parameters due to measurement data error.
6.3.1.1 Percentage errors
In practice, it is often useful to express an error in relation to the magnitude of the
quantity being measured. Expressing error as a percentage is important because it helps
predict the accuracy needed for individual measurements in order to achieve a given
percentage error in the final result [9]. Manufacturers mostly give the accuracy of a
given measuring device or sensor in percentage error.
The percentage error in a given measurement is given by
m t
e 100
t
= (6.30)
where e - percentage error
m - measured value
t - true value
Here e is the size of the error in percent when the measured value of a quantity is
compared with its true value.
139
Rearranging Equation (6.30), can give a measured quantity expressed in terms of the
percentage error and the true value as follows.
( ) m 1 e t = + (6.31)
6.3.1.2 Combining errors
The errors in two measured values are combined when the two measured values are
added, subtracted, multiplied or divided. Assume that the measured values for two true
quantities of a and b are a
m
and b
m
respectively. And assume the measurement errors for
the two measured quantities are e
a
and e
b
. Then the measured quantities can be
expressed as
m a
a a e = (6.32)
m b
b b e = (6.33)
When the two measured quantities are added, subtracted, multiplied or divided the
combined error in the result will be the sum of the individual errors [9].
6.3.2 Induction machine parameters for analysis of measurement error
For the purpose of analysis and simulation a hypothetical three-phase induction
machine with double-cage or deep-bar rotor is used. The values of the parameters were
chosen to have similarity with a real motor in the laboratory. To have a wider operating
region the parameters were selected for speeds starting from 1500rpm to 1500rpm,
where 1500rpm is the synchronous speed of a 4-pole, 50Hz induction machine.
The hypothetical three-phase induction motor that was selected for this simulation has
the following parameters:
X
1s
=2O
R
s
=1.5O
R
m
=800O
X
m
=100O
and the values of rotor leakage reactance and rotor resistance with respect to speed were
as shown in Fig. 6.15 below.
140
-1500 -1000 -500 0 500 1000 1500
0
1
2
3
4
5
6
Speed(rpm)
O
h
m
s
X
lr
R
r
Fig. 6.15 Values of rotor resistance, R
r,
and rotor leakage reactance, X
lr
The values of R
r,
and X
lr
are dependent on speed, as shown in Fig. 6.15. All parameters
in the per-phase equivalent circuit of the three-phase induction motor were substituted
with the parameter values given above. The advantage of using a hypothetical machine
for the sensitivity study is that the value of the parameters and parameter variations are
precisely known. These exact values can then be compared precisely with the new
values calculated as a result of the known error in the simulated measured data.
6.3.3 Statistical tools
The main problem in experimentally investigating parameters with the aid of measuring
systems is the properties of the systems on the measurement results. The source of
measuring errors could be an internal problem of the meter or the influence of external
disturbances. Whatever the cause of the disturbance or error in the measuring system a
statistical tool is needed to analyse the measurement error.
The preferred statistical tool for the analysis of this measurement error is the normal
(Gaussian) distributions. Normal distributions are continuous probability distributions.
141
They are frequently used as a population model because they provide a reasonable
approximation to the distribution of many different variables [10]. It is an assumption
that the distribution of error as measured from the true value follows the normal curve.
According to an Empirical Rule [10] if the distribution of error can be reasonably well
approximated by a normal curve, as shown in Fig. 6.16, then:
- Approximately 68% of the measured data are within one standard deviation of
the actual measured value
- Approximately 95% of the measured data are within two standard deviations of
the actual measured value
- Approximately 99.7% of the measured data are within three standard deviations
of the actual measured value.
error
Actual measurement
error
34% 34%
13.5% 13.5%
2.35%
2.35%
1o 1o
2o 2o
3o 1o 1o 1o
Fig. 6.16 Measurement error with a normal distribution
Any deviation from the actual measurement (mean), , is an error of measurement. The
measurement error can be expressed in terms of standard deviation (o) or percentage of
error. Using the normal curve approximately 99.7% of all the observations are within
three standard deviations of the mean. If the actual value is 100 and the standard
deviation is 0.333 then one standard deviation is 0.333 in the unit of measured value.
Therefore, a value three standard deviations from the mean will give an error of 0.999
which is about 1%. To simulate 1.5% error for an actual value of 100 the standard
deviation is 0.5 in the unit of measured value.
142
6.3.4 Simulation of parameter determination with measurement error
The hypothetical induction motor with the equivalent circuit and parameter values
specified above was considered when connected in a star configuration to a 50Hz
supply with line to line voltage of 415V.
The speed was allowed to vary from 1500rpm to 1500rpm. The corresponding values
of slip were 2 to 0. The synchronous speed was assumed to be 1500rpm (i.e. 4-pole).
The simulation was carried out at a constant supply voltage and then finding the actual
values of current and power for a given speed or slip. The given value of voltage and
the calculated values of current and power from the equivalent circuit model of the
induction machine represent the actual values of measurement. Any deviation from
these actual values would signify an error of measurement in a practical application of
the parameter determination technique. Parameters calculated from the actual values of
voltage, current and power correspond to the actual parameters of the motor and any
calculation with the introduction of some error in voltage, current and/or power gives
the values of parameters with a given percentage of error from the actual values.
Random generated measured values using MATLAB functions that fit to a normal
(Gaussian) distribution were used in the investigation. The measurement error was
expressed in terms of percentage error. The standard deviation of the true (actual)
measured value can be calculated from the percentage measurement error.
For the implementation of the measurement error analysis expressed in terms of the
standard deviation of the true value, a MATLAB program was developed with the
possibility of adjusting measurement error in the voltage, current, and/or power. The
percentage of error was selected in such a way that it would represent the practical
error, which could occur in a practical measurement setup. To simulate the different
measurement errors that can occur in the conventional meters and sensors for measuring
the values of current, voltage and power a MATLAB function given by
normrnd(mean, standard deviation, N
o
of columns, N
o
of rows)
was used to generated the hypothetical measured values with distribution as shown in
Fig. 6.17.
143
Hypotetical measured value
D
e
n
s
i
t
y
actual value
Fig. 6.17 Data generated for simulation of measurement error
In Fig. 6.17 since the values were generated randomly, at any point there is any
possibility of minimum or maximum error to represent the error in voltage, current or
power measurement. In general the possibility of measuring the precise value or a value
with maximum error is a matter of probability. Of course the extent of the deviation
from the true value is proportional to the percentage of measurement error or the
standard deviation. At each speed there are 200 hypothetical measured values generated
by the MATLAB program to represent the possibility of measurement error within a
given percentage error of the conventional meter reading.
As shown in Fig. 6.18 to Fig. 6.20 the maximum error in the rotor parameters increases
as the percentage measurement error in voltage, current and/or power increases. Close
to synchronous speed, when the slip approaches zero, there is greater error, in the
calculated values of X
lr
, but reduced error in the calculated values of R
r
.
144
-1500 -1000 -500 0 500 1000 1500
0
1
2
3
4
5
6
7
8
Speed (rpm)
O
h
m
s
R
2
X
2
(actual value)
(actual value)
Values with error
R
r
X
lr
Fig. 6.18 Error in rotor parameters due to 0.5% error in voltage current and/or power
-1500 -1000 -500 0 500 1000 1500
0
1
2
3
4
5
6
7
8
Speed (rpm)
O
h
m
s
R
2
X
2
(actual value)
(actual value)
Values with error
R
r
X
lr
Fig. 6.19 Error in rotor parameters due to 1% error in voltage current and/or power
145
-1500 -1000 -500 0 500 1000 1500
0
1
2
3
4
5
6
7
8
Speed (rpm)
O
h
m
s
R
2
X
2
(actual value)
(actual value)
Values with error
R
r
X
lr
Fig. 6.20 Error in rotor parameters due to 1.5% error in voltage current and/or power
The reason for the error close to synchronous speed is the magnitude of X
lr
is very small
compared to R
r
/s. The result shows that induction motor performance is very insensitive
to large errors in rotor leakage reactance close to synchronous speed.
From the result of the simulation, i.e. for 1% measurement error, the maximum error in
X
ls
is 6.4%, in X
m
1.5% and in R
m
2.2%. For the rotor parameters, as can be seen in
Fig.19, the maximum error in the motoring region for R
lr
is 3.3% and 11.4% for that of
X
lr
. Except near to synchronous speed, the absolute error of all values of X
lr
and R
r
is
almost the same.
The developed shaft torque for the hypothetical induction motor discussed above, using
constant rotor parameters and variable rotor parameters, is shown in Fig. 6.21. When
the rotor parameter variations are ignored there is a maximum error of 61%. Similar
percentage error is obtained for current and power [1, 2]. These differences in the
calculated results are enormous when compared with the errors introduced by the
inaccuracy in measurement. This vividly illustrates the need to take rotor parameter
146
variations into account for squirrel-cage machines which exhibit any significant degree
of rotor current displacement effect.
-1500 -1000 -500 0 500 1000 1500
0
20
40
60
80
100
120
Speed (rpm)
T
o
r
q
u
e

(
N
m
)
For variable R
r
and X
lr
Assuming constant R
r
and X
lr
Fig. 6.21 Simulated shaft torque for variable and constant rotor parameters
6.4 Summary
The results of an investigation into the variation of magnetizing reactance, iron loss
resistance and rotor parameters with temperature and supply voltage have been
presented. A monitoring system employing digital signal processing techniques was
used to observe these effects.
To date parameter identification methods over a large speed range have typically
required reduced voltages to prevent the motor overheating from excessive current at
high slips. The run up to speed test described enables the parameters to be determined at
supply voltages up to and including the rated voltage without damaging the motor under
test. A separate test was used to determine the iron loss resistance and magnetizing
reactance. This test incorporated two methods. Both methods were conducted at
synchronous speed, one using continuously varied supply voltage, the other at discrete
voltages. Agreement between the results was excellent. All tests were performed at
corresponding temperatures.
147
Parameter identification is especially important in the design of electrical drive systems.
It is obvious from these results that parameter variations cannot be ignored. It has been
shown that rotor parameter variations exist even for a single-cage induction motor.
The error in the values of induction motor parameters arising from measurement error in
voltage, current and power used to determine the parameters have been presented.
These three measurement quantities are essential for accurate parameter determination.
Even with measurement errors of 1.5% the determined parameters are still very
acceptable and considerably more accurate than if rotor parameter variations are
ignored. Only very close to synchronous speed is there substantial error in the value of
X
lr
determined, but this is almost irrelevant in any meaningful analysis of the inductive
machine, because close to synchronous speed the effect of X
lr
is swamped by the effect
of R
r
/s. When rotor parameter variations are ignored the percentage errors in the
current, power and torque are substantial for machines which exhibit a significant
degree of rotor current displacement effect.
6.5 References
[1] C. Grantham and H. Tabatabaei-Yazdi, "Rapid Parameter Determination for use in the Control
High Performance Induction Motor Drives" IEEE 1999 International Conference on Power
Electronics and Drive Systems, PEDS99, July 1999, Hong Kong, pp. 267-272.
[2] D. McKinnon, D. Seyoum and C. Grantham Investigation of the effects of supply voltage and
temperature on parameters in a 3-phase induction motor including iron loss, Proc. AUPEC02,
Melbourne, Sep29-Oct 2, 2002, ISBN0-7326-2206-9.
[3] D. H. Ware, Measurement of stray load losses in induction machines, Trans., AIEE, 1945, pp.
194-196.
[4] P. L Alger, The nature of Induction Machines, Gordon and Breach Inc., New York, 1965, pp.
111.
[5] C. Grantham and H. Tabatabaei_Yazdi, "Rapid Parameter Determination for use in the Control of
High Performance Induction Motor Drives, Proc. AUPEC 2001, Perth, 23 - 26 September. 2001,
pp. 31 - 36.
[6] J. E. Brown and C. Grantham, "Determination of the parameters and parameter variations of a 3-
phase induction motor having a current displacement rotor", Sept.1975, Proc. IEE, 122, No. 9, pp.
919-921.
[7] G. R. Slemon, Electric Machines and Drives, Addison-Wesley Publishing Inc., Ontario, 1992.
148
[8] C. Grantham, D. Seyoum D. Indyk and D. McKinnon, Calculation of the Parameters and
Parameter Variations of an Induction motor and the effect of measurement error, Proc.
AUPEC00, Brisbane, Australia, 2000, pp. 225-228.
[9] Donald T. Graham, Principles of radiological Physics, Edinburgh, UK, 1996, pp. 86-88.
[10] J. Devore and R. Peck, STATISTICS The Exploration and Analysis of Data, Duxbury press,
Belmont 1997, pp. 187 - 232.
CHAPTER 7
EXCITATION OF THREE PHASE INDUCTION
GENERATOR USING THREE AC CAPACITORS
7.1 Introduction
Any induction machine requires excitation current to magnetise the core and produce a
rotating magnetic field. The excitation current for an induction generator connected to
an external source, such as the grid, is supplied from that external source. If this
induction generator is driven by a prime mover above the synchronous speed, electrical
power will be generated and supplied to the external source. An isolated induction
generator without any excitation will not generate voltage and will not be able to supply
electric power irrespective of the rotor speed.
In general an induction generator requires reactive power for its operation. Three
charged capacitors connected to the stator terminals of the induction generator can
supply the reactive power required by the induction generator. Provided that the
conditions for self-excitation are satisfied the charged capacitors cause the terminal
voltage to build up at the stator terminals of the induction generator. When the charged
capacitors are connected to the terminals a transient exciting current will flow and
produce a magnetic flux. This magnetic flux will generate voltage and the generated
voltage will be able to build the charge in the capacitors. As the charge increases, more
exciting current is supplied to the induction generator. The magnetic flux continues to
increase hence producing a higher generated voltage. In this way voltage is built up.
However, if the capacitors are not charged, and a remnant magnetic flux in the core
exists, then a small voltage will be generated at the terminals of the induction generator
due to that remnant flux. This small voltage will charge the capacitor. The charged
149
CHAPTER 7 EXCITATION OF THREE PHASE INDUCTION GENERATOR USING THREE AC CAPACITORS
capacitor will now be able to produce a small exciting current. With time the exciting
current grows and produces magnetic flux more than the remnant magnetic flux and
voltage will be built up. This is similar to the way that current and voltage interact in a
resonant circuit. For the voltage to build up across the terminals of the induction
generator, there are certain requirements for minimum rotor speed and capacitance
value that must be met. When capacitors are connected across the stator terminals of an
induction machine, driven by an external prime mover, voltage will be induced at its
terminals. The induced emf and current in the stator windings will continue to rise until
steady state is attained. At this operating point the voltage and current will continue to
oscillate at a given peak value and frequency. The rise of the voltage and current is
influenced by the magnetic saturation of the machine. In order for self-excitation to
occur with a particular capacitance value there is a corresponding minimum speed.
Self-excited induction generators are good candidates for wind powered electricity
generation, especially in remote areas, because they do not need an external power
supply to produce the excitation magnetic field. Permanent magnet generators can also
be used for wind energy applications; however the generated voltage increases linearly
with wind turbine speed. An induction generator can cope with a small increase in
speed from its rated value because, due to saturation, the rate of increase of generated
voltage is not linear with speed. Furthermore when there is a short circuit at the
terminals of the self-excited induction generator (SEIG) the voltage collapses providing
a self-protection mechanism. Additional advantages of SEIGs include lower cost,
reduced maintenance, they are rugged with simple construction, and they have a brush-
less rotor (squirrel cage). Fig. 7.1 shows the SEIG driven by a wind turbine.
Wind direction
Induction
generator
Gear box
Rotor blade
C
C
C
Fig. 7.1 SEIG with a capacitor excitation system driven by a wind turbine
150
7.2 Model of self-excited induction generator
Basically the model of a SEIG is similar to an induction motor. The only difference is
that the self-excited induction generator has capacitors connected across the stator
terminals for excitation. The conventional steady state per-phase equivalent circuit
representation of an induction machine is convenient to use for steady state analysis.
However, the d-q representation is used to model the self-excited induction generator
under dynamic conditions. The d-q representation of a self-excited induction generator
with capacitors connected at the terminals of the stator windings and without any
electrical input from the rotor side is shown in Fig. 7.2 below.
The representation shown in Fig. 7.2 can be redrawn in detail, in a stationary stator
reference frame, with direct and quadrature circuits separately represented as given in
Fig. 7.3. The capacitance is labelled C in Fig. 7.3.
The capacitor voltages in Fig. 7.3 can be represented as
cq qs cqo
1
V i dt V
C
= +
)
(7.1)
cd ds cdo
1
V i dt V
C
= +
)
(7.2)
Where
cqo cq
t 0
V V
=
= and
cdo cd
t 0
V V
=
= are the initial voltage along the q-axis and d-axis
capacitors, respectively.
With L
s
= L
ls
+L
m
and L
r
= L
lr
+L
m
the rotor flux linkage is given by
qr m qs r qr qro
L i L i = + + (7.3)
dr m ds r dr dro
L i L i = + + (7.4)
Where
qro qr
t 0

=
= and
dro dr
t 0

=
= are the remnant or residual rotor flux linkages
along the q-axis and d-axis, respectively.
Then, with an electrical rotor speed of e
r
, the rotational voltage in the rotor circuit
along the q-axis is
( )
r dr r m ds r dr r dro
L i L i e e e = + +
( )
r dr r m ds r dr qr
L i L i K e e = + + (7.5)
151
i
qs
i
qr
i
dr
i
ds
Q-axis
Q-axis
e
r
Fig. 7.2 D-Q representation of self-excited induction generator
L
ls
R
s
i
qs
L
lr
i
qr
L
m
i
mq
R
r
qr
qs
V
cq
e
r
dr
+ -
C
S
(a)
L
ls
R
s
i
ds
L
lr
i
dr
L
m
i
md
R
r
dr
ds
V
cd
-e
r
qr
+ -
C
S
(b)
Fig. 7.3 Detailed d-q model of SEIG in stationary reference frame (a) q-axis circuit
(b) d-axis circuit
152
and the rotational voltage in the d axis of the rotor circuit is
( )
r qr r m qs r qr r qro
L i L i e e e = + +
( )
r qr r m qs r qr dr
L i L i K e e = + + (7.6)
where
dr r qro
K e = and
qr r dro
K e = are constants, which represent the initial induced
voltages along the d-axis and q-axis, respectively. The constants K
dr
and K
qr
are due to
the remnant or residual magnetic flux in the core. And e
r
is the equivalent electrical
rotor speed in radians per second. That is,
Electrical speed = number of pole pairs mechanical speed
The matrix equation for the d-q model of a self-excited induction generator, in the
stationary stator reference frame, using the SEIG model given in Fig. 7.3 and from
Equations (7.5) and (7.6), is given as:
1 0 0 0
0 1 0 0
0
0
qs cqo
s s m
s s m ds cdo
m r m r r r r qr
r m m r r r r
dr dr
i V
R pL pC pL
R pL pC pL i V
pL L R pL L i K
L pL L R pL
i K
e e
e e
qr
( ( + + ( (
( (
( (
+ +
( (
( (
= +
( (
( ( +
( (
( (
+
( ( (

(7.7)
Z I
V
V
V
Where Z is the impedance matrix, I
V
is the stator and rotor currents vector and V
V
is the
voltage vector due to initial conditions.
7.3 Analysis of self-excitation process
Whether it is a wound rotor induction machine or a squirrel cage induction machine
voltage will develop across the capacitors connected to the stator terminals when the
rotor of the induction machine is driven by an external prime mover. The voltage
developed across the capacitors is the terminal voltage of the self-excited induction
generator.
153
7.3.1 RLC circuit characteristics
The behaviour and analysis of the self-excited induction generator is similar to an RLC
circuit. Since the equations in the induction generator are complex, the principle of self-
excitation process will be explained first using a simple RLC circuit. For analysis
purpose the RLC circuit given in Fig. 7.4 will be considered. The plus sign at the
capacitor is the polarity for the initial capacitor voltage.
C
R L S
i(t)
+
V
co
Fig. 7.4 RLC circuit
Energy may be stored in an inductor or in a capacitor. A resistor is incapable of energy
storage. Switch S in Fig. 7.4 is closed at t = 0. In general the two initial conditions at t =
0
-
are: current might have been flowing in the inductor (provided that the inductor was
part of another circuit, not shown in Fig. 7.4, before t = 0
-
) or initial voltage exists in the
capacitor. If all initial conditions are zero then there will not be any transient or steady
state current flow.
In Fig. 7.4 assume that at the instant the switch is closed, the current is zero and the
voltage across the capacitor is v
c
= -V
co
. When the switch is closed, the voltage equation
in the RLC circuit is given by
1
0
co
di
Ri L idt V
dt C
+ + =
)
(7.8)
Introducing the p operator for d/dt Equation (7.8) can be rewritten as
1
( )
co
R pL i t V
pC
| |
+ + =
|
\ .
(7.9)
Then i(t) can be expressed as
2
( )
1
co
pV
i t
p L pR C
=
+ +
(7.10)
If the denominator of Equation (7.10) is equated to zero, then
2
1 p L pR C + + = 0 (7.11)
154
Equation 7.11 is called the characteristic equation because it contains the information
about the behaviour of the resulting current. The roots of the characteristic equation are
2
1
1
2 2
R R
p
L L L
| |
= +
|
\ .
C
and
2
2
1
2 2
R R
p
L L L
| |
=
|
\ .
C
2
(7.12)
Using the roots given in Equation (7.12) the complete solution for the current
expression in Equation (7.10) is
(7.13)
1
1 2
( )
p t p t
i t Ae A e = +
where A
1
and A
2
are determined from the initial conditions and circuit parameters, and
p
1
and p
2
are determined from the values of the circuit parameters R, L, and C. If the
voltage across the capacitor v
c
(t) is the output voltage of interest then
1
( )
c c
v t idt V
C
=
)
o
.
In Equation (7.12), if
2
1
2
R
L LC
| |
<
|
\ .
then the roots p
1
and p
2
of Equation (7.13) are
complex quantities which can be expressed as p
1
=o +je and p
2
=o -je. Relating these
expressions with Equation (7.12), the real part of the roots, o
,
is always negative
provided the resistance R is positive. As a result with positive R there will be a decaying
oscillation. o represents the rate at which the transient decays and e, the imaginary part
of the roots, represents the frequency of oscillation. In passive circuits, like the RLC
circuit mentioned above, all transient solutions have negative o, meaning that the
transient is reducing in magnitude with the progression of time and finally decays to
zero. However, if o is positive, this implies that the transient is growing with the
progression of time, and in theory would increase to infinity. o can be positive only if
the resistance R is negative. Negative resistance implies a power source whereas
positive resistance implies a power sink. Fig. 7.5 shows the current in the RLC circuit
when L = 0.1H, C = 100F, V
co
= -10V and the magnitude of resistance R is 1.2O with
positive value in Fig. 7.5(a) and negative value in Fig. 7.5(b). Close to t = 0 the
magnitude of the instantaneous current flowing in the RLC circuit in both cases is the
same.
155
0 0.2 0.4 0.6 0.8 1
-150
-100
-50
0
50
100
150
time (sec)
C
u
r
r
e
n
t

(
A
)
0 0.05 0.1
-0.3
0
0.3
0 0.2 0.4 0.6 0.8 1
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
time (sec)
C
u
r
r
e
n
t

(
A
)
(a)
A
detail of A
(b)
Fig. 7.5 Current in series RLC circuit (a) for R = 1.2O and (b) for R = -1.2O
Transients which grow in magnitude as shown in Fig. 7.5(b), with a positive value of o,
are very rare. There is no variation in any of the values of R, L or C and as a result the
current keeps on growing. Any current flowing in a circuit dissipates power in the
circuit resistance. If there is an increasing current that dissipates increasing power, there
must be some energy source available to supply the increasing power. This is in fact the
case in the self-excited induction generator. The example above of a very rare transient
is characteristic of a SEIG where the power source is a prime mover.
7.3.2 Conditions for self-excitation in induction generator
Basically an induction machine is modelled using RLC circuit parameters. Self-
excitation in an induction generator is the growth of current and the associated increase
in the voltage across the capacitor without an external excitation system. As in Fig.
7.5(b), transients that grow in magnitude (self-excitation), with a positive real part of
the root, can only happen if there is an external energy source that is able to supply all
the power losses associated with the increasing current. The self-excited induction
generator is able to have a growing transient because of the external mechanical energy
source that is driving the induction generator. The process of terminal voltage build up
continues in the manner described until the iron circuit saturates and the voltage
therefore stabilises. In terms of the transient solution considered above, the effect of this
156
saturation is to modify the magnetisation inductance L
m
, such that the real part of the
roots becomes zero; the transient then neither increases nor decreases and becomes a
steady-state quantity giving a continuous self-excitation.
The energy source, referred to above, which is necessary for this type of unusual
transient to occur, is provided by the kinetic energy (KE) of the rotor. If the rotor is
driven by an external prime mover, the KE of the rotor is maintained and self-excitation
and energy transfer continues permanently. The initiation of the process of self-
excitation is therefore a transient phenomenon and is better understood if analysed using
instantaneous values of currents and voltages.
Unlike the simple RLC circuit that has been discussed, the roots for the self-excited
induction generator which can be derived from Equation (7.7), are dependent on the
induction machine parameters, the capacitor connected at the stator terminals of the
induction generator and the rotor speed when coupled to an external prime mover.
Determination of the roots of the characteristic equation of the currents in the induction
generator is the key to finding out whether the induction generator will self-excite or
not.
Equation (7.7) can be re written as
1 0 0
0 1 0
qs cqo
s s m
s s m ds
m r m r r r r qr
r m m r r r r
dr dr
i V
R pL pC pL
R pL pC pL i V
pL L R pL L i K
L pL L R pL
i K
e e
e e
cdo
qr
( ( + + (
( (
(
+ +
( (
(
=
( (
(
+
( (
(
+
( ( (

(7.14)
The characteristic equation of the currents can be obtained from the expression for the
current vector in Equation (7.14). This characteristic equation for the currents can be
solved using a matrix partitioning, which gives a compact polynomial expression or a
direct matrix inversion, which produces an expression with a lower order polynomial
characteristic equation.
157
7.3.2.1 Using matrix partition
Equation (7.14) can be rewritten as
ss sr os s
rs rr or r
Z Z V I
Z Z V I
( (
=
( (

(
(
(7.15)
Where
cqo
os
cdo
V
V
V
(
=
(

, initial conditions on the stator side,
qr
or
dr
K
V
K
(
=
(

, initial conditions on the rotor side
qs
s
ds
i
I
i
(
=
(

, and
qr
r
dr
i
I
i
(
=
(

,
1 0
0 1
s s
ss
s s
R pL pC
Z
R pL pC
+ + (
=
(
+ +

,
0
0
m
sr
m
pL
Z
pL
(
=
(

,
m r
rs
r m m
m
pL L
Z
L pL
e
e
(
=
(

,
r r r r
rr
r r r
R pL L
Z
L R p
e
e
+
L
(
=
(
+

From Equation (7.15)
os ss s sr r
V Z I Z I = + (7.16)
or rs s rr r
V Z I Z I = + (7.17)
1 1
r rr or rr rs s
I Z V Z Z I

= (7.18)
Where
1
2 2 2
1
( )
r r r r
rr
r r r r r r r
R pL L
Z
L R pL R pL L
+ (
=
(
+ + +

e
e e
Substituting Equation (7.18) in Equation (7.16)
1
(
os sr rr or ss sr rr rs s
V Z Z V Z Z Z Z I
=
1
)
(7.19)
Equation (7.19) can be rewritten as
( )
( )
2 2 2 2 2
2 2 2
1 ( )
1 ( )
cqo qr r r r r
m
r r r r cdo dr
s s m r r m r r m r r
qs
ds
m r r s s m r r m r r
V K R pL L
pL
L R pL V K
R pL pC p L R pL pL L pL R
i
i
pL R R pL pC p L R pL pL L
+ ( ( (
=
( ( (
+ A

(
+ + + + A A
(
(
(
(
A + + + + A

e
e
e e
e e
2 2
(7.20)
where
2 2 2 2 2 2 2 2
( ) 2
r r r r r r r r r r
R pL L R pR L p L L e e A = + + = + + +
158
From Equation (7.19) the stator currents can be calculated as
( )
1 1 1
( )
s ss sr rr rs os sr rr or
I Z Z Z Z V Z Z V

