Thesis Fulltext

Transformer models for
electromagnetic transient studies

with particular reference to HV dc
transmission
Wade G. Enright

A thesis presented for the degree of
Doctor of Philosophy
III
Electrical and Electronic Engineering
at the
Univendty of Canterbury,
Christchurch, New Zealand.
February 1996
To my wife Jodi,
thallks for the encouragement.
ABSTRACT
In this thesis a. formulation for incorporating general magnetic equivalent circuits in
electromagnetic tran.sient programs is presented. The formulation is applied to cre-
ate new single-phase, and the first three-limb three-phase PSCAD-EMTDC program
transformer models. The new models proposed are validated with laboratory, factory,
and field data.
Problems associated with magnetically independent phase repl'csentations of three-
limb three-phase transformers are discussed. Considerable differences a rC shown ill the
behaviour of three-phase bank and three-limb three-phase transformers, particularly
for the case of ill-rush studies.
The new transformer models arc applied to: the first intervalidation of the l-1a1'-
monic Domain Analysis program with the widely accepted PSCAD-EMTDC program,
investigating the division of magnetising current harmonics between transformer wind-
ings, and the development of a new means to reduce HVde converter de side harmonics;
the smoothing transformer.
The first set of field measurements from a group connected generator HVdc con-
verter station are presented. A PSCAD-EMTDC model for the group collnection is
validated with the field data and used to investigate the transient behaviour of three-
phase bank and three-limb three-phase converter transformers. Considerable differences
are shown in the case of a single-phase grolll1d fault.
LIST OF PUBLICATIONS
Publications associated with this research work:
MACDONALD, S.J., ENRlCRT, W., ARRILLAGA, J. AND O'BRIEN, M.T. 'Test results
from Benmore HYde stat ion with generators operating in group connected mode'. Pa-
per presented to the ClGRE Joint working group 11 / 14-09, August 1994.
LISBOA, M.L.V., ENRlCHT, W. AND ARRILLAGA, J. (1995), 'Harmonic and time
domain simulation of transformer magnetisation non-linearities', Proceedings IPENZ
annual confe.rence (innovative technology), VoL 2, February, pp. 72-77.
MACDONALD, S.J. , ENRIGHT, W., ARRILLAGA, J. AND O'BRmN, M.T. ' Harmonic
measurements from a group connected generator HV de converter scheme', iEEE Trans-
actions on Power' Delivery, VoL 10, No.4, October 1995, pp. 1937-1943,
ENRIOHT, W. , ARIULLAOA, J ., WOOD, A.It. AND HIDALOO, F.P. (The smoothing
transformer, a new concept in dc side harmonic reduction of HVdc schemes'. Paper
presented at the lEEE/PES 1996 Winter meeting, BaltimOlc, Maryland, J anuary 1996.
ARRILLAGA, J ., ENRIOHT, W., WOOD, A.R. AND HIDALGO, F.P. tDC side harmonic
reduction in HVdc converters by direct ripple reinjection'. Paper accepted for presen-
tatioll at the lEE sixth international conference on AC and DC transmissioll, London,
April-May 1996.
LISBOA, M .L. V. , ENRICH'r, W. AND AnRU..LAOA, J. 'I mproved repesent ation of the
magnetic non-linearities of multi-limb transformers for use in power system harmonic
studies'. Paper accepted for presentation at t he 12th PSCC conference, Dresden, Ger-
many, 19-23 August 1996
ENRlOHT, W. , AnRlLLAOA, J ., WOOD, A.R. AND WATSON, N.R. 'Modelling multi-
limb transformers with an electromagnetic transient program'. Paper accepted for
presentation at ELECTRlMACS 1996, Saint-Nazaire, Flance, September 1996.
ACKNOWLEDGEMENTS
In this document these paragraphs al'e the last 1 write, With both happiness and
sadness 1 leave this work behind. For this thesis and the future prospects I now look
forward to, I acknowledge the following people.
'Ib my supervisor Professor J. ArriUaga, 1 wish to express my sincere thanks for
your loyalty, guidance, and friendship. I will not forget your advice Jos. Thank you also
to Dr P,S. Bodger, Dr A.H.. Wood and Dr S.J, Macdonald for your helpful discussions
during the course of this work.
1 also wish to acknowledge my mentor, Dr K.S. Turner, I am grateful for your
commitment an.d encouragement. Special recognition is due to K. Devine, N . .rl:ampton,
and the Electricity Corporation of New Zealand Ltd for the financial support of this
research. To TI:ans Power New Zealand Ltd thanks are due for technical support,
especially the help of P. Thompson.
I acknowledge the practical help and friendship given by Ken Smart. Thank you
Ken, Martha, and the boys, for offering me a home away from home.
Finally, to my postgraduate colleagues, family and friends who cared, I offer my
warmest thanks. To Tony Enright (senior) your advice throughout my university years
has been calming. To Tony Enright (junior) I hope you enjoy your years of research.
Charlie Stobcrt , thanks for being a good friend.
CONTENTS
ABSTRACT
LIST OF PUBLICATIONS iii
ACKNOWLEDGEMENTS v
GLOSSARY x ix
CHAPTER 1 INTRODUCTION 1
CHAPTER 2 A P SCAD-EMTDC MULTI-LIMB TRANSFORMER
FORMULATION 5
2. 1 Electromagnetic t ransient progra.ms 5
2.2 The state of t he ar t in component models 9
2.3 Modelling multi-limb transformers
2.4 Incremental permeance formulation
2.5 Conclusions
10
11
15
CHAP TER 3 THE LINEAR AND LINEARISED TRANSFORMER 17
3. 1 Incorporating a linear MEC in PSCAD-EMTDC 17
3. 1. 1 Interphase magnetic coupling verification 19
3.2 Incorporating a non-linear MEC in PSCAD-EMTDC 20
3.2. 1 MEC branch saturation characteristics 20
3.2.2 The linearised Norton equivalent 22
3.2.3 Comparing three-phase tnUlsformer representations 24
3.2.4 Field t ransformer saturation characteristics 26
3.3 Representing winding flux leakage 28
3.3. 1 Single-phase representation 28
3.3.2 HDA and PSCAD-EMTDC intervalidation: single-
phase 29
3.3.3 Three-limb three-phase representation 30
3.3.4 I-IDA and PSCAD-EMTDC intervalidation: t hree-
limb three-phase
3.4 Solving t he non-linear MEC
3.5 Conclusions
33
34
37
viii CONTENTS
CHAPTER 4 THE UNIFIED MAGNETIC EQUIVALENT
CIRCUIT 39
4.1 including winding flux leakage in t he MEC 39
4.2 Actual permeance formulation 41
4.2. 1 The UMBC open and short-circuit inductances 43
4.2.2 The UMEC self and mutual inductances 46
4.3 Incorporating a nOll-linear UMEC in PSCAD-EMTDC 47
4.3.1 Time step delay validation 49
4.3.2 Removal of per-unit representa.tion 50
4.3.3 The Benmore conver ter transformers
4.4 UMEC and PSCAD-EMTDC transformer model compar-
ison
4.4. 1 Laboratory verification
4.4.2 Harmonic comparison
4.5 Three-limb three-phase UMEC
4.5.1 UMEC factory data validation
4.6 Conclusions
CHAPTER 5 THE GROUP CONNECTION; MEASUREMENT
54
59
60
62
63
67
68
AND SIMULATION 71
5.1 Direct connected generator HVdc conversion 71
5.2 Ol"OUp versus unit connection 73
5.3 Harmonic tests 73
5.3.1 The Benmore HVdc Terminal 74
5.3.2 Measurement Equipment 74
5.3.3 Experimental Procedure 75
5.4 Generator rating 75
5.5 Cl"OUp connection simulation and fi eld data. comparison 76
5.6 Conclusion3 79
CHAPTER 6 THE SMOOTHING TRANSFORMER
6.1 HVdc converter de side fi lters
6.2 The smoothing t ransformer principle
6.3 Smoothing t ransfol'mer saturation
6.4 Realistic design considerations
6.4. 1 Test system
6.4.2 Steady-state performance
6.4.3 Transient performance
6. 5 Conclusions
CHAPTER 7 HVDC CONVERTER TRANSFORMER
COMPARISONS
7.1 A three-limb converter transformer
7.1.2 Trallsient comparison
81
81
82
83
85
90
91
93
94
97
97
98
100
CONTENTS ix
7.2 Benmore converter transformer comparisons 105
7.2.1 Steady-state performance 106
7.2.2 Transient comparison 106
7.3 Conventional converter transformer comparisons 108
7.3.1 Steady-state performance 108
7. 3. 2 Transient comparison
7.4 Conclusions
CHAPTER 8 FUTURE WORK AND CONCLUSIONS
APPENDIX A NORTON EQUIVALENT ADMITTANCE MATRIX
SYMMETRY
APPENDIX B PER-UNIT DYNAMIC SIMULATION
APPENDIX C LEAKAGE PERMEANCE CALCULATION
APPENDIX D TRANSFORMER RELUCTANCE/ PERMEANCE
PATHS
D.l Primary open-circuit reluctance
D.2 Secondary open-circuit reluctance
D.3 Primary short-circuit reluctance
D.4 Secondary short-circuit reluctance
D.5 Mutual inductance path permcance
APPENDIX E PER-UNIT </>-1 AND B-H SATURATION
REPRESENTATION
109
112
115
119
121
123
125
125
125
126
126
126
131
APPENDIX F LABORATORY AND SIMULATION PARAMETERS 133
APPENDIX GBENMORE GENERATOR AND
INTERCONNECTING TRANSFORMER
PARAMETERS 135
APPENDIX HMANAPOURI TRANSFORMER TECHNICAL
DOCUMENTATION 137
APPENDIX I BENMORE POLE 1A PARAMETERS 151
1.1 UMEC single-phase validation data
1. 2 UMEC converter transformer comparison data
1. 2.1 Delta-primary, 1l0kV fault
1. 2.2 Star-primary, 16kV fault
APPENDIX J SMOOTHING TRANSFORMER TEST SYSTEM
PARAMETERS
151
154
154
156
157
x
APPENDIX K CONVERTER TRANSFORMER MODEL
PARAMETERS AND DC LINE DATA
REFERENCES
CONTENTS
159
161
LIST OF FIGURES
2.1 Electromagnetic transient program Norton equivalents: (a) Inductor, (b)
Capacitor. 6
2.2 Simplified EMTDC main program.
2.3 PSCAD-EMTDC lineal' transformer model: (a) Mutually coupled coils,
(b) Steinmetz circuit.
2.4 Magnetic equivalent circuit branch.
3.1 Linear formulation implementation in PSCAD-EMTDC.
3.2 Laboratory transformer: (a) Three-limb three-phase core type, (b) Mag-
netic equivalent circuit.
3.3 Laboratory transformer single-phase excitation.
3.4 Single-phase excitation, primary winding voltagesj Solid= recorded, Dash-
dash=simulation, (a) Red-phase V" (b) Yellow-phase V
3
, (e) Blue-phase
V5
8
10
12
18
19
20
21
3
'
. 0 Non-linear formulation implementation in PSCAD-EMTDC. 23
3.6
3.7
Transformer magnetic core test system. 24
Three-phase transformers during transient operating conditions (top=three-
ph""c bank [xfmr-3p2w], bottom=three-limb [3limb-oc]), (a) and (e)
Steady state energisation and source removal, (b) and (d) Heavy satu-
ration following source reconnedion.
3.8 Field transformer core cluy:acteristics (left hand side=non-zero correc-
tion factor, right hand side=zero correction factor): (a) and (b) Satura-
tion curve, (c) and (d) Core incremental permeance.
3.9 A linear and linearised PSCAD-EMTDC single-phase transformer model:
25
27
(a) core flux paths, (b) magnetic equivalent circuit, (c) Norton equivalent . 29
3.10 Single-phase validation test system.
3.11 Single-phase bank intervalidation, red-phase comparisonj Solid=HDA,
Dash-dash= PSCAD-EMTDC: (a) primary current waveform, (b) pri-
mary voltage waveform, (c) primary current spectra, (d) primary voltage
spect ra.
30
31
xii LIST OF FIGURES
3.12 A lineal' and linearised PSCAD-EMTDC three-limb t hree-phase trans-
former model: (a.) core flux paths, (b) magnetic equivalent circuit, (c)
Norton equivalent (star-grounded/star-grounded, blue-phase only). 32
3.13 Three-limb validation test system. 33
3.14 Three-limb three-pha..')c bank intervalidatioll, red-phase comparison; Solid:::::HDA,
Dash-dash= PSCAD-EMTDC: (a) primary current waveform, (b) pri-
mary voltage waveform, (c) primary current spectra, (d) primary voltage
spect ra.
3. 15 Newton solution implementation.
3.16 MEC with winding-limb flux known.
3.17 Ma.gnetic equivalent circuit solutions (left hand side:::::non-iterative, right-
hand-side=iterative) j (a) and (b) Yoke flux, (c) and (d) Zero-sequence
34
35
36
flux. 38
4.1 A UMEC PSCAD-EMTDC single-phase transformer model: (a) core
flux paths, (b) unified magnetic equivalent Ci rcuit, (c) Norton equivalent. 41
4.2 Incremental and actual permcance.
4.3 Linear UMEC reduction: (a) bpen-circuit secondary, (b) simplified open-
circuit equivalent, (c) short-circuit secondary, (d) simplified short-circuit
equivalcnt .
4.4 Mutual inductance equivalent circuit.
42
44
46
4.5 Actual permeance formulation UMEC implementation in PSCAD-EMTDC. 48
4.6 Iteration compar ison test system. 50
4.7 Iterative ami nOll-iterative Single-phase UMEC model comparison; Solid=non-
iterativc,Dash-dot = iterative: (a) open-circuit current , (b) open-circuit
leakage flux (primaryL (c) short-circuit leakage flux (primary), (d) short-
ci rcuit leakage flux (secondary), (e) in-rush primary winding-limb, (f)
in-rush primary leakage flux.
4.8 Open-circuit single-phase UMEC solution comparison; Solid=non-iterative
t echnique, Dash-dot:::::i terative technique: (a) mmf loop sum, (b) node
flux sum.
4. 9 Real value UMEC implementation in PSCAD-EMTDC.
4.10 Benmore transformer core dimensions.
4. 11 Assumed B-H saturation characteristic.
4. 12 Single-phase field transformer test system.
4. 13 UMEC Bemnore single-phase transformer model: (a) open-circuit cur-
rent , (b) open-circuit yoke flux, (c) short-circuit voltage, (d) shor t-circui t
current, (e) rated-load. current.
51
52
53
55
56
57
58
LIST OF FIGURES
4.14 The PSCAD-EMTDC model and Steinmetz "approximate" circuit: (a)
Steinmetz Happroximated" equivalent, (b) PSCAD-EMTDC representa-
tion.
4.15 The single-phase Norton equivalent and Steinmetz "exact" circuit: (a)
core flux paths, (b) unified magnetic equivalent circuit, (c) Steinmetz,
equivalent (d) Norton equivalent.
4. 16 Laboratory experimental set up.
4. 17 Benmore generators and interconnecting transformer bank.
4.18 A UMEC PSCAD-EMTDC thl'ee-limb three-phase transformer model:
(a) core flux paths, (b) unified magnetic equivalent circuit.
XIll
59
60
60
63
65
4. 19 Three-limb three-phase Norton equivalent (blue-phase only, star-grounded/star-
grounded). 66
4.20 Manapouri transformer B-H saturation cilaracteristic; Solid=Savigliallo
data, Dash-Dash= Hypel'bola approximation.
4.21 UMEC Manapouri three-limb three-phase transformer model test sys-
tem: (a) open-circuit, (b) short-circuit.
5.1 HVde conversion: (a) conventional scheme, (b) unit connection.
5.2 Pole 1 of the New Zealand HVde link.
5.3 Harmonic comparison of recorded generator current with steady state
and dynamic simulation; Solid=recorded data, Dash-dash=simulation:
(a) 11th harmonic, (b) 13th harmonic, (c) 23rd hannonic, (d) 25th har-
monic, (e) 35th harmonic, (f) 37th harmonic.
5.4 Harmonic comparison of recorded generator voltage with dynamic simu-
lation; Solid= recorded data, Dash-dash=simul8otion: (a) 11th harmonic,
(b) 13th har roonic, (0) 23rd harmonic, (d) 25th harmonic, (e) 35th har-
monic, (f) 37th harmonic.
5.5 Group connected generator waveforms (3 machines, Idc= 405A); Solid= field
68
69
72
74
77
78
measurement, Da.'lh-dot=sirnulation: (a) currellt, (b) voltage. 79
6. 1 Ideal smoothing transformer operation.
6.2 Iron-core smoothing tra.nsformer de side harmonic voltage cancellation:
(a) linear case, (b) and (c) non-linear case n=15, (d) and (e) non-linear
case 0 = 600.
6.3 Impression of smoothing transformer section.
6.4 Harmonic voltage cancellation principle.
6.5 Primary winding self inductance; Dash-dash=NZ HV de Pole 2 de line
filter capacitor.
82
84
85
86
88
xiv LIST OF FIGURES
6.6 Inductive and capacitive harmonic cancellation solutions; (a) the positive
root, (b) the negative root. 89
6.7 Smoothing transformer gain and phase; (a) Inductive solut ion, (b) Ca-
pacitive solution.
6.8 Resonant fr equency variation; Dash-dash= L22 equal to the NZ HVdc
Pole 2 smoothing reactor inductance.
6.9 Fa.ult initiation and recovery, Pole 1A, NZ HVdc link DC line voltage;
(a) Case A, conventional system, (b) Case B, smoothing transformer
90
91
(k=0. 96). 93
6,10 Fault initiation and recovery, Pole l A, NZ HVdc link DC line current;
(a) Case A, conventiona.l system, (b) Case B, smoothing transformer
M
7,1 Benmore three-limb converter transformer core dimensions. 98
7.2 Benmore HVde converter transformer steady-state test system, 99
7.3 Benmore three-limb three-phase transformer steady-state tests; Solid=red-
phase, Dash-dash=ycllow-phase, Dash-dot=blue- phase: (a) open-circuit ,
primary winding current , (b) open-circuit , primar y winding-limb flux,
(c) short-circuit primary winding voltage, (d) short-circuit, primary wind-
ing current.
7.4 Benmore steady-state over-voltage t ransformer cnergisation; SoHd= rcd-
phase, Dash-dash= yellow-phasc, Dash-dot = blue-phase: (a) three-phase
bank primary winding current, (b) t hree-limb three-phase primary wind-
ing current,
7.5 HVde converter transformer transient test system,
7.6 Generator currents, 16kV red phase to ground fault, star-grounded/star-
grounded converter transformer; Solid= rcd- phase, Dash-dash=yellow-
phase, Dash-dot=bluc-phase: (a) three-phase bank test system, (b) three-
limb three-phase test system.
7,7 Star-grounded/star-grounded converter transformer red phase to ground
fault (left hand side=three-phasc bank, right hand side= three-limb three-
phase); Solid= red-phase, Dash-dash=yellow-phase,
(a) and (b) primary winding current , (c) and (d) primary winding leak-
age flux.
7. 8 St ar. grounded three-limb transformer flux distribution, red phase to
ground fault: (a) open-circuit core flux, (b) voltage, current and flux
phasor diagram,
100
101
101
102
103
104
LIST OF FlGURES
7.9 Star-grounded/stal'-gl'ouuded converter transformer fault recovery (loft
hand side=three-phase bank, right hand side=tm:ee-limh three-phase);
Solid=red-pha..<;e, Dash-dash= yellow-phase, Dash-dot = bluc-phase: (a)
xv
and (b) primary winding current , (c) and (d) pl'imary winding-limb flux. 105
7.10 Benmore group connection generator waveforms, DdO and Dyll three-
limb converter transformers, Idc=405Aj Solid=field measurement, Dash-
dot =simulation: (a) generator current, 3 machines, (b) generator voltage. 106
7.11 Benmore group connected HVdc converter transformer transient test
system.
7.12 DdO converter transformer, llOkV red phase to ground faultj Solid= red-
phase, Dash-dash=yellow-phasc, Dash-dot= blue-phase: (a) Three-phase
107
bank primary line cUl'l'ent, (b) Threc--limb three-phase primary line cUl'l'ent.108
7.13 Converter transformer line currents, fault recovel'y (left hand side= three-
phase bank, right hand side= three-limb three-phase); Solid= red-phase,
Da.<;h-dash=yellow-phase, Dash-dot=bluc--phase: (a) and (b) DdO pri-
mary current, (c) and (d) Dyll primary current.
7.14 DdO converter transformer, 1l0kV red phase to ground fault and recov-
ery; Solid=red-phase, Dash-dash=yellow-phase, Dash-dot=blue-phase:
(a) Three-pha.<;e bank primary winding-limb flux, (b) Three-limb three-
phase primary winding-limb flux.
7.15 Benmore group connection generator waveforms, Y yO and Y dll three-
109
110
limb converter transformers, I
dc
=405A; Solid=field measuremcnt, Dash-
dot=simula.tion: (a) generator current, 3 machines, (b) generator voltage.ll 1
7.16 Conventional converter transformer configuration transient test system. 111
7.17 Generator currents, 16kV red phase to ground fault; Solid=red-phase,
Dash-dash=yellow-phase, Dash-dot=bluc--phase: (a) three-phase bank
test system, (b) three-limb threc--phase test syfoitem.
7.18 COllverter transformer winding currents, 16kV red phase to ground fault
(left hand side= three-phase bank, right hand side= three-limb three-
phase); Solid= red-phase, Dash-dash= yellow-phase, Dash-dot= blue-phase:
112
(a) and (b) YyO primary current , (c) and (d) Ydll primary current. 113
7.19 Converter transformer winding currents, fault rccovery (left hand sidc= three-
phase bank, right hand side= threc-limb threc--phase); Solid= red-phase,
Dash-dash=yellow-phase, Dash-dot=blue-phase, (a) and (b) YyO pri-
mary current, (c) and (d) Ydll primary current . 114
8.1 Synchronous generator magnetic equivalent circuit. 118
C.l Mutually coupled coil equivalent circuit. 123
xvi
D.1 Mutual inductance equivalent circuits.
E.1 Saturation characteristic collventions.
K.l Benmore PIA de line filter design.
LIST OF FIGURES
127
131
160
LIST OF TABLES
3.1 Laboratory transformer parameters.
3.2 Test transformer hyperbola parameters.
3.3 Field tntnsformer hyperbola parameters.
3.4 Magnetising current spectra (p.ll.).
3.5 TI:ansformcr hyperbola parameters; Newton iteration and linear approx-
imation comparison.
4.1 Non-iterative and iterative test-system electric parameters.
4.2 Benmore transformer name-plate data.
4.3 Assumed B-H curve hyperbola parameters.
4.4 Single-phase UMEC model parameters.
4.5 Single-phase field transformer test-system electric parameters.
4.6 Measured and simulated primary and secondary 5th harmonic magnetis-
ing current.
4.7 Beumore magnet ising current division ratio.
19
24
26
28
37
50
54
55
56
57
61
62
4.8 Magnctising current division; Transformer represented with UMEC model. 63
4.9 Magnetising current division; 'I\'ansfonner represented with conventional
model. 63
4,10 Manapouri three-limb transformer UMEC model parameters, 68
4.11 Manapouri transformer B-H curve hyperbola parameters. 69
4.12 Manapouri three-limb transformer test system electric parameters. 69
4.13 Simulation a.nd factory data comparison. 70
5.1 Maclline stator current harmonics.
6.1 Ha.rmonic reduction at variable frequency.
6.2 Conventional filter and smoothing transformer harmonic current.
7. 1 Benmore three-limb UMEC transformer model parameters.
7.2 Benmore three-limb transformer test system electric parameters.
76
92
92
98
99
xviii LIST OF TABLES
J.l Per-unit smoothing transformer iron-core chru:acteristics. 158
.1.2 Per-unit smoothing transformer iron-core saturation characteristics. 158
GLOSSARY
The following are the main symbols used in this thesis:
B vector of branch mmf
J ~ magnetic equivalent circuit (MEC) resultant branch
magneto motive force (MMF)
Ok] MEC branch MMF drop
81.:2 MEC branch forcing MMF
intercept of branch saturation characteristic tangent
with MMFaxis
vector of nodal MMF
/-Lv permeability of free space
}1TI.: relative permeability
vedor of MEC branch flux
MEC branch flux
intercept of branch saturation characteristic tangent
with flux axis
yoke, leakage and zero-sequence flux vector
s winding-limb flux vector
harmonic voltage source angular frequency
Wo base angular frequency
wn
[AJ
[AJ"
smoothing transformer resonant frequency
branch-node connection matrix
node-branch connection matrix
AI.: core cross-sectional area
a transformer mtio
branch flux density
b
1
ordinates to the origin of saturation charactcrisitic
asymptote to Tnl
GLOSSARY
b
2
- ordinates to the origin of saturation characteristic
asymptote to m2
saturation characteristic correction term
Hk
branch magnetisillg force
[1]
identity matrLx
l ;nprim
-
primary magnetizing current harmonic
hnsec -
secondary magnetizing current harmonic

-
vector of winding current
'0
- capacitor current
' k
MEC winding cuncnt
"
inductor current
tml - red phase magnetising current
'm2
yellow phase magnetising current
'm3
blue phase magnetising current
inc
capacitor Norton equivalent current source
'n'
-
inductor Norton equivalent current source
n.
- previous time step transformer equivalent current injection
k
-
smoothing transformer coupling factor
Lk
core limb length
LLI - primary winding self inductance
L"
secondary winding self inductance
M12
two winding transformer mutual inductance
rnl
slope of saturation characteristic illlsaturated region
7n2 -
slope of saturation characteristic saturated region
Nk -
MEC branch turns number
P
k
- MEC branch incremental permeance
Pi.
UMEC bra.nch actual pel'meance
Rk
MEC branch incremental reluctance
R'k -
UMEC branch actual reluctance
R*prim
oc
single-phase open-circuit reluctance (primary)
R>I<Sec
"
single-phase open-circuit reluctance (secondary)
R prim -
transformer primary winding resistance
Rsec
transformer secondary winding resistance
R*prim
- single-phase short-circuit reluctance (primary)
"
R*sec
"
single-phase short-circuit reluctance (secondary)
R' - winding-limb reluctance
w
T.
generator annature rcsist;'Ulce
Vo
capaci tor voltage
Vd, -
dc transmission line voltage
Vdd -
dc transmission line harmonic voltage
Vh
converter dc side voltage harmonics
GLOSSARY
v, - rectifier wall bushing voltage
V
WblJ
- converter wall bushing voltage
V, smoothing transformer capacitor side winding voltage
V, smoothing transformer de line side winding voltage
Uk MEC winding voltage
v,
-
inductor voltage
v, primary voltage (red-phase)
v, secondary voltage (l'ed-phase)
V3 - primary voltage, yellow-phase
V4 secondary voltage, yellow-phase
V5 primary voltage, blue-phase
V6 secondary voltage, blue-phase
Xlprim. transformer primary leakage reactance
Xlscc transformer secondary leakage reactance
x" generator subtransicnt reactance
Yo
-
capacitor Norton equivalent admittance
Yt -
inductor Norton equivalent admittance
[Y"j
- mutually coupled transformer equivalent
admittance matrix
Zioad
load impedance
Zsource - generator impedance
Single-phase MEC
cPc core flux
primary winding leakage flux
secondary winding lcalm.ge flux
core incremental permeance
lea.kage admitta.nce
Three-limb tIme-phase MEC
1>", ,3
- winding-limb fhL"{
1>4,5
yoke flux
1>6,7,8
zero-sequence flux
cPl1 ,13,t5
cPl2,l4,l6
secondary winding leakage flux
Pl,2,3 -
winding"-limb incremental permeance
P4,5 - yoke incremental pcrmcance
P1,8,9
- zero-sequence incremental permeance
xxi
xxn
y, leakage admittance
yl mutual leakage admittance between primary coils
y/, mutual leakage admittance between primary and secondary
coils on different limbs
mutual leakage admittance between secondary coils
Single-phase UMEO
primary winding-limb flux
secondary winding-limb flux
yoke flux
<P4 primary winding leakage flux
<P5 secondary winding leakage flux
'Pi primary winding-limb permeanee
secondary winding-limb permeance
yoke permeance
P: primary winding leakage permeanee
P
s
secondary winding leakage permeance
Three-limb three-phase UMEO
<Pl ,2,'J
<P4,5,6
<P7,8, !J
<PlO,l l ,12
<P13,14
<P15,16,17
po
1,2,3
P4',5,6
p'
7,8,9
p'
10,11,!2
Pi3,14
p '
15,16,17
primary winding-limb flux
secondary winding-limb flux
secondary winding leakage flux
yoke flux
zero-sequence flux
primary winding-limb penneance
secondary winding-limb penneancc
primary winding leakage permeancc
secondary winding leakage permeance
yoke permeance
zero-sequence permeance
GLOSSARY
Chapter 1
INTRODUCTION
Converter transformer models in current lise with electromagnetic transient programs
are based on the Steinmetz Happroximate" equivalent circuit. A unifOl"lll flu.x is s ~
sumed tlu-oughout the core legs and yokes, individual winding leakages are combined
and the magnetising current is placed on one side of the resultant series leakage reac-
tance. Such arbitrary placement of rnagnctising current and lumped leakage reactance
representation is expected to produce inaccurate information of transient behaviour
and harmonics, where correct representation of winding flux leakage and division of
magnetising current between transformer windings is critical.
In the absence of better information a proposal was made by (Dommel 1975J for
using the coil closest to the core to calculate core flux, Considering the complexity of
modern transformer designs, the proximity criterion, even when construction details
are available, is difficult to implement in power systems programs.
Some HVdc schemes use the three-limb core converter transformer arrangement,
However, the Steinmetz equivalent circuits were not initially intended to model the
three-limb core and each transformer phase is commonly modelled as a magnetically
independent unit,
A three-limb three-phase t ransformer model has recently been proposed [Stuehm 1993]
which derives an electric circuit dual from a magnetic circuit equivalent that includes
air and steel flux paths. Although the core leg, yoke and zero-sequence paths are in-
dividually represented) t he individual winding leakages are lumped together and the
question of the subdivision of the magnetising components between the windings is not
resolved.
A new transformer model based on the unified magnctic equivalent circuit theory
is developed in this t hesis which removes the ambiguity of magnetising current sub-
division and permits any type of core construction. The influence of the new model
on the tra.nsient performance of HVdc converters is assessed by comparison with the
conventional PSCAD-EMTDC model. The transient behaviour of three-phase bank
and three-limb three-phase converter transformers is also compared.
Chapter 2 introduces the electromagnetic t ransient technique: the capaCitor/inductor
2 CHAPTER 1 INTRODUCTlON
Norton equivalents, and the power system nodal admittance matrix representation.
The historical development of as a specialised HYde electromag-
netic transient program, is given. The state of the art in PSCAD-EMTDC comp(}-
nent models is described, particular attention is given to the magnetically independent
transformer pha.')c representation. Present multi-limb multi-phase transformer models
are reviewed and a new magnetic equivalent circuit based electromab'11ctic t ransformer
representation proposed.
In chapter 3 the interphase magnetic coupling representation of the new three-limb
three-phase transformer representation is verified with laboratory results. The tran-
sient behaviour of the PSCAD-EMTDC single-phase bank, and new mabrnetic equiva-
lent Cil'cuit (MEC) tl'ansforrner models is compared. The new models are applied to the
first rigorous intervalidation of the Harmonic Domain Analysis progTam with PSCAD-
EMTDC. Newton's technique is applied to solve the non-linear three-limb three-phase
magnetic equivalent circuit at each time step. Iterative and nOll-iterative MEC trans-
former models are compared under steady-state over-voltage operating conditions.
Chapter 4 introdU(:es the unified magnetic equivalent ci rcui t (UMEC). A two wind-
ing transformer UMEC is used to investigate the effect of core, and fl ux leakage paths on
short , and open-circuit inductance respectively. Two winding transformer self and mu-
tual inductances are expressed in terms of branch penneance. UMEC model accuracy
is verified with laboratory data. A UMEC model of the Benmore valve-group con-
verter transformers is derived. The harmonic flows predicted by t he PSCAD-EMTDC
and UMEC transformer models are compared in a test system derived from the New
Zealand HVdc Benmore converter station. The single-phase UMEC is extended to
represent the three-limb three-phase core configuration. A model of the Manapouri
three-limb generator t ransformers is derived and verified with fac tory test data.
Cha.pter 5 introduces the group cOllnected generator HVde converter principle.
The first ever set of field measurements for such as scheme in operation are given. A
PSCAD-EMTDC model of the New Zealand HVdc group connected scheme at Ben-
more, inclusive of single-phase UMEC converter transformer representation, is verified
with field data.
Chapter 6 prcsents t he smoothing transformer as a ncw means for reducing HVdc
converter de side harmonics, at characteristic and non-characteristic frequencies, with
compact passive components. The smoothing transformer is investigated with a UMEC
transformer model, a realistic design is examined in the field data verified test system
for the Bellmore group connection. The steady-state and transient performance of the
smoothing t ransformer is compared with that of the existing de smoothing reactor and
filter ba.nk.
Chapter 7 derives a UMEC three-limb three-phase converter transformer model
that is equivalent to the validated single-phase representations. The responses of the
3
single-phase bank and tluce--limb three-phase converter transformers to a single-phase
fault are campaled. The field data verified PSCAD-EMTDC Benmore test system is
again applied and the conventional converter transformer winding configuration is used.
Finally, in chapter 8 the main conclusions to this thesis are given along with direc-
tions for future work.
Chapter 2
A PSCAD-EMTDC MULTI-LIMB TRANSFORMER
FORMULATION
In this chapter the electromagnetic transieut program underlying thcory and histori
M
cal development is presented. The state of the art in individual component models is
discussed with reference to the PSCAD-EMTDC program. Transformer models that
include core construction details are not presently represented in the PSCAD-EMTDC
prob'1'a,m. Therefore a new formulation for representing multi-limb multi-phase trans-
formers in an electromagnetic transient program is proposed.
2.1 ELECTROMAGNETIC TRANSIENT PROGRAMS
The electromagnetic transient program is now widely accepted as a tool for the simu-
lation of power system dynamics which arc sensitive to relatively fast electromagnetic
interactions between components. In addition to transient studies, the increasing num-
bel' of non-linear controllers and fast switclling components place a greater emphasis on
the application of electromagnetic transient analysis to the steady-state; here significant
waveform distortion results from continuous switching transients.
Gcneraireviews emphasising the many techniques categorised as electromagnetic
transient analysis, and the accompanying computer programs have recently been doc-
umented [Zavahir 19921 [Anderson 19951. In these theses an electromagnetic transient
program was modified to incorporate state variable analysis, and transient stability
respectively; detailed technique and computcr program historical reviews were nec-
essary to understand exactly why the new hybrid progra.ms improved ac/ de system
si mulation, The main aim of the present work is to improve modelling of transformer
non-linearities in electromagnctic transient progra.ms.
Using as a basis the algorithrn proposed by Dorrunel, m the early 1970's the
Bonneville Power Administration produced the ElectroMagnetic 'I\'ansient Program
(EMTP) for the simulation of traJlsient phenomenon in single and multi-phase networks
[Dommel 1969], 'l):apczoidal integration is used to transform inductive and capacitive
differential equations into equivalent admittances and current sources. The inductor
6 CHAPTER 2 A PSCAD-EMTDC MULTI-LIMB TRANSFORMER
differential equation
becomes
where
and
In'
v,(t) = L di ,(t)
dt
;,(t) = l'jv,(t) + i",
i ,(t - LH) + W;v(t - At)
The inductor Norton equivalent is shown in figure 2.1(a}.
ilr) i
1
(0
'1(1)1
L
Vdl)l r,
(a)
VI) iit)
",{} ,

V<(lf
y
<
(b)
(2. 1 )
(2.2)
i
,r
,
i
llc
Figure 2.1 Electromagnetic transient program Norton C<lui valents: (a) Inductor, (b) Capacitor.
The capacitor differential equation
becomes
where
and
2C
III
; (t) = Cdv,(t)
, dt
t,,, -;,(t - At) - 'f; v(t - At)
The capaci tor Norton equivalent is shown in figure 2.1(b) .
(2.3)
(2.4)
2.1 ELECTROMAGNET1C TRANSIENT PROGRAMS 7
Once inductive and capacitive branches are converted to Norton equivalents they
are combined with purely resistive branches, and the state of the system a.t time t is
described by the set of linear equations [Dommel 1969]
[Y]ii(t) = ;(t) - j (2.5)
where
[Y] nodal admittance matrix
v(t) node voltage vector
i( t) injected node current vector
I current source vector
If the vector of node voltages is partitioned into two sets, one with the unknown voltages
VA(t) and another with the known voltages (specified voltage sources) VB{t), then
equation 2.5 can be rewritten
(2.6)
The vector of unknown node voltages is found by solving
(2. 7)
EMTP is well suited for simulating large ac systems where dynamic behaviour
follows a small number of network topological changes. The program wa.s not initially
applied to frequent and variable switching case'> such as ac/dc conversion.
In 1976 Manitoba Hydro began developing the ElectroMagnetic Transient Di-
rect Current program (EMTDC). The resulting speCialised HVde simulation program
combined Dommel's technique with valve group and salient pole synchronous machine
models [Woodford et al. 1983]. A user written interface was included to process and
control voltage and current sources, and switch branch elements. Disconnected sub-
networks were utilised to maximise the speed of solution. In 1985 Manitoba HV dc
Research Center validated EMTDC simulation with transient field data from Bipoles 1
and 2 of the Nelson River dc transmission system [Woodford 1985]. Recently the Power
System Computer Aided De:sign (PSCAD) package was combined with EMTDC, this
graphical interface assists the user in setting up, running a.nd analysing simulations
[EMTDC models 1994].
The practical applications of the transformer formulations presented ill this work
are all specific to HVdc transmission. For this reason PSCAD-EMTDC has been se-
lected as the most appropriate program to implement the new modeL,> proposed. Figure
2.2 presents a simplified flow diagram for the PSCAD-EMTDC program.
8 CHAPTER 2 A PSCAD-EMTDC MULTI-LIMB TRANSFORt\1ER FORMULATION
"-
Start EMTDC.
)
Update history terms.
;
j
Run user defined
dynamics file. ii
j
Solve network
equation 2.5.
iii
j
Increment time
one time-step.
Interpolation.
;,
j
Run user defined
output file.
,
j
Write output.
,;
Does
time equal
No
finish time?
Yes
StopEMTDC.
F igul'e 2.2 Simplified EMTDC main program.
2.2 THE STATE OF THE ART IN PSCAD-EMTDC COMPONENT MODELS
2.2 THE STATE OF THE ART IN PSCAD-EMTDC
COMPONENT MODELS
9
Many individual HVde components are accurately represented by PSCAD-EMTDC
models [EMTDC user 1994]. The Bergeron and frequency-dependent distributed trans-
mission line models are used to represent overhead lines. Conductor properties and un-
balanced geometries are included in the transmission line models. The six pulse gract"
bridge model uses a phase locked oscillator for the complex co-ordination of valve firing
pulses. Time-step interpolation accurately determines individual switching instants. A
complete range of HVdc link controllers is available. FACTS devices such as Sta.tic
var compensators (SVC)) thyristor controlled reactors and thyristor switched capacitor
models have been added to the library of components.
The SVC, salient pole generator and squirrel cage induction machine models are
incorporated in state variable formi a current source and stabilising Norton resistance
provide a simple and effective interface with the electromagnetic transient program.
Electric circuits and continuous system modelling functions are interfaced. Me-
tering functions record instantaneous electrical, magnetic and mechanical variables, as
well as steady-state information on real and reactive power, phasor magnitude and
angle.
In comparison to the precision of other PSCAD-EMTDC component models, the
mutually coupled winding representation for transformcl's is simplistic and can result
in a degradation of simula.tion accuracy. The electrical equations for the two coupled
coils in figure 2.3(a) are
Mt2] []
L
di2(t)
22 dt
(2.8)
If winding resistance is included) equation 2.8 is easily manipulated into the form of
the Steinmetz "exact" equivalent circuit shown in figure 2.3(b L where
LI La - aM
I
2
and
2
a 22 - aM
1
2
This linear equ.ivalent circuit applies to individual phases only. This approxima-
tion is also appropriate for multi-phase banks of single-phase tl'ansformers where the
individual units are magnetically independent. Moreover in ac fault studies transform:
ers are usually represented with magnetically independent units [Ciesner 1971] [The
ElectriCity Council 1981] [Bickford and Heaton 1986] [General Electric Company 1987].
Non-linear magnetising and core-loss components of winding cunent are incor-
porated in the PSCAD-EMTDC transformer model with a shunt current source and
resistance respectively across one winding. Since the single-phase approximation does
10 CHAPTER 2 A PSCAD-EMTDC MULTI-LIMB TRANSFORMER FORMULATION
ilrJ
MI2
~ r )
W)!

1-" L,2
(a)
ill)
R,
L,
w!
aM'2
! r )
",
i,I')
0
OV
2W
!
Ideal
transrormer
! r )
(Rati o changer)
<>-----------'------- - --1 0:1 f----o
(b)
Figure 2.3 PSCAD-EMTDC linear transformer model: (3) Mut ually coupLed coiLs, (b) Steinmetz
circuit.
not incorporate interphase magnetic coupling, the magnetising current injection is cal-
culated at each time step independent of the other phases.
Accurate simulation of HVde eonvelter stations with multi-limb converter trans-
formers requires a new PSCAD-EMTDC transformer model. Three-limb converter
transformers are present in some existing HVdc schemes, as they can offer financial
and operational advantages. The eight converter transformers of the Villdhyachal
500MW back-to-back HVdc converter station (India) are three-limb three-phase units.
Although transportation limitations have traditionally restricted the presence of three-
limb three-phase transformers in remote HVdc systems, sealed winding site-assembled
multi-limb transformers can provide the lowest cost alternative to single-phase banks
[Flux 1971]. Moreover three-limb converter transformers have been shown to signifi-
cantly reduce the detrimental effect of dc exci tation caused by geomagnetically induced
currents [Takasu et ai. 1994].
2.3 MODELLING MULTI-LIMB TRANSFORMERS
An important contribution to t he modelling of power system tra.nsformers was made by
[Chen and Dillon 1974). Although they recognised the influence of the transformer core
and flux path leakage, t heir models were based on linear transformations. Non-linea.r
transformer magnetisation effects were given detailed consideration [Abu Nasser 1981],
but only for the case of magnetically independent phases.
The principle of duali ty has been used [Arturi 1991] [Stue1lln 1993] to take into
2.4 INCREMENTAL PERMEANCE FORMULATION 11
aCcOlmt the magl1ctising currents and core configuration when modelling multi-limb
transformers with the EMTP program. The duality based equivalents involve a large
number of components; for instance twenty-three inductances and nine ideal transform-
ers are required to represent a three-phase three-winding transformer. Recent attempts
to vaJidate duality based models [Stuehm 1993] with experimental results for a three-
legged stacked core transformer have shown considerable discrepancies (of up to 40%).
Moreover the duality method requires the use of true non-lineal' inductors in series;
this is not feasible in the EMTP [Donunelen 1987) and Stuelllll adds components to
the electric dual to isolate each true nOll-lineal' inductor.
Alternative proposals [Hatziargyriou et al. 1993] [Chen and Neudorfer 1993],
based on an equivalent inductance matrix, reduce the complexity of the equivalent
circuit. However, Hatziargyriou's method does not take into account the core non-
linearity under tl'8.nsient conditions and Chen's non-lineal' equivalent inductance matrix
requires regular updating during the EMTP solution; such updating reduces the speed
of electromagnetic transient programs.
[Leon and Semlyen1994] have recently proposed a Norton equivalent transformer
representation a.s a simple interface with an electromagnetic transient program; but do
not perform direct analysis of the magnetic circuit and select a combination of the
duality and leakage inductance representation.
Also based on the Norton equivalent transformer concept this chapter now de-
scribes a multi-limb transformer formulation derived directly from magnetic equivalent
circuit analysis. This formulation provides a method that allows detailed magnetic
equivalent circuits to be easily implemcnted in electromagnetic transicnt programs.
2.4 INCREMENTAL PERMEANCE FORMULATION
Thc representation of the magnetic equivalcnt circuit branch k, as displayed in fig-
ure 2.4, separates the branch reluctance and winding magneto-motive-force (MMF)
components. Individual magnetic characteristics can be applied to each branch. Each
transformer winding can be modelled as a lumped coil 01' on a. turn to turn bases with
single or multiple magnetic equivalent circuit branches respectively.
The non-linear relationship between branch flux and branch reluctance MMF
drop must be linearised. This non-lineal' relationship can be described by
(2.9)
The MMF of winding Nk is given by
(2.10)
12 CHAPTER 2 A PSCADEMTDC MULTILIMB TRANSFORMER FOIUvfULATION
~ ~
,
,
,
'"
'0
Figule 2.4 Magnetic equivalent circuit br anch.
Combining these two equations to give the resultant branch MMF Ok yields
Linearising equa.tion 2.9 a.bout (t - .6.t) in the time domain gives
where
fit -> 0
and
Solving equation 2.12 for the branch reluctance at the present time step gives
(2.11)
(2.12)
(2.13)
Equation 2.13 allows the linearised relationship between winding current and branch
flux to be formulated, it can be re-written as
(2.14)
where
(2.15)
2.4 INCREMENTAL PERMEANCE FORMULATION 13
Hysteresis or single valued magnetic characteristics can be represented with equation
2.14. Combining equation 2. 15 with equations 2. 10 and 2.11 gives
or
where the incremental branch permeance
1
k ~
Rk
(2.16)
(2.17)
(2.18)
The relationship in equation 2.17 is general and can be extended to any number of
branches in a multi-limb transformer magnetic equivalent circuit.
IPI (INI; - ii' - iin)
(2. 19)
Diagonal matrix [PI has individual elements of branch incremental permeance and
diagonal matrix [N] elements of branch winding turns. Diagonal elements of [N] are
zero for branches with no windings.
The lincariscd relationship between winding cunent and branch flux can be
extended to incorporate the magnetic equivalent circuit branch connections. Let the
node-branch connection matrix of the magnetic circuit be [Aj'f and the vector of nodal
magnetic drops be B
node
. At each node the flux must sum to zero, stated as
(2.20)
Application of the branch-node connection matrix to the vector of nodal magnetic drops
gives the branch MMF.
. .,
IAWnod, ~ e (2.21)
Combining equations 2. 19, 2.20 and 2.21 gives
(2.22)
Solving equation 2.22 for 9
nQde
and writing n = [PIOn gives
(2.23)
Substitution of equation 2.23 into 2.21 gives
(2.24)
Equation 2.24 is combined with equation 2.19 to give
4> = [PJ ([NJi - [AJ([Anp][AJr' [[Ar [p][NJi - [Ar nj -8n )
(11] - [P][AJ ([Ar [p][AJ) - I [Ar) ([P][NJi - n)
- [MJ ( [P][NJi - n) (2.25)
where
[M] = [IJ - [P][AJ ([AjT[P][AJ) - I [AF'
1 the branch connections of the magnetic circuit are such that two branches
carry the same flux, the matrix [M) is singular. If set of branches are partitioned into
two sets, ODe with the MMF sources r$$ and another with pure reluctance paths r$r,
then equation 2.25 can be rewritten
= [ [M"J
4>. [M"J
[M"J 1 ([ [P,J
[M"J [0]
[OJ 1
[p.J 0 <!>n.
(2.26)
The problem of possible [M] matrix singularity may be removed by inverting only
matrix partition [Mu]. Utilizing matrix par titioning equation 2.26 gives
(2.27)
Equation 2.27 relates winding current to branch flux, what is required next is to relate
branch flux to winding voltage. Winding voltage Vk is related to branch flux tPk by
Faraday's law of magnetic flux
(2.28)
Discretizing equation 2.28 with trapezoidal integration and generaiisillg for magnetic
branches with an MMF source gives
- - Cl.t I[ j
4>,(t) = 4>,(t - 6t) + "2[N,J- .,(t) + ;;,(t - 6t) (2.29)
The number of first order differential equation solutions required is equal to the number
of winding-limb branches. Residual flux can be simply included by setting each branch
flux to the residual value desired at time zero. Finally, combining equation 2.27 and
2.29, a.nd solving for winding current i" gives the Norton equi valent
;,(t) = [y"Jv,(t) +i .. (2.30)
where
2.5 CONCLUSIONS 15
[leu] = ([M,,][P,][N,]) - I ';'[N,]- '
and
'n.' = ([M,,][P,][N,]) - I (';'[N,] - lv,(t - M)+,(t-M)+[M,,]n,+[M,,]nc)
The Norton equivalent representation is a vector extension of the scalar inductor equa-
tion 2.2, it is in an ideal form for PSCAD-EMTDC implementation. The symmetric
admittance matrix [Y
ss
] is non-diagonal and thus represents mutual couplings.l Equa,..-
tions 2.9 to 2.30 are general and apply to any magnetic equivalent circuit consisting of
a finite number of branches as shown in figure 2.4.
Air as well as iron path magnetic characteristics can be applied to magnetic
equivalent circuit branches. Historically the transformer has been represented with
shunt magnetising and series leakage reactances [Steinmetz 1895J. This approximation
leads to difficulties in division of the transformer magnetising current up amongst the
excited windings [Dornmel1975J. A unified magnetic equivalent circuit will be described
in chapter 4 which overcomes this problem.
Winding copper loss representation can be included by placing series resistances
at the terminals of the Norton equivalent. Alternatively each individual transformer
winding can be modeled on a turn to turn basis with many individual maf:,'l1ctic equiv-
alent circuit branches, t his provides a more precise copper loss representation than the
lumped winding resistance.
The above formulation is based directly on a magnetic equivalent circuitj it is
specifically eoncerncd with the simulation of transients resulting from saturation of the
magnetic core and thus winding capacitance is not represented.
The formulation calculates operating flux within each branch of the magnetic
equivalent circuit at each time step. This flux is utilized to accurately determine an
equivalent resistance [SIemon 1966] that is added as a shunt across the terminals of the
Norton equivalent to represent magnetic core losses.
2.5 CONCLUSIONS
Although the need for suitable electromagnetic transient models of ac and ac/de compo-
nents has intensified, the effort spent on transformer models relative to other non-linear
elements (such as HVde converters) has been small. The PSCAD-EMTDC version of
the electromagnetic transient program has been selected to implement the proposed
transformer models because the main application intended relates to HVde transmis-
sion. The present phase independcnt PSCAD-EMTDC transformer models are suited
IThc matrix WuJ is impleulented in PSCAD-EMTDC as a network of admittances. Proof of matrix
symmetry is essential, and is given in appendix A.
to represent multi-phase banks of single-phase transformers, but not the multi-pha.<;e
multi-limb core construction.
An incremental permcance formulati on of the non-lincar behaviour of multi-
phase multi-limb transformers for use with an electromagnetic transient program has
been presented. The non-linear relationship between winding mmf and saturated
branch flux is linearised in the time-domain. The formulation utilizes direct analy-
sis of a magnetic equivalent circuit to provide a Norton equivalent electrical interface
at the transformeI" terminals. The Norton equivalent is easily implementable in the
PSCAD-EMTDC program. The transformer magnetic equivalent circuit is constructed
from a finite number of branches incorporating individual magnetic characteristics a.nd
windings.
Chapter 3
THE LINEAR AND LINEARISED TRANSFORMER
In this chapter the magnetic equivalent circuit (MEC) transformer model interphase
magnetic coupling representation is verified with laboratory results. The transient
behaviour of equivalent PSCAD-EMTDC three-phase bank and three-limb three-phase
transformer models are compared.
Single-phase and three-limb three-pbase MEC Time Domain equivalents to ex-
isting Harmonic Domain transformer representations are derived. The new models
are used to intervalidate the Harmonic Domain Analysis (HDA) and PSCAD-EMTDC
programs.
Finally, Newton's technique is applied to solve the non-linear three-limb three-
phase MEC at each time step. Iterative and nOll-iterative MEC models arc compared
for steady-state over-voltage operation.
3.1 INCORPORATING A LINEAR MEC IN PSCAD-EMTDC
The flow diagram of figure 3.1 describes the implementation of a linear MEC in PSCAD-
EMTDC simulation and forms the basis for the transformer models pl'esented in this
thesiS; it is placed in operation (ii) of the PSCAD-EMTDC main program shown in
figure 2.2.
The transformer core is divided into segments of length Lk and area A
k
. The
incremental permeance of each MEC steel bra.nch is constant and can be ca.lculated
from
J.LoJ.LrkAk )
P, = (3.1
L,
The relative permeability lJ,rk, which is fixed in a linear transformer model, is derived
from the core B-1I characteristics; typical J1.rk values for unsaturated grain-oriented
silicon-steel range from 2,000 to 70,000 [Turner 1977J. Once path permeallces are
known, the admittance matrix [YssJ can be calculated and placed in the nodal admit-
tance matrix [YJ of equation 2.5. In this case the admittance matrix is constant and
need only be calculated once during simulation. The current injections ins however
18 CHAPTER 3 THE LINEAR AND LINEARlSED TRANSFORMER
User input:
Core dimensions,
Winding turns.
Start user defined
dynamics file. ii
Time equal
to
zero?
Construct matrices
[PI and [N].
Calculate [ M I with
equation 2.25. 2
Calculate fYssl with
equation 2.30. 3
Meter previous time step
variables 1;(1-61). 4
Calculate 4>jt.l:lf) with
equation 2.27. 5
Meter previous time step
variables 6
CaJculate I"Jwith
equation 2.30. 7
Begin
network solution. ir
No
Figure 3.1 Linear formulation impLementation in PSCAD-EMTDC.
3.1 INCORPORATING A LINEAR MEC IN PSCAD-EMTDC 19
must be recalculated at each time step and applied to the current source vector 1 of
equation 2.5.
3.1.1 Interphase magnetic coupling verification
Before discussions are complicated by the addition of core non-linearity, the verification
of interphase magnetic coupling representation can be achieved with the linear imple-
mentation strategy. A three-limb three-phase transformer model was derived from the
laboratory transformer parameters of table 3. 1, and the MEC of figure 3.2{b ).1
,
.
, !
-

.
, !
-

Electric circuit parameters Magnetic circuit parameters
Rating 40 kVA
Awinding limb
0.0122 (m')
Configuration delta-star
Lwinding limb
0.350 (m)
Frequency 50 (Hz) AYQke
0.0122 (m')
Vprimary 440 (V) LYQke
0.180 (m)
Vsecondary
220 (V) 1',
2000
Nprimary
J08 (turns)
Nsecondarv
31 (turns)
Table 3.1 Laboratory transformer parameters.
,
"
. .
" !
f-
e, !
-
,
"', .
" !
.
" !
f-
-

(al (bl
,.
,
Figure 3.2 Laboratory transformer: (a) Three-limb three-phase core type, (b) Magnetic equivalent
circuit.
In this open-circuit transformer model, leakage flux paths arc ignored. The MEC
represents the core and zero-sequence magnet.ising component of transformer winding
current alonc. Permeances PI to p.." represent the transformer winding limbs, P4 and
P5 represcnt the yokes, and P
u
to Pa represent the transformer zero-sequence flux
paths. MMF sources iml to im3 represent the transformer magnetising current .
Single-phase excitation of the three-limb tI'ansfol'mer clearly demonstrates in-
terphase magnetic coupling. In figure 3.3 the transformer red-phase primary winding
ITo examine interphase magnetic coupling the labora.tory transformer is operated below rated
voltage, the effect!; of core saturation are removed. Under these conditions the zer()-sequcncc Oux
paths do 1I0t greatly influence experimental result!;, therefore has been arbitrarily set. to
O.0005P"';1I.di1l.9 hJ1lb
is energised. When a voltage Vcoupling is applied to only one transformer winding the
magnetic flux froIll the energised limb retmlls via the remaining two limbs.
q
v, ..