= (7.21)
and in detail
( )
( )
2 2 2 2 2
2 2 2 2
1 ( )
1 ( )
m r r s s m r r m r r
qs
ds
cqo qr r r r r
m
r r r r cdo d
R pL pC p L R pL pL L pL R
pL R R pL pC p L R pL pL L
i
i
V K R pL L
pL
L R pL V K
e e
e e
e
e
| | (
+ + + + A A
|
(
|
(
A + + + + A
(

|
=
(
|
O

|
|
|
\ .
| | + ( ( (
|
( ( (
|
+ A
\ .
2
(7.22)
Where
( ) ( ) ( )
2 2
2 2 2 2 2
1 ( )
R pL pC p L R pL pL L pL R e e O= + + + + A + A
When a balanced three phase system is transformed to a two axis system the stator
currents i
qs
and i
ds
have similar waveforms. The difference is that i
qs
lags i
ds
by 90
o
.
Hence it suffices to analyse only the expression for i
ds
. The numerator of Equation
(7.22) helps to determine the multiplying constants for the solution of the current in the
time domain and these constants are dependent on the machine parameters, capacitance
value, rotor speed and initial conditions. However, setting the denominator of Equation
(7.22) equal to zero, = 0, gives the characteristic equation because it contains the
information about the behaviour of the resulting current. If any of the roots of the
characteristic equation has a positive real part then there will be a growing transient
indicating that there will be self-excitation. If there is no positive real root then there
will not be any self-excitation.
O
Using Equation (7.22) the expression for i
ds
can be given as
( ) ( )
2 2
2 2 2 2 2
1 ( )
ds
U
i
R pL pC p L R pL pL L pL R e e
=
(
+ + + + A + A

(7.23)
where
159
( )
2 2 2 2 4 2 2 3 2 2 2
3 2 2 2 3 2 3 2
2
( ) ( )
( )
s s m r r m r
m r r cqo m r r qr m s s dr
r r cdo
p C R pL pC p C L R pL p C L R L V
p C L R V p C L L K p C L R pL K
U
pC
( A + + A A + A

A A + A +
=
A
e
e e
In Equation (7.23) U represents all the terms on the numerator and is dependent on the
initial charge in the capacitors, the remnant flux in the core, capacitance, rotor speed
and the machine parameters. U only has an effect on the coefficients of the partial
fraction expansion of (2), which determine the constants that will be multiplied with the
exponential current expression in the time domain, and does not affect the behaviour of
the current. The detail of the expression of U is long and it is not necessary to consider
when determining whether there is self-excitation or not. The analysis here is to identify
if there is self-excitation or not which is solely dependent on the expression in the
denominator of Equation (7.23).
Setting the denominator of Equation (7.23) to zero gives the characteristic equation
expressed by
( ) ( )
2 2
2 2 2 2 2
1 ( )
R pL pC p L R pL pL L pL R e e
(
+ + + + A + A =

0
( )
( )
2 2
2 3 2 2 2 2 2 2
2
( 1) ( )
0
pCR p CL p CL R pL p CL L p CL R
pC
e e ( A + + + +

=
A
( )
( )
( )
2
2 2 2 2 3 2 2 2 2
2
2 2
2
2 2 2 2 2
( ) ( 1) ( )
0
( 2 )
r r r r s s m r r m r r
m r r
r r r r r r
R pL L pCR p CL p CL R pL p CL L
p CL R
pC R pR L p L L
e e
e
e
(
+ + + + +

+
=
+ + +
(7.24)
Hence
( )
( )
2
2 2 2 2 3 2 2 2 2
2
2 2
( ) ( 1) ( )
0
r r r r s s m r r m r r
m r r
R pL L pCR p CL p CL R pL p CL L
p CL R
e e
e
(
+ + + + +

+ =
(7.25)
( )
2
2 2 2 2 2 2
2
2 2
3 2 2 2 2
( 2 )( 1)
0
( )
r r r r r r s s
m r r
m r r m r r
R pR L p L L pCR p CL
p CL R
p CL R pL p CL L
e
e
e
( + + + + +
+ =
(
+
(

(7.26)
Equation (7.26) can be written as
160
( ) ( )
2 2
4 3 2 2
0 Ap Bp Dp Ep F Gp + + + + + = (7.27)
where
2 2
( )
r s m r
A C L L L L =
2 2
( 2
r s r r s m r
) B C L R R L L L L = +
( )
2 2 2 2 2
2 ( )
r r s r r r s m r r r
D R L R R L L L L C e e = + +
2
L +
)
2 2 2
2 (
r r s r r r
E R L CR R L e = + +
2 2
r r
2
r
F R L e = +
2
m r r
G CL R e =
The expanded form of Equation (7.27) is
8 7 6 5 4 3 2
8 7 6 5 4 3 2 1 0
0 a p a p a p a p a p a p a p a p a + + + + + + + + = (7.28)
where
2
8
a A =
7
2 a A = B
B
E
E
F
3
2
6
2 a AD = +
5
2 2 a BD A = +
2 2
4
2 2 a AF BE D G = + + +
3
2 2 a BF D = +
2
2
2 a DF E = +
1
2 a E =
2
0
a F =
It is easy to find the roots of Equation (7.28), the characteristic equation, using
MATLAB or any mathematical equation solving program.
When Equation (7.28) is factorized it gives
1 1 1 1 2 2 2 2 3 3 3
4 4 4 4
( - )( - - )( - )( - - )( - )( - - )
( - )( - - ) 0
p j p j p j p j p j p j
p j p j
o e o e o e o e o e o e
o e o e
+ + +
+ =
(7.29)
161
If any of the roots in Equation (7.29) has a positive real value then there is self-
excitation. To determine the required capacitance value for an induction generator
running at a given rotor speed, the roots in Equation (7.29) are evaluated by increasing
the capacitance value until one of the real parts in the roots becomes positive.
7.3.2.2 Direct matrix inversion
The current expression for the equation representing a self-excited induction generator,
given in Equation (7.14) can be solved by applying Cramers rule or by finding the
inverse of the impedance matrix.
Cramers rule [1] is a mathematical tool for finding one of the variables in an unknown
vector in a matrix equation based on the calculation of determinants. Applying
Cramers rule to Equation (7.14) results in
1 0
0 0
1 0 0
0 1 0
s s cqo m
cdo m
m qr r r r
r m dr r r r r
ds
s s m
s s m
m r m r r r
r m m r r r r
R pL pC V pL
V p
pL K R pL L
L K L R p
i
R pL pC pL
R pL pC pL
pL L R pL L
L pL L R
e
e e
e e
e e
+ +
+
+
=
+ +
+ +
+
+
r
r
L
L
pL
(7.30)
Since the characteristic equation of the d-axis stator current is the determinant of the
denominator, only the denominator part of i
ds
will be expanded. The determinant of the
numerator will be represented by a variable U, which is dependent on the machine
parameters, initial conditions, capacitance and electrical rotor speed. U affects only the
magnitude of the current i
ds
and does not contain any information on the behaviour of
the resulting current. The determinants in Equation (7.30) can be evaluated to give
6 5 4 3 2 2 2
6 5 4 3 2 1 0
( )
ds
U
i
A p A p A p A p A p A p A C p
=
+ + + + + + /( )
(7.31)
where
( )
2 2 2 2 4
6
- 2
s r s r m m
A C L L L L L L = +
162
( )
2 2 2 2
5
2
2
s s r r s m r s r s r m
A C R L L L R L L L R L R L = +
( )
( )
2 2 2 2 2 2 2 2 2 4 2 2
4
2 2
( ) 4 2 2
2
s r r r r s r s r s r s m r m r s r m
s r r m
A C L L R L R L R R L L L L L R R L
C L L L L
e e = + + + +
+
( )
e
( )
2 2 2 2 2 2 2 2 2
3
2 2
s r s r r m s r r r s s s r s r r s r r m
A C L L R L L R L R R R L R C R L L L R R L e e
(
= + + + +

2 2 2 2 2 2 2 2 2 2 2
2
2 ( ) ( ) 2 4
s r r r s r r r r m r r r s r
A L C L R R C L R L CL L L R CR e e e = + + + + +
2 2 2
1
2 ( ) 2
s r r r r r
A R C L R R L e = + +
2 2 2
0 r r r
A L R e = +
Analysing the denominator of Equation (7.31) is sufficient to determine whether
initiation of self-excitation will occur. To determine if there is an onset of self-
excitation, or not, the denominator of Equation (7.31) is set to zero. That is
6 5 4 3 2
6 5 4 3 2 1 0
0 A p A p A p A p A p A p A + + + + + + = (7.32)
Equation (7.32) is a sixth order characteristic equation and it has six distinct roots which
are first order complex roots in the form of
1 1 1 1 2 2 2 2 3 3 3 3
( - )( - - )( - )( - - )( - )( - - ) 0 p j p j p j p j p j p j o e o e o e o e o e o e + + + =
r
(7.33)
If any of the roots has a positive real part, then at that given specific operating point
there will be self-excitation.
The current and voltage will grow until the magnetising inductance saturates and makes
the real part of the roots zero, which shows that there is a continuous oscillation
(Alternating Current and Voltage) as long as the prime mover is driving the induction
generator. The transient and steady state solution due to each of the roots can be
obtained by using partial fraction expansion.
The eighth order and the sixth order expressions given in Equations (7.28) and (7.32),
respectively, are solutions for the same variable in the same machine. The eighth order
ended up as a higher order because it has a term with a second order expressed as
that can be factorized in the numerator of Equation (7.24)
2 2 2 2 2
2
r r r r r
L p R L p L R e + + +
163
and cancelled with one of the expressions in the denominator. As the eighth order
polynomial is very long it is difficult to factorise and see the terms that can be readily
cancelled. At a specific operating point, i.e., for a given speed and capacitance two of
the roots in the eighth order polynomial due to the terms that would be cancelled can be
determined as
2 2 2 2 2
2 0
r r r r r r
L p R L p L R e + + + =
( )( ) 0 ( )) ( ))
r r r r r r r r
p R j L L p R j L L e e = + + + (7.34)
For any speed and capacitance values, the real part of the roots for the expression given
in Equation (7.34) is always negative and is given by -R
r
/L
r
. This means that whether
the 8
th
order or the 6
th
order characteristic equation is used the result is exactly the same.
The roots with positive real parts in both characteristic equations are exactly the same.
The additional roots in the 8
th
order have a negative real part which gives a damped
transient response.
7.4 Characteristics of magnetising inductance in induction machine
In the modelling of an induction machine used for motoring applications, it is important
to determine the magnetising inductance, L
m
, at rated voltage. In the SEIG the variation
of magnetising inductance is the main factor in the dynamics of voltage build up and
stabilisation. In this investigation the magnetising inductance is determined by driving
the induction machine at synchronous speed and taking measurements when the applied
voltage was varied from zero to 120% of the rated voltage with rated frequency. The
magnetising inductance is calculated, without approximation, using the parameter
determination method discussed in Chapter 6. Here, conventional high accuracy meters
are used for measurements of voltage, current and power, because the accuracy of the
voltage and current sensors in the fast measurement system, discussed in Chapter 5, are
not good for low values (close to zero) of voltages and currents. The computed power
will be erroneous if the accuracy of voltage and current measurements is poor. This is
especially important because the magnetising inductance for voltages and currents close
to zero is used in the calculation for the initiation of self-excitation process.
164
The variation of the magnetising inductance, measured at rated frequency, for the
induction machine used in this investigation is given in Fig. 7.6, where the dots are
experimental results and the curve is a fourth order curve fit given by:
11 4 8 3 5 2 3
1.62 10 2.67 10 1.381 10 1.76 10 0.23
m ph ph ph p
L V V V V

= + + +
h
(7.35)
Where V
ph
is the phase voltage.
0 50 100 150 200 250 300
0
0.05
0.1
0.15
0.2
0.25
0.3
Phase voltage V
ph
(V)
- Experimental
- Fourth order curve fit
M
a
g
n
e
t
i
s
i
n
g
i
n
d
u
c
t
a
n
c
e
L
m
(
H
)
A
B
C
Fig. 7.6 Variation of magnetising inductance with phase voltage at rated frequency
As can be observed in Fig. 7.6, L
m
starts from a small value then increases to reach its
peak value and finally starts to drop. This change in L
m
is due to the characteristics of
the magnetising curve and the fact that:
m
m
L
I
= (7.36)
Where
m
- rms flux linkage, in wb-turn
I - rms magnetising current, in A
165
It was found by conducting a test in the laboratory that the variation of magnetising
inductance in a transformer follows a similar pattern to the one shown in Fig. 7.6.
7.5 Minimum speed and capacitance for self-excitation
The induction machine used as the SEIG in this investigation is a three-phase wound
rotor induction motor with specification: 4 pole, 415V star connected, 7.8A, 3.6kW,
50Hz. The d-q model, shown in Fig. 7.3 is used because it provides the complete
solution, transient and steady state, of the self-excitation process. The parameters
obtained from parameter determination tests at rated values of voltage and frequency
are L
ls
= L
lr
= 11.4mH, L
m
= 181mH, R
s
= 1.6O, R
r
= 2.75O.
When the three capacitors are connected in star the voltage rating of each capacitor is
equal to the rated phase voltage. However, if the capacitors are connected in delta the
voltage rating of each capacitor should be equal to the line-to-line voltage. In delta
connected capacitors, even though the voltage rating of each capacitor is higher than the
rating of the capacitors in star connection by a factor of 1.73, the magnitude of the
capacitance is lower by a factor 3, i.e. 1/3 of the capacitance in the star connection.
When the induction machine, as shown in Fig. 7.3, with switch S closed, is driven by a
prime mover, voltage will start to develop at a corresponding minimum speed. The
minimum speed for the onset of self-excitation can be obtained by solving the roots of
the 8
th
order polynomial equation given in Equation (7.28) or the 6
th
order polynomial
equation given in Equation (7.32) and then searching if there is a positive real part in
the roots. The minimum capacitance required for a given rotor speed of the induction
generator can be found by fixing the rotor speed and then increasing the value of the
capacitance until one of the real parts of the roots changes from negative to positive,
passing through zero. The value of capacitance that makes the real part of one of the
complex roots greater than zero is the minimum value of capacitance required for self-
excitation. To have a smooth plot of the minimum rotor speed versus minimum
capacitance requirement, the capacitance was incremented by a small value. The detail
of this procedure, theoretical determination of the minimum speed and minimum
166
capacitance for a fixed speed by incrementing the capacitance, is given in the flow chart
of Fig. 7.7.
Another way of finding the minimum rotor speed and corresponding minimum
capacitance required for self-excitation is first to set the capacitance at a given value
and then increase the rotor speed until one of the real parts of the complex roots
becomes positive. This is a good way to find the minimum capacitance and its
corresponding minimum rotor speed in the experimental setup.
start
Read machine
parameters
C = 0
Increment
C
Minimum
e
r
=0
Increment
e
r
Find roots from
Equations (7.32) or
(7.28)
Any positive
real root?
Yes
No
Save values of
C and e
r
Maximum
e
r
?
Yes
No
Stop
No
Reset
C to zero
Fig. 7.7 Flow chart to determine the minimum speed and minimum capacitance for
SEIG at no load
In the experimental setup a DC motor was used as a prime mover. The rotor speed of
the SEIG was varied by varying the speed of the DC motor while the capacitance was
kept at a given value. It is possible to increase the capacitance for a given speed,
167
however it is not as convenient as varying the speed and it is difficult to find capacitor
values that will give a smooth variation of capacitance.
There are two important rotor speeds; the first is the point at which self-excitation
occurs and the second is where self-excitation is lost. For a given capacitance value the
speed of the test machine was increased until the SEIG started to generate voltage, this
is the normal way of achieving self-excitation. The capacitance value and the rotor
speed at which the self-excitation started were recorded. Another test was conducted
where the self-excited induction generator is already generating voltage and the speed is
reduced until the SEIG loses its self-excitation. For a particular capacitance value, the
minimum rotor speed for self excitation, determined by increasing the rotor speed from
zero, is greater than the minimum rotor speed obtained by decreasing the rotor speed
until the SEIG losses its self-excitation. Because the SEIG is always started in the
unexcited mode, that is, from zero speed, the correct minimum rotor speed is chosen
from the first test, even though the second test produces a lower minimum speed.
The minimum rotor speed and minimum capacitance for self-excitation calculated and
measured are given in Fig. 7.8. For the theoretical determination of minimum rotor
speed and minimum capacitance, the effect of magnetising inductance on the onset of
self-excitation using the rated value of L
m
(i.e. 0.18H), as used in motoring analysis, and
the unsaturated value of 0.23H is shown in Fig. 7.8. The different values of L
m
are
obtained from Fig. 7.6. If the rated L
m
, 0.18H, is used for determining the onset of self-
excitation there will be an error as shown in Fig. 7.8. It was found out that the minimum
rotor speed and minimum capacitance required for self-excitation are dependent on
magnetising inductance but not on rotor parameters variation.
The principle of finding the minimum capacitance and the minimum rotor speed for
self-excitation at no load can be approximated by neglecting the stator winding
resistance and stator leakage inductance so that the capacitive reactance and the
inductive reactance will be equal. Since the induction generator starts without load the
rotor speed is almost the same as the synchronous speed of the induction machine.
Hence the approximate minimum capacitance required for self-excitation can be
calculated using
168
2
1
min
r m
C
L e
= (7.37)
Where e
r
- the electrical rotor speed, in rad/s
L
m
- the value of magnetising inductance close to zero voltage, in H.
0 50 100 150 200 250 300 350
0
500
1000
1500
Capacitance micro F
Speed (rpm)
Speed (rpm)
Capacitance (F)
- measured values
- using L
m
=0.18H (saturated value)
- using L
m
=0.23H (unsaturated value)
Fig. 7.8 Values of minimum capacitance and rotor speed for self-excitation at no load
A given induction machine can operate as a motor and as a generator depending on
whether the power source is electrical or mechanical. Considering motoring operation
the capacitance given in Equation (7.37) is equivalent to the capacitance required to
have a unity power factor when the induction motor is operating at no load, neglecting
friction, supplied from a source with angular frequency equal to e
r
. In no load motoring
or generating operation the synchronous speed is almost equal to the electrical rotor
speed.
The approximate minimum capacitance calculated using Equation (7.37) is slightly
more than the actual minimum capacitance calculated using the exact solution. The
169
percentage error of capacitance calculated using the approximate method, given in
Equation (7.37), and the exact method, discussed in Section 7.3.2, is shown in Fig. 7.9.
The error introduced by using the approximate calculation reaches up to 4.5% which is
acceptable to find an approximate minimum capacitance value required for an onset of
self-excitation in the SEIG using a simple calculation.
600 700 800 900 1000 1100 1200 1300 1400 1500
0
1
2
3
4
5
error in capacitance value ( %)
Speed (rpm)
Fig. 7.9 Error in capacitance when calculated using the approximate method
The SEIG needs to be started at no load. Hence it can be generalised that the
capacitance required for the onset of self-excitation in the SEIG rotating at a rotor speed
of e
r
, is almost equal to the capacitance required to have a unity power factor in a
motoring application, as the induction motor operates at no load and with an angular
supply frequency of e
r
.
7.6 Magnetising inductance and its effect on stability of generated voltage
The magnetising inductance, measured at 50Hz frequency of excitation, varies with
voltage as shown in Fig. 7.6. At the start of self-excitation (point A), where the voltage
is close to zero, L
m
is close to 0.23H. Once self-excitation starts the generated voltage
will grow and then L
m
also increases up to point B. When there is an increase in
magnetising inductance it increases the value of the positive real root of the
170
characteristic equation and consequently the generated voltage grows faster. Beyond
point B, up to point C, L
m
decreases while the voltage continues to grow until it reaches
its steady state value determined by the L
m
value, capacitance and the rotor speed.
Referring to Fig. 7.6, the unstable region is between points A and B. If the SEIG starts
to generate in this region, a small decrease in speed will cause a decrease in voltage and
this will bring a decrease in L
m
, which in turn decreases the voltage, and finally the
voltage will collapse to zero. Once the voltage collapses there is no transient
phenomenon and there will not be voltage build up even if the speed increases once
again to its initial value as shown in Fig. 7.10. This condition can cause
demagnetisation of the core. When the core is demagnetised there will not be self-
excitation. In order to initiate self-excitation in the demagnetised core, the core should
be magnetised by running the generator as a motor or exciting the windings from a DC
supply. The other arrangement is to charge the exciting capacitors from a DC supply.
0 2 4 6 8 10 12
-200
-100
0
100
200
0 2 4 6 8 10 12
0
200
400
600
800
1000
G
e
n
e
r
a
t
e
d
V
p
h

(
V
)
S
p
e
e
d

(
r
p
m
)
time (sec)
(a)
(b)
Fig. 7.10 Measured unsuccessful self-excitation at C=60F (a) generated phase voltage
(b) speed
Between points B and C is a stable operating region. When the speed of the prime
mover decreases voltage will decrease and L
m
increases which enables the self-excited
171
machine to continue to operate at a lower voltage as shown in Fig. 7.11. An increase in
L
m
means an increase in the positive real roots of the characteristic equation which is
good for stable operation of the SEIG. The results given in Fig. 7.10 and Fig. 7.11
reflect the condition of a SEIG driven by a wind turbine when the wind turbine is
operating at low wind speed. This condition is mapped by the curves for low torque
versus rotor speed as given in Fig. 2.9 of Chapter 2. From the characteristic of a wind
turbine, the gradient of the torque versus rotor speed curve is gentle, so that a large
change in rotor speed brings a small change in output torque.
0 2 4 6 8 10 12
-400
-200
0
200
400
G
e
n
e
r
a
t
e
d

V
p
h

(
V
)
0 2 4 6 8 10 12
0
200
400
600
800
1000
1200
S
p
e
e
d

(
r
p
m
)
time (sec)
(a)
(b)
Fig. 7.11 Measured self-excitation at C=60F and lower speed
(a) generated phase voltage (b)speed
For the same capacitance value, if the rotor speed is increased the generated voltage
also increases. The rated voltage given in Fig. 7.12 can be generated by increasing the
rotor speed above the speed given in Fig. 7.11(b) to a speed close to the synchronous
speed. The result given in Fig. 7.12 shows the voltage build up for the case where the
speed is close to the synchronous speed and is typically of the result to be expected
172
from a SEIG driven by a wind turbine. The small drop in speed is due to the power loss
in the self-excited induction generator associated with the generation of voltage.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
-400
-200
0
200
400
G
e
n
e
r
a
t
e
d

V
p
h

(
V
)
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0
500
1000
1500
time (sec)
S
p
e
e
d

(
r
p
m
)
(a)
(b)
Fig. 7.12 Measured self-excitation at C=60F with speed and generated voltage close to
rated values (a) generated phase voltage (b)speed
7.7 Onset of self-excitation when the SEIG is loaded
The SEIG equivalent circuit shown in Fig. 7.3 can be loaded with a resistive load by
connecting a resistance R
L
across the capacitor, C. With resistive load Equation (7.7) is
modified to the following equation
0 0
1 0
0
0 0
1
0
0
L
s s m
qs cqo
L
L ds
s s m
L
qr q
m r m r r r r
dr d
r m m r r r r
R
R pL pL
i V
R pC
R i
R pL pL
R pC
i K
pL L R pL L
i K
L pL L R pL
e e
e e
(
+ +
(
( ( +
(
(
( (
(
(
( (
(
+ +
= +
(
( (
+ (
(
( (
(
(
+
( (

(
+
(

cdo
V
(7.38)
173
Equation (7.38) can be solved to obtain its characteristic equation using a similar
procedure to that used to solve Equation (7.7). Analysing in a similar way, as discussed
in Section 7.3.2, the curves in Fig. 7.13 are obtained for different load resistors.
0 100 200 300 400 500 600
0
500
1000
1500
Capacitance (F)
RL > 1000O
R
L
= 25O
Speed (rpm)
RL = 13O
RL = 11O
RL = 9.5O
Fig. 7.13 Required capacitance and speed for self-excitation with load, R
L
For a given capacitance value the wind speed can vary without warning. Without load
the SEIG requires only a minimum speed for self-excitation, but a loaded SEIG has a
minimum and maximum speed for self-excitation as shown in Fig. 7.13. When R
L
is
large the characteristic is similar to the no load self-excitation case. If R
L
is small, larger
load, there is a minimum and maximum speed to produce self-excitation at a particular
capacitance value.
The characteristics in Fig. 7.13 help to find the minimum and maximum speed set
points for a given capacitance value. Once the minimum and maximum speed points are
obtained, the speed range for a safe generating range can be identified. It is clear that a
loaded generator has different onset of self-excitation characteristics for different values
of load resistance. This also helps to determine the speed range for the steady-state
generating characteristic of the SEIG. At high load resistance or at no load the
maximum speed limit is so high that it is not necessary to take that into consideration.
7.8 Simulation of self-excited induction generator
174
In conducting an experimental analysis of a self-excited induction generator, it is
difficult to get continuous values of a capacitance and determine the corresponding
speed. Mostly the components are in discreet values. It is also hard to see the condition
of the self-excitation beyond the rated values of the machine as it can damage the
machine. Simulation is extremely useful in predicting the condition of self-excitation
within the rated values of the machine and/or beyond these rated values.
7.8.1 The modelling of self-excitation process
To find the dynamics of a self-excited induction generator a mathematical model is
developed. The solution of this mathematical model gives the complete characteristics
comprising of transient and steady state for the voltage, current, power and frequency of
a self-excited induction generator. With the help of this mathematical model the
dynamic values of voltage, current and power of the SEIG at any given time can be
evaluated. The mathematical model takes into account the initial conditions in the
induction generator, namely the initial voltage in the exciting capacitors and the initial
induced voltage due to remnant magnetic flux in the magnetic core.
7.8.1.1 Determination of initial conditions
The initial conditions required in the equation for the simulation of self-excited
induction generator given in Equation (7.7) can be determined from measurements
performed on the induction machine and the capacitors.
Using the self-excited induction generator model shown in Fig. 7.3 the remnant or
residual flux linkages along the d-axis
dro
and along the q-axis
qro
can be estimated
from the stator terminal voltage of the induction machine measured without the exciting
capacitors, i.e. when the switch S is open. Then the rotational initial voltages in the
rotor are calculated from the remnant flux linkages and the rotor speed using
qr r dro
K e = and
dr r qro
K e = .
The initial voltage in the exciting capacitor can be measured using a conventional
voltmeter while the switch S in Fig. 7.3 is open. The initial voltage in the capacitor
175
decreases with time because of the internal leakage and discharge through the internal
resistance of the voltmeter.
7.8.1.2 The dynamic representation of self-excitation at no load
Based on the general equation of the self-excited induction generator a mathematical
model is developed to represent the dynamic characteristic involved in the voltage build
up of the SEIG. The dynamic analysis has to demonstrate the transient and steady state
values of voltage and frequency developed by the induction generator at the no load
condition.
Using Equations (7.1) and (7.2) Equation (7.7) can be written as
0 0 0
0 0 0
0
0
qs cq
s s m
s s m ds cd
m r m r r r r qr q
r m m r r r r
dr dr
i V
R pL pL
R pL pL i V
pL L R pL L i K
L pL L R pL
i K
e e
e e
r
( ( + ( (
( (
( (
+
( (
( (
= +
( (
( ( +
( (
( (
+
( ( (

(7.39)
Equation (7.39) can be rearranged to give
0 0 0 0 0
0
0 0 0 0 0 0
0 0 0
0 0 0 0
qs qs cq
s m s
s m ds s ds cd
m r qr r m r r r qr qr
m r r m r r r
dr dr dr
pi i V
L L R
L L pi R i V
L L pi L R L i K
L L L L R
pi i K
e e
e e
( ( (
( (
(
( ( (
( (
(
( ( (
( (
(
+ +
( ( (
( (
(
( ( (
( (
(
( ( ( ( (

0
= (7.40)
The above equation can be simplified as
ApI
V
+ BI
V
+ V
V
= 0 (7.41)
where
0 0
0 0
0 0
0 0
s m
s m
m r
m r
L L
L L
A
L L
L L
(
(
(
=
(
(
(

0 0 0
0 0
0
0
s
s
r m r r m
r m r r r
R
R
B
0
L R L
L L
e e
e e R
(
(
(
=
(
(
(

qs
ds
V
qr
dr
i
i
I
i
i
(
(
(
=
(
(
(

cq
cd
V
qr
dr
V
V
V
K
K
(
(
(
=
(
(
(

176
Using Matrix inversion, Equation (7.41) can be written as
pI
V
= -A
-1
BI
V
-A
-1
V
V
(7.42)
Where
1
0 0
0 0
0 0
0 0
m r
m r
m s
m s
L L
L L
L L
L L
A
L L
L L
L L
L L
(
(
(
(
(
=
(
(
(
(
(
(

and
2
s r m
L L L L =
Then evaluating Equation (7.40) in a state space form gives
2
2
r cq m qr
r s m r m r m r r
qs qs
m r s s m r r m r
ds ds
qr qr
m s s r m s r s r r
dr dr
s r m m s s r r s r
L V L K
L R L L R L L
L L L L L L
pi i
L
L L R L L L R
pi i
L L L L
pi i
L R L L L R L L
L L L L
pi i
L L L R L L L R
L L L L
e e
e e
e e
e e
(
(
(
( (
(
( (
(
( (
= + (
( (
(
( (
(
( (