6)
l) v, <1
y
, .... l
"'- "'-

........ I
"'-
v. >
"'-

"'-
4I] <f
R
y
8
Figure 3.3 Laboratory transformer singl&-phase excitation.
The return path reluct ance for red-phase flux cPl via the yellow winding-limb is
less than via the blue winding-limb. The yellow-phase flux 2 is thus slightly greater
than and the blue-phase flux cP3 is slightly less than !1' The magnitudes of pri-
mary and secondary yellow-phase induced voltages are slightly gr eater than VcouptinYl
and V coupling respectively. The magnitudes of primary and secondary blue-phase
voltages are slightly less than VcouJlling and 21/1 Vcoupling respectively. Orientation of the
yellow and blue-phase windings phase shifts primary and seconda ry voltages by 1800.
Experimcntal and simulated results for a 240V excitation of the laboratory t rans-
former red-phase primary winding arc displayed in figure 3.4. As predicted, the yellow
and blue-phase voltages are phase shifted 180
0
and eac11 of them is approximately half
the magnitude of the red-phase.
The magnitude unbalance which exists between yellow and blue-phase voltages
is not obvious from the simulated results. hl this linear model the consta.nt relative
permeability Ij r=2000 results in a high branch permeance. The large winding-limb and
yoke permea.nces give only a small flux unbalance between the yellow and blue winding
limbs.
3.2 INCORPORATING A NON-LINEAR MEC IN
PSCAD-EMTDC
In t he linear model, the matrix [Y.u] was calculated once only. This is not the case for
the non-linear model. Each incremental permeance Pk changes as the magnetic bra.nch
moves around its saturation characteristic.
3.2.1 MEC branch saturation characteristics
Hysteresis will not be included in t he transfol'mer models proposed ill t his work, The
effcct of hysteresis is only a.pparent in the magnetising Cllrrents of small VA rated labo-
3.2 INCORPORATING A NON LINEAR MEC IN PSCAD-EMTDC
m
,.
,.

:0
1

,
-.
-,.
-,.
-
:0

;)
,. _lOCI
--0 0.0112 0.", 0._ 0.001 0.012 0.0" 0.011 0.01. o.ol
'lmG(I)
(a)

,.
,.
M

"
-- -
,
,
21
---
"-0 0._ 0-'101 0 __ UlI 0.01 UI2 UII 0.011 0.02
l imc(a)
... ..... ___ .. ... .... .... ..
limo (Il
(b) (c)
Figme 3.4 Single-phase excitation, primary winding voltages; Solid= recorded, Dash-
dnsh=simulatioll: (a) Red-phase VI, (b) Yellow-phase V3, (c) BhH.'-phase Vr;.
ratory transformers. Moreover it has been shown that the effects of detailed hysteresis
loop representation in HVdc converter transformers have little effect on the converter
behaviour during disturbances [Joosten et at. 1990] .
During transient conditions the transformer bra.:nclles may enter very high and
different levels of saturation. To accurately represent the transformer beha.viour dur-
ing such transients, precise representa.tion of the branch magnetising characteristics
is required. The hyperbola equation 3.2 provides a. preferred match to experimental
per-tmit characteristics in both the saturated and unsatma.ted regions [Medina and
Arrillaga 1992aJ.
where
1n] = slope of unsaturated region
m2 = slope of saturated region
(3.2)
b
1
= ordinates to the origin of asymptote to mJ
b2 = ordinates to the origin of asymptote to m2
= correction term
As correction term is increased from zero, the knee point of t he hyperbola saturation
representation is made less sharp. Solving equation 3.2 for ik and taking the positive
root gives
where
ik= - DA*(B - I'.)
A = 1nlm2
B = m, (I>, - ,p.) + m,(b, - </>k)
C = i - .(b, + I>, + E)
I'. = JB'- 4AC
DA
-_'
- 'A
The incremental bmllch reluctance is derived from equation 3.2
(3.3)
(3.4)
Although equations 3.3 and 3.4 will not represent third quadrant operating points, the
symmetrical nature of the saturation characteristic can be used for calculation in this
quadrant. The incremental branch reluctance R.k never equates to zero, and at the
origin it is equal to
(3.5)
3.2. 2 The linearised Nor ton equivalent
The flow diagram of figure 3.5 describes the per-unit implementation of a non-linear
MEC in PSCAD-EMTDC. The incremental penneance formulation modifications re-
quired for pcr-unit implementation are listed in appendix B. The previous time step
winding limb flux vector s(t - .6.t) can be ealculated from the vector vs(t - 6.t), and
stored vectors s(t - 26.t) and js(t - 2.6..t) with the trapezoidal integration
- - I'.t [ 1
</>,(t - I'.t) = ,(t - 21'.t) + 2 [N,) - ' o,(t - I'.t) + o,(t - 2M) (3.6)
The per-unit form of equat ion 3.6 is given in appendix B. Calculation of the flux in yoke
aJld zero-sequence paths ,.(t - 6.t) requires solving the non-linear MEC. [Medina 1992]
3.2 INCORPORAT1NG A MEC IN PSCAD.EMTDC 23
/
Start user defined
dynamics file. ii
Time equal
No
to
zero?
User input:
Per-unit brancb
Yes
sat urat ion curves.
/
Calculate 'Ff. for each branch Meter previous time step
from G\ initial.
I variables V;(r-6t) . 6
Construct matrix Calculate ll-6
1
) with
I PI
2 equation 3.6. 7
j
Calculate [M J with Calculate with
equation 2.25. 3 equations 3.7 to 3. 11. 8
j
Calculate [YJs I with Calculate 1i for each branch
equation 2.30.
4 with equation 3.1. 9
j
Calculate i,u wilh
equation 2.30.
5
j
/
Begin
"-
network solution. iii
Figure 3.5 Non-linear formulation implcmcntation in PSCAD-EMTDC.
24 CHAPTER 3 THE LINEAR AND LINEARISED TRANSFORMER
proposes a linear approximation to the true solution
9'>,(t - c.t) k,ko9'>, (t - c.t) - k2kOql,(t - c.t) - k3k09'>3(t - c.t) (3.7)
9'>5(t - c.t)
-
-k3k09'>,(t - c.t) - k2kOql,(t - c.t) + k,ko9'>3(t - c. t) (3.8)
9'>,(t - c.t)
-
Mt - c.t) - 9'>4(t - c.t) (3.9)
9'>,(t - c.t)
-
9'>,,(t - c.t) + <I>,(t - c.t) + 9'>5(t - c.t) (3.10)
q\g(t - c. t) - Mt - c.t) - 9'>5(t - c.t ) (3.11)
where ko=0.9, k,=0.508, k2=0.5 and k3=0.492.
Using the solved branch flux >dt - ~ t branch penneance PI.; is calculated di-
rectly from the branch saturation characteristic . Once the branch permcance is known
the per-unit admittance matrix [' sJ and current source vector in, can be calculated.
3.2.3 Comparing t hree-phase transformer repr esentat ions
The circuit as shown in figure 3.6 a.nd ideal circuit breaker action have been chosen
to highlight the differences between t he PSCAD-EMTDC three-phase bank and MEC
three-limb three-phase transformer representations.
F igul'e 3.6 Transformer magnetic core test system.
The MEC of figure 3.2(b) is again utilized to represent the open-circuit three-
limb tra.nsformer, but in this case all branches are non-linear. Table 3.2 lists the
hyperbola constants for the per-unit characteristics. The saturation characteristics for
the winding, yoke and zero-sequence paths ar c those suggested by [Medina 1992J.
Branch
ml(lincar) m2(sat)
b, b, e
Winding limb 292.38 0.67 0.0 1.03 0.0
Yoke 292.38 1. 67 0.0 1.03 0.0
Zenrsequence 6.00 1.44 0.0 0.22 0.0
Table 3.2 Test transfornler hyperbola parameters.
The three-limb three-phase model (3limb-oc) was compared with the present
PSCAD-EMTDC model of the three-phase, two-winding transformer (xfmr-3p2w) [EMTDC
user 1994] with saturation enabled. The circuit of figure 3.6 was subjected to three
sequential operating conditions specially selected to emphasize the magnetic behaviour
of the transformer. The resulting primary currents are shown in figure 3.7.
3.2 INCORPORATING A NON-LINEAR MEC IN PSCAO-EMTDC
M",--_-_--_--_--_-
. m
f\(\f
. '\ VV\ VC
. .. .--
limo (.)
(a)
"'r- ------------- --,
0.001
-
"O'OJ1,, '- .. .. ,-",,-
0. "
limo (.)
(c)
,
u

u
...

,

l- V

-,
0."

'.'
!

,

..

-,
0. ' 1
25
f\(\
\. V
"
O. '!Z o.:z.o 0.2'
lime (0)
(b)
o.u
(d)
Figure 3.7 Three-phase traDSfonners during transicnt operating conditions (top=three-phusc bank
[xfmr-3p2wl , bot tonl = tbrc(."-limb [3lilllb-oc]): (a) and (c) Steady state energisation and source removal,
(b) and (d) Heavy saturation following source reconneclion.
The first operating condition is t he steady-sta.te open-circuit energisation, which
is achieved after a O.06secs ramp up time. The resulting primary currents fm the
three-phase bank and three-limb three-phase models are shown in figure 3.7(a) and (c),
for a two aJ.ld a half cycle interval (time=O.06s-0.l1secs). The three-limb magnetising
currents show a substaJ.ltial unbalance, whereas the three-phase bank fails to show this
effect .
The second operating condition, of a two and a half cycle duration, started with
the opening of the circuit breaker (time=O.l1secs), the negative going zero crossing of
the red-phase voltage. During this interval some of the energy stored in the magnetic
circuit is dissipated in the primary load, with the primary currents decaying exponen-
tially. In the case of the three-phase bank representation, dl.le to the symmetry of the
waveforms at the point of switching, the yellow and blue cmrents decay in an identical
fashion. This is not the case however for the three-limb three-phase transformer.
At the negative-going zero crossing of the red-phase voltage (time=O.17secs) the
circuit bI'eaker is reclosed giving rise to the third operating condition. As the core de
flux has not decayed to ",ere, this case forces the transformer into deep saturation. The
phase currents are very large in comparison to the previous unsaturated state and the
results need to be displayed separately (figure 3.7(b) and (ci)). The three-limb three-
phase model experiences deeper saturation than the three-phase bank. Differences
appear also in the balance and shape of the saturated open-circuit current. Each cycle
is distorted to a greater extent than the three-phase bank.
3.2.4 Field transformer saturation characteristics
Care must be exercised when equation 3.2 is applied to represent field transformer
saturation characteristics. Magnetising currents at rated voltage are sensitive to small
changes in the hyperbola parameters, and the experimental characteristics are often
only available in graphical rather than tabular form.
The hyperbola constants selected for the transformer model comparisons of sec-
tion 3.2.3 give a two i:!lope piecewise linear saturation representation. With = 0.0 and
b
l
= 0.0 then the solutions of equation 3.2 are simplified to
(3.12)
in the unsaturated region, and
(3.13)
in the saturated region. This simple representation is not accurate enough to match
field transformer behaviour.
[Dick and Watson 1981] published pcr-unit saturation characteristics for the
winding-limb, yoke and zero-sequence flux branches of a 25MVA, 110/44/4 kV three-
limb transformer. The hyperbolas of figure 3.8{a) provide a reasonable match to this
transformer's saturation characteristics. The curve parameters are given in table 3.3
2
.
Branch
1nHlinear) 7n2(satl
b, b-,
[
Winding limb 292.38 0.076 0.0 1.280 0. 850
Zero-sequence 292.38 1.670 0.0 1.200 0.652
Yoke 6.00 1.440 0.0 0.345 0.010
Table 3.3 Field transformer hyperbola parameters.
The non-zero correction factor rounds the hyperbola knee region, and affects the
incremental permeance down to the origin. The transformer characteristics of figure
'l[Dick and Watson 19811 tabulated saturated branch penneances only. The unsaturated permeances
are a5Sllmcd to be as given by [Medina 1992]
3.2 INCORPORATING A MEC IN 27
3.8(a) arc used to show this effect. Setting the correction factors of the three curves each
to zero gives the piecewise linear characteristics shown in figure 3.8(b). The incremental
permeance of the figures 3.8(a) and 3.8(b) magnetic characteristics are shown in figures
3.8(c) and 3.8(d) respectively.

u
u
-;r
,e.,.t
<
,
p:::; 0.'
'.
,..

Y(llcc
Winding. limb
Z.m.sequence
.. , ... .. .. ..
i (p.u.)
(a)

,
00 0.2 0.' U
nUll (p.u.)
(e)

.. ,

, ..
, ..
"
u
1
,
..
! .

..
..
.,

-

,
."

!l. ,.

Y(lke
Windinglimb
Zero-sequence
.. ,

..

i (p,u.)
(b)
Zer(looSCqumce

..,
,
,
,
Yoke :
,
,
,
Windi ng limb

'.

0. 2 0.<0.' 0.. .2
(p.u.)
(d)
Figure 3.8 Field transformer core characterist ics (left hand sidc= non-zero correction factor, right
hand side=zero correction factor): (a) and (b) SatUIation cnrve, (c) and (d) Core incremental perme-
anee.
The field transformer piecewise linear (E = 0.0) and smooth (E 0.0) saturation
characteristics were each applied to the section 3.2.3 model (3limb-oc). The magnetis-
iug current spectra at rated voltage are given in table 3.4. The yellow-phase draws less
CUl'rent than the symmetrical red and blue-phases for both smooth and piecewise satu-
ration representations. However, the current drawn when the saturation characteristic
is smooth is ten times that with the piecewise charactcljstic. Moreover the piecewise
representation open-circuit current is undistorted, even at rated voltage, and a perfect
sinusoidal magl1etising current is unrealistic.
Harmonic number
1
I
3
I
5
I
7
Red-phase
{; - 0.0 0.006 0.0 0.0 0.0
{; -f 0.0 0.072 0.027 O.OlD 0.003
Ycllow phasc
{; -0.0 0.005 0.0 0.0 0.0
{; -f0.0 0.058 0.016 0.009 0.003
Blue-phase
{; 0.006 0.0 0.0 0.0
[ -f0.0 0.072 0.027 0.010 0.003
Table 3.4 Magnetistllg current spectra (p.Il,).
3.3 REPRESENTING WINDING FLUX LEAKAGE
3.3.1 Single-phase representation
Before describing the flux leakage in MEC three-limb three-phase transformer models
a single-phase description is first given. In the Harmonic Domain a. single phase model
has been proposed that superimposes the transformer linear leakage reactance network
on a linearised magnetising current Norton equivalent representation [Acha et at. 1989].
A Time Domain version of this Harmonic Domain model can be derived for
PSCAD-EMTDC. The formulation is applied to the calculation of the core magnetising
current only. In the simple MEC of figure 3.9(b) the incremental permeance formulation
reduces to the per-unit scalar form [Lisboa et al. 1995J
(3.14)
where
P
- l b.two
Yss = c 2
and
ills = p;: l - + is(t -
It has been suggested that the core flux c(t) should be calculated using the
voltage vs(t) from the winding closest to the core [Domme! 1975]. In the absence of
construction details the primary and secondary voltage average V,t
V2
is often used to
calculate core flux. The magnetising current i.,(t) is halved and placed on both sides
of the transformer leakage admittance network. The resultant linear and linearised
electromagnetic transient equivalent to the Harmonic Domain transformer model is
shown in figure 3.9(c).
/S.
v, 1

Magnetic circuits
e,

I,

", 1
<>'-
V
Single-phase transfonner
(a)
Magnetic equivalent circuit
(b)
Electric circuit
I,
EMTDC implementation
(c)
29
Figul'e 3.9 A linear and lincariscd PSCAD-EMTDC single-phase transfornlCf model: (a) cure nux
paths, (b) magnetic equivalent circuit, (c) Norton equivalent.
3,3,2 HDA and PSCAD-EMTDC intervalidation: single-phase
Considerable progress in steady-state analysis of the non-linear power system has been
achieved in the Harmonic Domain [Arrillaga ct at. 1995J. Although highly efficient for
harmonic analysis, the accuracy of the Harmonic Domain Analysis program (UDA)
is yet to be rigorously validated. [Acha 1988] compared field data from the Jaguarar
Tarquaril tra.nsmission circuit with early Har monic Domain simulat ion. However the
comparisons were presented as waveforms only, whereas spectra are required to gauge
the accuracy of t he harmonic solution.
The electromagnetic transient model presented in section 3.3. 1 is an exact equiv-
alent to the Harmonic Domain single-phase transformer model. If these two models
are applied to the same test system, and simula.ted with PSCAD-EMTDC and HDA
accordingly, then the Harmonic Domain solution can be compared with that of t he
well accepted electromagnetic tra.nsient technique. The test system of figure 3.10 was
selected for this intervalidation.
The star-groundedfstar-grounded t hree-phase bank of single-phase transformers
is fed from an infinite busbaJ. through an impedance Ts = 1.0 p.u. This large source
1;. =I.Op.u.
X, =0.084 p .lI.
Figure 3.10 Single-phase validation test system.
impedance introduces voltage distortion at the transformer terminals. The source volt-
age is set to 1.1 p. u. to force the transformer into deep satmation. The transformer
leakage reactance is xl =O.084 p.ll.) and the saturation characteristic is represented with
the winding-limb hyperbola of table 3.2.
Figure 3.1l(a) illustrates a full period of the magnetising current derived with
HDA and PSCAD-EMTDC simulations, and figure 3.11(c) the haJ:monic spectra. The
results show good agreement, the maximum magnitude difference being 0.002 p.u.
occurring at the third harmonic. The transformer primary voltage waveforms are il-
lustrated in figure 3. 11(b) ! the HDA and PSCAD-EMTDC waveforms almost match
perfectly. The harmonic spectra illustrated in figure 3.11(d) indicate very close results!
the maximum magnitude difference being 0.002 p.u. at the third harmonic.
3.3.3 Three-limb three-phase representation
[Medina and Arrillaga 1992bJ extended the single-phase Ha.rmonic Domain transformer
models proposed by [Acha 1988J to represent three-limb three-phase core configurations.
The MEC of figure 3.12(b) is linearised! converted into a Norton equivalent and super-
imposed on the [Chen and Dillon 1974J common core linear short-circuit admittance
matrix. A Time Domain version of the Harmonic Domain three-limb three-phase linear
and linearised model can also be derived for PSCAD-EMTDC.
The primitive electric parameters of figure 3.12(0.) are represented by
I,
YI - YI
YI
1/'
"
VI vV
V
,
h -YI YI
yj'
vt V:' vt
V,
1,
YI Y;' V, - YI
YI
V;I
V3
(3.15)
-
'"
, "
1111
1,
vI' V, - YI V, Y, Y,
V,
Is YI " VI
yi yl' VI
- v, Vo
10 yl'
'"
y,
" V, vt
-YI YI V6
where:
YI
leakage admittance,
YI
- mutuallea.ka.ge between primary coils,
V:
I
mutual leakage between primary and secondary coils on different limbs!
1111
Y,
mutual leakage between secondary coils.
3.3 REPRESENTING WINDING FLUX LEAKAGE 31

,

"
,
f\

,
"
-.
'.
0.002
.-

o .... s

1
!
o.cl
i': 0.01S
...
..
"
,
- IIDA
-- EMTDC
V
.-
om 0.012 0.01. O.olf
time(s)
(a)
l.et't_IIDA
RightEMTOC
I!
"
h:mllonic order
(c)

..

0.01. .m

.ro
,

"
,
"

..
.ro
,
" "
"
- IIDA
- - Et.-ffDC
G.OOl' G.OM 0._ 0.001 0,01 0.012 0.01. 0.011 0.011 O.Ol'
lime (5)
(b)
Left HDA
Risht-EMTOC
;
,
I
,
l
I
,
,
;
,
,
,
!
l
,
-

,

" " "
It.lnnonic oro.,
(d)
Figure 3.11 Single-phase bank intervaliclation, red-phase comparison; Solid=HDA, Dash-
dash=PSCAD-EMTDC: (a) primary current waveform, (b) primary voltage waveform, (c) primary
CUIrent spectra, (d) primary volt age spectra.
In the MEC of figure 3.12{b) the flux in the prirnary and secondary windings
on each transformer limb is the same. Therefor e each limbs per-unit primary and
secondary winding voltage is the same. The MEC Norton equivalent is thus
[i,,(t) ]
[YII
Y12
YI'] [VI(t)]
['n" 1
i,,(t)
- Y12 Y22
Y23 1l3{t)
+
(3.16)
i,,(t)
Y13 Y23
Y33 V5 (t) 'lnsb
[Yll
Y12
YI'] [V2(t) ] [in" ]
YJ2 Y22 Y23 V4{t) + lnsy
(3. 17)
Y I3 !/23
Y33 V6(t) insb
If the Norton equivalent is halved and placed 011 both sides of the leakage admittance
matrix, the linear and linearised three-limb three-phase transformer represent ation is
32 CHAPTER 3 THE LINEAR AND LINEARJSED TRANSFORMER
Three-phase throo-limb tr:msformcr
(,J
Magnetic circuits
'.
Electric circuit
EMTDC implementation
(,J
,
, ,

,
,
,

"
;

"
"
,.