(

(
(

r cd m dr
m cq s qr
m cd s dr
V L K
L L
L V L K
L L
L V L K
L L
(
(
(
(
+
(
(
(
+
(
(
(
(

(7.43)
Expanding Equation (7.43) gives four sets of first order differential equations as given
below
( )
2
1
qs r s qs m r ds m r qr m r r dr r cq m qr
pi L R i L i L R i L L i L V L K
L
e e = + (7.44)
( )
2
1
ds m r qs r s ds m r r qr m r dr r cd m dr
pi L i L R i L L i L R i L V L K
L
e e = + + + (7.45)
( )
1
m s qs s r m ds s r qr s r r dr m cq s qr qr
L R i L L i L R i L L i L V L K pi
L
e e + + + + = (7.46)
( )
1
s r m qs m s ds s r r qr s r dr m cd s dr dr
L L i L R i L L i L R i L V L K pi
L
e e + + = (7.47)
and from the equations of the capacitors
qs
cq
i
pV
C
= (7.48)
ds
cd
i
pV
C
= (7.49)
177
V
cd
and V
cq
are the no load output generated voltages and the three phase voltages can
be obtained by using the three-phase to two-axes transformation discussed in Chapter 3.
Since the mechanical time constant is much larger than the electrical time constant, in
equations 7.44 - 7.47 the rotor speed is assumed constant for small changes in the
voltages and currents.
The relationship between capacitance, rotor speed and generated voltage at the no load
condition is given in Fig. 7.14.
e
r1
C
1
C
2
Generated voltage
Exciting current
C
3
e
r2
e
r3
C
3
> C
2
> C
1
e
r3
> e
r2
> e
r1
O
Fig. 7.14 Relationship between capacitance value, rotor speed and
generated voltage at no load
The capacitor provides the exciting current required by the induction generator and the
induction generator charges the capacitor to increase the terminal voltage. An increase
in capacitor voltage provides an increase in exciting current to the induction generator.
In this way the voltage build up continues until the magnetising inductance decreases to
its saturated value and an equilibrium point is attained. The process of voltage build up
178
in a SEIG is similar to a shunt connected DC generator [2]. It can be assumed that point
O is the rated voltage in Fig. 7.14. The rated voltage can be generated using capacitance
C
2
at rotor speed e
r2
. To generate the same amount of voltage at lower rotor speed, e
r1
,
the capacitance increases to C
3
and the exciting current can exceed the rating of the
stator current of the induction machine. And at higher speed, e
r3
, the rated voltage can
be generated with a small capacitance value C
1
.
The magnetising inductance given in Fig. 7.6 applies only at rated frequency 50Hz. The
characteristic of magnetising inductance as a function of voltage at other frequencies is
shown in Fig. 7.15. The magnetising inductance can be expressed as a function of
magnetising current as shown in Fig. 7.16. However, since voltage is the output
parameter of interest in a SEIG, the magnetising inductance as a function of voltage
clearly indicates the generated voltage at a given operating point and also shows the
minimum voltage that can be generated without loss of self-excitation.
0 50 100 150 200 250 300 350 400
0
0.05
0.1
0.15
0.2
0.25
55Hz
45Hz
50Hz
35Hz
40Hz
V
ph
(V)
L
m
(H)
L
m
(H)
V
ph
(V)
Fig. 7.15 Variation of magnetising inductance with phase voltage at different
frequencies
179
From Fig. 7.15 the minimum stable generated voltage operating at low frequency,
corresponding to low speed (large capacitance value), is less than the minimum stable
generated voltage operating at high frequency corresponding to high speed (small
capacitance value). For example the minimum stable generated voltage at 35Hz is lass
than that of 50Hz.
In the simulation, the polynomial equations for the curves in Fig. 7.15, derived from the
polynomial curve for 50Hz, can be used. The current in a capacitor is calculated from
the voltage across the capacitor by applying differentiation and differentiation can be
represented by a high pass filter. Hence the rms current might need a low pass filter to
avoid unnecessary noise if a polynomial curve fit for L
m
as a function of the
magnetising current is used.
0 1 2 3 4 5 6 7 8 9 10
0
0.05
0.1
0.15
0.2
0.25
0.3
L
m
(H)
I
m
(A)
L
m
(H)
Magnetising current I
m
(A)
Fig. 7.16 Variation of magnetising inductance with magnetising current
The components of the magnetising current I
m
are calculated as
i
mq
= i
qs
+ i
qr
(7.50)
180
i
md
= i
ds
+ i
dr
(7.51)
and the rms value of the magnetising current I
m
is
2 2
2
mq md
m
i i
I
+
= (7.52)
In Fig. 7.16 the magnetising inductance as a function of rms magnetising current I
m
is
represented by two polynomial curve fits.
For I
m
< 1.157A
4 3 2
0 063 0 14 0 017 0 125 0 23
m m m m m
L . I . I . I . I . = + + + (7.53a)
and for I
m
> 1.157A
6 4 4 3 3 2
3 98 10 2 4 10 5 48 10 0 0605 0 3552
m m m m m
L . I . I . I . I .

= + + (7.53b)
In order to have a constant speed before and after self-excitation of the induction
generator the rotor was driven at constant speed using a DC motor with speed regulator
as shown in Fig. 7.17.
PI PI
Controlled
3 phase
rectifier
3 phase
supply
K
I
K
S
DC motor
Reference
speed
Speed
controller
Current
controller
Fig. 7.17 DC motor speed regulator
When the induction generator is driven using the DC motor with speed regulator, as
illustrated in Fig. 7.17, it is easier to compare the experimental results with the results
obtained from the simulation of the dynamic model. SIMNON [3] simulation software
was used to predict the generated voltage of a three-phase SEIG rotating at a given
speed with appropriate capacitors connected at the stator terminals.
181
With zero initial charge in the capacitors the experimental dynamic self-excitation
process with per-phase capacitance of 60F is given in Fig. 7.18. Since the initial
charge in the capacitors is zero, the initial condition for self-excitation is coming from
the remnant or residual flux in the iron core. Fig. 7.19 is the dynamic simulated result
with the same initial conditions, capacitance values and speed as the experimental
results shown in Fig. 7.18. In the experimental result of Fig. 7.18b there is a small
transient dip in speed during the self-excitation process because of the associated
electrical power loss in the induction generator. However this small dip in speed is
compensated for by the speed regulator.
0 0.5 1 1.5 2 2.5 3 3.5
0
2
4
I
s

(
A
)
time(sec)
(c)
0 0.5 1 1.5 2 2.5 3 3.5
-400
-200
0
200
400
V
a

(
V
)
0 0.5 1 1.5 2 2.5 3 3.5
1400
1450
1500
S
p
e
e
d

(
r
p
m
)
RMS phase
voltage
(a)
(b)
Experimental result for capacitance 60micrF
Fig. 7.18 Measured self-excitation at C=60F and with regulated speed (a) generated
phase voltage (b) speed (c) stator current
182
The experimental results in Fig. 7.18 and the simulated results in Fig 7.19 are in good
agreement. The rms current and voltage are calculated at each sampling time using the
algorithm developed in Chapter 3. The rms and instantaneous voltages are captured
during the transient condition. The algorithm for calculating rms values is based upon
instantaneous measurements, and therefore is effective for both steady state and
transient conditions.
0 0.5 1 1.5 2 2.5 3 3.5
0
2
4
I
s

(
A
)
time(sec)
(c)
0 0.5 1 1.5 2 2.5 3 3.5
-400
-200
0
200
400
V
a

(
V
)
0 0.5 1 1.5 2 2.5 3 3.5
1400
1450
1500
S
p
e
e
d

(
r
p
m
)
RMS phase
voltage
(a)
(b)
Simulation result f or capacitance 60micrF
Fig. 7.19 Simulated self-excitation at C=60F and with regulated speed (a) generated
phase voltage (b) speed (c) stator current
183
Since the experimental and simulated results given in Fig. 7.18 and Fig. 7.19,
respectively, are in good agreement it is possible to predict with confidence other
parameters which are not convenient to measure in the real induction generator. The
magnetising current, magnetising inductance and the stator flux-linkage are given in
Fig. 7.20, where the speed of the induction generator and capacitance values are the
same as that of Fig. 7.19. The dynamic magnetising inductance varies with the
generated voltage and/or exciting current during the self-excitation process. The
magnetising current and stator flux linkage grow with the generated voltage.
0 0.5 1 1.5 2 2.5 3 3.5
0.1
0.2
0.3
Simulated result for capacitance 60micrF
L
m

(
H
)
0 0.5 1 1.5 2 2.5 3 3.5
0
2
4
6
I
m

(
A
)
0 0.5 1 1.5 2 2.5 3 3.5
0
0.5
1
1.5
f
l
u
x

(
w
e
b
-
t
u
r
n
)
time(sec)
Fig. 7.20 Simulated self-excitation at C=60F and with regulated speed (a) magnetising
inductance (b) rms magnetising current (c) peak stator flux-linkage
The simulation results in Fig. 7.20 indicate that as the magnetising current increases
from zero the magnetising inductance increases, reaches its peak value, then starts to
decrease and finally reaches its saturated value.
184
Fig. 7.21 shows the build-up of d-axis stator flux-linkage and q-axis stator flux-linkage
during the self-excitation process as a function of time in a three dimensional plot. The
flux-linkages continue to grow until they reach their steady state values which are the
saturated flux-linkages. The peak values of the d-axis flux-linkage and the q-axis flux
linkage are equal.
-1.5
-1
-0.5
0
0.5
1
1.5
-1
0
1
0
0.5
1
1.5
flux-d (web-turn)
flux-q (web-turn)
t
i
m
e

(
s
e
c
)
Fig. 7.21 Three dimensional d-axis flux-linkage and q-axis flux-linkage as a function of
time during self-excitation process
The results of the dynamic self-excitation process given in the previous figures are
based on remnant or residual flux in the iron core providing the initial condition
required by the self-excited induction generator. When the initial conditions for self-
excitation are satisfied the flux grows and associated with the growth of flux linkage the
generated voltage also grows.
As discussed previously the self-excitation process can be also initiated with a charged
capacitor. A charged capacitor will provide magnetising current to the induction
generator and the flux and the terminal voltage will grow. When the self-excitation
process is started from a charged capacitor there is a step voltage at the moment the
185
capacitors are connected to the terminals of the induction generator and provides a
transient exciting current. The experimental and simulated results for the self-excitation
process initiated by a charged capacitor for a capacitance of 60F and rotor speed of
1480rpm are given in Fig. 7.22 below.
0 0.5 1 1.5 2 2.5 3 3.5
-400
-200
0
200
400
V
a

(
V
)
0 0.5 1 1.5 2 2.5 3 3.5
-400
-200
0
200
400
V
a

(
V
)
(a)
(b)
For capacitance 60micrF
For capacitance of 60F and speed 1480rpm
Fig. 7.22 Self-excitation process initiated by a charged capacitor of 60F and rotor
speed of 1480rpm (a) experimental result (b) simulated result
7.8.2 The dynamic representation of a loaded SEIG
The generated voltage in a self-excited induction generator grows and reaches a steady
state value where the peak of the generated voltage remains at a constant peak value.
Once steady state is attained a load can be connected to the SEIG. In a SEIG operating
without load the stator current and the capacitor current are equal. However, in a loaded
SEIG the stator current is divided into capacitor current and load current. The equations
used to analyse the loaded SEIG are modified from the equations representing the
unloaded SEIG, discussed in the previous section, and additional equations are
developed to take into account the relationship between the stator, capacitor and load
186
currents of the loaded SEIG. With a good model and implementing the model using
simulation software the dynamic characteristics of generated voltage, stator current,
capacitor current, load current, magnetising current, power, and electromagnetic torque
can be studied with confidence.
The model for the loaded SEIG is shown in Fig. 7.23.
L
ls
R
s
i
qs
L
lr
i
qr
L
m
i
mq
R
r
qr
qs
V
cq
e
r
dr
+ -
C
R
L
i
cq
i
Lq
(a)
L
ls
R
s
i
ds
L
lr
i
dr
L
m
i
md
R
r
dr
ds
-e
r
qr
+ -
V
cd
C
R
L
i
cd
i
Ld
(b)
Fig. 7.23 d-q model of a loaded SEIG in a stationary reference frame (a) q-axis circuit
(b) d-axis circuit
The differential equations given in Equations (7.44) to (7.47) are also used for the
loaded condition of the SEIG. For a resistive load the additional equations needed are:
cq
Lq
L
V
i
R
= (7.54)
cd
Ld
L
V
i
R
= (7.55)
cq qs Lq
i i i = (7.56)
cd ds Ld
i i i = (7.57)
cq
cq
i
pV
C
= (7.58)
187
cd
cd
i
pV
C
= (7.59)
The dynamic equations developed for the loaded SEIG are simulated using SIMNON
[3] simulation software. Using this simulation algorithm it is possible to determine the
characteristic of a loaded SEIG and the dynamic change in the parameters during
loading and unloading of a SEIG. The no load condition can be represented by
introducing a very large value of load resistance.
When the induction machine operates as a motor at a constant frequency, the speed of
the air gap rotating magnetic field is fixed. When this motor is loaded the rotor speed
will be varied relative to the synchronous speed to produce an output power equivalent
to the mechanical power demand. For a SEIG, with constant rotor speed, the speed of
the rotating magnetic field lags behind the rotor speed. When the load of the SEIG is
increased the magnitude of the negative slip also increases. In this case, as the rotor
speed is the input and is constant, the increase in slip is only due to a decrease in the
speed of the rotating magnetic field. The generated frequency and voltage are
proportional to the speed of the rotating magnetic field. For the same capacitance value
a decrease in the speed of the rotating magnetic field will inevitably decrease the
generated voltage and its frequency. This explanation is illustrated in Fig. 7.24.
When the SEIG is operating at no load, point A in Fig. 7.24, the slip is almost zero and
the generated voltage is equal to its rated value, curve 1. When the SEIG is loaded the
rotor speed remains at point A. However, the synchronous speed has to decrease to
point B in order to operate at a negative slip equivalent to the electrical load demand
which is equivalent to the electromagnetic torque at point O, curve 2. The speed of the
rotating magnetic field at point B is less than that of point A. Since the generated
voltage is proportional to the speed of the rotating magnetic field a decrease in the
synchronous speed decreases the generated voltage. Hence loading will decrease the
generated voltage and frequency of the SEIG. The simulation and experimental results
given in Figs. 7.25 to 7.30 confirm this.
188
The instantaneous phase voltage is measured using a data acquisition board. The
instantaneous frequency is calculated by applying a d-q transformation to two voltage
samples as discussed in Chapter 3 or it can be calculated from flux-linkages of the
induction machine using the differentiation technique [4].
0 500 1000 1500 2000 2500 3000
-100
-80
-60
-40
-20
0
20
40
60
Speed (rpm)
T
o
r
q
u
e

(
N
m
)
B
A
1
2
O
Fig. 7.24 Relationship between rotor speed and synchronous speed in a SEIG
The steady state variation of voltage and frequency are given in Fig. 7.29 and Fig. 7.30,
respectively. The voltage and frequency are dependent on the amount of loading. In Fig.
7.30 the frequency of the generated voltage, due to remnant magnetic flux in the core,
of the free running induction generator without capacitance and at no load, is given by:
P
nP
f
60
= (7.60)
where P
p
- the number of pole pair
n - rotor speed, rpm
189
0 0.5 1 1.5 2 2.5 3 3.5
0
100
200
300
V
r
m
s

(
V
)
0 0.5 1 1.5 2 2.5 3 3.5
0
1000
2000
3000
P
o
w
e
r

(
W
)
0 0.5 1 1.5 2 2.5 3 3.5
0
2
4
I
s
-
r
m
s
(
A
)
time (sec)
(d)
(e)
(f)
0 0.5 1 1.5 2 2.5 3 3.5
-500
0
500
Experimental result f or capacitance 60micrF and RL=55ohm
V
a

(
V
)
0 0.5 1 1.5 2 2.5 3 3.5
1400
1450
1500
s
p
e
e
d
(
r
p
m
)
0 0.5 1 1.5 2 2.5 3 3.5
0
20
40
60
f
r
e
q
u
e
n
c
y
(
H
z
)
(a)
(b)
(c)
Experimental result for capacitance 60F and RL = 55O
Fig. 7.25 Experimental loading of SEIG after the voltage has developed to its steady
state value (a) phase voltage (b) speed (c) frequency (d) rms phase voltage (e) generated
power (f) rms stator current
190
0 0.5 1 1.5 2 2.5 3 3.5
0
100
200
300
V
r
m
s

(
V
)
0 0.5 1 1.5 2 2.5 3 3.5
0
1000
2000
3000
P
o
w
e
r

(
W
)
0 0.5 1 1.5 2 2.5 3 3.5
0
2
4
6
I
s
-
r
m
s
(
A
)
time (sec)
(d)
(e)
(f)
0 0.5 1 1.5 2 2.5 3 3.5
-500
0
500
Simulation result f or capacitance 60micrF and RL=55ohm
V
a

(
V
)
0 0.5 1 1.5 2 2.5 3 3.5
1400
1450
1500
s
p
e
e
d
(
r
p
m
)
0 0.5 1 1.5 2 2.5 3 3.5
0
20
40
60
f
r
e
q
u
e
n
c
y
(
H
z
)
(a)
(b)
(c)
Simulation result for capacitance 60F and RL = 55O
Fig. 7.26 Simulated loading of SEIG after the voltage has developed to its steady state
value (a) phase voltage (b) speed (c) frequency (d) rms phase voltage (e) generated
power (f) rms stator current
191
0 0.5 1 1.5 2 2.5 3 3.5
0
2
4
6
I
s

(
A
)
0 0.5 1 1.5 2 2.5 3 3.5
0
2
4
I
c

(
A
)
0 0.5 1 1.5 2 2.5 3 3.5
0
2
4
6
I
L

(
A
)
time (sec)
For 60micrF and RL=55ohm
(a)
(b)
(c)
For 60F and RL = 55O
Fig. 7.27 Simulated loading of SEIG (a) rms stator current (b) rms capacitor current (c)
rms load current
0 0.5 1 1.5 2 2.5 3 3.5
0
0.1
0.2
L
m

(
H
)
0 0.5 1 1.5 2 2.5 3 3.5
0
0.5
1
1.5
f
l
u
x

(
w
e
b
-
t
u
r
n
)
0 0.5 1 1.5 2 2.5 3 3.5
0
2
4
I
m

(
A
)
time (sec)
For 60micrF and RL=55ohm
(a)
(b)
(c)
For 60F and RL = 55O
Fig. 7.28 Simulated loading of SEIG (a) L
m
(b) peak flux-linkage (c) rms magnetising
current
192
1150 1200 1250 1300 1350 1400 1450 1500 1550
80
100
120
140
160
180
200
220
240
260
280
300
Generated phase voltage (V)
Speed (rpm)
No load
RL=139ohm
RL=70ohm
RL=55ohm
1150 1200 1250 1300 1350 1400 1450 1500 1550
80
100
120
140
160
180
200
220
240
260
280
300
Generated phase voltage (V)
Speed (rpm)
No load
RL=139ohm
RL=70ohm
RL=55ohm
speed (rpm)
Fig. 7.29 Measured variation of generated voltage with load for a 60F
capacitance
1000 1100 1200 1300 1400 1500 1600
30
35
40
45
50
55
F
r
e
q
u
e
n
c
y

(
H
z
)
speed(rpm)
f=np/60- Vgen due to remnant flux (no C)
no load (C=60microF)
RL=139ohm (C=60microF)
Fig. 7.30 Measured variation of generated frequency with load for a 60F
capacitance
193
When the capacitors are connected to the induction machine, voltages and currents
build up and power is dissipated in the machine. The induction generator has to absorb
an equivalent amount of power from the prime mover, which makes it operate at a
synchronous speed which is marginally lower than the rotor speed. When the load of the
induction generator increases, its synchronous speed keeps on decreasing to produce the
required amount of slip at each operating point.
7.9 Characteristics of wind turbine and its effect on generator output
As discussed in Chapter 2, when the angular rotor speed of the wind turbine increases
from zero the torque and power produced by the wind turbine increases from zero value,
reach their peaks and then decrease. At zero rotor speed there is a small torque which is
enough to start and rotate the wind turbine. A typical wind turbine torque-rotor speed
characteristic is given in Fig. 7.31. This characteristic is modified from the original
characteristic given in Reference [5] to have an increased torque and decreased rotor
speed by changing the gear ratio.
There are different torque-rotor speed curves representing different wind speeds. The
torque produced by the wind turbine and measured at the shaft of the induction
generator can be represented by a function dependent on the linear wind speed and
angular rotor speed at the induction generator. However to simplify the representation
and dynamics of the torque produced by the wind turbine, it is assumed that the wind
speed is constant at 9m/s during the operation of the self-excitation and the torque
produced by the wind turbine will follow a single characteristic curve as shown in Fig.
7.31 represented by the solid line. The torque-speed characteristic is approximated
using a fifth order polynomial curve fit given by:
13 5 9 4 6 3 3 2
4 123 10 1 993 10 3 428 10 2 395 10 0 54 36
t
T . n . n . n . n . n

= + + +
(7.61)
where T
t
- wind turbine torque measured at the generator shaft, Nm
n - angular rotor speed, rpm
194
0 200 400 600 800 1000 1200 1400 1600 1800
0
20
40
60
80
100
Rotor speed (rpm)
W
i
n
d

t
u
r
b
i
n
e

t
o
r
q
y
e

(
N
m
)
5m/s
6m/s
7m/s
8m/s
9m/s
12m/s
11m/s
10m/s
Wind speed
Fig. 7.31 Wind turbine output torque as a function of rotor speed
Here the aim is not to operate at maximum output power or maximise the output torque
of the wind turbine but is simply to analyse the dynamic effect of variation in load and
variation in angular rotor speed on the characteristic of the SEIG. The analysis is
simplified when the wind speed is assumed constant. At constant wind speed variation
in the load connected to the SEIG will change the electromagnetic torque developed by
the induction generator. A change in the electromagnetic torque will vary the
mechanical torque demand from the wind turbine. The torque output from the wind
turbine is adjusted by changing the rotor speed. The equation that relates the mechanical
output torque produced by the wind turbine measured at the shaft of the induction
generator and the electromagnetic torque developed by the induction generator is given
by
m
t
d
T J D T
dt
m e
e
e = + + (7.62)
where T
t
- turbine output torque measured at the generator shaft, Nm
195
T
e
- induced electromagnetic torque in the induction generator, Nm
e
m
- the angular mechanical rotor speed, rad/s
D - friction coefficient referred to the generator shaft, Nm/rad/s
J - effective inertia referred to the generator shaft, Kg-m
2
Fig. 7.32 and Fig. 7.33 show the dynamic characteristics of different parameters of the
induction generator when driven by the wind turbine given in Fig. 7.31. At t = 0 sec a
capacitor with capacitance of 60F is connected at the stator terminals of the induction
generator without any load, i.e. R
L
= , and voltage is generated because of the
available rotor speed at t = 0. The generated voltage V
ph
, shown in Fig. 7.32d, is
expressed as an instantaneous value and rms value. At t = 2sec a 55O load resistor is
connected. The generated voltage, frequency, capacitor current (i
c
) and stator current
(i
s
) decrease. However, the load current (i
L
) and induced electromagnetic torque
increase.
At t = 6sec the capacitance is increased to 201F to compensate for the voltage drop.
The voltage rises to its no load value but the frequency of the generated voltage
decreases further. All currents, torque and output power increase.
At t = 10sec the resistance value is increased to 90O, which demands less power, and
the generated voltage rises to a value higher than the no load voltage. The stator current,
capacitor current, frequency, and speed increase. However the load current, output
power and torque decrease.
At t = 14sec the capacitance value is decreased to 126F to reduce the generated
voltage to its no load value. All currents, electromagnetic torque and output power
decrease. However the speed and frequency increase. As can be observed from the
results the variation of the frequency follows the variation of the rotor speed, which is
dependent on the characteristics of the wind turbine.
196
0 2 4 6 8 10 12 14 16 18 20
0
20
40
60
(
H
z
)
(e)
time (sec)
f
0 2 4 6 8 10 12 14 16 18 20
0
500
1000
1500
0 2 4 6 8 10 12 14 16 18 20
-500
-250
0
250
500
(
V
)
(c)
(d)

(
r
p
m
)
Vrms
S
p
e
e
d
V
p
h
0 2 4 6 8 10 12 14 16 18 20
0
20
40
60
80
100
0 2 4 6 8 10 12 14 16 18 20
0
50
100
150
200
250
(
m
i
c
r
o

F
)
(a)
(b)
(
o
h
m
)
R
L
C
Fig. 7.32 Simulated results for wind turbine with variable rotor speed (a) load resistance
(b) capacitance (c) rotor speed (d) phase voltage (e) frequency as a function of time
197
0 2 4 6 8 10 12 14 16 18 20
0
2000
4000
6000
(
W
)
(e)
time (sec)
P
o
w
e
r
0 2 4 6 8 10 12 14 16 18 20
0
2
4
6
0 2 4 6 8 10 12 14 16 18 20
0
20
40
60
(c)
(d)

(
A
)
i
L
T
e

(
N
m
)
0 2 4 6 8 10 12 14 16 18 20
0
5
10
15
0 2 4 6 8 10 12 14 16 18 20
0
5
10
15
(a)
(b)

(
A
)
i
S

(
A
)
i
C
Fig. 7.33 Simulated results for wind turbine with variable rotor speed (a) rms stator
current (b) rms capacitor current (c) rms load current (d) electromagnetic torque (e)
output power as a function of time
198
At a given speed, adjusting the excitation capacitance might compensate the variation in
voltage caused by loading. However the frequency is dependent on the rotor speed and
amount of connected load. Adjusting the capacitance does not have direct effect on the
frequency.
7.10 Effect of rotor parameters variation on self-excitation
The hypothetical three-phase induction machine discussed in Section 6.3.2 is used for
the purpose of analysis and simulation of a SEIG with rotor parameter variations. This
hypothetical induction machine is typical of a double-cage or deep-bar rotor where rotor
parameter variation is significant. The values of the parameters were chosen to have
similarity with a real induction machine in the laboratory with specifications 415V, 4-
pole, 50Hz.
The hypothetical three-phase induction machine that was selected for this simulation
has parameters L
1s
= 6mH, R
s
= 1.5O and L
m
= 318mH. The variation of magnetising
inductance with voltage at rated frequency, which was derived from Equation 7.35, is
given by
11 4 8 3 5 2 3
1.757( 1.62 10 2.67 10 1.381 10 1.76 10 0.23)
m ph ph ph
L V V V

= + + +
ph
V
(7.63)
where V
ph
is the voltage across the magnetising inductance.
The multiplying factor 1.757 was chosen from the ratio of that for L
m
at rated values so
that the variation of magnetising inductance will be similar to the real machine. When
the magnetising inductance is expressed as a function of rms magnetising current I
m
, it
is represented by two polynomial curve fits given by:
for I
m
< 0.6345 A
4 3 2
1 06 1 34 0 0894 0 387 0 404
m m m m m
L . I . I . I . I . = + + + (7.64a)
and for I
m
> 0.6345 A
5 4 3 3 2
7 01 10 2 357 10 0 0303 0 189 0 624
m m m m m
L . I . I . I . I .

= + + (7.64b)
199
The values of rotor leakage inductance L
lr
and rotor resistance R
r
with respect to the
magnitude of slip were determined from Fig. 6.15 and are expressed as
4 3 2
0 969 4 26 4 9 1 52 1 48
r
R . s . s . s . s . = + + + (7.65)
4 3 2
0 0032 0 0144 0 0252 0 027 0 0189
lr
L . s . s . s . s . = + + + (7.66)
In the calculation, the magnitude of the slip was taken because the rotor current
frequency is the same whether the slip is positive or negative.
If the rotor parameters are considered constant then L
lr
= 6mH and R
r
= 4.6O, which are
the values at s = 1. For variable rotor parameters the parameters start close to L
lr
=
20mH and R
r
= 1.4O, which are the values at s = 0, no load operation. The variations of
load resistance and excitation capacitance at the input of the SEIG, and the rotor speed
are given in Fig. 7.34.
0 2 4 6 8 10 12 14 16
0
50
100
0 2 4 6 8 10 12 14 16
0
50
100
0 2 4 6 8 10 12 14 16
0
1000
2000
s
p
e
e
d

(
r
p
m
)
C

(
m
i
c
r
o
F
)
R
L
(
o
h
m
)
time (sec)
(a)
(b)
(c)
Fig. 7.34 Input to the hypothetical SEIG (a) capacitance, F (b) load resistance, O
(c) speed, rpm
200
The capacitance is 40F for time t less than 10sec and then increased to 68F to make
the generated voltage equal to the no load voltage. The SEIG is operating at no load
(infinity load resistance) for time t less than 5sec and then loaded with load resistance of
45O. The rotor speed is fixed at 1500rpm during the operation.
The results for the hypothetical SEIG with both constant and variable rotor parameters
are given in Fig. 7.35 and Fig. 7.36 for the variation of excitation capacitance, load
resistance and rotor speed shown in Fig. 7.34. The simulation results for the variable
rotor parameters are indicated with black lines and for the constant rotor parameters
with grey lines.
It has been previously established that the minimum capacitance and minimum rotor
speed required for the initiation of self-excitation are not dependent on the variation of
rotor parameters. The onset of self-excitation is mainly dependent on the magnitude of
magnetising inductance.
The total rotor inductance is the sum of the magnetising inductance plus the leakage
inductance. The leakage inductance is very small compared to the magnetising
inductance and so any change in leakage inductance will have a negligible effect on the
rotor circuit time constant. However, the rotor circuit time constant is inversely
proportional to the rotor resistance. And as discussed in Chapter 6, the rotor resistance
of an induction machine with variable rotor parameters increases with an increase in the
magnitude of slip. Consequently, variation in the rotor resistance causes the rotor circuit
time constant to change. For an induction machine with constant rotor parameters,
however, the value of the rotor resistance which is obtained at slip equal to one (locked
rotor test), remains constant and hence the rotor circuit time constant does not change.
During the initiation of self-excitation the induction generator operates at slip close to
zero. An induction machine with variable rotor parameters has minimum rotor
resistance at slip close to zero. Hence if the rotor resistance of an induction machine
with variable rotor parameters is calculated only from the locked rotor test then this
rotor resistance will be higher than the value of rotor resistance that occurs at low slips
(close to zero).
201
The simulation of an induction machine with variable rotor parameters is contrasted
with the same machine considering only constant rotor parameters and is shown in Fig.
7.35 and 7.36. The results related to the variable rotor parameters are shown with a dark
line and indicated with variable. In Fig. 7.35a, the voltage build up of a SEIG when
rotor parameters are assumed constant is faster than the voltage build up when rotor
parameter variations are taken into account. When the SEIG with variable rotor
parameters is loaded at t = 5sec the dynamic performance of the generated voltage is
slow (large time constant) because of the small rotor resistance close to slip equal to
zero. However, if the rotor parameter variations are neglected, that is, the rotor
parameters are taken as constant, then the dynamic performance of the generated
voltage is fast (small time constant). This is because of the large rotor resistance close to
slip equal to one. When the excitation capacitance is increased at t = 10sec the variation
of rotor parameters affects the dynamics of the generated voltage as explained for the
case during loading conditions.
The dynamic performances of stator current (Fig. 7.35b), capacitor current (Fig. 7.35c)
and magnetising current (Fig. 7.35e) are affected by variation in rotor parameters during
the process of self-excitation, loading conditions and change in excitation capacitance.
The dynamics of the load current (Fig. 7.35d) during loading and changes in excitation
capacitance is similar to the dynamics of the stator and capacitor currents. The dynamic
variation of magnetising inductance is shown in Fig. 7.35f.
202
0 2 4 6 8 10 12 14 16
0
2
4
6
0 2 4 6 8 10 12 14 16
0
2
4
0 2 4 6 8 10 12 14 16
0
0.2
0.4
0.6
(d)
(e)
(f)
I
m

(
A
)
I
L

(
A
)
L
m

(
H
)
variable
time (sec)
0 2 4 6 8 10 12 14 16
0
200
400
0 2 4 6 8 10 12 14 16
0
5
10
0 2 4 6 8 10 12 14 16
0
2
4
6
(a)
(b)
(c)
I
s

(
A
)
V
r
m
s

(
V
)
I
c

(
A
)
variable
Fig. 7.35 Comparison of constant and variable rotor parameters performance in SEIG
(a) rms phase voltage (b) rms stator current (c) rms capacitor current (d) rms load
current (e) rms magnetising current (f) magnetising inductance
203
0 2 4 6 8 10 12 14 16
0
2
4
0 2 4 6 8 10 12 14 16
0
2
4
6
0 2 4 6 8 10 12 14 16
0
25
50
75
100
(d)
(e)
(f)
P
m
e
c
h

(
K
W
)
P
e
l
e
c

(
K
W
)
E
f
f
i
c
i
e
n
c
y

%
variable
time (sec)
0 2 4 6 8 10 12 14 16
0
20
40
60
0 2 4 6 8 10 12 14 16
-0.2
-0.1
0
0.1
0 2 4 6 8 10 12 14 16
0
100
200
300
(a)
(b)
(c)
s
l
i
p

s
f
r
e
q
u
e
n
c
y

(
H
z
)
T
o
r
q
u
e

(
N
m
)
variable
Fig. 7.36 Comparison of constant and variable rotor parameters performance in SEIG
(a) generated frequency (b) slip (c) electromagnetic torque (d) electrical generated
output power (e) mechanical input power (f) efficiency
204
The operating point of an induction generator is close to synchronous speed (slip equal
to zero). Taking the rotor parameter variations into consideration the rotor resistance
will have its minimum value in its operating region. However, when the rotor parameter
variations are ignored the value of rotor resistance will be high, because the rotor
resistance corresponds to a slip equal to one. As shown in Fig. 7.36 the effect of using
constant rotor parameters for an induction generator exhibiting variable rotor
parameters is the same as if external rotor resistance is added to the rotor circuit. When
constant rotor parameters are used, the frequency (Fig. 7.36a) and efficiency (Fig.
7.36f) drop. Also, the magnitude of the operating slip (Fig. 7.36b) increases. In general,
using constant rotor parameters means a larger rotor resistance than if the variable rotor
resistance is used, resulting in the synchronous speed and efficiency both decreasing. A
decrease in frequency reduces the generated voltage for the same magnitude of
magnetising current. For electromagnetic torque, generated power and required
mechanical power, they are dependent on the generated voltage and load current;
however their dynamic performance is slower when modelled with variable rotor
parameters.
It is noted that when the generated output power, calculated using constant rotor
parameters, is almost the same as that when calculated using variable rotor parameters,
the former requires greater mechanical power than the latter.
7.11 Summary
The use of the variation in magnetising inductance with voltage leads to an accurate
prediction of whether or not self-excitation will occur in a SEIG for various capacitance
values and speeds in both the loaded and unloaded cases. The characteristics of
magnetising inductance, L
m
, with respect to the rms induced stator voltage or
magnetising current determines the regions of stable operation as well as the minimum
generated voltage without loss of self-excitation. Once self-excitation has been initiated
and a steady state condition has been reached, the speed at which self-excitation ceases
is always lower than the speed to initiate self-excitation.
205
At a particular speed the capacitance required for self-excitation, when the machine
operates at no load, is less than the capacitance required for self-excitation when it is
loaded.
When an induction machine operates as a motor the speed of the rotating air gap
magnetic field is totally dependent on the excitation frequency. In synchronous
generators the frequency of the generated voltage is dependent only on the speed of the
prime mover, for a given number of poles. However, in the SEIG the frequency of the
generated voltage depends on the speed of the prime mover as well as the condition of
the load. Keeping the speed of the prime mover constant with increased load causes the
magnitude of generated voltage and frequency of an isolated SEIG to decrease. This is
due to a drop in the speed of the rotating magnetic field. When the speed of the prime
mover drops with load, then the decrease in voltage and frequency will be greater than
for the case where the speed is held constant.
unloaded conditions while the speed was varied or kept constant. Using the simulation
algorithm more results which are not accessible in an experimental setup have been
predicted. Increasing the capacitance value can compensate for the voltage drop due to
loading but the drop in frequency can be compensated only by increasing the speed of
the rotor. The variation of magnetising inductance follows the variation in terminal
voltage or magnetising current. Increasing the capacitance can compensate the
generated voltage, however it increases stator current. Hence care should be taken not to
exceed the stator rated current.
The dynamic calculated performance comparison between the machine with constant
and variable rotor parameters has been simulated and discussed considering a
hypothetical induction machine that represents a typical double-cage or deep-bar rotor
induction machine where rotor parameter variation is significant.
206
All of these characteristics are the basic tools required to develop a control system,
using power electronics, which will regulate the generated voltage and frequency for a
SEIG over a wide variation in speed.
7.12 References
[1] G. Strang, Linear Algebra and its Application, Harcourt Brace Jovanovich Publishers, San
Diego, 1988.
[2] A. E. Fitzgerald, C. Kingsley and S. D. Umans, Electric machinery, McGraw-Hill, London,
1992.
[3] SIMNON-Simulation of nonlinear systems, SSPA Systems, Gothenburg, Sweden, 1993.
[4] B. K. Bose, Modern Power Electronics and AC Drives, Printice-Hall, New Jersey, 2002.
[5] Z. Zhang, C. Watthanasarn and W. Shepherd, Application of a matrix converter for the power
control of a variable-speed wind-turbine driving a doubly-fed induction generator, in Proc. 1997
IEEE IECON97 Conference, pp. 906-911.
207
208
CHAPTER 8
MODELLNG OF AN ISOLATED SELF-EXCITED
INDUCTION GENERATOR TAKING IRON LOSS INTO
ACCOUNT
8.1 Introduction
As discussed in Chapter 4, the iron loss in induction machines arises due to hysteresis
and eddy current losses. Iron loss or core loss is represented in the induction machine
model using R
m
, a resistance value which has the same power loss as the total iron loss
in the induction machine. For induction generators with small iron loss, i.e. large R
m
,
neglecting R
m
in the machine model will make negligible difference to predicting the
performance of the induction machine.
The steady state analysis of the SEIG including iron loss has already been reported [1,
2]. However, the steady state analysis is not able to show the dynamics of the SEIG. In
all analyses reported in the literature based on the generalized machine theory and using
the D-Q axes model of the SEIG, iron loss has been neglected. It is important to note
that for stable operation of the self-excited induction generator, the machine has to
operate in the region of magnetic saturation. Therefore, iron loss should be included in
any accurate analysis. For small induction machines, the current associated with iron
loss has almost the same per-unit value as the magnetising current [3]. Neglecting the
iron loss in this case will cause a large error in the analysis. Few works have been
reported which include iron loss in the D-Q axes model of an induction motor [4-6].
In this chapter a novel analysis (to the best knowledge of the author) for the dynamics
of the self-excited induction generator driven by a variable speed prime mover and
taking iron loss into account is given and establishes the error introduced if iron loss is
CHAPTER 8 MODELLNG OF AN ISOLATED SELF-EXCITED INDUCTION GENERATOR TAKING IRON LOSS
INTO ACCOUNT
209
neglected. Iron loss is represented as resistance R
m
in the standard D-Q axes equivalent
circuits. The D-Q model of the induction generator discussed in this chapter takes into
account the actual non-linear variation of magnetising inductance L
m
as a function of air
gap voltage or magnetising current, and R
m
as functions of air gap voltage and predicts
the dynamics of the self-excitation process in the time domain. The complete study of
the stability behaviour of the SEIG was previously not possible; however, this new
model that includes the iron loss and variation in magnetising inductance provides the
tool to do so and is presented below in a simplified way.
8.2 SEIG dynamic model including R
m
To model the SEIG effectively, the parameters of the machine should be measured
accurately. The parameters used in the SEIG can be obtained by conducting tests on the
induction generator when it is used as a motor. The traditional tests used to determine
the parameters are the open circuit (no load) test and the short circuit (locked rotor) test
as discussed in Chapter 6. The induction machine used as the SEIG in this investigation
is a three-phase wound rotor induction motor with specification: 415V, 7.8A, 3.6kW,
50Hz, and 4 poles. The parameters given in the D-Q equivalent circuit shown in Fig. 8.1
are obtained by conducting parameter determination tests on the above mentioned
induction machine. As it is a wound rotor induction machine there is no variation of
rotor parameters with speed. The parameters obtained from the test at rated values of
voltage and frequency are L
ls
= L
lr
=11.4mH, L
m
= 181mH, R
m
= 1600O, R
s
= 1.6O,
R
r
= 2.75O.
For motoring application these parameters can be used directly. However, for self-
excited induction generator application the variation of L
m
with the induced stator
voltage or magnetising current, and variation of R
m
with the induced stator voltage
should be taken into consideration to find the correct voltage build up. Using the correct
parameters the dynamic currents, output power and induced electromagnetic torque can
be predicted accurately during no load and loaded conditions.
INTO ACCOUNT
210
L
ls
R
s
i
ds
L
lr
i
dr
L
m
i
md
R
r
dr
ds
V
cd
-e
r
qr
+ -
R
m
C
R
L
i
Ld
i
Cd
S
(a)
L
ls
R
s
i
qs
L
lr
i
qr
L
m
i
mq
R
r
qr
qs
V
Cq
e
r
dr
+ -
R
m
CC
R
L
i
Lq
i
Cq
S
(b)
Fig. 8.1 No load D-Q model of a SEIG including core loss represented by R
m
(a) d-axis (b) q-axis
8.3 Characteristics of L
m
and R
m
As discussed in Chapter 7 L
m
starts from a small value then increases to reach its peak
value and finally starts to drop [7]. The characteristic of L
m
is helpful for the stability of
generated voltage and to determine the minimum generated voltage without loss of self-
excitation. The value of R
m
also exhibits variation and in general increases with
generated voltage [2]. When the per-unit value of R
m
is almost the same as X
m,
neglecting R
m
will give rise to an error in the analysis. However if the magnitude of R
m
is much larger than X
m
neglecting R
m
will not have any significant effect on the results.
The variation of magnetising inductance, L
m
, used in this investigation is the same as
the one used in Chapter 7. The machine investigated in Chapter 7 is not a good example
to illustrate the effect of iron loss because R
m
is very large with that machine and its
effect is minimal. However, this is not the case for all practical machines and an
example to illustrate this is the machine investigated by Grantham [3]. The variation in
R
m
is modeled by the following curve fit [2]
R
m
= 3V
ph
+50 (8.1)
INTO ACCOUNT
211
where V
ph
is the phase rms voltage across R
m
in parallel with L
m
.
V
ph
can be calculated by subtracting the sum of the voltage drop in the stator winding
resistance and stator leakage inductance from the stator terminal voltage of the
induction machine while the induction machine is motoring. For generator application
V
ph
is the sum of the voltage drop in the stator impedance and the voltage at the
terminals of the stator.
8.4 Analysis of SEIG including R
m
Using the D-Q equivalent circuit model in Fig. 8.1, the equations for a SEIG including
R
m
can be derived and used to simulate the machine dynamics. These equations for a
self-excited induction generator, in the stationary stator reference frame are given as:
1 0 0 0
0 1 0 0
0
0
qs cqo
s sNEW N
s sNEW N ds cdo
N r m r rNEW r rNEW qr qr
r N N r rNEW r rNEW
dr dr
i V
R pL pC pL
R pL pC pL i V
pL L R pL L i K
L pL L R pL
i K
e e
e e
( ( + + ( (
( (
( (
+ +
( (
( (
= +
( (
( ( +
( (
( (
+
( ( (

(8.2)
Z I
V
V
V
Where
m m
N
m m
R L
L =
R +L p
sNEW ls N
L =L +L
rNEW lr N
L =L +L
cqo cq
t 0
V V
=
= and
cdo cd
t 0
V V
=
= , are the initial voltages along the q-axis and d-axis
capacitors respectively.
qr r dro
K e = and
dr r qro
K e = are constants, which represent the initial induced
voltages along the d-axis and q-axis respectively, and are due to remnant or
residual d-axis magnetic flux (
dro
) and q-axis magnetic flux (
qro
) in the core.
From Equation (8.2) it is given that
0 = [Z][I]+[V
o
] (8.3)
The self-excitation currents are obtained from Equation (8.3) in the normal way, i.e.
[I] = [Z]
-1
[V
o
] (8.4)
INTO ACCOUNT
212
from which, following a similar procedure as in Section 7.3.2.1, the stator self-
excitation current along the d-axis is given by:
8 7 6 5 4 3 2
d
U
i
Ap Bp Dp Ep Fp Gp Hp Jp M
=
+ + + + + + + +
(8.5)
Here,
U is a function of the machine parameters, capacitance C, rotor speed and initial
conditions. As U is in the numerator its detail is not important at this stage. However,
the terms in the denominator contain the roots which are important in the determination
of a growing or decreasing transient solution. In the denominator A, B, D, E, F, G, H, J,
and M are functions of the machine parameters. The full expression for each of these
coefficients is included in the Appendix C.
If one of the roots of the denominator in the expression for i
d
has a positive real part
then there is self-excitation. A positive real part of the root produces a growing transient
response until saturation of the magnetising inductance is reached. Hence to find the
roots, the denominator of Equation (8.5) is set to zero as in Equation (8.6) below.
8 7 6 5 4 3 2
0 Ap Bp Dp Ep Fp Gp Hp Jp M + + + + + + + + = (8.6)
When there is self-excitation at least one of the eight roots will have a positive real part.
During the initiation of self-excitation, as the generated voltage is close to zero, the
values of R
m
and L
m
should be selected corresponding to a phase voltage close to zero.
The curve for the onset of self-excitation is given in Fig. 8.2.
As can be seen in Fig. 8.2, when the magnitude of R
m
is almost the same as X
m
, for the
same rotor speed, the SEIG requires more capacitance to have an onset of self-
excitation. Or it can be said that to have self-excitation for a given capacitance value the
model including R
m
requires a higher speed than the model with R
m
neglected. If the
magnitude of R
m
is very much larger than the magnitude of X
m
there is no difference
whether R
m
is included or neglected.
INTO ACCOUNT
213
0 50 100 150 200 250 300
0
500
1000
1500
Capacitance (micro F)
S
p
e
e
d

(
r
p
m
)
|R
m
|=|X
m
|
neglecting R
m
Fig. 8.2 Values of capacitance and speed for self-excitation
with and without R
m
at no load
8.5 Simulation of dynamic self-excitation including R
m
The matrix equation given in Equation (8.2) represents a SEIG in the D-Q axes model.
Differential equations derived from this matrix equation are used to simulate the SEIG
including R
m
during loaded and unloaded conditions.
8.5.1 Simulation of dynamic self-excitation at no load
Rearranging Equation (8.2) of the SEIG gives a 2
nd
order differential equation
represented by
2
o 1 o 1
p I A pI A I B pV B V = + + + (8.7)
Where
qs
ds
qr
dr
i
i
I
i
i
(
(
(
=
(
(
(

cq
cd
qr
dr
V
V
V
K
K
(
(
(
=
(
(
(

INTO ACCOUNT
214
s m m m
ls m ls ls
s m m m
ls m ls ls
o
m m m r
r
lr lr m lr
m m m r
r
lr lr m lr
R R R R
0 0
L L L L
R R R R
0 0
L L L L
A
R R R R
0
L L L L
R R R R
0
L L L L
e
e
( | |
+ +
( |
\ . (
(
| |
(
+ +
|
(
\ .
=
(
| |
(
+ +
|
(
\ .
(
(
| |
+ + (
|
(
\ .
m s
m ls
m s
m ls
1
r m m r m m
r
lr m lr lr m
r m m m m r
r
lr lr m m lr
r
R R
0 0 0
L L
R R
0 0 0
L L
A
R R R R R
0
L L L L L
R R R R R
0
L L L L L
e
e
e
e
(
(
(
(
(
(
=
(
| |
( +
|
(
\ .
(
| |
(
+
|
(
( \ .

m
m ls
m
m ls
o
m
m lr
m
m lr
R
0 0 0
L L
R
0 0 0
L L
B
R
0 0 0
L L
R
0 0 0
L L
(
(
(
(
(
(
=
(
(
(
(
(
(

ls
ls
1
lr
lr
1
0 0 0
L
1
0 0 0
L
B
1
0 0 0
L
1
0 0 0
L
(
(
(
(
(
(
=
(
(
(
(
(
(

1
cq qs cqo
V i dt V
C
= +
)
,
1
cd ds cdo
V i dt V
C
= +
)
INTO ACCOUNT
215
Expanding Equation (8.7) gives four sets of second order differential equations which
are given below
2
( )
cq
m s m s m ls m m m m
qs qs qs qr cq
m ls m ls ls m ls ls
pV
R R L R R L R L R R
p i i pi pi V
L L L L L L L L
+ +
= (8.8)
2
( )
m s m s m ls m m m m cd
ds ds ds dr cd
m ls m ls ls m ls ls
R R L R R L R L R R pV
p i i pi pi V
L L L L L L L L
+ +
= (8.9)
2
( )
( )
m r m m r m r m lr m m
qr qs ds qr qr
lr lr m lr m lr
qr
r m m lr m m
dr r dr qr
m lr m lr lr
R R R R L R R L R L
p i pi i i pi
L L L L L L
pK
R L L R R
i pi K
L L L L L
e
e
e
+ +
= +
+
+ + +
(8.10)
2
( )
( )
r m m r m m lr m m r
dr qs ds qr r qr dr
lr lr m lr m lr
m r m lr m m m dr
dr dr
m lr m lr lr
R R R L L R R R
p i i pi i pi i
L L L L L L
L R R L R L R pK
pi K
L L L L L
e e
e
+
= +
+ +

(8.11)
Since the SEIG is operating at no load the stator current is equal to the capacitor
current. Hence the equations relating the capacitor currents and capacitor voltages are
qs
cq
i
pV
C
= (8.12)
ds
cd
i
pV
C
= (8.13)
The simulation was carried out using SIMNON [8] to solve the 1
st
order and 2
nd
order
differential equations given in Equations (8.8) to (8.13). The dynamic simulation of the
SEIG operating at no load is given in Fig. 8.3.
The error in the steady state no load developed voltage, shown in Fig. 8.3, is very small
such that it can not be seen with the given scale. From the simulation it was discovered
that if the SEIG is operating at no load or if the magnitude of R
m
is much greater than
X
m
then there is no significant difference in the analysed results whether R
m
is included
or neglected.
INTO ACCOUNT
216
0 0.5 1 1.5 2 2.5 3 3.5 4
0
50
100
150
200
250
300
V
p
h

(
V
)
time(sec)
With Rm
Without Rm Rm
Rm
Fig. 8.3 No load RMS phase voltage during self-excitation with and without R
m
8.5.2 Dynamics of SEIG during loading
Power will flow to the load when the switch S in Fig. 8.1 is closed. The load is
connected across the capacitors and the terminal voltage is the voltage developed across
the capacitors. Equations (8.8) to (8.11) used in the no load condition will be used in the
loaded case. The equations that relate the stator current, load current, capacitor current
and terminal voltage or capacitor voltage are the same as the equations given in
Equations (7.54) to (7.59).
Since R
m
is included, the magnetising current is not the sum of the stator current and the
rotor current. Some of the current is bypassed through R
m
. Hence from the magnetising
inductance branch circuit of Fig. 8.1 the q-axis magnetising current is
( )
m
mq qs qr
m m
R
i i i
R L p
= +
+
(8.14a)
and in integral form it can be written as
INTO ACCOUNT
217
( )
m
mq qs qr mq
m
R
i i i i dt
L
= +
)
(8.14b)
For the d-axis magnetising current
( )
m
md ds dr
m m
R
i i i
R L p
= +
+
(8.15a)
and in integral form it is written as
( )
m
md ds dr md
m
R
i i i i dt
L
= +
)
(8.15b)
As discussed in the induction machine model including R
m
in Section 4.6, the induced
electromagnetic torque or the mechanical torque required to drive the SEIG is given by
[9]:
3
2
e p m r
T P I =
JJG JG
(8.16)
Substituting
m
JJG
and
r
I
JG
in Equation (8.16) and rearranging using vector manipulation
gives:
( )
3
2
m m
e p qs dr ds qr
m m
R L
T P i i i i
R L p
=
+
(8.17)
The dynamic induced electromagnetic torque with R
m
included can be expressed in the
integral form as
3
2
m
e p m qs dr ds qr e
m
R
T P R i i i i T dt
L
| |
=
|
\ .
)
(8.18)
The results given below are from the simulation of the SEIG when it is driven at
constant speed (1500rpm) and variable capacitance and resistive load are connected at
the stator terminals. The variations of capacitance and resistance are given in Fig. 8.4.
The dynamic variation of voltage current, power and torque, for variations in resistive
load and capacitance, are shown in Figs. 8.5 to. 8.8. At no load the effect of R
m
is
insignificant. However, when the induction generator is loaded, neglecting R
m
will
result in an error. When R
m
is included, which depicts the actual situation, the generated
voltage, currents, and output power, are lower than for the case when R
m
is neglected.
INTO ACCOUNT
218
However, due to additional losses the electromagnetic torque developed in the model
with R
m
included is higher than that without R
m
.
0 2 4 6 8 10 12 14 16 18 20
50
60
70
80
0 2 4 6 8 10 12 14 16 18 20
40
60
80
100
time (sec)
C

(
m
i
c
r
o

F
a
r
a
d
)
R
L

(
o
h
m
)
Fig. 8.4 Variation of connected capacitor and resistor
0 2 4 6 8 10 12 14 16 18 20
0
50
100
150
200
250
300
V
p
h

(
V
)
time (s ec)
without Rm
with Rm
Fig. 8.5 The dynamic rms generated voltage with variation of load and capacitance
INTO ACCOUNT
219
0 2 4 6 8 10 12 14 16 18 20
0
5
10
0 2 4 6 8 10 12 14 16 18 20
0
2
4
6
0 2 4 6 8 10 12 14 16 18 20
0
5
10
with Rm
without Rm
without Rm
without Rm
with Rm
with Rm
i
L
(
A
)
i
c
(
A
)
i
s
(
A
)
time (sec)
Fig. 8.6 Dynamic currents in the load, capacitor and stator
with variation in load and capacitance
0 2 4 6 8 10 12 14 16 18 20
0
500
1000
1500
2000
2500
3000
3500
P
o
w
e
r

(
W
)
without Rm
with Rm
time (sec)
Fig. 8.7 The dynamic output power with variation in load and capacitance
INTO ACCOUNT
220
0 2 4 6 8 10 12 14 16 18 20
0
5
10
15
20
25
30
T
e