'"
+ ,.
+
Magnetic cquivi\lenl circuit
(bJ
Figure 3.12 A linear and linearised PSCAD-EMTDC three-limb three- phase transformer model: (a)
core flux paths, (b) (c) Norton equivalent (star-grounded/star-grounded,
blue-phase only).
i1 (t) VI (t) "=
2
i,(t) V2(t) "=
2
i3(t) V3 (t )
i
n3N
[Yt]
+
2
(3. 18)
i4 (t ) V4 (t)
i n y
2
i,(t) V, (t)
;...
2
is(t) V6(t)
;...
2
where
YI + -YI
'+ ' YI '21112 Y;' tit +
-YI YI +
tltl
11/ + 1
y/ 2YI2
y:'
tl+l y','
[YII y, ;Y12
YY vt + !V12 - YI Yt + !Y22 y:1
YI +
-y,
tit + 41123
yl'
Y;" + tY13
Yl'
"'+' y, 2'Y23
,
'+ ', y," ,'+', V',' ,+"M YI '2Y13 Y/ '2Y23 YI 21:1,,3 -Yt
V,I yt + !Y13 V? yt + -YI YI +
33
The PSCAD-EMTDC circuit implementation of equation 3.18 for the blue-phase
of a star-grounded/star-grounded transformer connection is shown in figure 3.12(c).
3.3.4 HDA and PSCAD-EMTDC intervalidat ion: three-limb three-
phase
The intervalidation of the HDA program and PSCAD-EMTDC performed in section
3.3.2 showed acceptable agreement for a simple test system. The liew Time Domain
three-limb three-phase transformer model is again equivalent to a Harmonic Domain
representation [Lisboa. et al. ]. A more detailed test system can be constructed with
the three-limb models, and a. more valuable intervalidation performed.
The test system of figure 3.13 has been selected to compare the flDA solution
with that of the PSCAD-EMTDC program. In this example the star-grounded/stal
grounded transformer is a thr ee-limb thrce-phase unit. Unlike the previous three-phase
bank case, the three-limb transformer magnetising cunent unbalance increases the
complexity of the steady-state solution. The supply inductance and shunt ca.pacitance
are tuned to resonate at the third harmonic, the shunt r esistance is added to provide
damping. Any third harmonic cmrent required to energise the transformer core must
fl ow in the damping resistor and will Significantly distort the transformer primary
busbar. The inductance X
s
, capacitance q and resistance rs values are 0.1 p.u., 0.9
",.={J.I p.u.
"'/ p.M.
Figure 3.13 Three-limb validation test system.
p.u., and 1.0 p.u. respectively. The transformer hyperbola sa.turation characteristic
parameters are given in table 3.2, leakage reactance xj=O.084 p.u. and the mutual
leakage terms are set to xi=x:'=x:lI=lOO p.u.
3
The source voltage is set to 1.1 p.u. to
force t he transformer into deep saturation.
Figure 3.14{a) illustrates the source current obtained with HDA and PSCAD-
EMTDC solutions, and 3.14{c) the harmonic content. The transformer primary side
3In the absence of further information the Dlutllnl if!a.(.'lge terms have been set as suggested by
[Medina and Arrillaga 1992a)
,
~
" ! .
13 .. 1.5
_I .t
- IIOA
--"MTOC
-20 o.oot 0,000 0._ 0._ 0.01 O,OIZ 0.<11. 0.01 ' 0.01 ' o.G2
time (I )
(a)
"', ------ - - - - - ----,
...
.. Left .. HDA
Righi _ EMTOC
I

"
" "
, . ,
"

harmonic order
(e)

. ..
."
';'0.111
E,
roo..
U N
. a
'M
'.O.
"
- I IDA
-- EM'llX
0. .00:1 0._ 0. _ O, OOt 0,0 ' 0. 012 0.0" 0.'" 0.0" 0.<12
time (.)
(b)
!
I 11
Left !IDA
Ri&bt .. EMTOC

,
II 13
" Ilarmonle ()I' d ...
(d)
Figure 3.14 Three-limb three-phase bank intervalidati oll, rcd-phase comparison; Solid= HDA, Dash-
dash= PSCAD-EMTDC: (a) primary cnrrent waveform, (b) primary volt age waveform, (c) primary
current spectra, (d) primary voltage spectra.
voltage waveforms and harmonic content are shown in figures 3.14( b) and 3. 14(d) (only
red-phase results are shown) . In this case, a significant level of voltage distortion is
apparent (the third harmonic volt age amplitude is nearly 10%). An almost perfect
mat ch between the voltage waveforms is achieved. Considering the diffi culties of the
test system analysed, t he results indicate very good agreement between the l-lDA and
PSCAD-EMTDC solut ions. The maximum magnitude differences a.re 0.003 p.u. for
the voltages and 0.009 p.u. for the currents, both occurring at the third harmonic.
3.4 SOLVING THE NON-LINEAR MEC
The equations governing the MEC solut ion are non-linear and precise calculation of
yoke and zero-sequence flux r (t - 6.t) from wi nding-limb flux s (t - 6.t) requires
an iterative procedure. Newt on's technique can be applied to t he MEC model to
iteratively find a t rue magnetic core solution at each time step. Figure 3.15 shows the
3.4 SOLVING THE NON-LINEAR MEC 35
modifications required to place Newton's technique in operation box 8 of figure 3.5.
Calculate Z/r-ill) with
equation 3.6. 7
Calculate 11 F( i,) 112 with
equati on 3.29. 8,
II IF,) II, <
No
E
Ye,
Calculate 3Jt- fl.t) from Calculate [J
I
J with
iT with equatio[\ 3.2. 8e equation 3.26.
8b
j
Calculate 11 for each branch Solve s" from
using C(luation 3.1. 9
equation 3.28.
8c
I
Update iT with
equation 3.27. 8d
Figure 3.15 Newton solution implementation.
As discussed in section 3.2.2 the previous time step flux in the winding-limbs
is calculated with trapezoidal integration. Figme 3.16 presents the three-limb
core MEC when the winding-limb fluxes 4>s(t - b..t) are known. Branch ampere-turns
i4 and i5 lepresent the yokes, and iGl i7 and is represent the zero sequence paths. If
the saturation characteristic representing the yoke paths is expressed as
(3.19)
and the zero-sequence paths as
(3.20)
+
..
+
.,
+
,
" \
p, ,
',I
p,
,
',I
p, ,
Figur e 3.16 MEC with winding-limb fl ux known.
then the non-linear system of equations that describe the MEC aTe
FJ (i4! i s! i6! i7! is) - >PI - /(i4) - 9(i
6
) = 0 (3.21)
f':2 (i4, is, i6. i7, is) >P2 + /(i4) + /(i5) - 9(i7) = 0
(3.22)
F3 (i41 is, i6! i 71 i s) - >P3 - /(i8) - g(i5) = 0 (3.23)
FIj (i <\ , i 5! in! i
71
i8) - i4 - i s + i7 = 0 (3.24)
Fs(i4, i5. is. i7. is) -
is - is + i7 = a (3.25)
6 for the five unknowns iT = [i4 is is i7 iaP' is

0
- '{Iio)
0 0
'"

-6/Pr, }
0
-';I
i
,)
0
" "
(Jd = 0

0 0
-6:Ji
lJ
)
(3.26)
"
"
1 0 - 1 1 0
0 1 0 1 - 1
The previous time step solution is an appropriate st arting point for the iterative
procedure required to solve the unknown variables ir . If the vector iT does not provide
a suitable solution to the system of equat ions FCi'T) = 6 then Newton's method states
that
":n+ l -:n - n
=t,. -8 (3.27)
will be a better solution, where sn is obtained from
(3.28)
and 11 indicates the iteration number at each time step. The new solution is tested
with t he two norm error function 11 2 and iterations cease if the error is less than
3.5 CONCLUSIONS 37
the threshold E. The error function is written as
(3.29)
Figure 3.17 displays one cycle of yoke and zero-sequence fiux for a three-limb
three-phase transformer excited at 1.1 (p.ll.) voltage. The winding-limb, yoke, and
zero-sequence path hyperbola saturation characteristic parameters are given in table
3.5
4
Branch
mJ(li7l.ear)
m'('"'i
&, b,
Winding limb 292.38 0.67 0.0 1.03 0.85
Yoke 292.38 1.67 0.0 1.03 0.001
Zero-sequence 6.00 1.44 0.0 0.22 0.001
Table 3.5 Transformer hyperbola parameters; Newton iteration and linear approximation compari-
son.
1n the Newton solution the yoke limbs saturate, the permeance of the saturated
yoke is less than that of the unsaturated zero-sequence path, and the winding limb
flux returns via the zero-sequence path. Therefore less yoke flux is predicted by the
linear approximation than by the Newton solution, this is shown in figures 3.17(a) and
(b) respectively. The linear approximation also gives more zero-sequence flux than
the Newton solution, as shown in figures 3.17(c) and (d) respectively. Inaccura.te yoke
and zero-sequence flux waveforms must eventually result in unacceptable transformer
magnet ising currents.
3.5 CONCLUSIONS
A magnetic equivalent circuit (MEC) of the three-limb three-phase transformer core
type has been developed. The linear transformer model derived from this MEC has
been validated with experimental results and the model interphase magnetic coupling
has been verified.
Saturation has been incorporated using a hyperbola that matches experimentally
obtained saturation characteristics. The steady-state energisation, flux decay, and
in-rush behaviour of the conventional PSCAD-EMTDC single- phase bank and MEC
three-limb three-phase transformers have been compared and considerable differences
found, particularly in the simulation of in-rush transients.
Winding flux leakage has been added to a single-phase MEC model in a manner
that is equivalent to the present Harmonic Domain Single-phase representation. The
4The hyperbola para.meters are those of table 3.2 with modified correction factors. The rounded
winding-limb, and piecewise linear yoke a.nd zero-sequence characteristics are lIseful in t.his p..xaUlple to
clearly distinguish bcLwcen each paths unsaturated and saturated regious.
38 CHAPTER 3 THE L1NEAR AND LINEARlSED TRANSFORMER
-,
0.01 O.il2 OJ" O.iU un o.n Q,in 0,1 O.H' a.tit

(a)
(e)
0.'
-,
o.n O,ut UN 0.I0Il1 O.H 0-"2 0,* O.in o.nl
ti.me(s)
(b)
<.,
< " " ... ",Cou--.:iio",,,c:;--,;,,,",,,,,,o:ii.--1 0_" 0.012 UN U I 8 Mit O.QII 0.= 0._ 0._ O.gog,
time(s)
(d)
Figure 3.17 Magnetic equivalent circui t solut ions (left hand sicle=non-itcrativc, right -hand-
side=it erative) i (a) and (b) Yoke flux, (c) and (d) Zero-sequence flux.
magnetising current has been linearised and the resultant Norton equivalent super-
imposed on the linear leakage admittance network. Intervalidation of the HDA and
PSCAD-EMTDC programs has been carried out and an acceptable match obtained for
over-voltage open-circuit transformer energisation.
The new formulation ha..c; also been applied to the calculation of the three-limb
three-phase transformer magnetising current. In this case the resultant Norton equiva-
lent was combined with a three-limb three-phase linear leakage network. This network
superposition is equivalent to the present Harmonic Domain three-limb three-phase
transformer modeL A more challenging intervaJidation was then presented using the
three-limb three-phase transformer equivalents and a resonant test system. Again an
acceptable match was achieved.
A MEC solution was obtained using Newton's technique and the linear solu-
tions inaccuracies in predicted yoke and zero-sequence flux were presented. The linear
approximation and Newton technique flux waveforms were dissimilar.
Chapter 4
THE UNIFIED MAGNETIC EQUIVALENT CIRCUIT
In this chapter the unified magnetic equivalent circuit (UMEC) is described. The
UMEO branches represent winding flux leakage as well as core and zero-sequence flux
paths. A linear two winding UMBO is applied to investigate the effect of core and
flux leakage paths on open and short-circuit inductances respectively. Self and mutual
inductances are expressed in terms of individual branch permeance in a similar manner.
The transformer Norton equivalent is rederived in terms of actual rather than
illcrementa.l permeance. Non-linear UMEC iterative and nOll-iterative solutions ar e
compared. The previous time step approximation of the present time step UMEO
solution is shown to be acceptable.
The per-unit system is then removed and a. model derived for the Benmore valve-
group converter transformers. Model accuracy is verified with experimental data. A
comparison is made between the harmonic flows predicted by PSCAD-EMTDC and
UMEC transformer models for a test system derived from the New Zealand HVdc
Benmore converter station. Finally the single-pbase UMEC is extended to represent
the three-limb three-phase core configuration. A model is derived for the Manapouri
generator transformers and is verified with factory test data.
4. 1 I NCLUDING WINDI NG FLUX LEAKAGE I N THE MEC
A century has passed since Steinmetz presented the theory of the general alternating
current transformer [Steinmetz 1895]. Now, text books describe transformers with the
shunt magnetizing and series leakage reactances of the Steinmetz "exact" 1 and (lapprox-
irnate" equivalent circuits [McPherson 1981]. The linear and linea.rised transformer
models presented in this thesis so far are examples of the Steinmetz "approximate"
equivalent circuit. For example, in the single-phase model, lUliform flux throughout
the core legs and yokes is assumed, individual winding leakages are combined and
half of the magnetising current is placed on each side of the resultant series leakage
reactance. Such ar bit rary placement of magnetising current and lumped leakage r p ~
resentation was well suited to the open-circuit examples of sections 3.2.3, 3.3.2, and
40 CHAPTER 4 THE UNIFIED MAGNETIC EQUIVALENT CIRCUIT
3.3.4. However, transformer magnetising currents can playa large part in both power
system t ransients and harmonic levels, under loaded as weil as open-circuit conditions;
UMEC transformer models will be shown to represent winding flux leakage and the
division of magnetising current between windings with improved accuracy.
Often in power system modelling the open-circuit admittance of a linear single-
phase transformer is split into two equal parts [Chen and Dillon 1974]. [Dommel 1975]
raised the question that, if uniform flux throughout the core legs and yokes is assumed,
which winding should be utilized to calculate core flux? In the absence of better in-
formation, a proposal was made for using the coil closest to the core. Considering
the complexity of modern t ransformer designs, the proximity criterion, even when con-
struction details are available, is difficult to implement in power system programs.
The most recent EMTP multi-limb transformer model derives an electric cir-
cuit dual from a magnetic circuit equivalent that includes air and steel flux paths
[Stuehm 1993J. Although the core leg, yoke aJld zero-sequence paths arc individually
represented, the individual winding leakages are lumped together and the question of
which side to place the magnetizing components is not aJlswered.
The need for a proximity criterion, the arbitrary division of magnetising current
as well as uniform core flux assumptions and lumped leakage reactance appr oximations
are removed using the UMEC concept. The single-phase transformer, shown in figure
4.1{a), can be represented by the (UMEC) of figure 4.1{b). MMF SOurces Nli1{t) and
N2i2(t) represent each winding individually. Winding voltages VI (t) and V2(t) are used
to calculate winding limb flux (h (t) and (!>2(t) respectively. Winding-limb flux divides
between leakage and yoke paths. Uniform core flux is not assumed.
Although single-phase transformer windings are not generally wound separately
on different limbs, each winding can be separated in the UMEC. Permeances PI and
'P
2
represent the transformer winding limbs. If the total length of core surrounded
by windings Lw has uniform cross-sectional area A
w
, then UMEC branches 1 and 2
have length LI = L2 = Lr and cross-sectional area Al = A2 = Aw. Permeance 'P3
represents the transformer yokes. The upper and lower yokes are assumed to have the
SaJne length Ly and cross-sectional area Ay. Both yokes are represented by the single
UMEO branch 3 of length La = 2Ly and area Aa = Ay. Leakage penneances arc
obtained from the open and short-circuit testing method described in appendix 0 aJld
therefore the effective lengths and cross-sectional areas of leakage flux paths are not
required to calculate leakage permeances 'P4 and 'Ps.
Using the UMEC of figure 4.1(b) equation 2.30 becomes
[
i1{t)] = [Yll
i,{t) Y'l
(4.1 )
The matrix [Ysal is symmetric and the Norton equivalent is implemented in PSCAD-
4.2 ACTUAL PERMEANCE FORMULATION
Magnetic circuits
<I>
/..
.,
ill:\

',1

',1

l-
I(-
.!-
l/
P,
+
Nil/I}
~ I J
<,
"' " Y>z(tJ J
P,
N l ~ / I J
Si ngle-phase transformer
(a)
Unified magnetic equivalent circuit
(b)
Electric circuit
EMTDC implementAtion
(0)
41
FiguL'e 4.1 A UMBC PSCAD-EMTDC single-phase transformer model: (a) core fl ux paths, (b)
unified magnetic equivalent ci rcuit , (c) Norton equivalent.
EMTDC as shown in figure 4.1(c). Unlike the linear and linearised representation of
cha.pter 3 all components of the transformer model are now derived from t he UMEC.
4.2 ACTUAL PERMEANCE FORMULATION
For modern grain-oriented core steel, the - 8 characteristic becomes steeper below
saturation, flaUm' above satura.tioD, and t he transition around the knee region becomes
sharper. Presently, core non-linearity is represented in incremental permeance form.
Equation 2.14 can be rearranged to
(4.2)
This magnetic operating point representation is displayed as line (a) of figure 4. 2. As
t he transformer core moves around the knee region , the change in incremental pet'me-
ance is much larger and more sudden (especially in the case of highly effi cient cores)
t han the change in actual permea.nce. Although incremental permea.nce reliance is ac-
42 CHAPTER 4 THE UNIFIED MAGNETIC EQillVALENT CIRCUIT
tPt,(f) ..
,
"
,
,
,
,
"
,
.-
/"--- (a) Slope 'P
t
= incremental permeancc
/ (b) Slope 1T = actual permcance
, ,
9
Figure 4.2 Incremental and actual permcance.
ceptable in the steady-state transformer models of the Harmonic Domain, [Chen and
Ncudorfer 1993] have recently shown modern transformer dynamic simula.tion requires
actual permeance based magnetic circuit representations.
If the UMEC branch flux is expressed using actual permcance,
(4,3)
as line (b) of figure 4.2, then equa.tion 2.17 becomes
(4:4)
which, in vector form,
J, = [P']([N); - ii') (4.5)
represents all the branches of a. UMEO. Applying the identical matrix manipulations
that gave equa.tions 2.20 through to 2.25, converts equation 4.5 into
J, = [M' ][P'IIN]i (4.6)
where 1
[M'] = [I] -[P'JlA] ([AjT[P' JlA]) - [AjT
Partitioning the vector of branch flux into the set that contains the branches associ-
ated with each transformer winding 4>., and using tra.pezoidal integration (2.29), gives
the Norton
4.2 ACTUAL PERMEAJ,'CE FORMULATION
where
= (lM;,lIP;lIN,l) - I
and
= (lM;.lIP;lIN,l) - I ( 'IN,l- I'-;,(t _ Llt) + 4>,(t _ Llt))
43
(4.7)
The admittance matrix is again symmetric. The structure of the [Y.:!] and [Yu]
(equa.tion 2.30) matrices is identical, only the permeance matrix element magnitudes
'Pii and Pii are dissimilar I and the appendix A symmetry proof can equally be applied
to
4.2. 1 T he UMEC open and short - circuit inductances
The Steinmetz "approximate
ll
equivalent cir cuit is based ou the assumption that ex-
perimentally measured open-circuit, and short-circuit inductances result only from core
fLux paths, and leakage flux paths respect ively. Before core non-linearity is added, the
UMEC can be applied to derive open and short -circuit inductances taking into account
both the core and leakage flux paths. Inductances are presented in eitber actual branch
permeance Pi; or reluctance 'Ri: form depending upon which representation yields the
simplest derivation.
Figure 4.3(a) displa.ys the UMEC if the seconda.ry winding is open-circuit. This
magnetic network can be simplified to that of figure 4.3(b) which includes a. single
MMF N)i
1
(t) and au equivalent l/Jt{tLpath reluctance
.. prim _ (Rj'R4: + nin; + n;'R4:)(n; + 'R.;) + ('Rj + (4.8)
n
oo
- (n; + n;)(n; + n;) + Rinl
Similarly, if the primary winding is open-cin.'llit and the secondary winding energised,
the UMEC can be reduced to a single MMF N2 i2 (t) and equivalent fl( t) path relucta.nce
11."" = (Rins + Rin; + n;n;)(n; + 11.4) + (Ri + n,)n;n; (4.9)
oc (7<.; + 1?.s)('Rj + 'R4:) + 'Rj'R4:
Equations 4.8 and 4.9 a.re derived ill appendix D, sections D.l and D.2 respectively.
In so far the UMEO is linear and each path reluctance 'R.; is time invariant. The
inductances measured from open-circuit primary and secondary open-circuit tests a.re
(4. 10)
and

'R;
",.
n; ____ _
(.) (b)
Open-circuit test
"
",,(Ij
"

"
",.,
'R;
",',(1}
"
(,) (d)
Short-circuit tast
Figure 4.3 Linear UMBO [eduction; (a) open-circuit se<;ondary, (b) simplified open-ci rcuit equiva-
lent, (e) short-circuit secondary, (d) simplified shorl-circuit equivaJent.
(4.11)
The linea.r single-phase UMEC winding-limb reluctances arc assumed equal,
'R.j = 'R.2 = 'R.:'. Although steel cored transformer leakage pa.th reluctances may
not be equal, they will be considera.bly la.rger than winding-limb and yoke path re-
luctances, ('R.l =1= n;) Rj). In a linear steel-cored single-phase transformer
model, equations 4.8 and 4.9 can therefore be simplified to
R*prim - n*8CC - 2n' + n'
oc-oc-w 3 (4. 12)
4.2 ACTUAL PERMEANCE FORMULATION 45
The primary and secondary open-circuit equivalent reluctances are equal and thus the
inductances and L!ce<! (referred correctly by the squa.re of the tra.nsformer turns
ratio) must also be the same.
In the case of air or pa.rtially cored transformers, clissimilal' leakage path reluc-
tances may be of the same order of magnitude as winding-limb and yoke path reluc-
tance (n: "# n
5
) (n:, nil. Under these conditions, even if winding-limb reluc-
tances equate 'R.i = n; = the open-circuit equivalent reluctances are not equal,
-=F In this case the leakage flux paths significantly influence the primary
and secondru'Y open-circuit tests. Inductances and (referred by the square
of the transfOl"mer turns ratio) are different.
When the secondary winding is shorted, figures 4.3{c) and (d) display the re-
sultant UMEC and simplified equivalent respectively. The simplified circuit contains
MMF N1i l (t) and equivalent l(t) path reluctance
(4. 13)
Similarly, when the primary winding is shorted and the secondary energised, the UMEC
can be reduced to the single MMF N
2
i2(t) and equ.ivalent tf>2(t) pa.Lb reluctance
(4. 14)
Equations 4.13 and 4. 14 are derived in appendix D, sections D.3 and D.4 respectively.
If the UMEe path reluctances are time invariant, the inductances measured from
primary and secondary short-circuit tests are
(4. 15)
and
LSec ;::: N?
S3 n'"'u
..
(4. 16)
In the linear steel cored transformp..T example, nr = n; = n:, and since (n
4
::fi
Rs) en:,. 'R.;), equation.q 4. 13 and 4.14 call be simplified to
R*,prim = n*8t..C = 'R.41?s
U S8 n'" +n*
5
(4.17)
The primary and secondary short-circuit equivalent reluctances a.re dependent upon
leakage flux paths only, and a.re equal. The inductances v,;im and L;:c (referred
correctly by the square of the transformer turns ratio) must also be the same.
In the case of ai r or pa.rtia.lly cored transformers, the short-circuit path reluct8.11ce
core terms can not be neglected. However, provided the winding-limb reluctances
equa.te, the primary and seconda.ry equivalent reluctances remain equal.
4.2.2 The UMEC self and mutual inductances
The linear UMEC can also be applied to derive self and mutual terms of the inductance
matrix transformer representation
[
"I (t) ] =
",(t)
MI'] []
L
.ill
22 de
(4.18)
Self inductance can be defined as the scaling factor that relates winding voltage
to cha.nges in winding current. The single-phase UMEC self inductances L11 and L 22
are equal to the primary and secondary open-circuit inductances
(4.19)
and
(4.20)
Mutual inductance can be defined as the scaling factor that rela.tes winding
voltage to the change in current of another mutually coupled winding. The figure
4.3(a) open-circuit UMEC can be simplified to that of figure 4.4 which includes single
MMF N1it{t) and equivalent l/J2{t) pa.t h permeance
p;p;p;
1"1 = (1'; + 1';)(1'; + 1'; + 1'4) + 1';(1'; + 1',)
(4.21)
</>,(1)
Figure 4.4 Mutual iuductance equivalent circuit.
Similarly, if the primary winding is open-circuit a.nd the secondary energised, the
UMEC can be reduced to a single MMF N2i2(t) and equivalent l{t) path perme-
aI1CC
'PiP;,?;
1'1' = (1'; +1';)(1', +1'; +1';) +1';(1'; +1';)
(4. 22)
4.3 INCORPORATING A NON-LINEAR UMEC IN PSCAD-EMTDC 47
Equations 4.21 and 4.22 are both derived in appendix D, section D.S.
If the UMEC is linear and time inva.riant then the mutual inductances arc
(4.23)
and
(4.24)
The mutual inductances M12 and M21 always equate.
The transformer model parameters of open and short-circuit, or self and mutual
inductance derived in sections 4.2.1 and 4.2.2 all assume time invariant pa.th reluctance.
Although the derivations a.re useful to gauge the contribution of individual UMEC
branches upon each equivalent inductance, transformer models based solely upon these
parameters do not include sa.turation. If sa.turation is considered even the short-circui t
equivalent reluctances and n::ec are time variant; [Thrner 1977] experimentally
measured the non-linear relationship between leakage flux and windi ng MMF. Therefore
core non-linea.rity must be added t.o the UMEC to improve transformer representation.
4.3 INCORPORATING A NON-LINEAR UMEC IN
PSCAD-EMTDC
The flow diagram of figure 4.5 descri bes the per-unit implementation of a non-linear
UMEC ill PSCAD-EMTDC. IT leakage flux path permeances are converted to per-unit,
the winding-limb and yoke branch saturation characteristics given in section 3.2. 1 can
be applied to the UMEC.
The shape of leakage 8Jld winding-limb flux waveforms are dissi milar, since leak-
age flux is proportional to winding enrrent while winding-limb flux is proportional to
the integral of winding voltage. The section 3.2.2 linear division of the winding-limb
flux , into outer branches r is now Ullacceptable. Calculation of UMEC branch flux
k requires the expansion of linearised equation 4.6.
= [[M:'.) [M:,)] [[1';)

, [M;,] [M;"] [0]
(4.25)
Winding-limb flux 3(t - .6.t) is calcu1ated from winding current using the upper par-
tition of equation 4.25
(4.26)
Yoke and leakage path flux r(t - .6.t) is calculated from winding current using the
Slllrt usordefincd
dynomics file .
"

No
........
. . .. -
uro'
--
---
---
f
U_ iupul;
\
P uolt branch
Ya
,
J
,
n lUlatioll c ..... u _
,
C. k:ulole rNeIl bn.ltIl preiOlll lima l tep
,
Moter pr_ lima step ,
r, ....... 4- inil io.!.
,
v:lriables ,,(I' M) .
,
variablu r, (rJ.
"
I
:
I
\
Construct InQIIU Cakulate

willi
,
'JjtJ with
WI
,
tcqu:tlioo4.26.
,
cqu:ttl0ll4.26.
"
I I
,
I
,
Dllculwo I M'l with Calculate i'/'-Al)
!
Dllcllble . ,(1) willi. wilh ,
equation 4.6.
,
equall"" 4.21.

\
C<lultlon4.l7_
"
I
...
CoI lculale IYsH witb CoIIcIlI.tte 1{foft:ICb brtDCb \
equation 4.1.

with 4.28 .

,
,
!
,
;
CoIlcuJate ,':"willl
eC(\l>li1lD 4.1. ,
,-
I
.-'
,
,
Solye network
"
-
EMTDC
tqll"ioo 2.5.

,
,-
ITERATIVE LOOP
-'
. -.-
.......
.. -.-......... -..................... -
.........
-
,
,
MM.J'loop.'l
: No
"
and nodal [Jill sums
'- equal UfO?
'-
-
-'-
-
-- -,
-
-"
... _---._--
- ' .-.. -.........
Ya
...... _--..
_.-..... -.....
--''''-
(
------..
-....... _--
Begin
inlespololion.
,. -
-"-"
Figure 4.5 Actual pcrmcance formulation UM.EC implcmcutatiou in PSCADEMTDC.
4.3 INCORPORATING A NON LINEAR UMEC IN PSCADEMTDC 49
lower partition of equation 4.25
(4.27)
Branch actual permeance P; is calculated directly from the hyperbola. saturation
approximation using solved branch flux rPk(t - Once [POOl is known the per-unit
admittance matrbc. and current source vector can be obtained.
4.3.1 T ime step delay validation
The Norton admittance matrix and current source vector are both derived from pre-
vious time step parameters. A time step delay is present between the winding current
vector '$, and UMEC branch flux vectors , and r' With this dela.y a true solution
to the UMEC can not be fOlmd; the flux sums at each node and MMF sums around
each loop will not equate to zero. Although the typical PSCAD-EMTDC time step is
small (50 1'8) the error resulting from the time step delay has not yet heen gauged.
The iterative loop shown in figure 4.5 is required to obtain a simultaneous non-
linear UMEC and PSCADEMTDC electric circuit solution at each time t. Each it-
eration takes the electric network solution, uses the winding current '! to update the
UMEC parameters , and ", and recalcula.tes the Norton equivalent . Unlike the iter-
ative technique of section 3.4 this loop includes the electric circuit solution. Iteration
ceases when the UMEC flux sums at each node and MMF sums around each loop are
below specified tolerances.
Iteration at each time step modifies the original electromagnetic transient pro-
gram proposed by [Dommel1969L and the PSCAD-EMTDC main program structure
shown in figure 2.2. Modifica.tions are required when the electromagnetic transient
program is applied to non-lineal' systems. For instance PSCAD-EMTDC modifies the
electromagnetic transient technique with the addition of interpolation to accurately
determine HVdc converter valve switching instants.
In order to quantify the effect of the previous time step approximation, com-
parisons were made between the non-iterative and iterative UMEC implementation
techniques. To cnsure that only the iterative loop differed in each technique, two elec
tromagnetjc transient programs were written from the flow diagrams of figmcs 2.2 I1lld
4.5. Each program was written to model only the figure 4.6 single-phase transformer
test system.
1
Comparisons were made for open-circuit, short-circuit and in-rush operating
conditions. Table 4. 1 lists the test system electric ci rcuit parameters for the three
scenarios, a 50 J.l.S time step was applied in each case. The single-phase transformer
llnscrtion of the iterative technique in PSCAD-EMTDC requires alteration of the complex main
program structure. Valid technique comparisons are LHore ea.,>ily achieved with the simplier case specific
programs.
50 CHAPTER 4 THE UNIFIED MAGNETIC EQtflVALENT CIRCUIT
): .. 0.084
Figure 4.6 Iteration comparison test system.
UMEC winding-limb and yoke hyperbola. sa.turation cbru:a.cteristic parameters, and
leakage reactance arc as listed in sections 3.4, and 3.3.2 respectively. The leakage
permeances P4 and Ps wC!'e set at 4.12e-2 (poll.) to give the 0.084 p.ll. short-circuit
impedance.
v, (peak poll.) r, (p.ll.) r, (p.ll.)
Open-circuit 1.0 0.001 1.0e+ 6
Short-circuit 0. 1 0.001 1.00-6
in-rush 1.0 0.001 1.00+6
Table 4.1 Non-iterative and iterative test-system electric parameters.
i terative and non-iterative UMEC solutions are compared in figure 4.7. In all
three scenarios the iterative and non-iterative solutions were indistinguishable. More-
over t he open-circuit wa.veforms shown in figures 4.7(3) and (b) verify that UMEC
winding leakage flux is proportional to winding current. Figures 4.7(c) and (d) reinfor ce
that the short-circui t condition restrkts flux from flowing in the secondary winding-
limb; t he primary winding-limb flux (10% ra.ted) halves and flows in each leaka.ge path.
Figures 4.7(e) and (f) show that a linear division of the winding-limb flux into outer
branches is not a.cceptablei the winding leakage waveform is not a scaled version of the
winding-limb flux.
With a 501-'8 time step the non-iterative UMEC solution errors are small, and
the previous time step approximation is satisfactory. For example, figur e 4.8 displays
the level of error introduced by this technique for the opell-circui t test c8.::Ie. The MMF
sum around the primary winding-limb a.nd leakage pa.th loop is shown in figure 4.8(a),
and the flux swn at the top UMEC node (labeled N l in figure 4. 1(b is shown in
figu,e 4.8(b).
4.3.2 Removal of per-unit representation
The UMEC based PSCAD-EMTDC transformer models can be further simplified if
the reliance upon per-unit representation is removed. Althougb the per-unit system
~ well suited to the linear and linearised models, a requirement of the HDA interval-
idation, the PSCAD-EMTDC couvcntion is to set node voltages in kV, node current
0.'
".