(
N
m
)
without Rm
with Rm
time (sec)
Fig. 8.8 The dynamic electromagnetic torque with variation in load and capacitance
8.6 Summary
In this chapter a novel D-Q axes model that includes the iron loss equivalent resistance
R
m
in the dynamic analysis and simulation of the SEIG has been described in a simple
and understandable way. In some instances it is possible to neglect the iron loss;
however its effect has to be shown to be negligible. This paper provides the tool to
reach such a decision. When R
m
is included, which depicts the actual situation, the
generated voltage, currents, and output power, are lower than that when R
m
is neglected.
However, due to additional losses the electromagnetic torque necessary to drive the
machine is higher when R
m
is included than when it is neglected.
Additionally, the method presented here is the first to include the effects of iron loss in
the more general dynamic analysis of induction machines and their high performance
drives. It is noted that this method is easily understood, having drawn on many familiar
concepts and using the standard terminology and nomenclature of D-Q unified machine
theory.
INTO ACCOUNT
221
8.7 References
[1] N. H. Malik and S. E. Haque, Steady state analysis and performance of an
isolated self-excited induction generator, IEEE Transaction on EC, Vol. 1, No.3,
1986, pp.134-139.
[2] T. F. Chan, Analysis of Self-Excited induction generators using an iterative
method, IEEE Transaction on EC, Vol. 10, No.3, 1995, pp.502-507.
[3] C. Grantham, Determination of Induction Motor Parameter Variations From a
Variable Frequency Standstill test, Electric Machines and Power Systems,
Vol.10, No.2-3, 1985, pp.239-248.
[4] S. D. Wee M. H. Shin and D. S. Hyun, Stator-flux-oriented control of induction
motor considering iron loss, IEEE Trans. on Industrial Electronics, Vol.48, No.3,
June 2001, pp. 602-608.
[5] J. W. Choi D. W. Chung and S. K. Sul, Implementation of field oriented
induction machine considering iron losses, IEEE- APEC '96. Conference
Proceedings, 1996, pp. 375-379.
[6] E. Levi M. Sokola A. Boglietti and M. Pastorelli, Iron loss in rotor-flux-oriented
induction machines: identification, assessment of detuning, and compensation,
IEEE Trans. on Power Electronics, Vol.11, 1996, pp. 698 709.
[7] D. Seyoum C. Grantham and F. Rahman, Analysis of an isolated self-excited
induction generator driven by variable speed prime mover, Proc. AUPEC01,
2001, pp.49-54.
[8] Simnon-Simulation of nonlinear systems, SSPA Systems, Gothenburg, Sweden,
1993.
[9] B. K. Bose Modern Power Electronics and AC Drives, Prentice-Hall, NJ, 2002.
222
CHAPTER 9
INVERTER/RECTIFIER EXCITATION OF A THREE-
PHASE INDUCTION GENERATOR
9.1 Introduction
The main drawback of using induction generators excited by three AC capacitors is
their inherently poor voltage regulation and uncontrollable frequency of operation. The
output voltage of a SEIG can be controlled by introducing an appropriate voltage
regulating scheme. A number of schemes have been suggested for this purpose [1-5].
However, the variation of the frequency of the SEIG with load and speed cannot be
regulated by static means. As a result the equipment supplied by the three-phase SEIG
discussed in Chapter 7 should be frequency insensitive (e.g. heater, water pump,
lighting, battery charging etc).
The scheme based on switched capacitors [1] finds limited application because it
regulates the terminal voltage in discrete steps. A saturable reactor scheme of voltage
regulation [3, 5] involves a potentially large size and weight, due to the necessity of a
large saturating inductor. In the short/long shunt configuration [4] the series capacitor
used causes the problem of resonance while supplying power to an inductive load.
In a three phase capacitor excited induction generator the value of capacitance should
be varied so that the terminal voltage remains constant at different rotor speeds. It is
also shown that the value of capacitance is influenced by the load as well as by the load
power factor [1]. The problem is further aggravated by the uncertainty of the machine to
re-excite after a short circuit unless some charge is provided [6]. Loss of self-excitation
could be disastrous in applications like aircraft power supplies. There should be a way
to avoid this problem. An isolated induction generator with an excitation system
CHAPTER 9 INVERTER/RECTIFIER EXCITATION OF THREE-PHASE INDUCTION GENERATOR
223
provided by a single capacitor on the DC link side of the inverter can re-excite even
after a short circuit. Since a battery is required to control the switching of the IGBTs of
the inverter, the same battery can be used for the initiation of voltage build up using
vector control.
The excitation provided by a single capacitor on the DC link side of the inverter is
reported in [7-9]. In these papers the mechanism of maintaining constant output DC link
voltage for variable rotor speed is not clearly indicated. When the excitation comes
from the DC side capacitor of the converter, as covered in this chapter, then varying the
current flowing to the generator by controlling the switching of the IGBTs varies the
flux in the generator. Due to the switching of the inverter/rectifier the single DC side
capacitor acts like a three-phase capacitor. When the fundamental switching frequency
of the converter is varied the reactive capacitance of the DC side capacitor will be
varied as seen from the induction machine side. Overall the single DC side capacitor
provides all the reactive current or the VAR required by the induction generator.
In a grid connected induction generator, the grid acts as a stiff voltage source so that the
generator control structure is similar to a standard drive with sinusoidal front-end
converter, i.e. by varying the modulation index the terminal voltage at the induction
generator can be varied with the rotor speed while the DC bus is maintained at constant
voltage.
For an induction generator operating in stand-alone mode there should be a system that
regulates the output voltage. The output voltage is the DC voltage and the control
system, which is implemented using vector control, is required to keep this DC voltage
at a constant level. The frequency of the AC voltage can vary with speed but the aim is
to have constant peak voltage and as a result to have constant DC voltage. Once a
constant DC voltage is achieved a DC load can use it directly or, if required, it is a
matter of having an inverter to produce a constant voltage and frequency AC output.
The electrical and mechanical connections for an isolated induction generator driven by
a wind turbine are shown in Fig. 9.1. To simplify the diagram the control system is not
included in Fig. 9.1.
224
In Fig. 9.1 the turbine rotor speed will be varied depending on the wind speed. The
system is loaded by a DC load connected at the terminals of the DC capacitor or an AC
load can be connected via a second inverter that adjusts the frequency and peak voltage
of the generated AC power supply. Induction generators can be excited from a single
DC capacitor by using an inverter/rectifier arrangement. The voltage build up process is
started from a small voltage in a charged DC capacitor or from a battery. During the
voltage build up process the DC capacitor gets its charge from the induction generator
via the rectifier and the capacitor in turn supplies the excitation current or the reactive
VA (VAR) required by the induction generator for its operation. This is implemented
using the well established field oriented vector control technique.
C
dc
Induction
generator
Gear box
Rotor blade
Wind direction
Fig. 9.1 Electrical and mechanical connections
9.2 Vector control
Since the introduction of the basic principles of vector control in the 1970s, vector
control of induction machines has been the standard means of achieving high
performance in induction motor drives. Vector control is a mathematical control method
based on the space vector theory implemented by controlling the phase angle and
modules of the current in the synchronous reference frame [10-14]. In these papers it
has been shown that with the help of field oriented vector control, induction motors can
have performance similar to separately excited DC motors, where the torque and flux
225
are controlled independently by the naturally decoupled armature current and field
current, respectively.
The technique of vector control established to control induction motor drives will be
used in the control of an induction generator to control the generated voltage for
different loading conditions and variable rotor speeds. In a DC generator the main flux
(due to field current) and the load current can be controlled independently because they
are fixed in space. The field current is fixed, corresponding to a given speed, to generate
rated voltage. The armature current increases without affecting the field current when
the DC generator is loaded. For an increase in armature speed the flux should be
decreased by decreasing the field current to maintain the generated voltage at the rated
value. The generated voltage in the DC machine is given by [15]
a a
E k|e = (9.1)
where k dependent on the armature winding
E
a
armature generated voltage
| flux per pole
e
a
angular speed of the armature
The DC generator will continue to operate based on a control system that adjusts the
field current inversely proportional to the change in armature angular speed. The
principle of decoupling the flux producing current and power producing current in an
induction generator will be used to produce performance similar to the DC generator.
The principle of field oriented vector control will be discussed briefly before its use in
the induction generator. Based on the orientation of the flux, the three field-oriented
control schemes are, rotor flux oriented, magnetizing flux oriented and stator flux
oriented. The magnetizing flux oriented vector control will not be discussed here as its
analysis is similar to the principles discussed in rotor oriented and stator oriented vector
controls schemes.
9.2.1 Rotor flux oriented vector control
As its name states, the rotor flux oriented vector control is produced when the alignment
of the d-axis current in the synchronously rotating reference frame is along the rotor
226
flux linkage space vector. This technique makes the q-axis rotor flux linkage to be equal
to zero and the d-axis flux linkage to be equal to the total rotor flux linkage in the
induction machine.
The vector diagram for the rotor flux oriented vector control of an induction machine is
shown in Fig. 9.2. In Fig 9.2 the rotor flux linkage, which is aligned along the d-axis of
the rotating reference frame, is rotating at synchronous speed, however the rotor speed
is lagging behind the rotor flux linkage space vector when the induction machine is
motoring and it is leading when the machine is generating, as discussed in Chapter 4.
R
o
t
o
r
a
x
i
s
dr
=
r
e
r
u
r
u
e
u
sl
e
e
e
sl
e
d
s
q
s
d
e
q
r
=
0
e
q
r
=
0
e
i
qs
e
i
ds
e
qr
s
dr
s
Fig. 9.2 Vector diagram for rotor flux oriented vector control
If the d-axis rotor flux linkage,
dr
s
, and the q-axis rotor flux linkage,
qr
s
, in the
stationary reference frame, are calculated, then the modulus or magnitude of the rotor
flux linkage will be
2 2
( ) ( )
s s
r dr qr
= + (9.2)
and the space angle of
r
is given as
1
tan
s
qr
e s
dr
| |
=
|
|
\ .
(9.3)
227
If the magnitude of the rotor flux linkage,
r
, is calculated in the rotating reference
frame the space angle of
r
cannot be evaluated from components of the rotor flux
linkage. In this case the space angle u
e
is obtained from the slip angle u
sl
and rotor angle
u
r
as illustrated in Fig. 9.2, and is given by
e sl r
u u u = + (9.4)
The slip angle u
sl
is negative for induction generators.
With rotor flux oriented vector control the space vector of the rotor flux linkage is
aligned along the d-axis of the synchronously rotating reference frame so that all the
rotor flux linkage will be produced by the d-axis current in the rotating reference frame.
The space angle u
e
is used for transformation from stationary reference frame to the
rotating reference frame and vice versa. The rotor flux oriented vector control is
implemented using two schemes, direct and indirect, to obtain the magnitude and space
angle of the rotor flux linkage space vector.
9.2.1.1 Direct (feedback) flux oriented vector control
With the direct rotor flux oriented control the modulus and space angle of the rotor flux
linkage are calculated from the measured currents, voltages and/or rotor speed using
different types of flux models, or the rotor flux linkage can be directly measured using
hall-effect sensors, search coils, etc. In this investigation the direct estimation of rotor
flux linkage from the measured currents, voltages and/or rotor speed was used.
a) Rotor flux linkage calculation utilising monitored stator voltages and currents in the
stationary reference frame
Once the stator flux linkage is calculated from the measured stator voltages and currents
then using Equation (4.5) the d-axis rotor current is expressed as
s s
s ds s ds
dr
m
L i
i
L

= (9.5)
and from Equation (4.6) the q-axis rotor current is expressed as
s s
qs s qs s
qr
m
L i
i
L

= (9.6)
228
Substituting Equation (9.5) in Equation (4.9)
( )
s s s r
dr ds s ds
m
L
L i
L
o = (9.7)
where
2
m
r s
L
1
L L
o =
And Substituting Equation (9.6) in Equation (4.10)
( )
s s s r
qr qs s qs
m
L
L i
L
o = (9.8)
Here the estimation of rotor flux linkage is dependent on the parameters of the induction
machine L
s
, L
m
, L
r
, and R
s
, where the effect of stator resistance is coming from the
estimation of stator flux linkage.
b) Rotor flux linkage calculation utilising monitored stator currents and rotor speed
evaluated in the stationary reference frame
With a shorted rotor circuit Equations (4.11) and (4.12), the rotor side voltage
equations, are rewritten as
0
s
s s dr
r dr r qr
d
R i
dt
e + + = (9.9)
0
s
qr s s
r qr r dr
d
R i
dt
e + + = (9.10)
From Equation (4.9) and (4.10) the rotor currents are expressed in terms of rotor flux
linkages and stator currents as
s s
s dr m ds
dr
r
L i
i
L

= (9.11)
s s
qr m qs s
qr
r
L i
i
L

= (9.12)
The rotor flux linkages can be expressed in terms of the stator currents by substituting
Equation (9.11) in (9.9) and Equation (9.12) in (9.10) and after simplifying gives
s
s s s m dr
dr ds r qr
r r
L
i dt
T T
e
| |
=
|
\ .
)
(9.13)
229
s
qr s s s m
qr qs r dr
r r
L
i dt
T T
e
| |
= +
|
|
\ .
)
(9.14)
Where
r
r
r
L
T
R
= is the rotor circuit time constant.
From Equations (9.13) and (9.14) the modules and space angle of the rotor flux linkage
can be estimated using Equations (9.2) and (9.3), respectively. However, the accuracy
of the estimation is dependent on the rotor inductance and rotor resistance. Hence any
variation in rotor parameters will affect the estimation procedure.
c) Rotor flux linkage calculation utilising monitored stator currents and rotor speed
evaluated in the rotor flux linkage oriented rotating reference frame
Using the D-Q representation of an induction machine in the reference frame rotating at
synchronous speed, as illustrated in Fig. 4.19, the voltage equations from the rotor side
are given by
( ) 0
e
e e dr
r dr e r qr
d
R i
dt
e e + = (9.15)
( ) 0
e
qr e e
r qr e r dr
d
R i
dt
e e + + = (9.16)
and the rotor flux linkages are expressed as
e e e
dr m ds r dr
L i L i = + (9.17)
e e e
qr m qs r qr
L i L i = + (9.18)
From Equations (9.17) and (9.18) the rotor currents are given as
( )
e e
dr m ds
e
dr
r
L i
i
L

= (9.19)
( )
e e
qr m qs
e
qr
r
L i
i
L

= (9.20)
To express the Equations (9.15) and (9.16) in terms of accessible and measurable stator
currents, Equations (9.19) and (9.20) are used to give
0
e
e e e dr r m
dr r ds sl qr
r r
d R L
R i
dt L L
e + = (9.21)
230
0
e
qr e e e r m
qr r qs sl dr
r r
d
R L
R i
dt L L
e + + = (9.22)
where
sl e r
e e e = is the slip frequency as illustrated in Fig. 9.2.
With rotor flux oriented vector control, shown in Fig. 9.2, the total rotor flux is aligned
along the d-axis of the rotating reference frame so that
e
dr r
= (9.23)
and
0
e
qr
= (9.24)
Then
0
e
qr
d
dt
= (9.25)
Substituting Equations (9.23) to (9.25) in Equation (9.21) and utilising T
r
= L
r
/R
r
gives
e r
r r m ds
d
T L i
dt
+ = (9.26)
Equation (9.26) shows that the rotor flux linkage can be evaluated from the d-axis
measured stator current in the rotating reference frame. It is implemented by applying a
low pass filter to
e
ds
i , with a cut off frequency of 1/T
r
and a gain of L
m
. Since the flux
linkage is dependent only on the d-axis current and not on the q-axis current the rotor
flux oriented vector control scheme gives decoupled control when implemented with a
current regulated source.
And substituting Equations (9.23) to (9.25) in Equation (9.22) will give
e m
sl qs
r r
L
i
T
e
= (9.27)
At steady state the rotor flux linkage is constant. Hence Equation (9.26) can be rewritten
as
e
r m ds
L i = (9.28)
and
231
1
e
qs
sl e
r ds
i
T i
e = (9.29)
Here also, the accuracy of the rotor flux linkage estimation is dependent on the rotor
inductance and rotor resistance. Any variation in rotor parameters will affect the
estimated rotor flux linkage and slip speed. The rotor resistance can vary with
temperature and the rotor inductance can be varied with the magnetising saturation level
of the iron core.
9.2.1.2 Indirect (feed forward) flux oriented vector control
In this scheme the reference d-axis current is computed from reference values in a feed
forward manner not from measured values. In the implementation of the indirect rotor
flux oriented control of an induction machine the space angle of the rotor flux linkage
space vector is obtained from the rotor angle and the computed slip angle.
The derivation is the same as given in Part c of Section 9.2.1.1. The main difference is
that in the indirect flux oriented scheme the rotor flux linkage is given as a reference
value and then the reference current is evaluated from Equation (9.26) given by
( ) 1
*
r r e *
ds
m
T p
i
L
+
= (9.30)
At steady state the derivative term in Equation (9.30) is eliminated because at a given
operating point the magnitude of the rotor flux linkage is constant.
The slip speed, from Equation (9.27) will be given as
e *
qs
m
sl *
r r
i
L
T
e
= (9.31)
This scheme is a simple one; however it needs a good prior knowledge of the
characteristics of the rotor flux linkage because there is no flux controller. As in Part c
of Section 9.2.1.1 the calculation of reference current and the slip speed is dependent on
rotor parameters.
232
9.2.2 Rotor flux oriented control with voltage as the controlled variable
The implementation of the rotor flux oriented vector control method discussed in
Section 9.2.1 is based on a current regulated source with the current controllers in the
stator stationary reference frame. However it is also possible to utilise stator voltage
equations in the synchronously rotating reference frame and express the stator voltage
equations using rotor flux linkages and stator currents. In this case the current
controllers are implemented in the synchronously rotating reference frame to produce
the voltage reference signal.
Using the D-Q representation of an induction machine in the reference frame rotating at
synchronous speed illustrated in Fig. 4.19 the voltage equations from the stator side are
given by
e
e e e ds
ds s ds e qs
d
V R i
dt
e = + (9.32)
e
qs e e e
qs s qs e ds
d
V R i
dt
e = + + (9.33)
and the stator flux linkages are given by
e e e
ds s ds m dr
L i L i = + (9.34)
e e e
qs s qs m qr
L i L i = + (9.35)
Substituting Equation (9.19) into (9.34) and Equation (9.20) into (9.35) and then
simplifying gives
e e e m
ds s ds dr
r
L
L i
L
o = + (9.36)
e e e m
qs s qs qr
r
L
L i
L
o = + (9.37)
where
2
m
r s
L
1
L L
o =
Substituting Equations (9.36) and (9.37) into Equations (9.32) and (9.33) and then
rearranging to give
233
e e
e e e e e m ds m dr
ds s ds e s qs qr s
r r
L di L d
V R i L i L
L dt L dt
e
e o o = + + (9.38)
e e
qs qr e e e e e m m
qs s qs e s ds dr s
r r
di d
L L
V R i L i L
L dt L dt
e
e o o = + + + + (9.39)
Since the stator voltage equations are to be expressed in the reference frame fixed to the
rotor flux linkage space vector, the total rotor flux linkage space vector is aligned along
the d-axis of the reference frame rotating at synchronous speed e
e
. Hence the conditions
given in Equations (9.23) to (9.25) will be substituted into Equations (9.38) and (9.39)
to give
e
e e e ds m r
ds s ds e s qs s
r
di L d
V R i L i L
dt L dt
e o o = + + (9.40)
e
qs e e e e m
qs s qs e s ds r s
r
di
L
V R i L i L
L dt
e
e o o = + + + (9.41)
In Equation (9.40) the q-axis term
e
qs
i appears in the d-axis voltage expression and in
Equation (9.41) the d-axis terms
e
ds
i and
r
(equal to
e
dr
) appear in the q-axis voltage
expression. These give rise to unwanted coupling and do not produce the ideal DC
machine like characteristics, where the d-axis stator current,
e
ds
i , is the rotor flux
linkage producing component and the q-axis stator current,
e
qs
i , is the active power
producing component. The stator currents
e
ds
i and
e
qs
i can only be independently
controlled if the stator voltage Equations (9.40) and (9.41) are decoupled. Then
controlling the stator voltages of the induction machine will indirectly control the stator
currents
e
ds
i and
e
qs
i .
Taking the rotor flux linkage as constant at a given rotor angular speed the stator current
components can be independently controlled if the decoupling rotational voltage
components are arranged in such a way that they will cancel the effect of the coupling.
The decoupling technique is implemented by adding the coupling terms to the voltage
signals obtained from the output of the controllers for the d-axis and q-axis stator
currents in the synchronously rotating reference frame.
The decoupling term to be added in the d-axis voltage is
234
e
ddeco e s qs
V L i e o = (9.42)
and the decoupling term to be added in the q-axis voltage is
e e m
qdeco e s ds r
r
L
V L i
L
e
e o = + (9.43)
Using the decoupling terms in the calculation of d-axis reference voltage and q-axis
reference voltages the currents that control the flux and the active power generated by
induction generator will be controlled indirectly.
9.2.3 Stator flux oriented vector control
In the stator oriented reference frame all variables are expressed in a reference frame
aligned with the stator flux linkage space vector. In stator oriented vector control the
stator flux linkages are determined from the measured phase voltages and phase
currents in the stationary (stator) reference frame, as illustrated in Fig. 4.6, and given by
( )
s s s
ds ds s ds
v R i dt =
)
(9.44)
( )
s s s
ds ds s ds
v R i dt =
)
(9.45)
Then the modulus or magnitude of the stator flux linkage will be calculated using
2 2
( ) ( )
s s
s ds qs
= + (9.46)
and the space angle of
s
is given as
1
tan
s
qs
e s
ds
| |
=
|
|
\ .
(9.47)
In the estimation of the stator flux linkage space vector the stator resistance is the only
machine parameter involved. Hence the accuracy of the estimated stator flux linkage
space vector is dependent on the stator resistance. Because of accessibility the stator
resistance can be measured with good accuracy and it is possible to predict its variation
with temperature. In the rotor flux oriented vector control, with the involvement of rotor
resistance, the rotor resistance variation becomes dominant due to temperature and skin
effect in squirrel cage induction machines as discussed in Chapter 6. Compensation of
this parameter is difficult because of inaccessibility, but it is easier to compensate stator
resistance [16].
235
The induction generator is controlled to generate almost constant terminal voltage
which is the rated voltage. It is reasonable to consider that the stator flux linkage can be
estimated accurately. However, the small effect of variation of stator resistance creates a
small error in the estimated flux, though this is seen as a reasonable compromise.
It can be shown that using trigonometric identities and derivative rules for
tan x u =
then
1
2
1
tan
1
d d dx
x
dt dt x dt
u
= =
+
(9.48)
Using Equations (9.44), (9.45), (9.47) and (9.48) the synchronous angular frequency
can be estimated as
1
2 2
tan
s
s
qs
ds
s
ds qs
qs
e
e s
ds qs ds
d
d
d d
dt dt
dt dt
u
e

| |
= = =
|
|
+
\ .
( ) ( )
ds qs qs s qs ds ds s
e 2 2
qs ds
v i R v i R
e

=
+
(9.49)
The vector diagram for stator flux oriented vector control is illustrated in Fig. 9.3
ds
=
s
u
e
e
e
e
d
s
q
s
d
e
i
ds
e
qs
s
ds
s
q
s
=
0
e
q
s
=
0
e
i q
s
e
i q
s
e
Fig. 9.3 Vector diagram for stator flux oriented vector control
236
Since in the stator oriented reference frame all variables are expressed in a reference
frame oriented to the stator flux linkage space vector, a mathematical model needs to be
developed to find the relationship between the stator flux linkage and the stator
currents. The rotor flux linkages in the synchronously rotating reference frame can be
expressed in terms of stator flux linkages and stator currents from Equations (9.36) and
(9.37) as
( )
e e e r
dr ds s ds
m
L
L i
L
o = (9.50)
( )
e e e r
qr qs s qs
m
L
L i
L
o = (9.51)
where
2
m
r s
L
1
L L
o =
Using the induction machine model in a reference frame rotating at synchronous speed,
as shown in Fig. 4.19, the stator flux linkages are expressed as
e e e
ds s ds m ds
L i L i = + (9.52)
e e e
qs s qs m qs
L i L i = + (9.53)
From Equations (9.52) and (9.53) the rotor currents can be expressed as
e e
e ds s ds
dr
m
L i
i
L

= (9.54)
e e
qs s qs e
qr
m
L i
i
L

= (9.55)
Equations (9.50), (9.51), (9.54) and (9.55) are substituted in the voltage equations (9.15)
and (9.16) to eliminate the rotor flux linkages and rotor currents from the induction
machine equations and to express the machine equations in the reference frame fixed to
the stator flux linkage space vector. The new expressions are given by
0
e e
e e e e ds ds
r ds r r s ds r s sl r qs sl r s qs
d di
R L R L i L L L L L i
dt dt
o e e o + + = (9.56)
0
e e
qs qs e e e e
r qs r r s qs r s sl r ds sl r s ds
d di
R L R L i L L L L L i
dt dt
o e e o + + = (9.57)
237
With stator flux oriented vector control, illustrated in Fig. 9.3, the total stator flux is
aligned along the d-axis of the synchronously rotating reference frame so that
e
ds s
= (9.58)
and
0
e
qs
= (9.59)
Then
0
e
qs
d
dt
= (9.60)
Using Equations (9.58) to (9.60) and utilising T
r
= L
r
/R
r
, Equations (9.56) and (9.57)
can be rewritten as
( ) ( ) 1 1
e e
r s r s ds sl r s qs
T p T p L i T L i o e o + = + (9.61)
( ) ( ) 1
e e
sl r s s ds s r qs
T L i L T p i e o o = + (9.62)
where p is the differential operator, i.e. p = d/ dt
In rotor flux oriented vector control with current as the controlled variable, the total
rotor flux linkage, as derived in Equation (9.26), is controlled by the d-axis current in
the synchronously rotating reference frame. That is the rotor flux oriented vector control
with current as the controlled variable gives a natural decoupling. However, as shown in
Equation (9.61), with stator flux oriented vector control, which is based on an impressed
stator current controller, the total stator flux linkage,
s
, and the q-axis stator current
e
qs
i , are coupled. This means that any change in
e
qs
i without changing
e
ds
i will cause
unwanted transients to occur in the stator flux linkage.
The undesirable coupling can be eliminated by utilising a decoupling circuit in the flux
linkage control loop. The decoupling circuit is implemented at the output of the stator
flux linkage controller. The stator flux linkage controller is a PI controller and its output
is designed to be the d-axis current required to produce the reference stator flux linkage
say
1
e
ds
i given by
( )
1
e *
ds s s
i G = (9.63)
where G is the transfer function of a PI controller.
238
However, due to the coupling problem,
1
e
ds
i does not have total control over the stator
flux linkage. The reference d-axis current,
e *
ds
i , with full control of the stator flux
linkage is expressed as
1
e * e
ds ds deco
i i i = +
( )
e * *
ds s s deco
i G i = + (9.64)
Substituting Equation (9.64) in (9.61) gives
( ) ( ) ( ) ( ) 1 1 1
* e
r s r s s s r s deco sl r s qs
T p T p L G T p L i T L i o o e o + = + + + (9.65)
For decoupled control with the help of
deco
i the last two terms of Equation (9.65) should
be cancelled, i.e.
( ) 1 0
e
r s deco sl r s qs
T p L i T L i o e o + = (9.66)
so that
1
e
sl r qs
deco
r
T i
i
T p
e o
o
=
+
(9.67)
The angular slip frequency is available from Equation (9.62) given by
( )
( )
1
e
s r qs
sl
e
r s s ds
L T p i
T L i
o
e
o
+
=
(9.68)
A differentiator is very sensitive to noise and the differentiator in the numerator of
Equation (9.68) can introduce noise in the current regulator. A low pass filter with
appropriate cut off frequency can be introduced to overcome the noise problem. In the
steady state, Equations (9.67) and (9.68) can be written as
e
deco sl r qs
i T i e o = (9.69)
( )
e
s qs
sl
e
r s s ds
L i
T L i
e
o
=
(9.68)
In stator flux oriented vector control the rotor speed can be estimated from the estimated
synchronous angular frequency and the estimated angular slip frequency given by
r e sl
e e e =
239
In induction generators the value of
sl
e and
e
qs
i are negative because power is being
converted from mechanical to electrical power.
The advantage of the stator flux oriented vector control is that the stator flux linkage
vector is derived by integration of the voltage behind the stator resistance and it is
sensitive to stator resistance only, which can be compensated somewhat easily because
of accessibility. The drawback of the stator flux oriented vector control is the coupling
effect which requires the introduction of the decoupling compensation
deco
i . In the stator
flux oriented vector control the induction generator behaves like a separately excited
DC generator by aligning the total stator flux linkage of the induction generator along
the d-axis of the rotating reference frame. Hence the stator flux linkage in the induction
machine is controlled by the d-axis stator current in the rotating (excitation) reference
frame.
9.3 System description
The system description for the implementation of vector control in an isolated induction
generator is shown in Fig. 9.4. The proposed system starts its excitation process from an
external battery V
b
or it can be started from a charged capacitor. The external battery V
b
helps to charge the capacitor and also to start the build up of flux in the core. When the
generated DC voltage rises to a value higher than V
b
then the diode blocks the back flow
of current to the battery. The diode can also be replaced with a switch that is operated
by comparing the value of the battery voltage and the magnitude of the generated DC
voltage in the capacitor.
240
Induction
Generator
C
Inverter/
Rectifier
V
b
Speed sensor
Voltage
and current
sensors
PWM generator
V
abc
I
abc
V
DC
S
Load
Wind
Turbine
PC with dSPACE
DS1102
e
r
Fig. 9.4 System description
As discussed in Chapter 7 a self-excited induction generator using three AC capacitors
and without any voltage control can start its voltage build up only from a remnant
magnetic flux in the core. The voltage build up starts when the induction generator is
driven at a given speed and an appropriate capacitance is connected at its terminals.
However, for a system with a single DC capacitor, as proposed in this Section, the
voltage build up process cannot start from the remnant flux in the core due to the
following reasons:
a) There is no way of initiating the injection of exciting current into the induction
generator via the inverter using vector control.
b) The terminal voltage is not continuous because of the PWM switching in the
inverter and the current is not sinusoidal.
c) The power loss in the machine, the switching power losses in the inverter and
the characteristic of all the instantaneous losses due to PWM switching.
d) The losses in the generator increase because of the harmonics.
As a result an initial voltage is required in the DC capacitor to start the vector control
and to allow the voltage build up process. The initial voltage can be obtained from a
previously charged DC capacitor or from a battery connected to the DC capacitor. The
easiest method is to utilise the battery that supplies power to the IGBT drivers. The
minimum initial voltage required in the DC capacitor is dependent on the components
used in the inverter/rectifier, their combined forward voltage drop in the converter
241
arrangement and the parameters of the induction generator. The simulation of the
system is implemented using Matlab/Simulink. From the simulation, the voltage build
up process can start for voltage as low as 10V. However, to have a good control and
faster voltage build up process 48V is used. 48V is appropriate to implement the
proposed scheme at the lowest speed and the 48V can also be obtained from connecting
several standard commercially available 12V batteries in series.
As discussed in Chapter 7 during the voltage build up process the magnetizing
inductance, L
m
, in an induction machine varies with the air gap voltage across it. The
implementation of the variation of magnetizing inductance in the model of induction
generator depicts the actual variation in the real system.
9.4 Establishment of reference flux linkage
Wind speed can vary at any time. Variation in wind speed affects the variation in rotor
speed of the induction generator. Then variation in rotor speed affects the generated
voltage unless there is a well-designed control system to regulate the generated voltage.
In motoring applications, all control schemes use constant flux for rotor speeds lower
than the rated speed. The flux will be reduced inversely proportional to the speed when
the induction motor is operated above its rated value. That is, the motor is being
operated in the flux weakening mode. For all speeds less than the rated value the control
scheme adjusts the voltage and frequency to regulate the desired reference speed or
torque while keeping the flux at a constant value. With an induction generator the aim is
to have a constant generated voltage. Of course the frequency of the generated voltage
is dependent on the rotor speed but once it is rectified the DC voltage depends only on
the magnitude of the peak AC voltage.
In general the no load generated terminal voltage, E, in an induction generator can be
approximated by
r
E Ke = (9.69)
where K - constant
e
r
- rotor speed in rad/s
242
- flux linkage in web-turn
From Equation (9.69) the product of the rotor speed and the flux linkage should remain
constant so that the terminal generated voltage does not change. Ideally this is perfect
for all variations of rotor speed. However, in practical applications there is magnetic
flux saturation of the core of the induction generator. Hence the minimum rotor speed
corresponds to the maximum allowable flux linkage, which is the saturated flux linkage.
The maximum speed is limited by the mechanical rating of the mechanical system of the
induction generator. The minimum flux linkage corresponds to the maximum generator
rotor speed.
Once the allowable variation between maximum and minimum rotor speeds is defined it
is required to find the corresponding minimum and maximum (saturated) flux linkages
respectively. Graphically the variation of flux linkage with generator rotor speed is
shown in Fig. 9.5.
max
min
e
r_min
e
r_max
e
r
Fig. 9.5 Relationship between generator rotor speed and flux linkage
Since the maximum value of flux linkage is determined by the saturation level of the
core, the flux linkage required at any speed is calculated based on this maximum flux
linkage, the minimum set speed and the speed of operation. Hence,
r _ min
max
r
e

e
= (9.70)
243
When the rotor speed decreases to a value lower than e
r_min
, theoretically the flux
linkage should increase to a value higher that
max
. However, in an induction machine
once the saturation level is reached the controller forces more direct axis current in the
synchronously rotating reference frame to produce more flux linkage. The magnitude of
this exciting current can exceed the rated current of the machine without approaching
the required reference flux linkage. As a result the magnitude of the generated voltage
drops for speeds lower than the minimum set value which is e
r_min
.
From Fig. 9.5, the maximum speed, e
r_max
, is determined by the mechanical limitation
of the induction generator and the wind turbine. A minimum exciting d-axis current in
the synchronously rotating reference frame that will be enough to produce the minimum
flux linkage,
min
, corresponding to the maximum speed, e
r_max
, is provided by the
controller.
9.5 Details for the implementation of vector control
The detail for the generation of d-axis current and q-axis current will be discussed in
this Section. The d-axis current in the synchronously rotating reference frame is
generated from the flux linkage controller or from the reference flux linkage as in the
indirect method. The reference flux linkage is varied according to the change in rotor
speed. The q-axis current in the synchronously rotating reference frame is generated
from the DC voltage controller. When there is an increase in loading, the DC voltage
will decrease and this will be interpreted with an increase in q-axis current in the
synchronous reference frame. This q-axis reference current in the synchronously
rotating reference frame is negative in induction generators and was discussed in
Chapter 4.
In Chapter 2 it was shown that at a particular wind speed the output power and the
output torque of a wind turbine decreases at high turbine rotor speeds. Hence at no load
the induction generator operates at high speed. When the induction generator driven by
the wind turbine is loaded, the generator rotor speed decreases. The operating points of
the output power and output torque characteristic of the wind turbine follow the output
power and the induced electromagnetic torque of the induction generator. The reference
244
flux linkage of the induction generator varies inversely proportional to the variation in
angular rotor speed. For different loading conditions the generator rotor speed can
remain constant provided that there is variation in wind speed. In this case the reference
flux linkage remains constant and any change in loading will only affect the DC voltage
controller in which its output is the q-axis reference current in the rotating reference
frame.
9.5.1 Implementation of direct rotor flux oriented vector control
The detail of the direct rotor flux oriented vector control implemented in an isolated
induction generator is shown in Fig. 9.6. The flux linkage estimation can be achieved
using one of the schemes for direct rotor flux oriented vector control.
Induction
Generator
C
Inverter / Rectifier
V
b
Speed sensor
r
*
PI
Voltage
and current
sensors
-
+
PWM generator
V
abc
I
abc
-
+
V
DC
V
DC
*
r
i
ds
e*
i
qs
e*
u
e
S
Load
PI
i
a
*
i
b
*
i
c
*
i
a
i
c
i
b
Wind
Turbine
e
r Flux linkage
estimation
d
e
q
e
/ abc
-
+
Current
controllers
-
+
-
+
e
r_min
.
r_max
Numerator
Fig. 9.6 Implementation of direct rotor flux oriented vector control with current
controlled PWM VSI
In Fig. 9.6 the current controllers are implemented using hysteresis controllers or
proportional integral (PI) controllers. With the utilization of hysteresis controllers the
actual current continuously tracks the reference or command current within a given
hysteresis band. For a hysteresis band of h, if the measured current is less than the
reference current minus the hysteresis band then an output signal 1 is generated to
increase the actual current by turning on the upper power switch of the inverter. When
the measured current is less than the reference current plus hysteresis band then an
245
output signal -1 is generated to decrease the actual current by turning off the upper
power switch of the inverter. The implementation of a hysteresis controller is very
simple. However, the drawbacks are uncontrolled and high switching frequencies when
a small hysteresis band is used or high ripple when the hysteresis band is larger. The
selection of the hysteresis band needs some compromise for the magnitude of the ripple
and switching frequency. The output signals -1 and 1 are chosen instead of 0 and
1 because the range of the value -1.0 to 1.0, when using the function for PWM
generation in the DSPACE DS1102 DSP card, represents a duty cycle of 0 to 100%.
Three similar hysteresis controllers are used for the three phases.
When the current controller is implemented using a PI controller the error between the
reference current and the actual current will be fed to the PI controller. The PI controller
converts the error signal to the sinusoidal reference voltage. This sinusoidal reference
voltage signal is then fed to the converter as a PWM signal. The PI controller is simple
and the switching frequency can be predicted, however due to its limited bandwidth
there will be a phase lag and a magnitude error between the reference current and the
actual current [12]. The phase lag and magnitude error are detrimental and they will
increase with rotor speed because at high rotor speed the frequency of the current will
be high.
As shown in Fig. 9.6 and which will be also given in the implementation of other
schemes of vector control, the control mechanism involves cascaded control structures
with inner and outer closed loop systems. From control theory the cascaded control
structure can only work under the assumption that the bandwidth of the control
increases towards the inner closed loops [14, 17]. The most inner loop of the cascaded
control structure should be the fastest and the most outer loop the slowest. The inner
loop can perform well if there is enough time to execute the command given by the next
outer loop. The fact that each feedback variable can be limited by limiting the reference
signal of importance is a major advantage of cascade control.
9.5.2 Implementation of indirect rotor flux oriented vector control
The detail of the indirect rotor flux oriented vector control implemented in an isolated
induction generator is shown in Fig. 9.7. The command d-axis current in the
246
synchronously rotating reference frame,
e *
ds
i , is calculated from the magnitude of the
rotor flux linkage divided by the magnetising inductance as given in Equation (9.30).
Induction
Generator
C
Inverter/
Rectifier
V
b
Speed sensor
PWM generator
-
+
V
DC
V
DC
*
i
ds
e*
i
qs
e*
u
e
S
Load
PI
i
a
i
b
*
i
c
*
i
a
i
c
i
b
Wind
Turbine
e
r
u
e
estimation d
e
q
e
/ abc
-
+
Current controllers
-
+
-
+
1
r
m
T p
L
+
r
*
e
r_min
.
r_max
Numerator
i
qs
e*
i
ds
e*
Fig. 9.7 Implementation of indirect rotor flux oriented vector control with current
controlled PWM VSI
The previous explanation for the types of current controllers discussed in the direct
rotor flux oriented vector control also applies in this indirect rotor flux oriented vector
control scheme.
9.5.3 Implementation of rotor flux oriented vector control with voltage as a
control variable
The implementation of rotor flux oriented vector control discussed in Sections 9.5.1 and
9.5.2 utilises a current controlled PWM voltage source inverter with fast closed loop
current control of the stator currents and with sinusoidal reference currents in the
stationary reference frame [11]. However if the stator currents cannot be impressed by
fast control loops, it is necessary to utilise a scheme with stator voltage as a control
variable as discussed in Section 9.2.2. This method is more complicated than the current
controlled rotor flux oriented vector control scheme because it involves decoupling
terms and there are more PI controllers and more magnitude limiters at the output of the
PI controllers.
247
In this section the outputs of the d-axis and q-axis current controllers in the
synchronously rotating reference frame are added with their respective decoupling
signals to produce d-axis and q-axis reference voltages in the synchronously rotating
reference frame. Then using the transformation from two-axes in the synchronously
rotating reference frame to three-axes in the stationary reference frame the three-phase
stator voltages in the stationary reference frame are obtained. The overall
implementation is shown in Fig. 9.8.
Fig. 9.8 shows only the direct rotor flux oriented voltage control scheme. However the
same figure can be used to implement the indirect rotor oriented vector control provided
that the evaluation for the d-axis current is changed.
Induction
Generator
C
Inverter/
Rectifier
V
b
Speed sensor
e
r
r
*
PI
Voltage and
current sensors
PWM generator
V
abc
I
abc
V
DC
V
DC
*
r
V
ds
e*
i
qs
e*
u
e
S
Load
PI
Wind
Turbine
e
r
Flux linkage, e
e
and u
e
estimation
d
e
q
e
/ abc
e
e
e
r_min
.
r_max
Numerator
abc / d
e
q
e
i
ds
e
i
qs
e*
-
+
oL
s
V
ddeco
-
+
-
+
oL
s
L
m
/ L
r
+
+
i
ds
e*
PI
-
+
-
+
V
qs
e*
PI
+
+
u
e
v
a
*
v
b
*
v
c
*
r
Fig. 9.8 Implementation of direct rotor flux oriented vector control with stator voltage
as a control variable
9.5.4 Implementation of stator flux oriented vector control
The stator flux linkage is estimated in a direct way from the monitored stator terminal
voltages and stator currents using integration of the back emf along the d-axis and q-
axis in the stationary reference frame as discussed in Section 9.2.3. The implementation
of stator flux oriented vector control is illustrated in Fig. 9.9.
248
C
Inverter/
Rectifier
V
b
Speed sensor
e
r
s
*
PI
Voltage
and current
sensors
-
+
PWM generator V
abc I
abc
-
+
V
DC
V
DC
*
s
i
ds
e*
i
qs
e*
u
e
S
Load
e
e
+
+
decoupling
i
qs
e
i
ds
e
d
e
q
e
/ abc
i
dcou
PI
i
b
*
i
c
*
i
a
i
c
i
b
Induction
Generator
Wind
Turbine
Current controllers
e
r_min
.
r_max
Numerator
-
+
-
+
-
+
i
a
*
Flux linkage, e
e
and
u
e
estimation
Fig. 9.9 Implementation of stator flux oriented vector control with current controlled
PWM VSI
The current controllers in Fig. 9.9 can be hysteresis or PI controllers. The detail in using
these two types of controllers is discussed in Section 9.5.1.
The stator flux oriented vector control can be implemented with stator voltage as a
control variable; however it requires additional decoupling in the stator voltage
equations expressed in the synchronously rotating reference frame and is not studied in
this work.
9.6 Results
The simulation for rotor oriented vector control and stator oriented vector control are
implemented using MATLAB/SIMULINK. The features in the Power Systems Blockset
are used to model an inverter, rectifier and all circuit components. The induction
machine model in the Power Systems Blockset is modified to include speed as an input
and to update the variation of magnetising inductance as the voltage builds up during
self-excitation.
249
The results obtained in using direct rotor oriented vector control and the stator oriented
vector control are similar. It is simply a matter of tuning the PI controllers in the DC
voltage controller and flux linkage controller.
The dynamics of the DC voltage build up process, at a rotor speed of 1500rpm for a
capacitance value of 1000F and 1500F, is shown in Fig. 9.10. When the capacitance
is large it takes a longer time to reach its steady state value. If the capacitance is too
small there will not be enough exciting current and as a result there will not be any
voltage build up.
In Fig. 9.11 the frequency of the generated voltage is estimated using Equation (9.49).
For a rotor speed of 314rad/s the AC voltage build up process starts with a low
frequency and then rises until it reaches its steady state value of 311 rad/sec. The small
slip is required to overcome all power loses in the induction machine and in the inverter
that supplies the exciting current and the load current.
From Fig. 9.12 it can be observed that the value of the flux linkage in the machine
varies inversely proportional to the rotor speed of the induction generator. Fig. 9.13
shows the generated voltage.
The loading for the induction generator is shown in Fig. 9.14. The no load resistance is
simulated by a large resistance. The load resistance is decreased from its large value at
no load to an effective resistance of about 120O and the rotor speed is varied from
1500rpm (157rad/sec) to about 2400rpm (250rad/sec). In response to these variations in
load and rotor speed the variations in DC voltage and induction generator flux linkage
are shown in Fig. 9.14c and Fig. 9.14d. Fig. 9.14e shows the variation in d-axis stator
current in the rotating reference frame and Fig. 9.14f shows the variation in q-axis stator
current in the rotating reference frame.
DC current in the load output electrical power, operating slip of the induction generator
and the induced electromagnetic torque in the induction generator are also given in in
Fig. 9.14
250
0 0.5 1 1.5
0
100
200
300
400
500
600
700
time (sec)
V
d
c

(
V
)
DC voltage @ n=1500rpm
C=1000micr F
C=2000micr F
Fig. 9.10 Generated DC voltage for different capacitance value
0 0.5 1 1.5
50
100
150
200
250
300
350
time (sec)
w
e

(
r
a
d
/
s
)
w
r
C=1000 micro F
C=2000 micro F
estimated excitation freqency w
e
@ w
r
= 314rad/s(electrical)
Fig. 9.11 Rotor speed and angular frequency of the generated voltage for different
capacitance value
251
0 0.5 1 1.5
0
0.2
0.4
0.6
0.8
1
1.2
1.4
time (sec)
F
l
u
x

l
i
n
k
a
g
e

(
w
e
b
-
t
u
r
n
)
Estimated Fluxlinkage for C=1000micro F
Rotor speed=1500rpm
Rotor speed=1900rpm
Fig. 9.12 Flux linkage at different rotor speeds of the induction generator for 1000F
0 0.5 1 1.5
-800
-600
-400
-200
0
200
400
600
800
time (sec)
V
a
b

(
V
)
Generated voltage V
ab
@ n=1500rpm and C=1000micr F
Fig. 9.13 Generated line to line voltage at the terminals of the induction generator
252
0 2 4 6 8 10 12 14
0
200
400
600
800
0 2 4 6 8 10 12 14
0
0.5
1
1.5
F
l
u
x

(
w
e
b
-
t
u
r
n
)
V
d
c

(
V
)
(c)
(d) time (sec)
0 2 4 6 8 10 12 14
0
50
100
150
0 2 4 6 8 10 12 14
0
1000
2000
3000
S
p
e
e
d

(
r
p
m
)
R
L

(
o
h
m
)
(a)
(b)
253
0 2 4 6 8 10 12 14
0
2
4
6
8
0 2 4 6 8 10 12 14
-10
-8
-6
-4
-2
0
I
q
e

(
A
)
I
d
e

(
A
)
(e)
(f) time (sec)
0 2 4 6 8 10 12 14
0
2
4
6
0 2 4 6 8 10 12 14
0
1
2
3
4
P
o
w
e
r

(
K
W
)
I
d
c

(
A
)
(g)
(h) time (sec)
254
0 2 4 6 8 10 12 14
-0.06
-0.03
0 2 4 6 8 10 12 14
-30
-20
-10
0
T
o
r
q
u
e