.-. -
'"
o.Oll
"" ""

Ii. IDO (0)
(a)
,.

."
,

..... -'f.ot

. -
'"
om. 0.014 .m o.olf

(0)

,

V
'.
'"
.. "
, .
.. "
" . '."
...
(0)
0

';i

..
;
,

,XI.'"

.,
4
i Mel u .. ...
..,
""
0 . " .m
... "
...
,imo(.)
(b)

..,

. ..
u ,
* .. -
. *
... Mil 0.0,. . m

..... (.)
(d)
o.g,
0.01

' .0

...
'"

'---' '---'
..... T,

(I)
Figure 4.7 Iterative and non-iterative singic-llhase UMEC model comparison; Solid=non-iterative,
Da:;h-dot= iterative; (a) open-circuit current, (b) open-circuit leakage nux (primary), (c) short-circuit
leakage lux (primary), (d) short-circuit leakage nux (secondary), (e) in-rush primary winding-limb, (f)
in-rush primary leakage flux.
,XI.-"
,
i 0 ... .. .
. ,
.,
.,
... .... ... , .. .. ...
limo (0)
-I.. U.. O.Moo 0.0.. 0.011 0.07 Ml't '.0'/1 u rn 0 .
-"
(a) (b)
Figure 4.8 Optl1--cin;uit !lingle-phase UMEO solution comparison; technique,
Dash-dot=itcrative technique: (a) mill! loop sum, (b) node ilux sum.
in kA, and admittances in mhos. As a direct interface with other PSCAD-EMTDC.
componcntl:l the Norton equivalent of figure 4, 1 must also be scaled to this convention.
Figure 4.9 presents the new UMEO implementation strategy. The leakage flux
branch permeances a1'e constant (described in section 4.1), and the core branch satura-
tion characteristic is the steel flux density-magnetising force (B-H) curve. If requested,
manufacturers will provide the B-H characteristic for a. transformer core, individual
branch per-unit characteristics a.re not a conventional specifica.tion.!!
Core dimensions, branch length Lk and crO$S-scctional area. Ak, are required to
calculate rcal value permeances
p. _ Ji-Q/JrkAk
k - Lk
Branch flux k(t - .6.t) is converted to branch flux density
B,(t _ tit) = 1>.(t - tit)
A.
(4.28)
(4.29)
The hyperbola Can be applied to represent the core B-H characteristic; branch perme-
ability J.LQJ.'rk is calculated from flux density Bk(t - .6.t) using
(4.30)
Figure 4.9 also shows that winding-11mb flux <$;s(t - 6.t) is now calculated with
trapezoidal integration (3.6) rather than linearised equation 4.26. Trapezoidal integra-
tion requires storage of vectors 3(t - 2.6.t) and us(t - 26.t) . Matrices [M;,l and [Pssl
Appendix E details the relationship betwef'.ll D-H and per-unit -i characteristics for the UMEC
winding-limb alld yoke branches.
4,3 INCORPORATING A NON-LINEAR UMEC IN PSCAD-EMTDC 53
(
Start user defined
dynamiCS file.
il
Time equal
No
10
zero7
/user input:
Core B-H characteristic,
Winding turns number, y"
Core dimensions,
Calculate core
" Leakage reactance.
MeIer previous time step
initilll, using
1 variables . 7
equations 4.28-4.30.
! I
Calculate leakage ',,*" Calculate cJt-tJ) with
Appendix C.
,
equation 3.6.
,
I
Construct matrix Calculatc [ P,"[ with
[ pO I ,
3 equations 4.28-4.30.
,
I
Calculate [ M "1 with Meter previous time step
equation 4.6. 4 variables I,(I-!JJ) . 10
I
Calculatc [Yu1 with Calculate with
equation 4.7. ,
equations 4.31-4. 33. II
I
Calculate ;;';with
Calculate 'I}" for each
equation 4.7.
yoke using
"
,
equations 4. 28-4. 30.
I
Begin
;, )
network solution.
FigUl'e 4.9 Real value UMEC impleOlentation in PSCAD-EMTDC.
54 CHAPTER 4 THE UNIFIED MAGNETIC EQUIVALENT CIRCmT
must be stored to apply equation 4.26. Although [P:.f] is diagonal, 1M:,, ] is full and
therefore, in the latter method, element storage increases w1th t he square of the UMEC
winding-limb branch number.
The elements of r(t - ilt) can be calculated using magnetic circuit theory. The
MMF around the prima.ry winding-limb and leakage branch loop must sum to zero,
stated as
(4.31)
The MMF around the secondary winding-limb and leakage branch loop must sum to
zero, stated as
(4.32)
FinaJly, the flux at node Nl must sum to zero, stated as
(4.33)
Yoke branch actual permeance Pi; is calculated directly from solved branch flux
(/>k(t - t:::..t) using equations 4.28 to 4.30. Once [P*] is known, the real valued admittance
ma.trix and current source vector can be obtained.
4.3.3 The Benmore converter transformers
The final UMEC implementation strategy, presented in the previous section, can be
a.pplied to create practical transformer models. Figure 4.10 displays the core
sions, and table 4.2 gives the name-plate data, for the Benmore valve-group converter
transformers. This informa.tion was supplied by Trans Power New Zealand Ltd, the
transformer owners. In this section, single-pbase transformer parameters have been
given for a connection.
Electric circuit parameters
Rating 62.5 MVA
Configuration single-phase
Frequency 50 (Hz)
VI
16 (kV)
7:i
V, (kV)
X'eakCl e
0.113 (p .u.)
Table 4.2 Beumore transformer name-plate data.
Winding turns number, core dimeusions and B-H characterist ics arc seldom speo-
ified as part of a transformer purchase. This example is DO exception and also there
is not sufficient information to calculate individual winding leakage permea.nces, This
I
J.JJ",
1
Figure 4.10 Benmore transformer core dimensions.
problem of insufficient data. must be solved either with accepta.ble assumptions applied
at the modellees discretion or by obtaining a satisfactory parameter schedule from the
transformer manufacturer. l<1.uther in this chapter, for the case of a three-limb three-
pha::;e field transformer, the latter option is exercised, In this case the former is applied
beca.use the problem of insufficient data is common, and the assumptions required can
be presented.
In the absence of better information the core satura.tion curve is represented
with the general B-H characteristics [Say 1983] presented in figure 4.11. Table 4.3 gives
hyperbola. consta.nts.
ml m2 bl b2 &
0.021 3.00-5 0.0 1.63 l.9
Table 4.3 Assumed B-I-I curve hyperbola parameters.
The peak flux density that the transformer winding-limbs experience under
steady-sta.te rated conditions is
J2v,
(4.34)
where Wo is the base angular frequency. In this example winding turns number must
be calcula.ted assuming a reasonable peak flux density. The transformer manufacturer
must trade off the winding volt-pcr-turn design and core cross-sectional area., against
the position of Br:
ak
with respect to the saturation curve knee region. As the availa.ble
transformer data gives only the core cl'oss-scctional area, a. reasonable peak operating
56
CHAPTER 4 THE UNIFIED MAGNETIC EQUrvALENT CIRCUIT
Silicon sleel saturation curve
1 . _ _____.
1.,
1.'
1.2
0.2
.. ... .. ..
H (Ampere-turns/meter)
FigUl'e 4.11 Assumed B-H sat.urat.ion characteristic.
flux of 1. 4T has been selected; the winding t urn numbers arc thus 65 and 450 turns
for the and kV windings respectively. Finally, the name-plate data does not
specify primary and secondary open-circuit current, the flux leakage path permeance
is assumed to be equally divided amongst the pri mary and secondary windi ngs. This
assumption allows leakage path permeances 'P4 and 'Ps to be calcula.ted in accor dance
with appendix C. The specified and assumed parameters required to model tbe Ben-
more transformer arc summarised in table 4.4.
Electric circuit parametcl'S Magnetic circuit parameters
Rating 62.5 MVA
Awinding limb
0.454 (m')
Configuration single-phase
Lwinding limb
3.59 (m)
Frequency 50 (Il z)
Ayoke
0.454 (m')
Vprimo.rll
16 (kV) L
yoke
2.66 (m)
;Ts
Vsuondo.ry 7a (kV) Nprimo.ry
65 (turns)
X'e4ko.ge
0.113 (p.u.)
NJeconrJa'1l
450 (tuTns)
'P. 5.S5e-S
'P5 5.850-8
TAble 4.4 Single-phase UMEC mode1 llarameters.
An PSCAD-EMTDC transformer model was created from the table 4.4 parame-
ters, placed in the figure 4. 12 test system, and subjected to open-circuit cnergisation
at full rated voltage, short-circuit energisa.tion a.t 10% rated volta.ge, and full rated
load. operation a.t unity power factor. Table 4.5 lists tbe test system electric circuit
parameters for the t hree sccnarjos, a. 50J.'s ti me step was applied in each case. Figure
4.13 displays a select few simulation waveforms produced D.-om the three scenarios.
Figure 4.12 Single-phase field transformer test systelu.
V, (kV) r, (0) r, (0)
Open-circuit

0.lD7 1.00e+6
Short-circuit 8. 000-3 7.50 .. 3
Rated-load
it
8. 00 .. 3 64.5
Table 4.5 Single-phase -field t ransformer test-system el ectric parameters.
The open-circuit current, shown in figure 4.13(80), exhibits the typical non-
sinusoidal magnetising waveform that lags the terntinal voltage by 90.3 The mag-
netising current magnitude appears reasonable, it equates to O.047kA, and when nor-
malised by the rated transformer base current l blue=6.77kA, cquuls 0.0069 p.u. The
open-circuit yoke flux, shown in figure 4.13(b), is correctly in phase with the magnotis-
ing current at a. peak flux level t/>3=O. 639Wb. The corresponding yoke flux density
B3=1.4T is in accordance with the previously described winding turns design at rated
voltage.
Figures 4.13(c) and (d) present shott-circuit primary voltage and current respec-
t ively. The short-circuit voltage vf eak=L31 kV and current if
eak
=8.42kA, give a leakage
impedance Xjeakage=O.156fl. When normalised by the impedance base Zba.se =1.37U,
Xl
eo
kage=O.1l4 p.lI., which agrees with the name-plate leakage specification. Finally!
figure 4.13(e) displays the primary current when a 1.0 p.ll. (64.50) resistive load
is placed at the transformer secondary terminals. The simulated ptimary current
1, =6. 80kA matches the rated primary current l /Jo,e ' The small pbase shift between
primary voltage and current results from the combined transformer leakage reactance
and winding resistance.
The Single-phase Norton equivalent provides a. reasonable representation of the
aeomore valve-group converter transformers. However assumptions were required to
derive model saturation curve, winding turns design and individual winding leakage
permeancc, therefore further validations are required to certify the performa.nce of
UMEC based transformer models.
3In all three cases the reference source voltage positive going zero-crossing occurs at time 0.08 secs.

.. UI
....t .. _ 0111. ._W 0.011 0.09> ._ 0_ 1.1" t.'
Iimo(.)
(a)

-.
OM .... t 0.0" 0.0 .. '.on 0." 0.0112 0_ 0,"" 0_ 0.'
"","(0)
(e)

""":.01 0.012 ..... 0.01t G.... 0." 0.0" 0 .... ' .001' ._ U
tim< (a)
(b)
-'- ,--
. ;;- MO. 0.... ... .. un I." "ot;! o.OM ..ott ..on G.'
time (0)
(d)
,.'r-------"C""---------------r---r-------,
-.
-.
. .... ..... 0.0" 0.0.. 0.011 U .2 . _ .,." . on
limo (ol
(e)
F igur e 4.13 UMEC Benmore single-phase transformer modd: (a) open-circuit current, (b) open-
circuit yoke flux, (c) short-circuit voltage, (d) short-circuit current, (c) rated-load current.
4.4 UMEC AND PSCAD-EMTDC TRANSFORMER MODEL COMPARISON
4.4 UMEC AND PSCAD-EMTDC TRANSFORMER MODEL
COMPARISON
59
Prcsent HDA a.nd PSCAD-EMTDC transformer models separate leakage, and core flux
into electric circuit series leakage reactance, and shunt magnetising branches respec-
tively. In FillA the single-phase two winding transformer magnetising current is halved
and placed on either side of t he lumped primary and secondary leakage reactance. The
present non-linear PSCAD-EMTDC transformer model implements the magnetizing
current as a source across one winding of an "ideal" transfor mer, as shown in figure
4. 14. In this context an tt jdeal" t ransformer is one that exhibits a leakage reactance
but zero magnetizing current. With reference to HVdc converter transformer!!, and
considering the relatively large impedances a,q seen from the converter side winilings,
it can be acceptable to place the magnetizing ClU'l'ents on the ac system side of the
transformer leakage reactance.

Steinmetz "approximate" equi vaJent
(a)

Conventional EMTDC non-linear trans former equivalent
(b)
Fig-ul'l" 4.14 The: PSCAD-EMTDC model and Sl.eilllilctz circuit: (a) Steinmetz "ap-
proximated" equivalent, (b) PSCAD-EMTDC representation.
The UMEC single-phase model can be applied to examine the consequences of
the conventional representation. H the core loss component of transformer magnetizing
current is neglected the single-phase t ransformer can be represented by the Steinmetz
ucxa.ct" equivalent circuit shown in figure 4. 15(a) and (c). Alternatively, the single-
phase transformer can be redrawn as the UMEC shown in figure 4.15(b). The Norton
equivalent implemented as shown in figure 4. 15{d), is a. model for the Steinmetz "exact"
equivalent circuit.
tl'llnsfor"""
(.)
Steinmetz '"xact" equivalent
(,)
Magnetic circuits

Electric ci rcuits

Unified mI'Ignetk e<Juivalomt cirC\lit

j
-,.
F.MTOC impJemenlillion
(d)
Figul"e 4. 15 The single-phase Norton equivalent and Steinmetz "exact" circuit: (a) core {lux paths,
(b) unified magnetic equivalent circuit, (c) Steinmetz, equivaleot (e1) Norton equivalent.
4.4.1 Laboratory verification
To highlight the effect of the UMEC model on transformer magnetising current har-
monic distribution, the laboratory set up of figure 4. 16 was utilised. An undistorted
sine-wave generator supplied the test transformer loaded with a series resonant cir-
cui t j the resonant point was tuned to the fifth harmonic. The sine-wave generator)
test transformer) and series resonant load parameters are given in appendix F. The
magnetising current distribution is dependent on the transformer individual leakage
reactances, winding resistances and the system impeda.nce at each frequency.
Sine wave Test transfomler -----t-=ne:nt
generntor

,_ri:"_m jX_'Prim---"---J 1 [1s '-
Figure 4. 16 Laboratory experimental set lip.
The pnmary and secondary impedances of the Steinmetz Hexact" equiva.lent
4.4 UMEC AND PSCAD-EMTDC TRANSFORMER MODEL COMPAIUSON
circuit are
zexact Z + R +'X
prim = source -'-"Prim J lprim
Z
-ao," 2(X + R Z- )
sec = a J /sec sec + load
61
(4.35)
(4.36)
where the is used to indicate parameters referred to the generator (prima.ry) side.
The magnctising current injection (drawn as the solid line current source of figure
4. 16) will divide up between primary and secondar y windings according to
1
1- I I Zo<o" I
K Stmz(exact) = _msec = _ prim
1
1m . I Iz<xa,J' 1
prIm sec
Similarly Steinmetz "approximate" equivalent circuit impedances are
Z
- ap'/JTOX Z- R
prim = source + prim
Z
- approxX 2( 'X l ' Z')
sec = J lprim + a J !sec + t.scc + load
(4.37)
(4.38)
(4.39)
Thus the magnetising current (drawn as the dashed line current source of figure
4.16) will divide up according to
I
- I IZapproxl
KStmz(approx) = prim
I I
- IZ',o'!prox*1
l mprim L",
(4.40)
Table 4.6 displays the laboratory measured and simulated fifth harmonic using
the UMEC and conventional transformer models. Placing the circuit parameters of
Experimental UMEC Conventional
Il mpriml
4.00 (rnA) 2.42 (mA) 1.11 (rnA)
- .
Il msecl
48.0 (mA) 27.0 (rnA) 5.54 (rnA)
Ilmsrcl
12.0 11.2 4.98

Table 4.6 Measured and simulated primary and secondary 5tb harmonic magoel.ising current.
Appendix F into equations 4.37 and 4.40 gives K Stmz(r:xacL) = 11.2 and =
4.97 respectively.
The UMEC model predicts the same current division as the Steinmetz Uexact"
equivalent and the current division ratio is similar to tha.t of the experimenta.l test
results. The conventional transformer model predicts a rnagnetising current division
ratio equal to that of the Steinmetz "approximate
ll
equivalent and does not match
62 CHAPTER Ij THE UNIFIED MAGNETIC EQUIVALENT CIRCUIT
the experimental results. Because hysteresis and corc loss has not been included, the
actual levels of fifth harmonic predicted by the UMEC model do not exactly match the
experimental results.
4.4.2 Harmoni c comparison
In the previous example the ratios KStm::(e:.:actj and KStmz(approx) are dissimilar. In
many practical cases, where the transformer individual leakage reactances are murll
JellS than the system impedance at each frequency, the ratios will be similar. However,
when tr ansformers are connected to generators, transmission lines or filter banks the
system impedance at each frequency may vary widely.
Recently a PSCAD-EMTDC model for the Benmore converter station was ver-
ified with field data [Macdonald et at. 1995J; this will be discllssed in greater depth
in chaptel' 5, In this system three 112.5 MVA generat ors arc connected to a 270MW
twelve-pulse mercury-arc valve converter, via two 187.5MVA, 16/110kV converter
(ormer banks. The 16kV bus-bar is connected to the South Island ac system via a
three-winding 200MVA, 16/33/220kV interconnecting transformer bank. The converter
iilters are connected to the 33kV windings of the interconnecting transformer bank,
and each filter includes individual circuits tuned at the 5th, 7th, 9th, 11th and 13th
harmonic. These filters provide a low impedance path to the respective transformer
currents as well as the converter harmonics.
Occasionally the converters are operated with the generators disconnected from
the South Island ac system and the filters left in service. In till!'! mode, if the mercury-
arc valves and convertel' are removed fl.'om service the generators can not
absorb t he capacitive VAR flow from the 33kV filter banks, and the 16kV bus-bar
voltage dses. The voltage rise ceases when the transformer magnetization balances the
extra capacitive VAR flow. Under such load rejection the generator bus-bar may rise
to 1.46p,ll [Arrillaga 1983], Figure 4.17 illustrates this scenario, the parameters for the
generators, transformer and filters a.re given in a.ppendix G.
Table 4.7 presents the magnetising current division ratios for this test system
calcula.ted from equations 4.37 and 4.40.
Harmonic
5 7 9 11 13

3.9 4.5 9.7 4.7 4.2
KStmz(appro1l)
1.8 1.9 3.0 1.8 1. 7
T-d.ble 4.1 Bellmore current eli vi sion ratio.
The magnetismg Clll"rent division ratios for each harmonic are different. In tills
test system, the predicted level of harmonic load on the generators and filter banks is
4.5 THREE-LIMB THREE-pnASE UMBO 63
Interconnecti ng transfonncr T2
Filter bank
Synchronous generators Gl-G3
Figure 4.17 Benmorc generators aud interconnecting transformer baok.
dependent upon the equivaJent circuit applied. Tables 4.8 and 4.9 present simulated
current harmonic magnitudes and division ratios when a. 1.46 p.u. overvoltage occurs
on the 16kV bus-bar for the UMBe, and convelltional transformer models respectively.
Harmonic current (ARMS)
5 7 9 11 13
ITnprim 56.4 5. 12 4.91 5.33 0.554
- .
m ~ c 221 22.8 47.3 24.8 2.35
K U M ~ U
3.92 4.45 9.63 4.65 4.24
Table 4.8 Magnetising current division; Transformer represented with UMEC model.
Harmonic cuncllt (ARMS)
5 7 9 11 13
Imprim 106 18.8 9.67 11.82 2.03
Im;ee 188 34.5 27.7 21.5 3.45
KCV"
1.77 1.84 2.86 1.82 1.70
Table 4.9 Maglletising current division; 'Ira..usformer represented with conventional model.
The levels of harmonic elU'rent predicted by each model are different. The con-
ventional transformer model creates a harmonic load on the generator greater that
that of the UMEC model. Again the UMEC model provides an excellent ma.tch to the
Steinmetz "exact" equivaJent and the conventional model an excellent match to the
Steinmetz "a.pproximate equivalent.
4.5 THREE-LIMB THREE-PHASE UMEC
Au extension of the single-phase UMEC concept to the three-phase tra.nsformer, shown
in figure 4.18(.), leads to the UMEC of figure 4. 18(b). Unlike the three-limb model,
64 CHAPTER.{ THE UNIF1ED MAGNETIC EQUIVALENT CIRcmT
presented in section 3.3.3, the magnetising cUJ'rent is not halved and placed on each
side of the linear short-circuit admittance matrix. Instead there is no need to specify in
advance the distribution of magnetising current components, which have been shown
to be determined by the transformer internal and external circuit parameters.
MMF sources Nt i
l
(t) to N6i6(t) represent each t ransformer winding individually,
and winding voltages Vl (t) to V6(t) are used to calculate winding-limb flux l (t) to 6 (t)
respectively.
Pel'meances Pi to 7>"6 represent the transformer winding limbs. If the total length
of each phase winding Drob Lw has uniform cross-sectional area AWl the UMEC branches
1 to 6 have length 4r and cross-sectional area Aw. Pcrmeances Pi3 and Pi4 represent
the transformer left and right hand yokes respectively. The upper and lower yokes are
assumed to ha.ve the same length Ly and cross sectional area Ay. Both left and right
hand yokes are represented by lIMEC branches 13 and 14 of length L
13
= L14 = 2Ly
and a.rea AI3 = AI4 = A". Zero-sequence permeances Pir; to 'Pi7 a.re obtained from
in-phase excitation of all three primary or secondary windings [Stuehm 1993].4
UnHke the three-limb linear and linearised model, which required acquisition of
non-conventional transformer parameters YI, V:, V;', and Vr' (section 3.3.3), the UMEC
can be derived from conventional transformer manufa.cturer specifications. Leakage
permeances al'e obtained from open and short-circuit tests and therefore the ciTective
length and cross-sectional areas of UMEC leakage branches 7 to 12 are not required to
calculate P; to Pi 2'
The UMEC circuit of figure 4.1S(b) places the actual permeance formula.tion in
the real value form
'1 (t) Vn
Vl'
V13 Y14 V15 Vl6
Vl (t) Ins 1
"(') 1/21 V22 Y23
V"
V25 !h6 v'(')
I n32
'3(t) Val V32 V33 Y34 Va5 Y36
.3(t)
+
i
n33
(4.41) =
i.(t)
V.l V12 Y43 Y44 Y45 Y46
v.(t) 1ns4
'5(t) V5! V52 Y53 V54 V5r; Y56 V5(') t n85
io(t) VOl V52 Vos y" V65 Y66
V6(t)
n . ~ 6
The matrix [Y:uj is symmetric and this Norton equivalent is implemented in PSCAD-
EMTDC as shown in figure 4.19, only the blue phase network of a star-grounded/star-
grounded transformer is shown.
The flow diagram of figure 4.9 describes the Uuce-limb three-phase UMEC im-
plemeJltation in PSCAD-EMTDC with only slight modifications. The trapezoidal in-
tegration equation 3.6 is now applied to the six transformer windings to calculate the
winding-limb flux vector 3(t - 6.t). Equations 4.28 to 4.30 al'e utilized to calculate
the permeances of the winding branches. Once the previous time step winding current
4IfzCJ"o-sequence test tlata is not available, P i ~ to Pi7 can be set equal to leakage path pernleance.
,
4.5 THREE-LIMB TIffiEE-PHASE UMEC 65
'"
."
/,
."
./
."
./
.,
'\ ,
I,
,

" 1
" ',1 ',1
lu
V
l-
V
llj
.,,-

.,,-
,
't

'\
1
'\

I.,
to"

V
."
(a)
1];

NJ

"'
N>

" .