(
N
m
)
S
l
i
p
(i)
(j) time (sec)
Fig. 9.14 Loading of the induction generator (a) R
L
(b) rotor speed (c) V
DC
(d) flux
linkage (e)
e
ds
i (f)
e
qs
i (g) Idc (h) Output power (i) Slip (j) Electromagnetic torque
In Fig 9.14 when the speed increases the flux linkage decreases and the d-axis current in
the rotating reference frame decreases also. This d-axis current is not affected by the
load. Loading affects the magnitude of q-axis current in the rotating reference frame. At
no load there is a small slip and induced electromagnetic torque to overcome the power
loses in the induction generator and in the converter.
The vector control can be implemented in real time using commercially available DSPs,
microcontrollers and PLCs. The versatile commercially available DSP from Texas
Instruments is TMS320LF2407 and an advanced one is the TMS320LF2812. This
project is, however, not concerned with the implementation of the proposed techniques
in appropriate real time hardware.
9.7 Summary
The voltage build up process of an induction generator with a single capacitor on the
DC side of the inverter using vector control is discussed. Since the induction generator
255
operates at a frequency away from the DC frequency, any integration offset error is
easily removed by offset adjustment, as discussed in Chapter 10, when stator flux
linkage estimation is used.
With stator flux oriented vector control the total flux linkage is aligned to the d-axis of
the stator flux linkage and with rotor flux oriented vector control the total flux linkage is
aligned to the d-axis of the rotor flux linkage in the excitation reference frame. A
decoupling signal is generated to cancel the effect of the q-axis current on the d-axis
flux for the stator oriented vector control. The main advantage of stator flux oriented
vector control is that the error in the estimated flux depends only on the stator
resistance. Unlike the rotor resistance the variation of stator resistance depends mainly
on temperature. If the variation of stator resistance causes a significant error in the
estimated flux then a compensation block can be added in the model.
The advantage of rotor oriented vector control is that there is no need to generate a
decoupling signal because the flux linkage is dependent only on the d-axis current and
not on the q-axis current. The rotor flux oriented vector control scheme gives a
decoupled control when implemented with a current regulated source.
When the load is varied
e
qs
i varies to provide the current demand of the load. When the
rotor speed is varied
e
ds
i is varied to provide the required amount of flux linkage
corresponding to the rotor speed in a similar way as in flux weakening operation of
induction motors. In this way the DC voltage will remain constant as the speed of the
wind turbine varies.
In a similar way the different flux oriented vector control schemes discussed here can be
used in an automotive combined starter/generator system which is an area likely to be
very typical in the future due to the expected change from 12V to 36V electrical
systems.
256
9.8 References
[1] N. H. Malik and A. H. Al-Bahrani, Influence of the terminal capacitor on the performance
characteristics of a self-excited induction generator , IEE Proc C., Vol. 137, No. 2, March 1990,
pp. 168-173.
[2] V. N. Nandakumar, K. Yadukumar, T. Sureshkumar, S. Ragupathi and R. K. Hegde, A wind
driven self-excited induction generator with terminal voltage controller and protection circuits,
IEEE Power Conversion Conference, 1993, pp. 484-489.
[3] H. C. Rai and A. K. Tandan, Voltage regulation of self-excited induction generator using passive
elements, 6th International Conf. on Electrical Machines and Drives, 1993, pp. 240-245.
[4] S. S. Murthy, C. Parabhu, A. K. Tandon and M. O. Vaishya, Analysis of series compensated self-
excited induction generators for autonomous power generation, IEEE Conference on Power
Electronics, Drives and Energy Systems for Industrial Growth, 1996, pp.687-693.
[5] S. M. Alghuwainew, Steady-state analysis of a self-excited induction generator self-regulated by
shunt saturable reactor, IEEE International Conf. on Electrical Machines and Drives, 1997,
pp.101-103.
[6] L. Shridhar, B. Singh and C. C. Jha, Transient performance of the self regulated short shunt self
excited induction generator, IEEE Transactions on Energy Conversion, Vol.10, No.2, June 1995,
pp. 261-267.
[7] S.N., Bhadra, K.V., Ratnam, and A, Manjunath, Study of Voltage Build up in a Self-Excited,
Variable Speed Induction Generator/ Static Inverter System with D.C. Side Capacitor,
International Conference on Power Electronics, Drives and Energy System, Vol. 2, 1996, pp. 964-
970.
[8] M. S., Miranda, R.O, Lyra, and S.R., Silva, An Alternative Isolated Wind Electric Pumping
System Using Induction Machines, IEEE Transactions on Energy Conversion, Vol. 14, No. 4,
December 1999, pp. 1611-1616.
[9] R., Cardenas, R., Pena, G., Asher, and , J. Clare., Control Strategies for Enhanced Power
Smoothing in Wind Energy Systems Using a Flywheel Drive by a Vector-Controlled Induction
Machine, IEEE Transaction on Industrial Electronics, Vol.48, No. 3, June 2001, pp. 625 635.
[10] F. Blaschke, The principle of field orientation as applied to the new Transvektor closed loop
control system for rotating field machines, Siemens Review, Vol. 34, 1972, pp. 217 220.
[11] Peter Vas, Sensorless Vector and Direct Torque Control, Oxsford University Press, New York,
1998.
[12] B. K. Bose, Modern Power Electronics and AC Drives, Prentice-Hall, NJ, 2002.
[13] T. Matsuo, V. Blasko, J. C. Moreira and T. A. Lipo, Field oriented control of induction machines
employing rotor end ring current detection, IEEE Trans. on Power Electronics, Vol. 9, No. 6,
November 1994.
[14] Werner Leonhard, Control of Electrical Drives, Springer-Verlag Berlin Heidelberg, New York,
1996.
257
[15] G. R. Slemon, Electric Machines and Drives, Addison-Wesley Publishing Company, New York,
1992.
[16] E. D. Mitronikas, A. N. Safacas and E. C. Tatakis, A new stator resistance tuning method for
stator-flux-oriented vector-controlled induction motor drive, IEEE Transactions on Industrial
Electronics, Vol. 48 Issue: 6, Dec. 2001, pp. 1148 1157.
[17] W. S. Levine, The Control Handbook, CRC press with IEEE press, New York 1996.
258
CHAPTER 10
FLUX LINKAGE ESTIMATION AND COMPENSATION
IN INDUCTION MACHINES
10.1 Introduction
Flux information is needed in induction machine control for the purpose of synchronous
angle and synchronous speed estimation, flux regulation and torque regulation.
Accurate flux estimation is very crucial in the control of induction motor drives and
induction generators if vector control or direct torque control (DTC) is used.
The stator flux is calculated by integrating the back electromotive force (emf), which is
the terminal voltage minus the voltage drop in the stator resistance. This method is
preferred because it requires knowledge of only one parameter of the induction
machine, namely the stator resistance. The stator resistance can be easily obtained from
measurement. If there is variation of stator resistance due to change in temperature it
can be readily compensated [1]. The back emf is calculated from the stator terminal
phase voltage and from stator phase current. When a pure integrator is implemented in
discrete form such as that used when controlling an induction machine from a DSP
system, an error can arise. This error comprises a drift produced by the discrete
integrator and also a drift produced by the offset error in the back emf. The offset error
in the back emf is due to the use of analog components in the sensor and amplifier
circuits for the measurement of voltages and currents. A small DC offset in the
measured signal, no mater how small it is, can drive the pure integrator into saturation.
The integration error associated with the implementation of the integrator is constant
and appears as an offset in the initial integrated value. From the signal at the input of the
integrator it is not easy to know whether the integrated signal will have an offset or not.
259
The constant offset available at the output of the integrator can be explained as follows.
Generally, when a sine signal (sinet ) is applied to the input of an integrator with an
integrator gain of e then a negative cosine (-coset ) signal is obtained at the output of
the integrator. In a discrete integrator this is only true when the input sine wave is
applied at its positive peak, sin(et+t/2), or negative peak, sin(et -t/2). If the sine signal
is applied at any other initial condition, a constant DC offset will appear at the output of
the integrator. This DC offset is an undesirable error and it distorts the output of the
integrated signal. When a sine input signal (sinet) is applied to the integrator, the
maximum constant DC offset that can occur is equal to the peak value of the sine signal.
Attempts have been made to modify the pure integrator by implementing it using a low
pass filter [2]. A low pass filter will produce errors in magnitude and phase angle,
especially if the excitation frequency is lower than the cut off frequency of the low pass
filter. It has also been reported that the pure integrator can be replaced by a
programmable low pass filter which is implemented by tuning the angle and changing
the magnitude of the output vector of the low pass filter according to the calculated
error [2-7]. In all these applications the cut off frequency of the low pass filter is
decided from the estimated excitation frequency. However the accuracy of the
calculated excitation frequency depends on the estimated flux. As a result the
magnitude and phase of the estimated flux is dependent on the calculated cut off
frequency of the low pass filter.
The measurement-offset error will appear as a ramp signal at the output of the
integrator. The modified integrator proposed in this chapter has feedback to cancel the
DC offset value due to the integrator output initial condition and the ramp due to
measurement offset error.
10.2 Theory of Integrator
As discussed in Chapter 9 the stator flux linkage is estimated by integrating the terminal
voltage minus the voltage drop across the stator winding resistance. To analyse the
principle of integration and the practical difficulty in the implementation of an
integrator, a theoretical signal is applied and the output is analysed. If a signal X is fed
260
to an integrator then an integrated output signal Y will be obtained. This can be written
as:
Y Xdt =
)
(10.1)
for ( ) sin
m
X A t e = , then
( ) ( )
( )
0
1
cos cos
m m
t
Y A t A t e e
e
=
= +
( ) ( )
1
cos
m m
Y A t A e
e
= + (10.2)
However, if
( )
sin cos
2
m m
X A t A t
t
e e = + = then
sin
m
A
Y t e
e
= (10.3)
If the signal X has an offset where X is represented by:
( ) sin
m dc
X A t A e = + then
( ) ( )
( )
0 0
1
cos cos
m m dc dc
t t
Y A t A t A t A t e e
e
= =
= + +
( ) ( )
1
cos
m m dc
Y A t A A t e
e
= + + (10.4)
However, if
( )
sin cos
2
m dc m dc
X A t A A t A
t
e e = + + = + then
sin
m
dc
A
Y t A t e
e
= + (10.5)
In Equation (10.2) there is a constant offset error A
m
/e at the output of the integrator
due to the initial value of the limit of integration, i.e. taking t = 0 as the starting point.
The input and output signals are displayed in Fig. 10.1. However in Equation (10.3)
there is no error in the integrated output, which is the expected type of signal in the
determination of flux linkage. These signals are shown in Fig. 10.2. Depending on the
phase shift of the sinusoidal function the value of the integration error can vary from
zero to A
m
/e.
From Equation (10.5) the ramp error A
dc
t at the output of the integrator is due to a DC
signal or offset in the signal that is fed to the input of the integrator and is shown in Fig.
261
10.3. However, in Equation (10.4) there are two types of errors involved: the first one is
similar to that expressed in Equation (10.2) and the second one is similar to the error
expressed in Equation (10.5). The ramp error signal A
dc
t keeps on increasing with time.
This type of error is displayed in Fig. 10.4.
Y Xdt e =
)
( ) sin X t e =
Fig. 10.1 Offset error equal to A
m
as a result of the integration initial condition
( )
sin
2
X t
t
e = +
Y Xdt e =
)
Fig. 10.2 No integrator error
262
( )
0 08
2
X sin t .
t
e = + +
Y Xdt e =
)
Fig. 10.3 Error produced due to measurement offset
Y Xdt e =
)
( ) sin 0.08 X t e = +
Fig. 10.4 Error produced due to measurement offset and integration initial condition
In order to obtain a sinusoidal integrated signal a technique is needed to eliminate the
effect of A
m
/e, appearing as a constant value and A
dc
t , appearing as a ramp signal at the
output of the integrator. With time the ramp signal will easily drive the integrator into
263
saturation irrespective of the non-zero magnitude of offset A
dc
occurring in the input
signal.
Figs. 10.1 to 10.4 show the different conditions that the input AC signals can have and
the different conditions the integrator output can exhibit. These results represent the
conditions that can happen to the back emf at the input of the integrator and the
computed flux linkage at the output of the integrator in any experimental flux linkage
estimation. The output of the integrator is multiplied by a constant, which is the angular
frequency of the signal to be integrated, so that the peak to peak value of the AC signal
at the input and output remain constant. Then the input and output signals can be
compared easily.
10.3 Numerical integrator
The integrator discussed in Section 10.2 can be implemented as a numerical integrator
for application in a discrete time system. With a sampling time of T
s
, the numerical
integrator is given by:
k
k n s
n 0
Y X T
=
=
_
(10.6)
This can be represented by the following figure:
+
+
T
s
Z
-1
X Y
Fig. 10.5 Numerical integrator representation
10.4 Proposed integration offset adjustment
In general the dc offset in the voltage and current sensors must be minimized to prevent
saturation of the integrator. In the experimental setup, calibrating the sensors every time
the system is started minimizes the dc offset in the measured signals.
The average of a pure sinusoidal signal in a full cycle is zero. This information is used
to correct the offset present at the output of the integrator. Every cycle the average value
is calculated and then subtracted from the integrator output. This method is dealt with in
264
Strategy I below. The measurement offset error is almost constant for a given duration
of time. However the integration error due to the offset at the input of the integrator
keeps on increasing with time. If the measurement offset error present in the signal of
interest is small then the ramp build up within a cycle is insignificant compared with the
integrated signal. Strategy II deals with a technique of minimizing the offset of the
signal at the input of the integrator.
Since the mechanical time constant of induction machines is large, the change of peak
current within a given cycle is small. The main advantage of the proposed technique is
that a pure integrator is used, as it is, without any approximations.
10.4.1 Strategy I - without input offset minimization
This strategy, as illustrated in Fig. 10.6, is applicable when the offset generated in the
hardware sensor and signal conditioning error is small. If the offset at the input of the
integrator is small then the increase of the ramp at the output of the integrator in one
cycle will also be small. Hence adjustment of the ramp every cycle will not distort the
signal of interest. Using a moving average makes it possible to track changes that occur
before the end of a full cycle.
+
+
T
s
Z
-1
X
Y
-
Once in a
cycle
Output error
adjustment Average in a
period
Fig. 10.6 Proposed offset adjustment in a numerical integrator
In Fig 10.6 the period of one cycle used in the proposed offset adjustment can be
obtained from the measured voltage or current using the zero crossing points. The time
between two zero crossing points is equal to half the period of the signal. The period
can be also calculated from the frequency of the signal to be integrated.
265
10.4.2 Strategy II - with input offset minimization
This strategy is applicable when the offset error in the hardware sensor and signal
conditioning is large. The measurement offset error at the input of the integrator can be
reduced to an acceptable value by introducing an offset adjustment at the input of the
integrator. This can be realized using a low pass filter as illustrated in Fig. 10.7.
The offset at the output of the integrator generated due to the initial condition, as given
in Equation 10.2 and Fig. 10.1, cannot be removed using any adjustment at the input of
the integrator. The offset generated from the integrator due to integration initial
condition is constant and once it is adjusted it will not appear again during the rest of
the procedure.
+
+
T
s
Z
-1
X Y
-
Average in a
period
Once in a
cycle
Output error
adjustment
+
-
Low Pass Filter
Fig. 10.7 Proposed integrator with input offset adjustment
A sinusoidal signal, shown in Fig. 10.8a, with an offset of 0.008 and a peak value of 1
unit is applied to the proposed integration error correction system. The integrator output
without any integration error compensation and with the proposed integrator error
compensation is shown in Figs. 10.8b and 10.8d respectively. The magnitude of the
error adjustment signal generated every cycle to compensate the integration error at the
output of the integrator is shown in Fig. 10.8c.
10.5 Stator flux linkage estimation with the proposed method
Flux estimation is the important part in induction machine control. The stator flux
linkage can be estimated from the measured terminal voltage and current [8]. As
discussed in Chapter 9, once the stator flux linkage is available it is easy to calculate the
rotor flux.
266
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
0
1
2
3
4
x 10
-3
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
-0.01
0
0.01
(d)
(c)
time (sec)
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
-2
-1
0
1
2
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
0
0.002
0.004
0.006
0.008
0.01
(b)
(a)
Fig. 10.8 Detail for integration error compensation (a) input X = sin(et )+0.008 (b)
output Y Xdt =
)
without output error adjustment (c) output error adjustment signal
value (d) output with output error adjustment
267
In the estimation of stator flux linkage the voltages and currents in the three axes model
are transformed to voltages and currents in the two axes model using the method
discussed in Chapter 3 and then all the analysis is done in the two axes model [9].
The integration given in Equations (9.44) and (9.45) is implemented in the numerical
integrator using the proposed integrator described in Section 10.4. The implementation
of the d-axis flux linkage estimation and q-axis flux linkage estimation using the
proposed scheme are illustrated in Fig. 10.9.
+
- V
ds
i
ds
R
s
Proposed
integrator
ds
+
- V
qs
i
qs
R
s
qs Proposed
integrator
(a)
(b)
Fig. 10.9 Stator flux linkage estimation using the proposed method (a) d-axis flux
estimation (b) q-axis flux estimation
10.6 Summary
The method of flux estimation proposed in this chapter is new and effective. It
eliminates the error produced by the measurement offset error and integrator output
error due to initial integration in a continuous time integrator or numerical/discrete time
integrator. The numerical integrator can be used in a discrete system such as in digital
signal processing applications. The integrator output offset due to the initial condition is
not dependent on frequency.
If the integration ramp output due to the existence of measurement offset error is large
then subtracting the output of a low pass filter of the signal from the signal to be
268
integrated minimizes the offset. A signal with small input offset will have a small
increment of ramp that will appear at the output of the integrator. As the time increases
the ramp keeps on increasing and eventually the distortion in flux will be unacceptable.
However, if the ramp is eliminated every cycle, the flux distortion due to the offset
correction at the output is insignificant.
Overall a simple method is proposed to compensate the error produced due to the initial
limit of integration as well as measurement error at the input of the integrator.
10.7 References
[1] E. D. Mitronikas, A. N. Safacas and E. C. Tatakis, A new stator resistance tuning method for
stator-flux-oriented vector-controlled induction motor drive, IEEE Transactions on Industrial
Electronics, Vol. 48, Issue: 6, Dec. 2001, pp. 1148 1157.
[2] K. D. Hurst, T. G. Habetler, G. Griva and F. Profumo, Zero speed tacholess IM torque control:
simply a matter of stator voltage integration, IEEE Transactions on Industry Applications, Vol.
34, No. 4, July/August 1998, pp. 790-795.
[3] B. K. Bose and N. R. Patel, A programmable cascaded low-pass filter-based flux synthesis for
stator flux-oriented vector-controlled induction motor drive, IEEE Transaction on Industrial
Electronics, Vol.44, No.1, February 1997, pp.140-143.
[4] J. Hu and B. Wu, New integration algorithms for estimating motor flux over a wide speed range,
IEEE Transaction on Power Electronics, Vol. 13, No. 5 September 1998, pp.969-977.
[5] M. H. Shin, D. S. Hyun, S. B. Cho and S. Y. Choe, An improved stator flux estimation for
speed sensorless stator flux oriented control of induction motors, IEEE Transaction on Power
Electronics, Vol. 15, No. 2, March 2000, pp.312-318.
[6] N. R. N. Idris, and A. H. M. Yatim, An improved stator flux estimation in steady-state operation
for direct torque control of induction machines, IEEE Transactions on Industry Applications, Vol.
38, No. 1, January/February 2002, pp. 110-116.
[7] M. Hinkkanen, and L. Luomi, Modified integrator for voltage model flux estimation of induction
motors, IECON01: The 27
th
Annual Conference of the IEEE Industrial Electronics Society, 2001,
pp.1339-1343.
[8] X. Xu, R. D. Doncker and D. Novotny A stator flux oriented induction machine, in Conf. Rec.
IEEE-PESC, 1988, pp. 870-876.
[9] P. Vas, Sensorless Vector and Direct Torque Control, Oxford University Press, New York, 1998.
269
CHAPTER 11
CONCLUSIONS AND SUGGESTION FOR FUTURE
WORK
11.1 Conclusions
Many important and interesting aspects of an isolated self-excited induction generator
have been discussed and presented in this thesis. The study comprises theoretical
analysis, simulation and experimental results related to induction generators. The
modelling and characteristics of induction machines in general has also been presented
to provide an overall perspective of induction generators.
For an isolated self-excited induction generator driven by a wind turbine the
characteristics of the output power and torque of the wind turbine are important. The
output power and torque of a wind turbine drop at high turbine angular rotor speed. Any
change of electrical load connected to the induction generator will be transferred to the
wind turbine and as a result the wind turbine operating point will be adjusted by a
change in angular rotor speed.
For a grid connected induction generator all the power generated can be supplied to the
grid as the grid can absorb theoretically an infinite amount of power. However for an
isolated induction generator the maximum power available in the wind might not be
utilised by a constant electrical load connected to the induction generator. If an isolated
self-excited induction generator is supplying a constant load then the operating points of
the output power and output torque of the wind turbine will be adjusted as per the
operation of the induction generator.
CHAPTER 11 CONCLUSIONS AND SUGGESTION FOR FUTURE WORK
270
The literature related to isolated induction generators and wind turbines is reviewed in
Chapter 1. This has involved clarifying the strengths and limitations of the previous
works and highlighted the advantages of the research covered in this thesis.
A detailed explanation about wind as a power source and the mechanism of conversion
of wind power to mechanical power is presented in Chapter 2. The general definitions
of wind and wind as a source of power have been presented in this chapter. The analysis
of power absorbed by a wind turbine is based on the horizontal axis wind turbine. The
mechanism of production of force from wind that causes the rotor blades to rotate in a
plane perpendicular to the general wind direction at the site has been discussed in detail.
The importance of having twisted rotor blades along the length from the base to the tip
is given. The variation of the torque produced by the wind turbine with respect to the
rotor angular speed has been presented and associated with this output torque the power
output from a wind turbine is evaluated.
Power absorbed by a wind turbine is proportional to the cube of the wind speed. Wind
turbines are designed to yield maximum output power at a given wind speed. In case of
stronger winds it is necessary to waste part of the excess energy of the wind in order to
avoid damaging the wind turbine. Different ways of power control to protect the
machine have been presented. The economics and growth of wind powered electric
generation is given and the projection for the future is also discussed.
The three-axes to two-axes transformation presented in Chapter 3 is applicable for any
balanced three-phase system. It has been discussed that the three-axes to two-axes
transformation simplifies the calculation of rms current, rms voltage, active power and
power factor in a three-phase system. Only one set of measurements taken at a single
instant of time is required when using the method described to obtain rms current, rms
voltage, active power and power factor. Furthermore from measurements taken at two
consecutive instants in time the frequency of the three-phase AC power supply can be
evaluated.
Existing electrical measuring methods, such as the Fast Fourier Transform, require
many samples from a significant period of the measured waveforms cycle to be
processed using elaborate computation techniques in order to evaluate rms or peak
271
magnitudes of AC currents and voltages. These traditional methods are unable to obtain
peak values in less than one quarter of a cycle. Therefore, the measurement system
presented here supersedes the traditional methods for the monitoring of dynamically
changing quantities in a three-phase system.
As preparation for the modelling and analysis of an isolated induction generator a
detailed model of an induction machine using the conventional or steady state model
and the D-Q or dynamic model are explained and analysed in Chapter 4. The voltage,
current and flux linkage in the rotating reference frame and their phase relationships in
the motoring region and generating region are presented.
For the same stator terminal voltage of an induction machine the magnitude of the
electromagnetic torque in the generating region is higher than the electromagnetic
torque in the motoring region. The reason for the difference in electromagnetic torques
is that during motoring all the electrical losses in the induction machine are supplied by
an external electrical power source and the electromagnetic torque is the output of the
system. However, in the generating region the electromagnetic torque is equivalent to
the external mechanical input torque and all the electrical power losses in the induction
machine are indirectly supplied by the external mechanical power source and the
terminal voltage is the output of the system. Hence to overcome all the internal power
losses in the induction machine and have the same terminal voltage as in the motoring
region the electromagnetic torque in the generating region must be higher than the
electromagnetic torque for motoring.
The D-Q axes induction machine model has been improved to include the equivalent
iron loss resistance, R
m
. This improved model was developed by D. McKinnon and C.
Grantham in collaboration with the author and is presented here in a simple and
understandable way. Using this model the dynamic current, torque and power can be
calculated more accurately.
The data acquisition system which is used for the measurement of voltages, currents,
angle and speed with their appropriate sensors is explained in Chapter 5. The outputs of
the voltage and current measurement circuits are calibrated so that the voltage signals of
272
the sensor outputs will not exceed the ADC input rating of the DS1102 DSP board. The
sensors for current and voltage are Hall-Effect devices. Anti-aliasing filters are
introduced in the analog signals of the sensor outputs to prevent the high frequencies
appearing as a low frequency when the analog signal is digitised in the A/D converter.
Speed and angle measurements for the induction machine are taken using an optical
incremental encoder. The resolution of angle and speed for a given encoder is derived.
For a given encoder the resolution of the angle measurement is constant; however the
resolution of speed measurement is dependent on the sampling period used. High
sampling period (low sampling frequency) gives small resolution in speed (discrete
steps) producing less error in the measurement of the speed. The advantage of digital
signal processing is discussed. Different types of filter design are presented. Digital
filters are used in the simulation and experimental set up with the results presented in
the thesis.
Induction machine parameter determination and the results of an investigation into the
variation of magnetizing reactance, iron loss resistance and rotor parameters with
temperature and supply voltage have been presented in Chapter 6. Most of the
experimentation was carried out by D. McKinnon. A monitoring system employing
digital signal processing techniques was used to observe these effects. To date,
parameter identification methods over a large speed range have typically required
reduced voltages to prevent the motor overheating from excessive current at high slips.
The run up to speed test described enables the parameters to be determined at supply
voltages up to and including the rated voltage without damaging the motor under test.
A separate test was used to determine the iron loss resistance and magnetizing
reactance. This test incorporated two methods. Both methods were conducted at
synchronous speed, one using continuously varied supply voltage, the other at discrete
voltages. Agreement between the results was excellent. All tests were performed at
corresponding temperatures.
Parameter identification is important in the modelling of and control of induction
machines. It is obvious from the results in this thesis that parameter variations cannot be
273
ignored. It has been shown that rotor parameter variations exist even for a single-cage
induction motor.
The error in the values of induction motor parameters arising from measurement error in
voltage, current and power used to determine the parameters have been presented.
These three measurement quantities are essential for accurate parameter determination.
Even with measurement errors of 1.5% the determined parameters are still very
acceptable and considerably more accurate than if rotor parameter variations are
ignored. Only very close to synchronous speed is there substantial error in the
determined value of X
lr
, but this is almost irrelevant in any meaningful analysis of the
inductive machine, because close to synchronous speed the effect of X
lr
is swamped by
the effect of R
r
/s.
When rotor parameter variations are ignored the percentage errors in the current, power
and torque are substantial for machines which exhibit a significant degree of rotor
current displacement effect.
The modelling, analysis and dynamic performance of an isolated three-phase induction
generator excited by three AC capacitors connected at the stator terminals is presented
in Chapter 7. The use of the variation in magnetising inductance with voltage leads to
an accurate prediction of whether or not self-excitation will occur in a SEIG for various
capacitance values and speeds in both the loaded and unloaded cases. The
characteristics of magnetising inductance, L
m
, with respect to the rms induced stator
voltage or magnetising current determines the regions of stable operation as well as the
minimum generated voltage without loss of self-excitation.
Once self-excitation has been initiated and a steady state condition has been reached,
the speed at which self-excitation ceases is always lower than the speed to initiate self-
excitation. At a particular speed the capacitance required for self-excitation, when the
machine operates at no load, is less than the capacitance required for self-excitation
when it is loaded.
274
When an induction machine operates as a motor the speed of the rotating air gap
magnetic field is totally dependent on the excitation frequency. However, in the SEIG it
is shown that the frequency of the generated voltage depends on the speed of the prime
mover as well as the condition of the load. Keeping the speed of the prime mover
constant with increased load causes the magnitude of generated voltage and frequency
of an isolated SEIG to decrease. Modelling and experimental results show that this is
due to a drop in the speed of the rotating magnetic field. When the speed of the prime
mover drops with load, then the decrease in voltage and frequency will be greater than
for the case where the speed is held constant.
unloaded conditions while the speed was varied or kept constant. Using the simulation
algorithm more results which are not accessible in an experimental setup have been
predicted.
Increasing the capacitance value can compensate for the voltage drop due to loading but
the drop in frequency can be compensated only by increasing the speed of the rotor. The
variation of magnetising inductance follows the variation in terminal voltage or
magnetising current. Increasing the capacitance can compensate the generated voltage,
however it increases stator current. Hence care should be taken not to exceed the stator
rated current.
The variation of the generated voltage and frequency for a self excited induction
generator driven by a wind turbine at constant and variable speeds has been
investigated. The dynamic calculated performance comparison between the machine
with constant and variable rotor parameters has been simulated and discussed
considering a hypothetical induction machine that represents a typical double-cage or
deep-bar rotor induction machine where rotor parameter variations are significant.
A novel D-Q axes model that includes the iron loss equivalent resistance R
m
in the
dynamic analysis and simulation of the SEIG has been described in Chapter 8 in a
simple and understandable way. It is noted that this method is easily understood, having
275
drawn on many familiar concepts and using the standard terminology and nomenclature
of D-Q unified machine theory. In some instances it is possible to neglect the iron loss;
however its effect has to be shown to be negligible. This study provides the tool to
reach such a decision. When R
m
is included, which depicts the actual situation, it has
been shown that the generated voltage, currents, and output power, are lower than that
when R
m
is neglected. However, due to additional losses the electromagnetic torque
necessary to drive the induction generator is higher when R
m
is included than when it is
neglected.
Using vector control the voltage build up process and terminal voltage control in an
isolated wind powered induction generator using an inverter/rectifier excitation with a
single capacitor on the DC link is discussed in Chapter 9. Due to the fundamental
switching of the IGBTs of the inverter, controlled by the vector control principle, the
single DC capacitor appears as if three capacitors were connected across the three stator
terminals of the induction generator.
Different types of vector control techniques are developed to control the excitation and
the active power producing currents independently. That is, the current control scheme
causes the currents to act in the same way as in a DC generator where the field current
and the armature current are decoupled. When the speed of the prime mover is varied
the flux linkage in the induction generator is varied inversely proportional to the rotor
speed so that the generated voltage will remain constant.
Since the torque produced by a wind turbine drops at high turbine rotor speed the
induction generator will run at high generator rotor speed when loaded with a small load
and the rotor speeds decrease with an increase in load. As the turbine rotor shaft and the
generator rotor shaft are connected via a gear box, both rotor speeds will increase and
decrease proportionally for a constant gear ratio. The flux linkage of the induction
generator is controlled by controlling the d-axis current in the synchronously rotating
reference frame. All vector control strategies presented are either rotor flux oriented
vector control or stator flux oriented vector control.
276
It is shown that the estimation of rotor flux linkage is more dependent on the induction
machine parameters whereas estimation of stator flux linkage is dependent only on the
stator resistance. If the variation of stator resistance is causing a significant error in the
estimated flux then a compensation block can be added in the model, however it is
difficult to compensate the variation in rotor parameters, especially rotor resistance. The
dynamic results for generated voltage, current, power, torque and slip for different rotor
speed and load have been given.
When the load is varied the q-axis current in the synchronously rotating reference frame
varies to provide the current demand of the load. Hence the q-axis current in the
synchronously rotating reference frame is varied when there is variation in load. When
the rotor speed is varied the d-axis current in the synchronously rotating reference frame
is varied to provide the required amount of flux linkage corresponding to the rotor speed
in a similar way as in flux weakening operation of induction motors. In this way the DC
voltage will remain constant as the speed of the wind turbine varies and also when the
electrical load changes. Once a constant DC voltage is achieved a DC load can use it
directly or, if required, it is a matter of having an inverter to produce a constant voltage
and frequency AC output.
The estimation of stator flux linkage using integration of the voltage behind the stator
resistance is the easiest way of estimating the flux linkage. However, there are problems
associated with the integration. The method of flux estimation proposed in Chapter 10 is
an effective method to compensate the error produced by the integrator. The cause of
the error in an integrator is illustrated in detail by using a sinusoidal signal as an input to
the integrator and then solutions are proposed in a simple and understandable way.
The compensation method eliminates the error produced by the measurement offset
error and integrator output error due to the initial integration in a continuous time
integrator or numerical/discrete time integrator. The numerical integrator can be used in
a discrete system such as in digital signal processing applications. The integrator output
offset due to the initial condition is not dependent on frequency.
277
If the integration ramp output due to the existence of measurement offset error is large
then subtracting the output of a low pass filter of the signal from the signal to be
integrated minimizes the offset. A signal with small input offset will have a small
increment of ramp that will appear at the output of the integrator. As the time increases
the ramp keeps on increasing and eventually the distortion in flux will be unacceptable.
However, if the ramp is eliminated every cycle, the flux distortion due to the offset
correction at the output is insignificant. Overall a simple method is proposed to
compensate the error produced due to the initial limit of integration as well as
measurement error at the input of the integrator.
11.2 Suggestions for future work
The analysis and explanations presented in this thesis provide a good foundation for
further research in the area of isolated induction generators driven by a wind turbine.
Some of the topics recommended for future work are:
- Implementation of vector control to regulate the output voltage and frequency of
an isolated self-excited induction generator and compare with the simulation
results.
- Implementation of the vector control technique for an isolated induction
generator with variable rotor parameters. As the contribution of rotor leakage
inductance is small compared to the total rotor inductance the focus is mainly in
the updating of the rotor resistance variation. This variation in rotor resistance
can also be explored and extended to include variation of rotor resistance due to
the operating temperature of the rotor conductors.
- Development of a model that takes into account the variation in stator resistance
or a way of compensating the stator resistance in the rotor resistance. Stator
resistance is important in the estimation of stator flux linkage.
- Technical and economic comparison between series connected inverter/rectifier
(discussed in Chapter 9 of this thesis) and shunt connected inverter/rectifier
associated with a separate rectifier to supply the load for an isolated induction
generator.
- Use of Direct Torque Control instead of Vector control in an isolated induction
generator. The development of a control system based on the balance of power
278
between the output power and the available power in the prime mover is to be
envisaged.
278
APPENDICES
APPENDIX A
DETERMINATION OF INERTIA AND FRICTION
COEFIENT OF THE INDUCTION GENERATOR SYSTEM
The induction generator system consists of an induction machine and a DC motor. Here
the DC motor is the prime mover and is always coupled to the induction machine.
Therefore the total inertia and friction coefficient determined here consists of the sum of
the inertia in the induction machine and in the DC motor.
The dynamics of the motor is given by
r
e r L
d
T J D T
dt
e
e = + + (A.1)
where
T
e
electromagnetic torque (Nm),
T
L
load torque (Nm),
e
r
rotor or shaft speed (rad/sec),
D friction coefficient (Nm/rad/sec) and
J rotor inertia (Kg-m
2
).
To determine J and D the induction generator system was driven by the DC motor.
During the test the terminals of the induction machine were open. Hence no load torque,
i.e T
L
=0.
From the voltage equation of the DC motor, the back emf in the armature can be
calculated as:
a a a
E V I R = (A.2)
APPENDIX A DETERMINATION OF INERTIA AND FRICTION COEFFIENT
279
where
E
a
back-emf (V),
V applied voltage (V),
I
a
motor current (A) and
R
a
Armature winding resistance (ohm)
The output power, amount of power converted to mechanical power, is given by
a a
P E I = (A.3)
and the electromagnetic torque is the mechanical power divided by the shaft speed,
which is calculated as:
a a
e
r
E I
T
e
= (A.4)
At steady state the electromagnetic torque developed by the DC motor, for T
L
= 0, is
given by:
e r
T De = (A.5)
T
e
is determined from (A.5) and then D is calculated as.
e
r
T
D
e
= (A.6)
For the determination of D, the voltage and current measurements at a given speed were
taken by giving some time until steady state speed was attained. In this part the
measurement was taken manually making sure that there was no change of speed at a
particular value of speed. The torque versus speed graph to determine D is given in Fig.
A.1. A line is drawn using Matlab linear curve fit and the slop of this line is the friction
coffiecient..
From (A.6) D is the slope of the line in the plot Te versus e
r
. From the curve fit the slop
of the line is 0.0027, which is the average gradient.
Therefore
D=0.0027 Nm/(rad/sec)
280
Torque, T
e
(Nm)
Speed or e
r
(rad/sec)
Fig. A.1 Electromagnetic torque versus motor speed at steady state
The effect of J appears when there is change in speed. To calculate J the variation of
speed with time should be recorded accurately. In this part of the test a fast data
acquisition system (dSPACE DS1102 DSP card) was used. The speed of the DC motor
reaches at a given point and then the DC supply to the motor is switched off.
If the supply to the DC motor is switched off, while the motor is running at a specific
speed, then the dynamics of the motor is given by:
0
r
r
d
J D
dt
e
e = + (A.7)
The transient solution of (A.7) is given by
D
t
J
r ro
e e e
= (A.8)
where e
ro
is the initial speed, which is the speed of the motor when the DC supply was
switched off.
The magnitude of
D
J
can be calculated by solving the equation in (A.8) using curve
fitting or based on the value of speed and time obtained from measurement.
281
Or J can be calculated by rearranging the basic equation given A.7 [1] as
r
r
D
J
d
dt
e
e
= (A.9)
The DC supply to the armature of the motor and to the field winding of the motor
should be switched off at the same time. The variation of motor speed recorded when
the DC motor field supply was on and off is shown in Fig. A.2. Of course the armature
supply of the DC motor was switched off.
Speed or e
r
(rad/sec)
(a)
(b)
Fig. A.2 Variation of speed with time (a) DC motor field supply on (b) DC motor field
supply off
When the supply is switched off at t = 0 there is transient due to the collapsing of the
field and this transient filed will enable the machine to generate a transient condition
Close to t = 0 the curves with field supply on and with field supply off appear as if they
overlap. Hence the inertia for the test with field supply off appears as if there is smaller
inertia than the rest of the time.
Using curve (a) in Fig. A.2 can lead to a wrong answer because the DC machine field
supply was on during the test. If the DC machine field supply is on, while the motor
armature supply is off, the DC motor will act as a generator and it will generate an open
282
circuit voltage. Due to the generation of open circuit voltage it will have some internal
electrical losses. These losses will decelerate the speed faster. The effect of faster
deceleration is reflected on the value of the J in the system and it will appear as if it has
a smaller value than the J obtained from the test with the DC motor field supply off. J
calculated from curve (a) in Fig. A.2 is 0.025Kg-m
2
and J calculated from Fig. A.2
curve (b) is 0.045Kg-m
2
/rad.
The correct value of the inertia J is the one calculated from the speed versus time
variation when the supplies to the field winding as well as the supply to the armature are
switched off at the same time. The curve that satisfies this condition is curve (b) given
in Fig. A.2. Hence the correct value of inertia, J, for the induction generator system
used in the lab is 0.045Kg-m
2
/rad.
Reference
Werner Leonhard, Control of Electrical Drives, Springer-Verlag Berlin Heidelberg, New York, 1996
283
APPENDIX B
MEASUREMENT AND CONTROL HARDWARE SYTEMS
The interconnection between the different measurement and control hardware systems
is given in this appendix.
Voltages and
currents sensors
8 to 4
multiplexer
DS1102
DSP card in PC
Incremental
encoder
Encoder interface
and 5V encoder
power supply board
Cross over protection board
DC motor speed
reference
Optocoupler
board
MITSBUSHI
PM50RSA120
Optocoupler
board
MITSBUSHI
PM50RVA120 IPM
Fig. B.1 Interconnection of hardware system
Fig. B.2 DSPACE DS1102 DSP controller board
Multiplexer
board
DSP
dSPACE DS1102
card
5
1
3
28
7
60
IOP0
IOP1
GND
Red
Orange
Black
IOP0
IOP1
GND
CON2
CON3
1
2
3
ADC1
ADC2
ADC3
Single core coaxial cable Output1
CON4
4
ADC3
Single core coaxial cable
Output2
Output3
Output4
CON1
Fig. B.3 Multiplexer board control to dSPACE DS1102 DSP card connection
APPENDIX B MEASUREMENT AND CONTROL HARDWARE SYTEMS
284
BNC socket
for speed
feed back
DSP dSPACE
DS1102 card
46
47
DAC4
GND
Coaxial cable
Fig. B.4 DAC output for DC motor speed control
Dead time generator
for IGBTs gate
drivers in the
Inverter
(Cross over
protection board)
dSPACE
DS1102 DSP
card
3
2
1
6
5
4
7
54
55
56
57
CMP0
CMP1
CMP2
CAP3
CMP3
CAP4
IOP4
52
53
30
digitalGND
blue
green
yellow
Orange
Red
Brown
White
8
PWM1
PWM2
PWM3
PWM5
PWM4
PWM6
Break
PWM1
PWM2
PWM3
PWM5
PWM4
PWM6
0V
GND
Break-to reduce DC bus voltage
Black
15
Fig. B.5 Dead time Generator board and DS1102 DSP card connection
Incremental
rotary encoder
HEIDENHAIN
ROD 426
5000line
Encoder
interface
board with
5V supply
dSPACE
DS1102
DSP card
U
a1
U
a2
U
ao
5V(sensor)
5V(U
p
)
0V(U
N
)
0V(sensor)
U
a1
U
a2
U
ao
5V
5V
0V
a2
U
0V
1
2
3
4
5
6
7
8
14
15
1
2
3
4
5
6
8
U
a1
a1
U
U
a2
U
ao
a2
U
ao
U
0V
20
41
19
40
21
42
60
Phi0 1
Phi90 1
Index1
Index1
Phi90 1
Phi0 1
Digital GND
brown
green
gray
pink
red
black
blue
White
Brown/green
White/green
blue
yellow
gray
orange
green
Red
Black
a2
U
a1
U
a1
U
a0
U
a0
U
Fig. B.6 Incremental encoder DS1102 DSP card connection
285
~240V
Diode
bridge
GBU6K
220F,
400V
Mitsubishi
M57120L-01
DC-DC
converter
12
11 12
1
+
+
6
6 5
2
10
5
+
20V
330F,
50V
4
3 2 1
Mitsubishi
M57140-01
415V isolated
DC output
+ 33F,
50V
22KO
VN1
+ 10F,
50V
22KO
VUP1
+ 10F,
50V
22KO
VUP1
+ 10F,
50V
22KO
VWP1
+15V
+15V
+15V
+15V
0V
0V
0V
0V
8
7
9
11
14
13
Fig. B.7 Four isolated 15V Power supply for optocoupler circuit
286
Fig. B.8 Optocoupler to Mitsubishi PM50RVA120 IPM
287
LF398
+1
-15
10nF
Input
LF398
+1
-15
10nF
Input
2
LF398
+1
-15
10nF
Input
LF398
+1
-15
10nF
Input
LF398
+1
-15
10nF
Input
LF398
+1
-15
10nF
Input
LF398
+1
-15
10nF
Input
LF398
+1
-15
10nF
Input
8
Analog
Switch
DG403
S1
S3
S2
S4
16
4
9
5
1
D1
D3
3
8
D1
D3
6
V
+
V
-
V
L
11 14 12
13
GND
+15 -15V+5
15 10
IN1 IN2
Analog
Switch
DG403
S1
S3
S2
S4
16
4
9
5
1
D1
D3
3
8
D1
D3
6
V
+
V
-
V
L
11 14 12
13
GND
+15V -15V +5V
15 10
IN1 IN2
9 pin socket
Connector
to DS1102
DSP card
1
5
3
To
DS1102
DSP card
ADC
inputs
4
3
2
1
+5V
+5V
9.1V
9.1V
9.1V
9.1V
1K O
1K O
1K O
1K O
+15
Offset adjust
2
+15
Offset adjust
2
+15
Offset adjust
+15
Offset adjust
2
+15
Offset adjust
2
+15
Offset adjust
2
+15
Offset adjust
2
+15
Offset adjust
2
L. Ref
L. Ref
L. Ref
1
V
+
4
V
-
3
Input
6
C
h
7
5
Output
Logic
8
2
L. Ref
L. Ref
L. Ref
L. Ref
L. Ref
1K 24K
1K 24K
1K 24K
1K 24K
1K 24K
1K 24K
1K 24K
1K 24K
IOP0
IOP1
ADC4
ADC3
ADC2
ADC1
1K O
1K O
1
V
+
4
V
-
3
Input
6
C
h
7
5
Output
Logic
8
2
1
V
+
4
V
-
3
Input
6
C
h
7
5
Output
Logic
8
2
1
V
+
4
V
-
3
Input
6
C
h
7
5
Output
Logic
8
2
1
V
+
4
V
-
3
Input
6
C
h
7
5
Output
Logic
8
2
1
V
+
4
V
-
3
Input
6
C
h
7
5
Output
Logic
8
2
1
V
+
4
V
-
3
Input
6
C
h
7
5
Output
Logic
8
2
1
V
+
4
V
-
3
Input
6
C
h
7
5
Output
Logic
8
2
Fig. B.9 8 to 4 multiplexer with Sample and Hold
288
+5V
To
DSP card
Dead time
Generator
IXYS
IXDP630
Br
PWM4
PWM5
PWM6
PWM1
PWM2
PWM3
7
6
5
4
8
2
1
3
GND
T
S
R
5
3
1
9
1
1
2
3
7
AND
SN74LS08N
+5V
4.7pF
20KO 47KO
18
Vcc
2 ENAR
4 ENAS
6 ENAT
8 RESET
14 SL
13 TU
12 TL
11
10
17 RU
16 RL
15 SU
+5
+5V
11
10
U
P
2
U
N
5
V
P
3
V
N
6
W
P
4
W
N
7
To optocoupler
IGBT gate
driver
4.7pF
20KO
47KO
18
Vcc
2 ENAR
4 ENAS
6 ENAT
8 RESET
14 SL
13 TU
12 TL
11
10
17 RU
16 RL
15 SU
+5
+5V
11
10
U
P
2
U
N
5
V
P
3
V
N
6
W
P
4
W
N
7
To optocoupler
IGBT gate
INHIBIT 14
INHIBIT 14
T
S
R
5
3
1
9
To Single IGBT for
DC voltage control
15
15
Break 8
7
OUTEN
OUTEN
7
+5V
+5V
+5V
Dead time
Generator
IXYS
IXDP630
1KO
1KO
1KO
1KO
1KO
1KO
1KO
1
4
2
9
12
5
10
13
3
6
8
11
7
14
3
6
14
7
+5
1
4
2
9
1
5
1
1
3
6
8
1
7
1
3
6
1
7
+5V
+5V
4
2
1
5
+5
4
2
1
5
74LS00
NAND
74LS00
NAND
74LS00
NAND
74LS00
NAND
+5V
+5V
+5V
Fig. B.10 Cross over protection board (dead time generator)
289
APPENDIX C
DETAILS IN INDUCTION MACHINE MODELLING
C.1 Introduction
In this part of the appendix the relationship between the induction machine parameters
in the conventional steady state equivalent circuit model and in the dynamic equivalent
d-q model is explained. The expanded form of the equations representing the dynamic
d-q model of an induction machine including R
m
are given.
C.2 Relationship of parameters in steady state model and d-q model of
induction machines
The parameters of an induction machine are obtained with the help of parameter
determination method using the conventional steady state equivalent circuit. The
conventional steady state model gives only steady state solution however the d-q model
gives the complete solution which includes the transient and steady state solutions
based on the steady state parameters. To find the relationship between the parameters in
the conventional steady state model and d-q model, the d-q model will be analysed
based on its steady state conditions.
0 0
0 0
0
0
qs
qs s s m
s s m ds
ds
m r m r r r r qr
r m m r r r r
dr
i
v R +L p L p
R +L p L p i
v
=
L L p L R +L p
i
(
( (
(
( (
( ( (
(
( (
(
( (
( ( (

(C.1)
The steady state form is obtained by substituting p by je
e
and e
r
by (1-s)e
e
where e
e
and s are the excitation angular frequency and rotor slip respectively
0 0
0 0
(1 ) (1 )
0
(1 ) (1 ) 0
qs
qs s s m
s s m ds
ds
m m r r r qr
m m r r r
dr
i
v R +jX jX
R +jX jX i
v
=
jX X s R +jX -X s i
X s jX X s R +jX
i
(
( (
(
( (
( ( (
(
( (

(
( (

( ( (

(C.2)
APPENDIX C DETAILS IN INDUCTION MACHINE MODELLING
290
The rotor current terms can be eliminated easily by partitioning the matrix equation and
rewriting as
11 12 1 1
21 22 2
0
Z Z I V
Z Z I
( ( (
=
( ( (

(C.3)
where
( ) ( )
1 11 12 21
22 1 2
1
22
2 2
0 0 (1 )
0 0 (1 )
(1 )
(1 )
(1 )
1
(1 )
(1 )
qs s s m m m
s s m m m ds
qs qr r r r
r r r ds dr
r r r
r
r r r
v R +jX jX jX X s
V Z Z Z
R +jX jX X s jX v
i i R +jX -X s
Z I I
X s R +jX i i
R +jX X s
Z
X s R
R +jX X s
( ( ( (
= = = =
( ( ( (

( ( (
= = =
( ( (
=

+ r r
+jX
(
(

From Equation (C.3)
1
2 22 21 1
I Z Z I
= (C.4)
( )
1
1 11 12 22 21 1
V Z Z Z Z I
= (C.5)
Expanding Equation (C.5)
0 0 (1 ) (1 )
1
0 0 (1 ) (1 )
qs qs s s m m m r r r
s s m r r r m m ds ds
v i R +jX jX jX X s R +jX X s
R +jX jX X s R +jX X s jX v i
| | ( ( ( ( ( (
=
| ( ( ( ( ( (
A
\ .
(C.6)
Where
( ) ( )
2 2
(1 )
r r r
R +jX X s A = +
Using mathematical equation a
2
+ b
2
= (a + jb) (a jb)
( )( ) (1 ) (1 )
r r r r r r
R +jX jX s R +jX jX s A = +
( )( ) (2 )
r r r r
R jX s R +jsX A = + (C.7)
Simplifying Equation (C.6)
2
0 (1 ) 1 0 (1 )
0 0 1 (1 ) (1 )
qs qs s s r r r m
s s r r r ds ds
v i R +jX R +jX X s j s jX
R +jX X s R +jX s j v i
| | ( ( ( ( ( (
=
| ( ( ( ( ( (
A
\ .
2 2
0 (1 )
0 (1 ) (2 )
qs qs s s r r r m
s s ds ds r r r
v i R +jX jR s X R s jX
R +jX v i R s sX s jR
| | ( ( ( (
=
|
( ( ( (
|
A +
\ .
291
( )
2 2 2
2 2
( ) (1 )
(2 ) (1 )
m r r m r
s s
qs qs
ds ds m r r m r
s s
jX jR s X jX R s
R +jX
v i
v i jX sX s jR jX R s
R +jX
(
(
( (
A A
(
=
( (
( +

(
A A
(C.8)
The d-axis stator voltage
ds
v in Equation (C.8) can be expressed as
( )
2 2
(2 ) (1 )
m r r m r
ds qs s s ds
jX sX s jR jX R s
v i R +jX i
| | +
= +
|
|
A A
\ .
(C.9)
As discussed in Chapter 3 the q-axis current lags the d-axis current by 90
o
.Hence
substituting
qs ds
i ji = in Equation (C.9) gives
( )
2
(2 )
m r r
ds s s ds
X jsX s sR
v R +jX i
| | +
= +
|
|
A
\ .
(C.10)
Substituting Equation (C.7) in (C.10)
( )
( )( )
2
(2 )
(2 )
m r r
ds s s ds
r r r r
X jsX s sR
v R +jX i
R jX s R +jsX
| | +
= +
|
|
+
\ .
( )
2
m
ds s s ds
r r
sX
v R +jX i
R +jsX
| |
= +
|
|
\ .
(C.11)
Bur X
s
and X
r
are the total stator reactance and rotor reactance respectively. Expressing
the reactances with the mutual and their leakage reactances
s ls m
X X X = +
r lr m
X X X = +
Then Equation (C.11) can be rewritten as
r
m lr
ds s ls ds
r
lr m
R
jX jX
s
v R +jX i
R
+jX jX
s
| | | |
+
| |
\ .
|
= +
| | |
+
| |
\ . \ .
(C.12)
Equation (C.12) represents the steady state per-phase equivalent circuit which can be
derived from Fig. 4.3. Hence at steady state the D-Q model converges to the
conventional steady state model.
C.3 Expanded equations for induction machine modelling including R
m
292
The matrix equation representing a stator supplied induction machine, from Equation
4.25, including R
m
is
0 0
0 0
0
0
qs
qs s sNEW N
s sNEW N ds
ds
N r m r rNEW r rNEW qr
r N N r rNEW r rNEW
dr
i
v R +L p L p
R +L p L p i
v
=
L L p L R +L p
i
(
( (
(
( (
(
( (
(
( (
(
( (
( ( (

(C.13)
where
m m
N
m m
R L
L =
R +L p
sNEW ls N
L =L +L
rNEW lr N
L =L +L
The expanded form of Equation (C.13) is given in Equation (C.15).
For the D-Q model of a self-excited induction generator the matrix equation including
R
m
, as given in Equation 8.1, is,
1 0 0 0
0 1 0 0
0
0
qs cqo
s sNEW N
s sNEW N ds cdo
N r m r rNEW r rNEW qr qr
r N N r rNEW r rNEW
dr dr
i V
R pL pC pL
R pL pC pL i V
pL L R pL L i K
L pL L R pL
i K
e e
e e
( ( + + ( (
( (
( (
+ +
( (
( (
= +
( (
( ( +
( (
( (
+
( ( (

(C.14)
The expanded form of Equation (C.14) is given in Equation (C.16).
293
0 0
0 0
0
0
m m m m
s ls
m m m m
qs m m m m
s ls
m m m m
ds
m m r m m m m m m
r lr r lr
m m m m m m m m
r m m m m
r lr
m m m m
R L R L p
R L p
R L p R L p
v R L R L p
R L p
R L p R L p
v
R L p R L p R L R L p
- R L p - L
R L p R L p R L p R L p
R L p R L p
L
R L p R L p
e
e
e
e
| |
+ +
|
+ +
\ .
| | (
+ +
|
(
+ +
\ .
(
=
(
| | | |
( + + +
| |
+ + + + (
\ . \ .
+
+ +
qs
ds
qr
dr
m m m m
r lr
m m m m
i
i
i
i
R L p R L
R L p
R L p R L p
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(

(
(
| | | |
+ + (
| |
+ +
(
\ . \ .
(C.15
1
0 0
1 0
0 0
0
0
0
m m m m
s ls
m m m m
m m m m
s ls
m m m m
m m r m m m m m m
r lr r lr
m m m m m m m m
r m m m m
r l
m m m m
R L R L p
R L p
R L p pC R L p
R L R L p
R L p
R L p pC R L p
R L p R L p R L R L p
- R L p - L
R L p R L p R L p R L p
R L p R L p
L
R L p R L p
e
e
e
e
| |
+ + +
|
+ +
\ .
| |
(
+ + +
|
(
+ +
\ .
(
=
(
| | | |
+ + + (
| |
+ + + +

\ . \ .
+ +
qs cqo
ds cdo
qr q
dr d
m m m m
r r lr
m m m m
i V
i V
i K
i K
R L p R L
R L p
R L p R L p
(
(
(
(
( (
(
( (
(
( (
+
(
( (
(
( (
(
( (

(
(
| | | |
+ + + (
| |
+ +
(
\ . \ .
(C.16)
294
From Equation (8.5) the d-axis current for the SEIG including R
m
is given by
d
8 7 6 5 4 3 2
U
i
Ap Bp Dp Ep Fp Gp Hp Jp M
=
+ + + + + + + +
(C.17)
The roots in the denominator (poles) contain the behaviour of the solution for the d-axis
current. To find the roots of the denominator
8 7 6 5 4 3 2
Ap Bp Dp Ep Fp Gp Hp Jp M 0 + + + + + + + + = (C.18)
A, B, D, E, F, G, H, J, and M are coefficients which are functions of the machine
parameters, exciting capacitance and rotor speed. The detail these coefficients are given
as follows.
2 2 2 2
ls lr m
A C L L L =
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2( 2 )
ls r m lr m ls lr m s ls lr m m ls lr m m ls lr m
B C L R L L R C L L L R C L L L R C L L L R C L L L = + + + +
2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2
2 2 4 2
4 2
4 4 4
ls r m ls lr m m ls lr m m ls r lr m m ls r m
ls r lr m s ls r m lr m ls m m m lr m s lr m
m s ls lr m m s ls lr m m ls r m
D C L R L CL L L R C L L L R C L R L L R C L R L
C L L L R C L R L L R C L L R C L L R R C L L
R R C L L L R R C L L L R C L R L L
e
= + + + +
+ + + + +
+ + +
2 2 2 2
2 2 2 2 2 2 2 2 2 2
2 2
lr s lr m
m ls lr m m ls lr m ls lr m
R C L L
R C L L L R C L L R C L L L
+
+ + +
2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 4 2 4 4
2 4 2 2
2 2 2 2
m lr m ls r m lr s lr m m ls lr m m ls lr m
s r m lr m s r m lr s ls r m s ls r lr m
m ls r lr m m ls r lr m m s lr m m s lr m
E CR L L CL R L L R CL L R CL L L R CL L L
R C R L L R R C R L L R C L R L R C L L L
R C L L L R C L L L R R C L L R R C L L
e
e e
= + + + +
+ + + +
+ + + +
2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2
2 2
2 2 2 2
8 4 2 2
2 4 2 2
4
m s ls lr m s ls m m s lr m m s lr m
m s ls r lr m m s ls r m m ls r m m ls r m
m ls r lr m m s ls lr m m m r lr m ls r m
m l
R R C L L R R C L L R R C L L R R C L L
R R C L R L L R R C L R L R C L R L R C L R L
R C L L L R R C L L L R C L R L R C L R L
R C L
e
+ + + +
+ + + +
+ + + +
+
2 2 2 2 2 2
2 2
s r lr m m ls r lr m ls r m
R L L R C L R L R C L R L + +
2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2
2(
)
r m m r lr m m r lr m m m s r m r m lr m s r lr m
r m m m s r lr m s r
J R R L L L R L R L R R C L R R L +2R R C L L
+R R L +R R C L +R R CR
e e e e
e
= + + + +
2 2 2 2 2 2 2 2 2 2
2
r m r m m r lr m r lr m m
M R R R L L R L R L e e e = + + +
295
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2
4 2 2 2 2 2 4 2
4 8 4 4 4
lr m m r m lr m lr m m lr m ls r m m ls lr m ls m m ls lr m ls r lr m
s r m lr m ls r lr m m ls r m m s lr m m s lr m s r m s r
F L L CR R L L CR L L CR L L CL R L R CL L R CL L R CL L L CL L L
R CR L L R CL R L L R CL R L R R CL L R R CL L R C R L R C
e
e
= + + + + + + + +
+ + + + + + +
2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2
2
2 4 2 2 4 2
4 4 4 4
lr m m s r m
m s r lr m m s r lr m m s r m m s m m s r m m s r lr m m ls r lr m
m s ls r r lr m s ls r r m m s ls r lr m m s l
L L R R C R L
R R C L L R R C R L L R R C R L R R C L R R C R L R R C R L L R C L L L
R R C L R L R R C L R L R R C L L L R R C L
e e
e e e
+
+ + + + + + +
+ + + +
2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2
2 2
2 2
s r lr m m s ls r m m ls r lr
m ls r lr m m ls r m ls r m m s lr m m s lr m m r lr m m r
m ls r lr m m ls r m
L L R R C L R L R C L L
R C L L L R C L R R C L L R R C L L R R C L R C L L R C L R
R C L L L R C L R L
e e
e e e
e
+ +
+ + + + + + +
+ +
G=2*Rr*Lm^2*Llr+2*Llr^2*Rm*Lm+2*Llr*Lm^2*Rm+2*c*Rm*Rr^2*Lm^2+2*c*Rm^2*Rr*Lm^2 +2*c*Rm*wr^2*Llr^2*Lm^2
+2*Rs*c*wr^2*Llr^2*Lm^2+2*Rm^2*Rs*c*Lm^2+4*Rm^2*c*Lls*Rr*Llr+4*Rm^2*c*Lls*Rr*Lm+2*Rs*c*Rr^2*Lm^2
+4*c*Rm^2*Rr*Llr*Lm+2*Rm^2*Rs*c*Llr^2+4*Rm*c*Lls*wr^2*Llr^2*Lm+4*Rm*c*Lls*wr^2*Llr*Lm^2+4*Rm^2*Rs*c*Llr*Lm
+4*Rm*c*Lls*Rr^2*Lm+8*Rm*Rs*c*Rr*Llr*Lm+4*Rm*Rs*c*Rr*Lm^2+2*Rm^2*Rs*c^2*wr^2*Llr^2*Lm+2*Rm^2*Rs*c^2*Rr^2*Lm
+2*Rm^2*Rs*c^2*wr^2*Llr*Lm^2+2*Rm*Rs^2*c^2*wr^2*Llr^2*Lm+2*Rm*Rs^2*c^2*Rr^2*Lm+2*Rm*Rs^2*c^2*wr^2*Llr*Lm^2
+2*Rm^2*Rs*c^2*Lls*wr^2*Llr^2+4*Rm^2*Rs*c^2*Lls*wr^2*Llr*Lm+2*Rm^2*Rs*c^2*Lls*Rr^2+2*Rm^2*Rs^2*c^2*Rr*Llr
+2*Rm^2*Rs^2*c^2*Rr*Lm+2*Rm^2*Rs*c^2*Lls*wr^2*Lm^2
H =wr^2*Llr^2*Lm^2+4*Rm^2*Rs*c*Rr*Lm+4*Rr*Rm*Llr*Lm+Rm^2*Lm^2+2*Rr*Lm^2*Rm+Llr^2*Rm^2+2*Llr*Rm^2*Lm
+2*c*Rm^2*wr^2*Llr*Lm^2+Rm^2*Rs^2*c^2*wr^2*Lm^2+4*Rm*Rs*c*Rr^2*Lm+2*Rm^2*c*Lls*Rr^2+4*Rm^2*Rs*c*Rr*Llr
+2*c*Rm^2*Rr^2*Lm+Rm^2*Rs^2*c^2*Rr^2+2*c*Rm^2*wr^2*Llr^2*Lm+4*Rm*Rs*c*wr^2*Llr^2*Lm+4*Rm^2*c*Lls*wr^2*Llr*Lm
+4*Rm*Rs*c*wr^2*Llr*Lm^2+2*Rm^2*c*Lls*wr^2*Llr^2+2*Rm^2*Rs^2*c^2*wr^2*Llr*Lm+2*Rm^2*c*Lls*wr^2*Lm^2+Rr^2*Lm^2
+Rm^2*Rs^2*c^2*wr^2*Llr^2
296
APPENDIX D
LIST OF PUBLICATIONS
This appendix contains the lists of authors publications during the course of this study.
Journal Publications
[1] D. Seyoum, C. Grantham and M. F. Rahman, The dynamic characteristics of an
isolated self-excited induction generator driven by a wind turbine, scheduled for
publication in the IEEE Transactions on Industry Applications, Vol. 39, No. 4,
July-Aug. 2003, pp. 936 -944.
[2] D. Seyoum, M. F. Rahman and C. Grantham, Improved flux estimation in
induction machines for control application, accepted for publication in the special
issue of Journal of Electrical and Electronic Engineering Australia (JEEEA),
Institute of Engineers Australia (IEAust), Vol. 22, No. 23, 2003, pp. 243-248.
[3] 3 D. McKinnon, D. Seyoum and C. Grantham, Novel dynamic model for a three-
phase induction motor with iron loss and variable rotor parameter considerations,
accepted for publication in the special issue of Journal of Electrical and Electronic
Engineering Australia (JEEEA), Institute of Engineers Australia (IEAust), Vol. 22,
No. 23, 2003, pp. 219-225.
[4] C. Grantham, M.F. Rahman and D. Seyoum, A regulated self-excited induction
generator for use in a remote area power supply, International Journal of
Renewable Energy Engineering, Vol. 2, Curtin University of Technology, 2000,
pp. 135-140.
Conference Publications
[5] D. Seyoum, M. F. Rahman and C. Grantham Inverter Supplied Voltage Control
System for an Isolated Induction Generator Driven by Wind Turbine, accepted in
IEEE Industry Applications Society 38th Annual Meeting, The Grand America
Hotel, Salt Lake City, Utah, USA, October 12 - 16, 2003.
APPENDIX D LIST OF PUBLICATION
297
[6] D. Seyoum, D. McKinnon, M. F. Rahman, C. Grantham and H. P. To, A novel
account of iron loss in the analysis and modeling of an isolated self-excited
induction generator, accepted in the 6
th
International Power Engineering
Conference, Singapore, 22-24 May, 2003.(due to the spread of SARES rescheduled
to Nov. 27-29, 2003).
[7] D. Seyoum, M. F. Rahman and C. Grantham, Simplified flux estimation for
control application in induction machines, IEEE- International Electric Machines
and Drives Conference (IEMDC), June 1st - 4th, 2003, Madison, WI, USA, pp. 691
-695.
[8] D. Seyoum, M. F. Rahman and C. Grantham, Terminal voltage control of a wind
turbine driven isolated induction generator using stator oriented field control,
IEEE-Applied Power Electronics Conference and Exposition, Miami Beach,
Florida, USA, February 9-13 2003, pp. 846 -852.
[9] D. Seyoum, C. Grantham and F. Rahman F., "The Dynamic Characteristics of an
Isolated Self-Excited Induction Generator Driven by a Wind Turbine", Proceedings
IEEE- IAS 2002 Annual Meeting Pittsburgh, USA, 2002, October 13-18, pp 731-
738.
[10] D. Seyoum, F. Rahman and C. Grantham, Terminal voltage control of a wind
turbine driven isolated induction generator, Proc. AUPEC02, Melbourne,
Australia, Sep29-Oct 2, 2002, ISBN 0-7326-2206-9.
[11] D. Seyoum, C. Grantham and F. Rahman, A Novel analysis and modelling of an
isolated self - excited induction generator taking iron loss into account, Proc. of
Australasian Universities Power Engineering Conference (AUPEC 2002),
Melbourne, Australia, Sep29-Oct 2, 2002, ISBN 0-7326-2206-9.
[12] D. Seyoum, F. Rahman and C. Grantham, An improved flux estimation in
induction machine for control application, Proc. AUPEC02, Melbourne,
Australia, Sep29-Oct 2, 2002, ISBN 0-7326-2206-9.
[13] D. McKinnon, D. Seyoum and C. Grantham Investigation of the effects of supply
voltage and temperature on parameters in a 3-phase induction motor including iron
loss, Proc. AUPEC02, Melbourne, Australia, Sep29-Oct 2, 2002; ISBN 0-7326-
2206-9.
[14] D. McKinnon, D. Seyoum and C. Grantham, "Novel Dynamic Model for a Three-
Phase Induction Motor With Iron Loss and Variable Rotor Parameter
298
Considerations", Proc. of Australasian Universities Power Engineering Conference
(AUPEC 2002), Melbourne, Australia, 29 Sept.-2 Oct. 2002, ISBN 0-7326-2206-9.
[15] D. Seyoum, C. Grantham and F. Rahman, An Insight into the Dynamics of
Loaded and Free Running Isolated Self-Excited Induction Generators,
Proceedings IEE-PEMD 2002, University of Bath, UK, 2002, pp. 580-585.
[16] D. Seyoum, C. Grantham and F. Rahman, Analysis of an Isolated Self-Excited
Induction Generator Driven by Variable Speed Prime Mover, Proc. AUPEC01,
Perth, Australia, 2001, pp. 49-54.
[17] D. Seyoum, C. Grantham and F. Rahman, The Dynamics of an Isolated Self-
Excited Induction Generator Driven by a Wind Turbine, Industrial Electronics
Society, IECON2001, the 27th Annual Conference of the IEEE, Denver, USA,
2001 pp. 1364-1369. (received student award)
[18] D. McKinnon, D. Seyoum and C. Grantham, Fast Algorithm for Rapid
Determination of RMS Values in a 3-Phase System, Proc. AUPEC01, Perth,
Australia, 2001, pp. 360-365.
[19] D. McKinnon D. Seyoum and C. Grantham, Rapid Determination of Fundamental
and Harmonic RMS Quantities in a 3-Phase System, Proc. AUPEC01, Perth,
Australia, 2001, pp. 73-78.
[20] C. Grantham F. Rahman, D. Seyoum, A self-excited induction generator with
voltage regulator for use in remote area power supply, IEEE Power Electronics
and Motion Control Conference, PIEMC 2000, China, Vol.2, 2000, pp. 710 - 715.
[21] C. Grantham, D. Seyoum, D. Indyk and D. McKinnon, Calculation of the
parameters and parameter variations of an induction motor and the effect of
measurement error, Proc. AUPEC00, Brisbane, Australia, 2000, pp. 225-228.
[22] C. Grantham, D. Seyoum. and H. Tabatabaei-Yazdi, Very fast and accurate
electrical measurements, Proc. AUPEC /EECON 99, Darwin, Australia, 1999,
pp. 99-103.
[23] C. Grantham, M. F. Rahman and D. Seyoum, A regulated self-excited induction
generator for use in a remote area power supply, Proc. AUPEC /EECON 99,
Darwin, Australia, 1999, pp. 438-443.
299
300
Fig. D.1 Student award

Whole PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Whole PDF

Uploaded by

Copyright:

Available Formats

THE DYNAMIC ANALYSIS AND CONTROL OF A

SELF-EXCITED INDUCTION GENERATOR DRIVEN

- wind pressure just after the wind turbine, N/m

- Volume flow rate of air per unit time, m

, remains constant, which means the product AV

is the wind at atmospheric pressure. The turbine first causes

Fig. 4.19 D-Q representation of induction machine in the excitation (e

You might also like