'I;'
1J:
N,jl/I
+
NJiJ/1
q,{IJ 4'J(1'1
1J; "'
'I;;
"'
'I};
N'
li'l IiI)
I,(t):eo
,

.p,'
'If;
1];
N
4
i,(t
(b)
FigUl'e 4.18 A UMEC PSCAD-EMTDC three-limb three-phase transformer model; (a) core flux
paths, (b) unified magnetic equivalent circuit.
66 CHAPTER <1 THE UNIFIED MAGNETIC EQUIVALENT CIRCUIT
;5(1)
Figure 4.10 Thrce-limb three-phase Norton equivalent (blue-phase only, star.grounded/star.
grounded).
vector i,(t - ~ t is fanned, the dux leakage elements of r(t - .6.t) can be calculated
using
",,(t - <'.t) - P; (NliJ (t - <'.t) - f, (, - <'.t)7I.i) (4.42)
f.(t - <'.t)
-
P; (N,i,(t - <'.t) - ",,(t - <'.t)7I.') (4.43)
o(t - <'.t) - Po ( N, is (t - <'.t) - >,(t - <'.')71.;) (4.44)
fJO(t - <'.,) - P;o (N,i'(' - <'.,) - ' - <'.t)7I.;) (4.45)
fll (t - <'.t)
- Pi, (N1i,(t - <'.t) - f,(' - <'.,)71.;) (4.46)
4.5 THREE-LIMB THREE-PHASE UMEO 67
(4.47)
The zero-sequence elements of ~ t - At) a.re calcula.ted using the MMF loop sum
around the primary to secondary winding-limb and zero-sequence branch, stated as
4>15(t - t:.t) = Pi, ( N, i, (t - t:.t) + N,i,(t - t:.t) - <Pt(t - t:.t )Rj
- q,,(t - t:.t)1<2)
4>,,(t - t:.t) - 1';, (N, i
3
(t - t:.t) + N,i
3
(t - t:.t) - Mt - t:.t)R;
- 4>,(t - t:.t)R')
4>,,(t - t:.t) Pi, (N,i,(t - t:.t) + N,i,(t - t:.t) - Mt - t:.t)R;
- 4>,(t - t:.t)R')
(4.48)
(4.49)
(4.50)
Finally the yoke flux is obtained using the flux summa.tion at nodes N 1 and N2, stated
lIS
4>,,(t - t:.t) = 4>,(t - t:.t) - 4>,(t - t:.t) - 4>,,(t - t:.t)
4>,,(t - t:.t) - 4>,(t - t:.t) - 4>" (t - t:.t) - 4>,,(t - t:.t)
(4.51)
(4.52)
Yoke branch permeance Pi3 and Pi4 is again calculated directly from solved
branch flux 4>13 and tP14 using equations 4.28 to 4.30. Once [p.} is known the real
valued admittance matrix [Yss] can be obtained.
4.5.1 UMEC factory data validation
A model for the Manapouri l05MVA, 13.8/220kV three-limb three-phase generator
transformer was constructed to valida.te the extended UMEC representation. The core
B-H cha.racteristic, winding turns number, core dimensions and leakage reactance pa.-
rameters were obtained directly from the transformer manufacturer) Savigliano. The
conncction matrix for the DyU transformer winding connection, and the complete
technical documentation provided by Savigliano are given in appendix H. The pa.rame-
tel'S required to model the Mana.pouri transformer are summarised in table 4.10. The
core B-H characteristics, presented in figure 4.20, is constructed from the hyperbola
consta.nts of table 4.11.
A PSCAD-EMTDC three-limb three-phase transformer model was created from
the parameters of tables 4.10 and 4. 11, placed in the figure 4.21 test systems. Table
4.12 lists the test system electric circuit parameters for each test system, a. 50j.ts time
step was applied in each case.
68
CHAPTER 4 THE UNIFIED MAGNETIC EQUIVALENT CIRCUIT
Electric circuit pa.rameters Magnetic circuit parameters
Rating 105 MVA
Awindin.g lim&
0.5555 (m')
Configuration delta.-star
Lwi-nding limb
1.932 (m)
.F\"eqllellCY 50 (Hz) Ayoh:
0.5635 (m')
Vprimltry
13.8 (kV)
Lyoke
4.001 (m)
Vsccondary
220 (kV)
Nprimary
78 (turns)
Xlcakage
0. 1126 (p.u.)
Nsecondary
718 (turns)
'P7,9,11
1.67e-7
PS,lO,12
1.67e-7
PIS 1(i 17
1.67e-7
Table 4.10 Manapouri three-limb transformer UMBC model parameters.
TnUlsfonner core ciJaracterislic
'"
-.-
1.6
,
'"
\,2
:<i:
I
g
co O.B
'.6
,.4
0.2 .
,
,
100 150 200
'"
'"
350
H (ampere/meter)
Figure 4.20 Manapouri transformer B-H saturation characteristic; Solid=Suvigliano data, Dash-
Dash=Hyperbo)a approximation.
The factory test data is given in tabular form (appendix H), table 4.13 DO-
load ratio, open-circuit, and short-circuit factory test results with simulation. Under
these operating conditions the three-limb threephase UMEC provides an
acceptable match with the factory data..
4.6 CONCLUSIONS
Winding leakage flux representation has been shifted from the transformer electric
network to the unified magnetic equivalent circuit (UMEC). Each UMEC MMF source
now represents total winding current, rather than the ma.gnctisillg current alone.
The PSCAD-EMTDC nonlinear transformer model is based on the Steinmetz
"approximate" equivalent circuit. A uniform flux density is assum.ed throughout the
,
4.6 CONCLUSIONS 69
ml m2 hi b2 E
0.0616 3.8.-4 0.0 1.69 0.750
Table 4.11 Manapouri transformer B-H curve byperbolli. Ilarameters.
(,)
IJ.8t.V x,=().1l26 220kV
,
"
,
rv
6
~
-
- (b)
Figur e 4.21 UMEC Manapouri three-l imb three-phase transformer model test system: (a) open-
circuit, (b) shorl-circuit.
core legs and yokes, individual winding flux leakage reactances are combined and the
magnetising current is placed on one side of the resultant series leakage reactance.
The core flux is calculated using the voltage of only one transformer winding. Labo-
latory and simulation results were used to show that this representation wiU produce
inaccUl"ate division of magnetising currcnt bctween the tra.nsformer windings.
The dependence of core flux 011 a single tra.nsformer winding, the arbitrary place-
ment of rn.a.gnetising current, the uniform o r ~ flux a..'>Slunption, and lumped leakage
reactance approximations can be eliminated with the use of the UMEC. The single-
phase UMEC model has been shown to be representative of the Steinmetz "exact')
equivalent circuit. The simulated division of magnetising current between the UMEC
transformer model windings was verified with the labora.tory rcsults.
A test system derived for the New ZcalaJ..Id HVdc converter station was used to
V;ineto merm .. (k V)
r, (!J.) r, (!J.)
No load ratio 13.8 0.00458 1.000+ 7
No load 13.8 0.00458 1.0007
Short-circuit 20.16 3.00e-3 3.ooe-3
Tl:Ible 4.12 Mao.apouri three-limb transformer test system eilJctric parameters.
J?actory measurement UMEC PSCAD-EMTDC simula.tion
No load ratio
9.218 (#Jv.)
9.215
3)VL
No load. red-phase current 23.8 (A) 23.2 (A)
Short-circuit leakage reacta.nce 11.3 (p.u.) 11.7 (p.u.)
Table 4.13 Simulation and factory data comparison.
compare the steady-state beha.viour of the PSCAD-EMTDC and UMEC single-phase
models. In this practical test system the PSCAD-EMTDC model led to significant
errors in predicting the magnetising CULTent distribution between transformer windings.
A model for the Benmore single-pbase converter transformers bas been pre-
sented and verified with name-pla.te specifications. The single-phase UMEC model was
extended to represent three-limb three-phase transformcl's. A model for the Manapauri
l05MVA, 13.8j220kV three-limb three-phase transformers was created and valida.ted
with fa.ctory test data.

Chapter 5
THE GROUP CONNECTION; MEASUREMENT AND
SIMULATION
A recent CIeRE document published in ELECTRA has described the potential benefits
of a direct <..'QDnectiOD of generators to HVde converters [Arrillaga e.t al. 19930.]' While
many theoretical contributions have been made, no practkal test data has become
a.vailable so far. This chapter reports 011 ha.rmonic tests carried out at the Benmore end
of the New Zealand HVde link operating as a. group connected scheme. The field data
obtained provides an excellent bench mark to further validate the UMEC transformer
model.
5.1 DIRECT CONNECTED GENERATOR HYDC CONVERSION
The direct connection of generators to HV de converters Wa.<I first proposed a little over
20 yearS ago [Calverley et at. 1973). It is envisaged that direct connection schemes
eould be strong contenders where generating stations exclusively feed HVDe convcrtcr
terminals [eIGHE J oint Working Group 11/ 14-09 1993]. The main attraction is a
grea.t simplification of the sending end system, as illustra.ted in figure 5. 1 where the
intermediary filtered ac bus bar is removed. Feasibility studies have shown that the
exclusion of generator transformers, ac filters and associated switchyards would reduce
the station cost by about ODe third [Ingram 1988]. As well as the initial capital saving
fuxther savings accrue as the low number of components should reduce maintenance
costs and improve reliability. The generators are completely independent and variable
turbine speed operation is possible to maximize overall generation efficiency [Arrillaga
et ai. 1992j,{Naidu and Mathur 19891. Furthermore, generator self excitation, sub
harmonic oscillations and tra.nsient ovcrvoltage phenomena arc eliminated.
It has been suggested [Arrilln.ga. et ai. 1991J that, while the lmit connection ca.n
be designed to provide any specified nomina.! power, the absence of filt ers may limit the
operational capability at larger currents levels and thus rcd.uce the ability of the link to
provide temporary power increases. In the absence of filters the generator must supply
all the reactive power of the converter. Other disadvantages are that characteristic
72 CHAPTER 5 THE GROUP CONNECTION; MEASUREMENT AND SIMULATION
,.,
,.,
Figure 5.1 HVde cOllversiou: (a) conventional scheme, (b) unit connection.
5.2 GROUP VERSUS UNIT CONNECTION 73
harmonic currents flow in the generator statOr windings causing additional losses and
parasitic torques, and the ha.rmonic voltages imptessed upon the sta.tor windings cause
extra insula.tion stresses.
Issues relating to the design and rating of generators a.rc becoming increasingly
the focus of unit connection studies. Hence while the steady state capability, opera-
tional aspects a.nd transient beha.viour are well documented from a. theoretical point of
view, there is no practical information on the operation of such schemes and particularly
the harmonic content and required ratings of the generator units involved.
This chapter reports on recent harmonic tests carried out at the BClUnore end
of the New Zealand HVdc link operating in the group connected lllode. The field data
obtained is used to show that the UMEC Norton equivalent representation is an ideal
interface to realistic synchronous generator a.nd valve group models. So far the UMEC
transformer models have only been validated in test systems with linear idealised source
and load representations.
5.2 GROUP VERSUS UNIT CONNECTION
One possible variation of the conventional unit connection concept is the group con-
nection and there are important operational differences between these two schemes. In
the unit connection each machine is directly connected to a 12 pulse converter, forming
a unit, and the de line voltage is raised by adding a number of these units in series. Re-
duced dc power levels may dicta.te the removal of generating units and this will reduce
the dc line voltage and therefore the transmission efficiency. Moreover, the inverter
end will then operate with larger value of /, increasing the reactivc power rcquirements
[Arrill.g. et al. 1993c).
In the group connection the generators a.re connected to a common bus and
therefore the disconnection of generators does not affect the dc line voltage. The gen-
erators require some form of joint speed control and drcuit bleakers are also necessary
to take machines in and out of service.
Regarding generator design there is practically no difference between the unit
and group connections and, therefore, the group connected harmonic test results should
also be a.pplicable to unit connection schemes.
5.3 HARMONIC TESTS
The New Zealand HVdc link transmits surplus hydro-power from the South Isla.nd
to major load centres in the North Island. The recently upgraded link consists of a
parallel arrangement of two 12-pulse mercury-axe valve converters on pole 1 and a-
new thyristor-valve 12-pulse converter on pole 2. The older pole 1 mercury-arc valve
equipment a.nd the new thyristor pole 2 are rated for 270kV and 350kV respectively.
74 CHAPTER 5 THE GROUP CONNECI'IONj MEASUREMENT AND SIMULATION
The nominal line current is 2kA and hence the overall maximum continuous power
transfer is 1240MW from Soutb to North.
5.3.1 The Benmore HVde Terminal
Figure 5.2 extends the figure 4.17 test system to display the Benmore pole 1 terminal.
When either of the interconnecting transformers, T2 01' T5, are removed from service
poles lA or lB can operate in group connected mode.
Figul"e 5.2 PoLe 1 of the New Zealand HVde link.
By arrangement with Tra.ns Power New Zealand Ltd the system was available
for harmonic tests in the group connection (i.e. with no filters and the ac system
diSCOlmected) during a. recent period of maintenance for transformel' T2. The maximum
power tl'a.nsmitted under this condition was 212 MW (80 % of full power) with three
generator units (Gl-G3) connected on pole lA.
5.3.2 Measurement Equipment
The harmonic measurements were made with a Hewlett Packard data aquisition unit
HP-3565 (DAQ) which forms part of the Benmore instrumentation to capture system
5.4 GENERATOR RATING 75
transients. The usua.l DAQ traces are derived from :filtered control system waveforms
and are consequently of no use for ha.rmonic studies.
In the group connection the voltage and current waveforms at the generator 16kV
bus-ba.r are severely distorted. This bus-bar is the most effective point in the Berunore
test system to compare simulated and measured wa.veforms. Three extra cha.nnels werc
installed to record the generator terminal voltage in one phase and the corresponding
currents in each of the two transformers of the 12-pulse converterj these two currents
were subsequently added to derive the total generator stator current.
It was assumed that the three phase waveforms were balanced. A thorough
harmonic nualysis would entai l measuring the three phase generator voltage and current
quantities; however this option was expensive in terms of technicia.n cost and computer
storage of all the sampled data. Other existing channels of interest were the dc voltage,
the pole lA dc current and the converter firing angle.
5.3.3 Experimental Procedure
The pole lA mercury-arc valve converter was deblocked and a steady operating point
of 350 Amps, 260kV at 15 firing angle was achieved. Each waveform was sampled at
8192 samples per second for a 5 second interval. Further half-pole de current increments
were performed, and after allowing sometime for the system to stabilise, the waveforms
were again sampled. In practice it was difficult to fix. the frequency at 50Hz especially if
the machines were lightly loaded. Small f1:eqncncy devia.tions were experienced over the
sampling period and eonsequenUy care was exercised when the spectra of the waveforms
were computed. In future tests it is recommended to use a. higber sampling rate and
more accura.te FFT processing taking into account the variation of the fundamental
frequency during the test, such as provided by the CHART hardware [Miller 1992].
The dc current range varied from 350 to 785 amps. In all cases t he uncltarac
teristic harmonic levels were insignificant. Moreover, the presence of de side filters also
made the de voltage harmonics negligible. This chapter therefore discusses only the
generator characteristic barmonic content.
5.4 GENERATOR RATING
The possibility of a direct connection was envisaged in the design of the Benmore
genera.tors although rest rictions were imposed on the nominal power rating (lowered to
102.5 MVA) and the current harmonic content under such operating conditions. The
ratings are listed in table 5. 1 toget ber with the actual maximum current harmonics
monitored during tbe tests. Clearly none of the specified maximum limits were exceeded
at the power levels involved. The maximum total harmonic distortion (THD) of the
generator current was determined to be 5.4%.
76 CHAPTER 5 THE GROUP CONNECTION; MEASUREMENT AND SIMULATION
Harmonic rating (pu) recorded (@212 MW)
11th 0.0744 0.0422
13th 0.0595 0.0267
23rd 0.0193 0.0067
25th 0.0154 0.0037
Table 5.1 Machine stator current harmonics.
If the dynamic simulation results are extrapolated in the high CUl'l'ent region
to the rated half- pole power of 270 MW, the rated harmonic currents are still not
exceeded.
Conventional generators normally have extra thermal capacity to sustain 10%
of negative sequence loading [Uhlman 1975J; however negative sequence is practically
absent in the unit connected mode and the extra capacity compensates for the harmonic
loading. This was verified for the Bcnmore generators after the original coIIlIIlissioning
in the mid 1960's, when the rated machine power output (90MW) was confirmed for
group connected 12-pulse operation.
5.5 GROUP CONNECTION SIMULATION AND FIELD DATA
COMPARISON
The ability of dynamic simulation to predict the harmonic content of the group con-
nection was assessed by comparison with the experimental results. Each converter
transformer within the deltardelta and delta-star three-phase banks were represented
with the new single-phase UMEO model. The simulation parameters of the UMEC
transformer, synchronous generator, and converter models are given in appendix 1.
Simulations were run for sufficient time to reach the steady state. The resulting voltage
and current waveforms were then processed, using the FFT, to derive their harmonic
spectra.
Figures 5.3 and 5.4 display the machine current and voltage harmonic level
comparisons. Overall the harmonic current levels are quite close to the test results, as
they arc less influenced by the generator operating conditions and more by the dc link
firing a.nd current control (which were fixed during the tests). The experimental and
simulated voltage trends are consistent, their magnitude discrepancies being sensitive
to uncertainties in the available generator parameters.
To give some perspective to the difference between the harmonic estimations,
for a dc line current of 405A, figures 5.5(a) and (b) compare a single cycle of simulated
generator current and voltage respectively with the measured data. Although there
are differences in the plotted per-unit values of figures 5.3 and 5.4, the time domain
waveforms seem to be in reasonable agreement.
5.5 GROUP CONNECTION SIMULATION AND FIELD DATA COMPARISON 77
"r----.-----.-----------------------
...
--

.m,
...
0.005
u u . 7 ..
do <Un'<"' (WI.J
(a)
.m'r----.----------------------.-----,
'2 .00
. .
!
...
~
~ ~ ~ ~
----
----- -------
.. .. ..
.,
4e c ...... , (kA)
(0)
..
.m',----._--____ --__ ----_. ______ ----,

i
li . m

'"
..
---- ---- ------ ------ .-----
----
... M
u ..,
dCOQ[l<.'M)
(e)

..=
-;; c.""
..
!
.m
MIS
.. ,
(b)
'r--------------.----.----------

. =
0.0'
IIA U 0.0 u
dC<IllJ"'(lcA)
(d)
. ~
.m,
. .. ,.
...
(I)
F igUl'e 5.3 Harmonil; comparison of recorded generator current with steady state and Jynamk sim-
ulation; Solid=recorded data, Dash-dash=SLmlliation: (a) 11th harmonic, (b) 13th harmonic, (c) 23rd
harnlOllic, (d) 25th harmonic, (e) 35th harmonic, (f) 37th harmonic.
78
,.
..

,
.M

..

1
o.

om
\,
..
,.
,.

o.

1

..
..

G,.'
\,
,.
o.

j OM

..
J
..

'M

\,
CHAPTER 5 THE GROUP CONNECTION; MEASUREMENT AND SIMULATION
---
---

l
t
J
"
.. ..
"
..
ok"""".'<kA)
(a)
-;;

t
1
----
-,

M .. .. .. ..
..
(0)

t

.. ---_.-
---
o. .. ..
'.'
.0
de ..... A'(kA)
(e)
..

om
.. "

, .
u .

. ..

.M

.. "
OM
O
...
I.
..

o

.M

'M
..
G."'
\.
-0 _
---
---
O. ..
"""""III(tA)
(b)
..

.. .. ..
"
..

(d)

o. .. .0 .. ..
dcCIItIUI'{V.)
(f)
Figul'e 5.4 Harmouic comptu:i.soll of recorded generator voltage with yuamic sim\l lation;
Soli d=rcoorded elata, Dash-dash=simulation: (a) 11th harmonic, (b) 13t.h harmonic, (c) 23r har-
monic, (dl 25th harmoni c, (c) 35th harmonic, (f) 37th harmouic.
5,6 CONCLUSIONS
o . o o ~ 0.* 0._ 0.001 . , MI. 0.014 Ult Mit ~
limo (I)
(a)

"
, .

j 0
: _$
-,
79
-0 0.002..... ..000 ..... U I 1,111 ..... 0.'.. I IIIt 0.11
~ ( I )
(b)
Figul'e 5.5 Group collllected geI1Cfator waveform.'1 (3 machines, Idr;=405A); Solid=fidd mr;asurc-
ment, Dash-dot =simulation: (a) current, (b) voltage.
5,6 CONCLUSIONS
The first set of harmonic measurements ever a.vaila.ble of a direct connected generator
BV dc converter scheme have been collected and discussed in this chapter, The har-
monic current levels derived from the tests were all well below the specified maximum
ratings of the generators and therefore the results provide the much needed evidence
of the feasibility of the unit connection.
The test results were also used to investiga.te the performance of the UMEC
transformer model when connected to the PSCAD-EMTDC synchronous machine and
converter representations, Even in the highly distorted case of the group connection
the simulated test system, inclusive of the UMEC transformer model, provides realistic
generator bus bar waveforms. The Norton equivalent representation correctly pbase
shifts and scales tbe non-sinusoidal converter at currents to the transformer primary
windings.
The experience gained with these tests will be extremely useful when the next
opportunity to opera.te Benmore in group connectioll mode a.riscs. In fnture tcsts it is
recommended to use a highcr sampling rate aJld more accurate FFT processing taking
into account the va.riation of the fundamental frequency during the test .
Chapter 6
THE SMOOTHING TRANSFORMER
This chapter descri bes an alternative concept to conventional de filters for reducing
the voltage harmonic contellt on the de side of an HV de converter. The scheme uses
a. transformer instead of a smoothing reactor, but the connection is such tha.t the
conventional transformer model of the PSCAD-EMTDC program can not be used to
simulate the schemes performance. Instead the UMEC model developed in chapter 4
had to be used to simulate the iron core saturation.
6.1 HVDC CONVERTER DC SIDE FILTERS
Unlike ac filters, which provide reactive power compensa.tion to the convclter , de filters
are used exclusively to reduce the de line harmonic content. Therefore the design of de
side filters, although a complex exercise [futi 1994], permits greater flexibility in the
selection of filter components to try and minimise the cost of the COllverter planL.
Compacting the HV de tel'miuai has been investigated as a means for cost re-
ductions [Asplund et al. ]. The outdoor valve [Astrom et al. ] has been suggested
to significantly reduce the dimensions of the valve housing, and active dc filters are
proposed to decrease the area needed for the dc switchyard.
The direct connection offers the advanta.ges of compact HVdc as well as variable
generator opera.ting frequency. If the ac frequency supplying the converter is variable
then the effectiveness of tuned dc side filters reduces. Recently it has been shown that
the variation of harmonic voltages in the de system will cause the conventional filters
to function ineffectively [pang and Tarnawecky 1995] and that act ive or on-line tuning
filters are required.
Active dc filters have been considered as an alternative to the conventional pas-
sive component design [Cheuksun ct al. 1989J. The active filter injects harmonic cur-
rent back into the converter to "neutralise" the harmonic currents produced by the
converter. Active filters are more complex than passive filtcrs and they require ac-
curate detection of harmonic voltage magnitude and phasc, at the transmission line
82 CHAPTER 6 THE SMOOTHING TRANSFORMER
terminals. Also, the current rating of the active harmonic sOurce must be large enough
to compensate the harmonics produced by the converter.
This chapter presents the smoothing transformer principle and describes initial
feasibility investigations. The primary aim of the smoothing transformer is a reduction
of the de side harmonic content at nominal and off nominal a.c system frequencies in a.
compact configuration.
6.2 THE SMOOTHING TRANSFORMER PRINCIPLE
The smoothing transformer operation is easily described with reference to the ciJ:cuit
of figme 6.1, using ideal components. During steady-state the de side capa.citor Cis
charged to Vde. If G is infinitely large then all harmonic voltages Vii produced by the
converter aJ:C developed across the primary winding of the smoothing transformer (the
capacitor winding). If the transformer ratio is set to unity and the coils connected
oriented as shown, then the harmonic voltages Vh are developed across the secondary
winding of the smoothing transformer (the de line side winding). The resulting voltage
at the terminal of the smoothing transformer is the dc component of the rectifier
terminal voltage Vdc.
Vh

Vd,l
L$
!V, =Vh+VdO
Vh
Vdol I C
Figure 6.1 Ideal smoothing transformer opera.tion,
If the smoothing tra.nsformer is ideal, with perfect coupling and no magnctising
current, then no current will Row in the capacitor circuit and the Volt-Ampere rating of
the primary winding and the capacitor is minimal. The converter ha.rmonic voltages are
cancelled out, rather than filtered from the transmission circuit. The ideal smoothing
transformer will eliminate all de side harmonics because, unlike conventional filters, it
is not tuned to fixed frequencies.
6.3 SMOOTHING TRANSFORMER SATURATION 83
6.3 SMOOTHING TRANSFORMER SATURATION
An unsaturated iron-cored smoothing transformer will exhibit the high level of coil
coupling required for successful dc side harmonic voltage cancellation. However, the
secondary winding of the smoothing transformer carries the full dc line current . Unless
the transformer core cross-section is unrealistically large this cunent will for ce the iron
CQ1'e deep into saturation, and reduce the coil coupling severely.
The new UMEC transformer model is better suited than .the PSCAD-EMTDC
single-phase repl'csentation to determine the effects of smoothing transformer satu-
ration. The magnetising component of the non-linear two-winding PSCAD-EMTDC
transformer model must be injected. across either the primary or secondary winding.
However, in t he case of the smoothing transformer, the magnetising current can not
be injected acrOSs the capacitor winding because dc component present in the mag-
netising current will charge the dc blocking capacitor to an unrealistic voltage. The
only remaining option is to inject the magnetising current across the dc line side wind-
ing. However, if the inverter and dc transmission line are represented by a constant dc
Cm-rent source (as was the case for the Benmore group connection validations) the ac
component s of the magnetising current can not flow in the dc Dne side winding. The use
of the UMEC transformer model avoids these problems since the magnetising current
will correctly distribute itself amongst thc capacitor and dc line windings according to
the system impedances at the transformer terminals.
The test system for the PSCAD-EMTDC solution consist s of a three-phase ideal
source directly connected to a six-pulse controlled converter feeding 8. purely resistive
load. The non-Dnear UMEC transformer model (in the smoothing transformer con-
figuration shown in figure 6.1) is connected between the convcltel' and dc load.. The
source, converter, smoothing transformer and load parameters are given in appendix
J.
The behavioUl' of a hypothetical Dnear iron-core smoothing transformer was first
examined. lolgure 6.2(a) displays the wall bushing, and de Due voltages for this test
case. Without saturation, even when a. dc cunent flows in the transformer winding, the
coil coupling remains high and the smoothing transformeJ' almost completely cancels
the wall bUShing voltage harmonics.
To asses the effect of saturation two extreme operating conditions were selected.
In both examples the iron-core winding-limb saturation knee point resides at a. flux
level of 1.03 (p.u.). In the first case the convcrter firing angle is set to 15 and the
iron-core smoothing transformer is operated deep in saturation. Figures 6.2(b) and (c)
present ~ h dc harmonic voltage cancellation and winding-limb flux. respectively. The
voltage waveforms show timt the smoothing transformer still Significantly reduces the
dc line harmonics, and the flux waveforms verify that this cancellation occurs entirely
in the saturated region.
84
~
11,0.7
i
..
"

(b)
'\
~
\
.. --" -',
~
"'" .' "
. '. " "'.
, ,
,

1\
'--"
.. "'.
CHAPTER 6 THE SMOOTHING TRANSFOR.\tIER
\
_ . J :._
(a)
'.'

'.'
' .
J.t. ' . t ..... t ,'"
OMI .... . .... ._ __ . _
limo (oj
(e)
r ~ ~ - - - - - - - - - - - ,

...
1, ...
i1

ttl ,.Nl UU ' .11'
'M'
.. un ' ..... .HI U ..

' .MI I.'" . ... , I ,flt ' .M I .m , .... ' .... , .... t
... "
(0) (e)
Figure 6.2 Iron-core smootbi.ng t ransformer de side harmonic voltage cancellation: (a) linear case,
(b) Ilnd (e) non-linelle elISe 0'=15", (d) and (e) llon-Iinear r.a.se 0:=60".
6.4 REALISTIC DESIGN CONSIDERATIONS 85
Deep in saturation the smoothing transformer magnetisillg current is linearly
proportional to winding volta.ge. Therefore, although a less likely operating condition,
the second case investigates the level of cancellation for operation that traverses the
saturation knee region. A 60 converter firing angle was used to obtain a greater flux
va.riatioll , and the dc current was lowered to 0.18 (p.u.) to bring opel'ation below the
knee region. Simulation time was lengthened to a.llow the de side waveforms to stabilise
lroder the lightly loaded conditions. Figures 6.2(d) and (e) display this operating
condition. Again the smoothing transformer significantly reduced. the wall bushing
voltage harmonics.
The wa.veforms of figure 6.2 provide only a qualitative assessment of the perfor-
mance of the smoothlng transformer. Thcy show that substantial voltage cancellation
still occurs when saturation reduces coil coupling. Altbough the leakage paths sepa.rate
the prima.ry and secondary winding-limb flux by a significant dc component, the har-
monic flux magnitudes are similar. Also, the primary and secondary winding voltages
are simila.r even if the coil coupling reduces and thc smoothing transformer experiences
very little harmonic load.
If a saturated iron*core smoothing transformer provides reasonablc voltage can-
cella.tion, an air-core design should also provide an effective solution. Since the air-core
design is linear, the PSCAD-EMTDO mutually coupled coil representation is equally
suitable as the UMEC model for linear smoothing transformer representation.
6.4 REALISTIC DESIGN CONSIDERATIONS
Figure 6.3 displays how the conventional air-core smoothing reactor may be modified
with the addition of a low current primary winding to create the compact smooLhiug
transformer.
Primary winding
Secondary winding J
Figure 6.3 Impression of smoothing transformer section.
The viability of the smoothing transformer principle is dependent on its steady-
state and transient performance using rca.listic components. Sizing the capacitor will
affect performance. The capacitor is thc most costly element of conventional de side
filters [Testi 1994J and is restricted in size. Even if the smoothing transformer capacitor
can bc designed to pass minima.l harmonic cllrrcnt it is stressed by the total dc voltage
in the same manner as the conventional filter. In line with present HVde filter designs
the value of capacitance is limited to one micro-farad [Arrillaga 1983J.
Oue coil of the smoothing transformer carries the full de line current. As shown
in the previous section this current would force a. steel core transformer deep into
saturation and ma.y not provide suitable coil coupling for acceptable harmonic voltage
cancellation. An air cored transformer will not saturate, but will offer poorer coil
coupling. As the smoothing transformer essent ially operates on no load, poorly coupled
coils may be sufficient to provide harmonic voltage cancellation. For this example the
self inductance of the coil carrying the de line current is set to a.161f, this is equaJ to
the NZ HV de Pole 2 smoothing reactor inducta.nce.
L12
( ' y
Ln L22
- - - ~ - - - - ~ r ~ - - - - - - 0
1
VIl

NI N2
vet C
T
Figure 6.4 Harmonic voitage cancellation principie.
As the capacitor value is finite and restricted in size, the transformer ratio cannot
be sct to unity. 'rhe design for the ca.pacitor winding is found from a. steady-state
analysis of the circuit of figure 6.4. The transformer secondary is open-circuit and
applying Kirchoff's voltage law around the primary circuit gives
(6.1)
where Wh is the angular frequency of the harmonic voltage source.
Under no load cond.itions, the transformer prima.ry voltage is related to the
secondary voltage by
(6.2)
where a is the transformer turns ratio.
Rearranging equation 6.1 to solve for the capaci tor circuit CUlTent and multiply-
ing both sides by jWhLll gives
87
(6.3)
Utili7,ing Vt = ijwhLll and equation 6.2, equation 6.3 can be rewritten as
(6.4)
.1"01' harmonic voltage cancella.tion to occur
(6.5)
Substitution of equa.tion 6.5 into 6.4 gives
(6.6)
The transformer turns ratio a can be defined in terms of primary and secondary self
inductance.
(6.7)
Substitution of equation 6.7 into equation 6.6 gives 8. quadratic equation in terms of
the primary will(liug self inductance Lll
(6.8)
where the two solutions of equation 6.8 (ie q=O in figure 6.5) provide possible values
for the self inductance of the smoothing transformer primary winding. With the self
inductance of the secondary winding L22 set at O.16H, the quadra.t ic equation 6.8 is
plotted in 6.5 for various capacit or values.
The two solutions for the quadra.tic equa.tion 6.8 are given by
(6.9)
In the case of the positive root , < w/tL
Il
and the current i will lag the
voltage Vh by 90
0
(the inductive solution). The phasor diagram for this solution is
shown in figure 6.6(a), LII will alwa.ys be greater than L 22 and the transformer ratio
a will be greater than unity. As C -+ 0, it is simple to show that L11 -+ 00 and as
C -+ 00, L11 -+ L22

O.OB
0.06
0.04
0.02
-0.04
- 0.06
_(1.06
Capacitive solution Lu=O' 199H
c..O.1!o1F
-O.1
0
L
--70-----c07---:'O'--- 07- --''-e------'
0.2 0.4 0.6 0.8
self inductance (H)
Figure 6.5 Primary winding self inductance; Da.,h-dash=NZ HVde Pole 2 de linc filter capacitor.
In the case of the nega.tive root, w:c > WhLll and the current i will lead the
voltage Vii by 90 (the capacitive solution) . The phasor diagram fOl" this solution is
shown in figure 6.6(b), as C O,L11 -? 00 and as C -? 00, L11 --7 O.
Utilizing the parameters G=O.923I-tF, L
22
=0.1611 and Wh sd to the angular fre-
quency of the NZ HVdc link 12th harmonic (50Hz=rated frequency), the inductive
solution give.'> Lll =O.292GH > "2, and a=1.35 > 1; the capacitive solution gives
L
u
=O.0199H < L
22
, and a=O.353 < 1.
Once the capacitor and the self inductance of the secondary winding are selected,
the choice remains between the capacitive or inductive solution, Both solutions provide
tota.l cancellation of a single harmonic voltage at angula.r frequency W1
1
, Rearranging
equation 6.4 gives
(6.10)
:ror harmonic voltage cancellation to occur the gain of equation 6.10 1&1 must
be unity and the phase 0
0
or 180") depending on the orientation of the secondary coil.
The solutions of quadratic equation 6.8 guarantee the correct gain and phase for a
single harmonic voltage at angular frequency Who Consideration must be given to the
smoothing transformer operation at other frequencies; reinforcement of charactedstic
converter voltage harmonics would be unacceptable. Figweg 6.7(a) and 6.7(b) display
the gain and phase plots for the inductive and capacitive designs using the NZ HVdc
Pole 2 parameters. The angular frequency Wh is set at the 12th harmonic.
6.4 REALISTIC DESIGN CONSIDERATIONS 89
Yh
Y
e
I)

V1
(a)
Y1
I)
Yh

Ve
(b)
Figw'e 6.6 Induct ive and capacitive harmonic cancellation solutions; (a) the posit ive root, (b) the
negative root.
Although in hoth cases at the 12th harmonic the gain is unity and the phase
either 0" 01" 180
0
, the characteristics a.re very different. In both solutions equation 6.10
shows that as WI
1
-t 00 the gain approaches ;. For the parameters used to create figure
6.7, i=0.74 and 2.83 for the inductive and capacitive solutions respectively. In this
example, the capacitive solution is clearly not an acceptable design.
The gain of the inductive solution for harmonics above the 12th remaiDs close
to unity, and at worst becomes equal to the inverse of turns ra.tio, 0.74. This implies
tha.t a smoothing transformer with this design would remove 74-100% of the harmonic
voltage content above the 12th.
Both cases exhibit a resonance point. In the inductive solution resonance occurs
at a non-ebaracteristic frequency just abovo the sixth harmonic (300Hz) . The angular
resonant frequency is calculated from
(6.11)
Combining equation 6.9 with 6.11 shows that as C -to the resonant frequency wn -t WI..
Fjgure 6.8 displays the variation in inductive solution resonant frequency with different
values of capacitance.
(a) (b)
Figure 6.7 Smoothing transformer gain and pha:;c; (a) Inductive solut ion, (b) Capacitive solution.
The resonant frequency can be altered with small variations in de blocking capac-
itance. If the resonant point of the smoothing transformer excites a non-characteristic
harmonic to an unsatisfactory level a series resistance R$ can be added to the primary
circuit to alter the gain of the smoothing transformer. Equation 6.10 is then replaced
by
V2 1 jWhLn
V
h
= -;;. jWhLu - w;c + R&
(6.12)
The gain and phase plots for the smoothing transfOl'rocr with Rs = 2100 are shown by
the dash-dash line of figure 6.7. This gain redudion ensures that harmonic reinforce-
ment will not occur, but the harmonk voltages on the secondary side of the smoothing
transformer are phase shifted and harmonic cancellation at frequencies of interest is
less effective.
6.4.1 Test system
ResonaJ.lce and coil conpling effects are best quantified with simulation of a. smoothing
transformer in a realistic HVdc test system. The chapter 5 field data verified PSCAD-
EMTDC model for group connected operation of the New Zealand HYdc link has
been selected for analysis of the smoothing transformer principle; all example solutions
presented above are relevant to NZ HYde link parameters and the group connection
offers variable frequency operation. The de filters operating on Pole lA and Pole
2 arc identicaL The Pole lA smoothing reactor is of the old iron-cored type and
the Pole 2 reactor the newer air-cored design. This feasibility study aims to include
the latest COmpOIlf'Jlt defdgns and for this reason the Pole 2 smoothing reactor has
been included ill the Pole lA test system. The section 5.5 synchronous generator and
converter model parameters are re-applied, tbe PSCAD-EMTDC converter-transformer
91
600
500

,

,

,
u
,
"
,
u
Ln=O.OIH
0
10300
.t:
"

ZOO

'" J. cO.8H
__ L22=O.16H
..o.3H ---- ___ _
-- -- -
- ----
"'"
. .
Capacitance (F) x 10""'
Figure 6.8 Resonant frequency variation; Dasb-dash=Ln equal to t ht! NZ HVdc Pole 2 smoothing
reru;;tor inductance.
and transmissioll line pru:ametcrs are given in appendix K,
The smoothing transformer design selected in the previous section is used with
the te.<;t system, that is Ll1=O.2926H, L22=O.16H and C=O.923 J.1F. The ideal cou-
pling factor (k=1.0) used for best harmonic cancellation is unrealistic for a transformer
with a.n air core, The coupling factor selected at k=O,96 uses as a basis a.n air cored
laboratory transfOl'mer that was tested to exhibit a coupling factor of this order.
6.4.2 Steady-state p erformance
The Benmore test system was run to steady-state a.t full Pole l A dc line voltage (270kV)
and current (lOOOA) using PSCAD-EMTDC. Harmonic voltages were calculated at
the rectifier wall bushing and at the start of the de transmission line for two cases.
Case A contains tile Benmore smoothing reactor and the dc side filters (the design
for the Benmore Pole 1A dc linc filters are given in appendix K). Case B utilizes
the smoothing transformer design selected above with the coil coupling factor k=D.96.
Table 6.1 presents the conventional filter aud smoothing t ransformer performance for
Cases A and B a.t 35Hz, 50Hz (rated frequency) and 55Hz geupxatol" frequenc.y. The
performance mea.'mre fol' eaeh harmonic is given by
% harmonic reduction
_ 100(1 _ V"",)
V
wbv
(6.13)
92
where
Vdd
CHAPTER 6 THE SMOOTHING TRANSFORMER
The harmonic voltage present at the start of
the de transmission line
The harmonic voltage present at the
converter wall bushing
35Hz 50Hz 55Hz
Case Case Case Case Case Case
A B A B A B
(%) (%) (%) (%) (%) (%)
V"
-55 43 97 97 85 92
V'4
83 78 95 75 93 73
V"
95 76 93 79 87 73
Table 6.1 Harmonic reduction at variable frequency.
The smoothing transformer has reduced the characteristic voltage harmonics
produced at the rectifier wall bushing. At the 12th harmonic for 35Hz operation
(420Hz) the smoothing transformer cancellation falls to 43%. This operating condi-
tion is near the primary cir cuit resonance point (306Hz) and in this example prima.ry
circuit resistance Rs has not been implemented to control the smoothing tranl:iformer
gain. In comparison at this operating point the conventional filter results in a 155%
reinforcement due to its own resonance between the smoothing reactor and dc blocking
capacitor.
Over aJ.l operating frequencies examined the performance of the smoothing r n s ~
former is either equal or superior at the 12th harmonic and slightly poorer at the 24th
and 36th as compared with the conventional filter . However the 12th harmonic is by
far the dominant dc side component of HV dc converters a.t all generator operating
frequencief>.
As postulated previously, the harmonic current ratings of the smoothing trans-
former components can be less than tha.t of the conventional filter. Table 6.2 presents
the rIUS current flowing in the conventional filter and smoothing transformer primary
circuit for the variable frequency operation.
Conventional filter Smoothing transformer
current (Arm., ) primary current (Arm!)
35 (Hz) 12.21 10.05
50 (Hz) 9.05 8.73
55 (H,) 10.30 9.71
Table 6.2 ConventionaL filter and smoothing transformer harmonic current.
6A REALISTIC DESIGN CONSIDERATIONS
93
6 .4.3 Transient performance
The inclusion of the smoothing transformer on the de side of a converter must not
impair the transient performance of the HVdc link. A dc line-to-ground fa.ult and
resta.rt has been selected to exa.rn.ine this performance. Dc faults occur due to broken or
contaminated insula.tors, icing insulators, failure of cables and flashover due to lightning
strike and internal over voltages [Hingorani 1961J.
The fault is applied at the de line side of the smoothing reactor or smoothing
transformer of the NZ HV dc model. This scenario provides the worst case disturbance,
the highest level of fault current in the l"Cctifier terminal [peterson et al. 1969). The
resulting fault current is larger than an ac line fault or an internal converter fault [Ando
et al. 1985).
The time to extinguish this fault and restart the link is dependent upon the
energy stored in the conventional smoothing reactor [Thrner 1980] or smoothing tranS-
former. The dc line to ground fault is initiated at time 3.0s. The fault resistance
affects the fault current decay a.nd damping. A zero impedance fa.ult is unrealistic.
Although an accura.te fault representation must include the tower footing resistance
(typicnily 200: [Hingorani 1970)) and complex arc resistance, for this comparison fl.
fault resistance of 50n has been selected (Heffernan et al. 1981).
Utilizing a simila.r fault detection delay to that of [Heffernan ct al. 1981} the fault
control begins a.t 3.02s, one cycle after the fault initiation. ill typical control action
[Arrillaga et at. 1993bJ the rectifier firing angle is advanced to 120
0
and the invertel
left on constant current control, with a 100 limit to the firing angle advance.
At tillle 3.5s a. restart is ordered to the rated prefault conditions. Figures 6.9
and 6.10 display the simulated rectifier end dc line volta.ge and current respectively.
(a) (b)
Figure 6.0 Fault initiation and recovery, PoLe lA, NZ HVde link DC line voltage; (It.) Case A,
convllnt ional system, (b) Case B, smoothing traJ.Isformer (k=O.96).

"
11_(0)
(0) (b)
Figure 6.10 Fault initiation and recovery, Pole lA, NZ HVde link DC li ne current; (a) Case A,
convent ionnl system, (b) Case B, :;moothing transformer (k=0.96).
The replacement of the smoothing reactor and cODventional. de filters with the
smoothing transformer does not degrade the transient performance of the de link under
a. de line-to-ground fault. In each simula.tion pea.k fault currents and decay times are
almost identical. The link restart to r a.ted conditions is not slowed by the presence
of the smoothing transformer. Moreover the smoothing transformer does not induce
overvoltages or oscillations that are dissimilar from the conventional system at the
converter wall bushing or on the de line.
6.5 CONCLUSIONS
The UMEO transformer model has been used in a. preliminary investigation of a new
design concept, the smoothing transformer, as a means to r educe cost and increase the
operational fl exibility of an HVdc terminaL In this design the smoothing reactor and
conventional filters arc replaced by a single smoothing transformer and de blocking ca-
pacitor. Unlike the conventional filters, the smoothing transformer achieves substantial
harmonlc cancellation at off nominal funda.mentaJ frequencies, an essential requirement
in the direct connected schemes with adj ustable speed. The UMEC representation was
required to asses the feasibility of an iron-core smoothing transformer because the
ventional non-linear PSCAD-EMTDC transformer model could not be applied to such
an investigation. Investigations show that even smoothing transformer operation en-
tirely in the core saturated region provides acceptable dc side harmonic reduction. The
iron-core can therefore bc removed and a linear air-core smoothing transformer design
postulated.
simulation studies using the NZ HVde group connected test system
have shown satisfactory performa.nce of the air-cOl'c smoothing trallsformcl" These
6.5 CONCLUSIONS 95
studies also indicate that the current harmonic ratings of the smoothing transformer
de blocking capacitor are less than those of the capacitor of a conventional. filter.
The transient response of the dc link to a de short-circuit, both in terms of peak
fault current and recovery voltage, have been shown to be practically the same as those
of a conventional scheme.
Chapter 7
HVDC CONVERTER TRANSFORMER
COMPARISONS
Unlike transmission lines, where the interphase mutual coupling is often incorporated
in fault studies, the strong mutual coupling betwccn three-limb transformer phases bas
been largely neglected. The literature reports on the effect of such coupling in relation
to the overvoltagcs present on disconnected phases when three-limb transformers arc
energised by asynchronous circult-bl'cakel' closing [Smith ct ai. 1975] ICeJ.man and
Davies 1981J [Dijk 1986J [Lin et al. 1995J .
In HVde transmission studies, transformers are normally represented with mag-
netica.l.ly independent single-phase units. However the response of a three-phase COD-
velter transformer bank to a. single phase to ground fa.ult is expected to be different to
that of the three-limb three-phase transformer. A single phase to ground fault is used
in this chaptcr to compare the transient behaviour of HVde converter tra.nsformers.
7.1 A THREE-LIMB CONVERTER TRANSFORMER
To demonstrate the importance of incorporating core construction details in PSCAD
EMTDC transformer models, the proposed UMEC representation is used to compare
the beha.viour of equivalent three-phase bank and three-limb converter
transformers. For a fair comparison, it is important that the transformer types draw
similar magnetising currents in the steadystate. Unforhmately, it is difficult to achieve
a perfect match, because three-limb transformers require less core steel and each phase
is energised ill an unbalanced manner.
The parameters of an equivalent Benmore three-limb UMEC transformer model
have been selected to achieve steady-state magnetising currents very close to those of
the single-phase model described in section 4.3.3.
Initially the windings are assumed connected in stargroundcdfstargrounded
configuration. Primary and secondary winding turns munber are set to limit the max
imUlli core flux density to 1.4T, at rated volta.ge, The hyperbola constants for the
three-limb tra.nsformer core BH characteristics are as given in table 4.3.
98 CHAPTER 7 HVDC CONVERTER TRANSFORMER COMPARISONS
-=
~
~
v, v,
, ,
<
,
"
.-
"
v, v,
<
,
<
-=
I
l.JJm
1
133m
,I
Figure 7.1 Benmore three-limb converter transformer core dimensions.
Figure 7.1 displa.ys the three-limb UMEC model core dimensions
l
and table 7.1
gives the UMEC parameters. The primary and secondary winding lealrage permeances
arc set equal to those of the Benmore single-phase units. In a similar manner to the
Manapami three-limb UMEC models (chapter 4), the permeanccs of the zero-sequence
paths are set equal to the leakage path permeances.
Electric circuit parameters Magnetic circuit parameters
Rating 187,5 MVA
Awitlding limb
0,454 (m')
Configuration star-star
Lwinding tim& 3.59 (m)
Frequency 50 (Hz)
Ayoke
0.454 (m')
Vprimo.ry ~ (kV) Lyoke 2.66 (m)
V, ,,,conr/.aTY ~ (kV) Nprimary
65 (turn.,)
X/eakaye
0.113 (p.u.) Nsea:mdnry 450 (turns)
P7,9,ll 5.85e-8
PS,10,12
5.850-8
P l 5 16 17
5.850-8
Table 7.1 Benmore three-Limb UMEC transformer model parameters.
7.1. 1 Steady-state performance
A three-limb three-phase UMEC transformer model was created from the ta.ble 7.1
parameters. To verify the close similarity of the three-phase bank and three-limb
' The three-Umb core is twice as high as the single-phase core of figure 4.10. Althougb such COJl-
st.Iuction is unrealistic, it must be applied in this example for a dose magnetising currp.nt mll.tr.h.
7.1 A THREE-LIMB CONVERTER TRANSFORMER 99
three-phase UMEO models under steady-state operating conditions the two alternative
constructions were subjected to open-circuit, short-circuit and over-voltage compar-
isons. The test system is shown in figure 7.2 and the circuit pa.rameters are listed in
table 7.2 for the three test cases.
2
V me
groun
rm$ (kY)
r. (fl)
n
fl
Open-circuit
,
0.107 1.00,+6

Short-circuit 8.000-3 7.500-3
Over-voltage

0.107 1.00e+6
Table 7.2 Benmore tbree-Iinlb transformer test system electric parameters.
,
, 16kV x ""OJ /3 //OkV
\
F igure 7.2 Benmore HYdc convener transformcr steady-state test system.
Figures 7.3(a) and (b) display the three-limb UMEO open-circuit magnetising
(.'urrent a.nd primary winding-limb flux: waveforms respectively. As expected the mag-
netising current of the yellow-phase in the three-limb core is lower than those of the red
and blue phases. The red, yellow and blue currents reach rIDS magnitudes of O.043kA,
O.037kA and 0.043kA respectively; these currents are close to the matched red, yellow
and blue three-phase bank magnetising currents of 0.047kA. The three-phase mag-
netising currents and winding-limb flux waveforms are in phase, and correctly lag the
winding voltage by 900.
3
The short-circuit three-phase primary winding voltage and current waveforms
are shown in figure 7.3(c) and (d) respectively. The three-limb transformer red, yellow,
and blue-phase leakage reactances equa.te to 0. 1530, O.156n, and 0,153n respectively.
The short-circuit results compare favow:ably with the single-phase leakage reactance
of 0.1560. The three-phase short-circuit currents all correctly lag the winding voltages
by 90.
The three-Hmb short-circuit reactances are slightly unbalanced. The yellow-
phase is slightly greater than the equal red and blue-phases. Such a result supports
the section 4.2. 1 conclusion (although explicit formula were only derived for the single-
phase linear case) that short-circuit inductances are dependent upon core as well a.,.<;
2For the single-phase UMEC tests only the over-voltage scenario required simulation; the open
circuit and short-circuit simulations wcre performed in section 4.3.3
3In all three scenarios the refereoce red phasc to groUlld voltage positive going zero crossing occllrs
at time O.08secs.
rHE L18fiAR'r
URNERS1TY OF CP,NTEA6UAl
N.2.
100 CHAPTER 7 HVDe CONVERTER TRANSFORMER COMPARISONS

O.Q< ,.-
,
,
,
"
-I
..0.01
,
,
.
.'
;'
'.
.'
/
", ,
/
...
O,OR . OM 011" G.OII oM .em 0 . 11 ..... ...... .
tiJno(.)
(a)

,''.
\,
6.S \
,
,

,
,
,
,
,
,
.,
..
..
. .
,
./
\.
r. \
.
,
,
,
, ,
i
/
,

'I
-,
, "
. ,
.

,'.
,:' \.
"
!
/
.
"
,
,
. ':"" . .... 0.0" U" . ... ._ ........... 0.090 . 1
lim< (,)
(0)
E
.'l

,

,
,

,
'.
\.
'.
' .
,
,
!
./
./
,

""' t... o.oU '.06< ..... M il .... o.ooa 0.... .......... 0.1
...... (.)
(h)
,
,
.'
' .
.,
,
,
,

...
,

,
, ,
\,
,
,

i
, ,
' .
,
,
,

,
,
,
, ,
,
i ,
!

-.
,
.
,

,
,
\
,
.'
,
.'
...
,
"
./
V
,.
, . i
-.
. , .
, ,
\
,
,
,./
, , ,
,
,
,
-
,
---
--
0.01 OJllt 0 .... M I. 0. ....... 0_ 01110 O1IH . .... ..,
limo (I)
(d)
Figure 7.3 Benmore three-limb three-phase transformer steady-state tests; SoLid=red-phase, Dash-
dash= ycllow-pbase, Dash-dot= blue-Vhase: (a) open-circuit, primary winding current, (b) open-circuit,
primary winding"limb flux, (c) short-circuit primary winding voltage, (d) short-circuit, primary winding
Cllucnt.
flux leakage paths.
The over-voltage test results for both the three-limb and three-phase bank UMEC
models arc shown in figures 7.4(a.) and (b) respectively. The three-limb red, yellow and
blue magnctising currents are O.078kA, O.067kA and O.078kA respectively Clli compared
with O.086kA for the three phases of the independent transformer banks. This is con
sidered sufficiently close for such excursion into saturation.
7.1.2 Transient comparison
The waveforms produced by the three-phase bank and three-limb three-phase trans
former models in the previous section are similar under open-circuit and shortcircuit
operating conditions. Such similarity, however, is not sufficient to justify the common
practice of ignoring the complexities involved in modelling the core. This
7.1 A THREE-LIMB CONVERTER TRANSFORMER
"
."

i

,
i
i '.
i \
;
..
".
", '",,-
, .' ....,\
, ,
, ,
, ,
'. '
, :
, ,
, ,
, ,
, ,
"
,
\/
"
, '
,
,
,
,
,
,
,
!
i
,
..
/
,
,
,
,
MIZ 0.0'" OJ " 0 . .,., 9,.' OAt;: .... "' oK 0.0" 0.1
Iimo(.)
(a)
. U
..
..... 15
,
i',
i .,
/ \
i "'
" ,
'.
,
".
,
, ,

"
, "
, "
, ,
, ,
, ,
, ,
'.'
,
,
"
..
" , ,
, ,
! ' , ,
, ,
101
.... 0_ o.,*, 0.... .... .... 0 .... "" 0.0" 0.1
limo (s)
(b)
Figw-e 7.4 Benmore steadystate over.voltage transformcr encrgisation; SoJid=redphase, Dash
dash=yellow-phasc, Dashdot=blue-phase: (a) three-phase bank primary winding current, (b) three-
limb three-phase pr imary winding current.
section shows thc errors that such an assumption can causc,
The fictitious system of figme 7.5 has been selected to highlight the diffel'ences
between three-phase bank and three-lirob converter transformers during transient con-
ditions. It consists of a synchronous generator, represented as an ideal source of im-
pedance equal to the machine subtransient rea.cta.nce x" [Arrillaga and Arnold 1990]
and armature resistance rll.' The generator parameters are given in appendix G. The
star-grounded/star-grounded (YyO) conver ter transformer configuration has been se-
lected. The use of a delta. winding connection would complicate presentation of the
three-limb core interphase magnetic coupling effects during unbalanced operating con-
ditions. The six-pulse mercury-a.rc valve converter parameters are as given in appendix
I.
Figure 7,5 HVdc converter transformer transient tc..<;t system.
A single-phase is applied to the test system 16kV red-phase
bus-bar at a nega.tive going zero-crossing of the primary winding red-phase volta.ge
(timc=0,83secs), The O.OHl purely resistive fault is removed (reset to l.OOMn) after
three-t:ycles, at the next OCCUlTing fault current zero-crossing.
A generator circuit-breaker has not been included; this is cornmon practice with
ac connected generator-transformer units where circuit-breakers are only prescnt at
the high voltage transformer terminals. In the latter case, opening of the high voltage
during a fault on the genera.tor bus-bar will give opera.ting conditions
similar to tha.t presented in figure 7.5 (once the converter 1mB blocked).
At the onset of the fault the de line current is ramped down to zero in ODe quarter
of a cycle [Joosten et ai. 1990]. In order to highlight the behaviour of each transfonncr
type the converter is not restarted after the fa.ult is cleared.
The generator currents for the three-phase bank and three-limb collverter trans-
former test cases are shown in figures 7.6(a) and (b) respectively. Although the peak
magnitude of red-phase generator current (82.7 kA) is identical for both converter
transformer types, in the three-limb example the generatol' feeds yellow and blue fault
currents as well. 1n both cases the generator red-phase the fault t hrough the
mainly inductive generator internal impedance; the generator red-phase current lags
the intemal emf by 900.
4

A
f\
f\
__
f\
/\ fI '\
'\

::... ..
"
''-'

"
V
V
V
V

V V

u. .M U'

OM U O.
..... (. j .... (.j
(.) (b)
Figure 7. 6 Generator currents, 16kV red phase to grollnd fault , star-grounded/st.ar-grounded con-
vener t ransformer; Solid=rcd-phasc, Da3h-dash=yellow-phase, Dash-dot=blue-phase: (a) three-phase
bank test system, (b) three-limb three-phase test system.
The primary winding currents for the t hree-phase bank and three-limb three-
phase UMEC models are shown in figures 7.7(a.) and (b) respectively. At the ooset of
the fault the three-phase bank primary current ramps to zero through the conducting
phases. In the three-limb case this ramp is superimposed upon significant fault current
(of peak 23.0kA) on aU t hree transformer windings.
Once the converter currents have ramped to U f O the transformer is effectively
open-circuit. Consider only the transformer primary windings shown in figure 7.8(a)
and the phasor diagram of fi gure 7.8(b). The reference phasor diagram is the three-
limb transformer primary red-phase voltage V
1
prior to the fault. During the faul t
instead of the nominal flux in the red-phase winding-limb, the shorted red-phase
t The reference red phase to ground generator internal emf positive goi ng zero crossing occurs at
time O.8secs.
7.1 A THIillE-LIMB CONVERTER TRANSFORMER 103
"
"
"
"'1",---,,,,- -...,--,;:;---,,,-,,--.
HZ .... .... 0.11 o.n
lin Is)
(a)
""r ----------------,

...
(.0.
om

(0)
"
,.
l.
.lIT] "J TTI J ;
[_'LUJJ

.. c_"'''''-,,_-. uz 0... 0... O.n o,n
1iI:no(')
(b)
.. ,
"
..

"
,

. ,

_1. _1 J-!
..
V
'I.
. "
... UI . ..

..
(d)
Figm'e 7.7 Star-grouncled/star-grounded converter trausformer red j)hase to ground fault (left hand
side=thrce-phase bank, right hand sidc=thrce--limb three-phase); Soli.d=rcd-phase, Dash-dash=ycllow-
phase, Dash-dot=bluc--phase: {a} and (b) prilllarywinding C;UInmt, (c) and (d) primary winding leakage
ft=,
phase causes the sum of the yellow and blue-phase flux 3 and s respectively to pass
into tbe red-pbase winding-limb. This flux is equal in magnitude, and 180
0
out
of phase, to the nominal flux l. Near rated voltage VI'" is induced in the red-phase
winding during the fault , and the fault current ij flows from the generator yellow and
blue phases through the mainly inductive transformer primary winding short-circuit
impedance. The transformer red-phase fault current lags the lnduced voltage by 90
0
and is in-phase with the fault current in the generator
The mmf sum around each loop of the three-limb core must balance. The fault
currents in the three-limb transformer yellow and blue-phases are therefore in phase
with the red-phase current. Moreover the core flux levels are not increased by the over-
current in the transformer prima.ry windings. The nature of the primary fault
currents cause the transformer zero-sequence and primary flux-leakage path flux lev-
els to significantly increase. The primary leakage path flux waveforms are displayed in
104 CHAPTER 7 HVDe CONVERTER TRANSFORMER COMPARJSONS
I,
., .,
'"
,
I,

, 1
"'-
.,1
0'-
v, 1
"'-
(a)
~ ~ ________ ~ ~ __________ v,
(b)
Figure 7.8 Star-grounded three-limb transformer flux distribut ion, red phase to ground fault: (a)
open-circuit core flux, (b) voltage, current and flux phasor diagram.
figures 7.7(c) and (d) for the three-pbase bank and three-limb transformers respectively.
Figmes 7.9(a.) and (b) display the three-phase bank and three-limb three-phase
transformer primaxy winding currents for five complete cycles following fault removal.
In both transformer types, the fault is l'emoved but the rcd-pJ;lase winding-limb fluxes,
shown in figure 7.9(c) and (d), have not decayed to zero. The resultant de flux offset
causes in-rush current to flow in the red-phase windings.
Upon fa.ult removal each tra.nsformer type a.gai n exhibits dissimila.r beha.viour.
In the three-limb test case the red-phase fault current ig (figure 7.5) is not exactly in
phase with the red-phase primary winding current it. Therefore at the fault removal,
fault current zero crossing, a significant cllrrent (point A) is present in all the three-
limb primary windings. The in-rush current is superimposed upon this zero-sequence
current, and a maximwn current of 2.46kA flows in the transformer neutl'a.!. The
zenrsequellce winding current does not affect three-limb winding-limb flux waveforms
(figure 7.9(d)) . Instead the zero-sequence and primary leakage path fluxes increase.
7.2 BENMORE CONVERTER TRANSFORMER COMPARISONS 105

,

! .

.... 'v"
,
i' "
'-"",r v .-r...... . .. . ',-;""
.' - , ,
I . ..

. ... . ..
Iimo(.)
(a)

0.' 1\
... . ;
r. i\ .''./\
I, 'i
l
I' ,...
,',,' ,
,-, !;
" i' ,
' .t I
, " , ,I
' . , .' . I i '
to \, i: \: i :
'\: ;'!
a ..... :, i
'
V
1 i
..... !\. 1\:
,II I I'! i
..... '.' \ i 'J

;
\ ,
, '
" \;
,
, .
, " '
"; .
\1 .. '
" "
;'
i \
"
\ i
"

:\ I
" I!
" \I

'.n . ...
Oms{.)
(e)
I
"
"
; ,
!

" '
1.
"
I I ,
i : "
\, i'
"
)
"
, 1\ !
': \}
...
,

y

...

"i'
V'
r'

<1 ... :'---'-,.0.'-- ,.,:.;-, " .
lime (.)
...

0.' f\
... '
(b)
(-.. .' ,
L
1
"
"
"
,
"
. .

, ,
......
"

,
,
...
'; .
.. '
I; !
I
, ' 1 .
\ 'I I
I .;

< .
-,';----;1.--,;';;----;;:0--,.;--,.;--- :
O.t 0.12 OoM 0." OoM
......(.)
(d)
FiSure 1.9 Star-grounded/star-grounded converter traosforDlC'.( fault recovery hand side=three-
bank, tigbt hand side=thrt""limb three-phase); Solid= rcd-phmie, Dash-dash=yellow-phase,
Dash-dot=bLue-phase: (a) and (b) primary winding clUTent, (c) aud (d) primary wincling-limb flux.
7.2 BENMORE CONVERTER TRANSFORMER COMPARISONS
The previous section has clearly shown the different behaviour of three-phase bank and
t hree-limb three-phase transformers. However the results are not applicable to HYde
converters, which do not include a star-grounded/star-groWldcd transformer.
In chapter 5 a PSCAD-EMTDC model of the Benmore group connected HVdc
converter station was verified with field data. In this practical test sYl:item the converter
transformers were represented adequately with single-phase UMEC models connected
in the delta/ delta (DdD), delt a/star (Dyll) configura.tion. For a realistic HVde test
system, the Benmore transformer connections are used in this section together with the
three-limb three-phase UMEC transformer of Mction 7.1, and !.he transient behaviour
is compared with corresponding results from the three-phase bank transformer model.
106 CHAPTER 7 HVDC CONVERTER TRANSFORMER COMPARISONS
7. 2. 1 Steady-state p erformance
Open and short-cir cuit tests of the UMEC single-phase and three-limb Benmore con-
verter transformer models have already been presented in sections 4.3.3 and 7. 1. 1 re-
spectively. The purpose of this section is t o verify that the three-limb core configuration
does not alter the steady-state operation of the HVdc converter.
Figures 7.10 (a), and (b) compare field data with the simulated three-limb three-
phase converter transformer Benmore test system. The simulated and measured time-
domain waveforms are in close agreement . The generator and converter parameters ar e
as applied in section 7. 1. 2. The par ameters of the DdO and Oyll three-limb three-phase
converter transformers arc given in appendix 1, and the Benmore eonvel'ter is operating
on group connection with a de line current of 405A.
-,
o.oot 9,000 MOO u ~ 0.0' o.o,t G..,. 0.010 0.011 o.oz
llime,,)
(.)
-.
-1$0 . ou 0.000 UN MO. 0.0, 0.012 "" . 0"' 0.011 0.01
'U>')
(b)
Figure 7.10 Benmore group connection generator waveforms, DdO and Dyll three-limb converter
transformers, Id<:,= <I05A; Solid=field lllcMurement, Dash-dot=silllulation: (a) generator current, 3 ma-
chines, (b) generator voltage.
7. 2.2 Tra nsient comparison
A red phase to ground fault has been selected to investigate the transient behaviour
of the Benmore converter transformers. The fault is applied to the llOkV side of the
DdO converter transformer (time-=l.Osecs), and the conver ter is blocked and the bypass
valves fired two cycles later (time=1.04secs). The circuit-breaker opens two and a half
cycles after the fault (t ime=l.05secs). The cir cuit-breaker reclases after O.03secs and
the converter is deblocked three cycles after the r eclose (time= 1.14secs) .
Figure 7. 11 displays the test system. The Pole 1 Benmore converter transformers
are situated at the generator power-house, and connected to the synchr onous machines
by a 16kV enclosed bus network. Therefore t he fault is more likely to occur on the
1l0kV overhead line connecting a converter transformer to the valves. The fault is
7.2 BENMORE CONVERTER TRANSFORMER COMPAIUSONS 107
placed on the DdO, rather than the Dyll converter transformer. This selection sim-
plifies the test system transient behaviour by ensuring that fault current flows only
through one of the Rix pulse valve groups.
The paramet.ers fol' the generator, couverter tra.nsformers, and converter are the
same as in section 7.2.1. The circuitbreaker, bypass valve, and fa.ult pa.rameters are
given in appendix I.
"
," ,
\ r - - ~
" . 0_113 llOtV
Figure 7.11 Bcnmorc group connected HVdc converter transformer transient test system"
Figures 7.12(30) and (b) present the three--phase bank and tbree--limb three-phase
converter transformer primary line currents respectively for the DdO connectioD prior
to the circuit-breaker reclose. Both transformer representations produce similar wave-
forms. The secondary windings of the DdO converter transformer are all electrically
connected to the fault and the interphase magnetic coupling of the three-limb three-
phase transformer has little effect.
Between the fault initiation and converter block the unidirectional fault currents
are superimposed on the converter currents. Once converter blocking is ordered and the
bypass valve fired, the blue-phase current naturally commutates off (paint A) whereas
conduction of the red and yellow-phases is extended until the fault current passes
through zero (point B).
Figures 7.13(a) and (b) compare the three-phase bank and three-limb three-
phase primary line currf'.nts respectively for the DdO convertcl' transformcr. The samc
comparison is made for the Dyl1 converter transformer in figures 7.13(c) and (d). These
waveforms display operation following the circuit-breaker reclose (time= 1.08sccs) and
in-rush follows.
The three-phase bank and three-limb three-phase converter transformer fault
recovery waveforms are almost indistinguishable. However the maximum in-rush cur-
rent peaks of the three-phase banks (points A and B) are slightly smaller than the
three-limb three-phase tranSfOl"mel' (points G and D).
When the generator circuit-breaker opens (time=l.05secs) the converter trans-
former primary windings are open-circuit. The converter snubber capacitance dis-
108 CHAPTER 7 HVDe CONVERTER TRANSFORMER COMPAIUSONS
< " T- " ,.. r. "'.
'i 0 :::... _. .'\: d I,;; C, ,\---1
!
' ..,:, i
'
" , , ,
, I, , , ., . '" ,
, II i , I,! ,
10 , ;. , , , J
, I' ' I! I , , . , , ,
I U 'i \
\ Ii j \ :i! I I
I II; , ,i! '
, I " . , I \ I'
'i \' ./
-"
-"'l.. .,00 1.01 1.02 ull ,.(>4 LOS Lot 1.07 1.01
Iimo(.)
(a)
"
,-"" ,- r -- I ----I
._, _, ' .. ' . I., _ -'_,
i -'.-_. ,\:.1-'. :
l: \ \:; \ \!! \ :
-I' \ ;' I , , ,
I ; , I' , . , , I
,! , . "
I! I ,,, I "
-20 " ',i ! 'J
V'" \!i/ \/
Ut T.OI 1.02 .03 1M LOS I.ot 1.07 1.01
limo(.)
(b)
Figure 7.12 DdO couv(>.rter transformer, llOkV red pbase to ground fault; Solid=red-phasc, Dash-
dnsh=ycllow-pbase, Dash-dot=blue--phasc: (a) Three-phase bank primary line current, (b) Three-limb
three-phose primary line current.
charges through the la.rge non-linear magnetising inducta.nce of the transformer sec-
ondary windings. Figures 7.14(a) and (b) compare the three-phase bank and thl'ce-lirnb
three-phase primary winding-limb flux respectively for the DdO converter transformer;
the Dyll transformer is connected to the same busbar, the primary winding-limb flux
waveforms arc therefore identical to the DdO. Snubber circuit discharge occurs in the
period time= l.05-1.08secs. The three-phase bank and three-limb snubber discharge
waveforms are dissimilar because the magnetising inductances of each core type a,1'e Dot
identical. At the circuit-breaker reclose, the resulting three-limb red-phase winding-
limb flux magnitude is slightly greater than that of the tbrce-pba..<;e ba.uk.
7,3 CONVENTIONAL CONVERTER TRANSFORMER
COMPARISONS
The Benmore DdO, Dyll converter transformers 3J:e not typical of conventional HVdc
configurations. The assessment of converter transformer transient behaviour must
therefore he extended t9 include the more common star-grounded/star, star-grounded/delta
converter transformer configuration.
The three-phase bank and three-limb three-phase converter transformers are now re-
in the star-grounded/star (YyO), star-grounded/delta (Ydll) configuration.
The reconnected transformer parameters are given in appendix 1, the dc line current is
set to 405A.
Figures 7.15 (a), and (b) compa.re the field test wavefornu; with the results of
the simula.tion using the three-limb three-phase converter transformer equivalent of the
7.3 CONVENTIONAL CONVERTER TRANSFORMER COMPARlSONS
f\
l.

'-/" ,W. ,.I t) \
Ii '\ ' ; " ,7'-';; , , i i f'
II , " ; ';' ; ' I, ; "
-, ,,'
, !;
,! ;
"
;'
"
; ,
,
,

.
..

,
, j'
, '.
" '
, :
,.: \\It !fl
_f '/,j ",'...'" 'J,! , I
, , ' II
,-'
\!d ;
.. ' o! i
, "
"!/
I;
,
il
'i
- ,." ..of 1.' 1. 11 1.12 ' .13 1. 1. 1.11 ' .11 1.11
-.
101
"
, , ,;
, . , .
\1/
,})
0, ...
-(.)
(a)
'. 11 '.a 1.1'
_(0)
(e)
, ...
: I
, 0
.' ,
,
, , '
. ,
, .
..
, ,
f) :
,
hI"; !
1\ !
\ !
\J
.. -}. i
, i ;
1,14 , ... I. ..
"
' .j
1.17
";."
o
J
f\
A.
I
1M ' .1 I.I f 1.1' 1,1' 'M IJ 7
(b)
-too o ...
..,.
.... I.u ... . 1.1. ..11 ..,7
"'w
(d)
109
Figure 7.13 Converter t ransformer line CUl"l"flllts, fault recovery (left hand sidc=three-phase bank,
right hand side=three-limb three-phase); SoLid= red-phasc, Dash-dash=yeltow- phase, Dash-dot=blue-
phase: (a) and (b) DdU primary current, (c) and (d) Dyll primary current.
Benmore units. As expected, reconnection of the transformer windings in the YyO and
Y dll configura.tion does Dot affect the steady-state waveforms of the Benmore group
connected scheme. measured and simulated time domain steady-state waveforms are
shown to be in close agreement.
7.3.2 Transient comparison
In conventional HVdc terminals the converter transformers are positioned with the
secondary transformer bushings extending into the enclosed valve-hall and a fault to
ground on the secondary side the converter tra.llSformers is unlikely. A phasc to ground
fa.ult is most likely to occur on thc converter transformer primary side; these windings
are commonly connected to an outdOOI' ac switchyard. Therefore the transient behav
iour of the three-phase ba.nk and three-limb three-p'hase COIlverter transformers is a.gain
compared, this time using a. red phase to ground fault on the 16kV bus-bar.
..
, '
ur, "I''' ,I i "" ,
fY1'''' I", '1; "'"
:'rL lV\: fl1 f \ l i ;e "j, ,I I' I' ' ,; \- , ,. "
.". '.,1 .,,' " '--.'" i' " ,, , ,I
, " 1 ., I .. " .' " ," I I
",-0 I I I I ." 01, I ;," i," " ,
I I ' . III , I. , i, l , ", I' ,<I,
.... '. 'i " ,\ ! ,I, i 1 " ! , ;'
/.,
"i' '" 'I ' " r, . 'I
....... 1 I, ! ' I I 'I i J. ! ..,
<.'
-,

! i !; !i \ i ! I
!! \! " i ' 1 ,
\/ \} V ',: \!
1M lJ 1.11 I.U 1.1' lJI
limo (a)
(al
,
" ,:\ ,!; "
"r\n:"\ IIi .0'
- .': ,,'(':r:
1
: \ \ : \ \ :
- ,I" , I' I , """
" .: . ;' . i' I I., , " I
i'i' !. " !oI
"
I. I.;:';':' it II !I "
!i "r . , . , .,', .1 I ' .j I , . ' .,
a...., 1 j I I , .!' i!' i.;' \, in,
.... 1 ;\ .\ /\.,i Pl: f\\:
,,;,." :' i 'l ,:l ::, il;
,! ".,')
, i .. , ii'
i' ',! ! i. ,'
.,

'. i ' r ' :
'.,' ! . '" . , .
',.i ! i ', ' J ,-j ,!
'. .' "
1,1 1.11 I.U ' JI IJ'
.w..(.)
(bl
Figure 1.14 DdO converter transformer, llOkV red phase to ground fault anu recovery; Solid=red-
phase, Dash dash=yellow-phase, Dash.dot=hlue-phase: (a) Three-phase bank primary winding-limb
flux, (b) Three-limb three-phase primary winding-limb fltu:.
Figure 7.16 displays the new HVdc converter transformer test system configura-
tion. The parameters for the generator, COllvertcr transformers, and valve groups arc
as applied in the previous steady-state field data verification (section 7.2). Details of
the circuit-breaker, bypass valve, and fault parameters arc given in appendix 1. The
sequence and timing of the fault, circuit-breaker, and converter controls arc identical
to those ofthe previous llOkV red phase to ground faul t study (section 7.2.2).
Figures 7.17(a), and (b) present the generator current in the three-phase bank,
and three-limb three-phase converter transformer test systems respectively, prior to
the circuit-breaker reclose. The red-phase generator fault currents are similar, but
the yellow and blue-phase currents of the three-limb converter transformer test system
exceed those of the three-phase bank.
The three-phase bank and three-limb three phase YyO converter transformer
primary currents itl/II are shown in figures 7.18(a), and (b) respectively. In this winding
configuration the single-phase bank is not affected by the presence of the fa.ult. However
significant fa.ult currents flow in the primary winding of the three-limb three-phase
converter transformer. The YyO three-limb converter transformer transient behaviour
is similar to that of the star-groundedfstar-grounded unit explained in section 7.1.2.
ra.ult currents flow in both the three-phase bank, and three-limb three-phase
Ydll transformer primary currents ityd, shown in figures 7.18(c), and (d) respectively.
This behaviour is caused by the delta secondary winding connection. Although the
red-phase primary is shorted, the three secondary Windings arc energised (due to the
delta connection) and fault current flows in all the transformer windillgs.
Between fa-ult initiation and converter blocking the fault current is superimposed
on the converter currents. Once the blocking is ordered, the bypass valve is fired, and
7.3 CONVENTIONAL CONVERTER TRANSFORMER COMPARISONS 111
-"
-"
. 000 . 001 . 001 . 01 0.012 0.0 .011 0.0,. =
llimo (Ij
-250 0.002 0.00. 0.0" 0.001 0.01 0.0" ..... 0.010 1.011 0.12
lim< (oj
(aJ (bJ
Figure 7.15 Benmore group connection generator waveforms, YyO a.nd Ydll three-limb converter
transformers, I",,,=405A; Solid=1'ielcl measurement, D<l,\;h-dot=simulati.on: (a) generator current, 3 ma-
chines, (b) generator voltage.
X, .. I).I/J III)W

"
," ,

D

;/

J...

R.d phlu ,,,,,,,,",,d foull
Figure1. lfl Conventional converter transformer configurati.on transient test system.
the red, yellow, and blue valves commutate off. After the circuit-breaker is opened
conduction does not cease until tbe fault current in each phase passes through zero.
The waveforms of figure 7.18 are continued in figure 7.19 for the period following
the circuit-breaker reclose. Again the three-pbase bank and three-limb three-phase
converter fa.ult recovery waveforms are almost indistinguishable. The maximum in-
rush cunent peak of the YyO three-phase ba.nk (point A) is greater than that of the
three-limb three-phase equivalent (point B). The maximum in-rush current peak of
the Y dll three-phase bank (point C) is less than that of the three-limb three-phase
equivalent (point D). When the converter is deblocked (time=1.14secs) the in-rush
currents are super-imposed on the converter currents.
Threephase phaoe-\o-,ound fault

f\

f\ fI
" "
f\
"
"
-.
v V V
-'8... 0." 1.01 1.0.! ' .000 '-"50 ,.00 UI7 1.01 ""'3'1 .. .. .. ,", .,;:: ",,",," -.1"
..... (0) lim< (.)
(a) (b)
Figure 7.17 Generator currents, 16kV red phase to fault; Solid=red-phase, Dash-
dash= yeUow-phase, Dash-dot = bluc.-phasc: (a) three-phase bank test system, (b) three-limb three-phase
test system.
7.4 CONCLUSIONS
In this chapter equivalent UMEC three-phase bank and three-limb three-phase HVde
converter h'ansformcr models have been compared under steady-sta.te and transient
operating conditions. Under steady-state open-circuit, short-circuit and over-volta.ge
conditions both transformer types produced acceptably similar waveforms. A single
phase to ground fa.ult was applied to the primary bus-bar of a star-grounded/star-
grounded converter transformer. This tf'.st system included a single six-pulse HYde
valve group. Significant fault currents flowed in all primary phases of the three-limb
thl'ee-phase converter transformer, yet no fault currents flowed in the three-phase bank.
The behaviour of each converter tra.usformer type was also dissimilar after the fault
was cleared.
Assessment of three-phase bank and three-limb thl'ee-phase converter trans-
former behaviour was made in a practical test system. A red phase to ground fault was
applied to the converter side of the delta/ delta. converter transformer of the Benmore
group connected test system. In this example the behaviour of the three-phase bank
and three-limb three-phase converter transformers were almost indistinguishable. The
faulted delta winding electrically, rather than magnetically, connected all phases to the
fault.
Finally, three-phase bank and three-limb three-phase pairs of star-grounded/ star-
grounded, star-gTounded/delta convertel' transformers were placed in the Benmore test
system and again verified with steady-state field data. A red phase to grOlmd fault was
applied to the C01lverter transformer primary side, In this typical winding configura.-
tion fault currents were present on all three-phases of the three-limb star-grounded/star
converter transformer. The single-phase equivalent failed to show this effect.
, '
7.4 CONCLUSIONS
"
"
Iimo{. )
(a)
"
'"
(e)
.,
, .
/1 \
"
.,
.

I
'." .. " ".
113
..
"
. "-T-.,,:;;-,
w..o(s)
... 1.01
(b)
1.<12 I.'" 1.0<1
-(->
(d)
. M ..
,.,
'"
Figure 7.18 Converter transformer winding cw:rents, 16kV red pbase to ground fault (left hand
side= three- pbase bank, r ight hand sidc=thrcc---li.mb three-phase); Solid=red-phase, Dash-dash=yeUow-
phase, Dash-clot= blue-phase: (a) and (b) YyO primary current, (c) and (d) Ydn primary CUlrflut.
114 CHAPTER 7 HYDO OONVERTER TRANSFORMER COMPARISONS
--r- ;
, " '
, 'i'
V \!
-'f.l ... ,,,,,-.,,,,.c,,,,.-.,,,.,-.,,"
liIno (I)
(a)

/
,
,
, '
, " ' I
, "' . , I, ,.
\.\-i
c
,\
,., ,;
i " ! , ;, !
\).J
, '
,i "
,\ ' .
, - I ;
\ ,.
',' '_I
t .
.'
. ,
",\ 1
"
\['
,,,,,-.,.,,,
1iI ... (.)
(e)

.,
\ '. "
,'. 'i
, '.1 ,
\ I I
\" ,
, '.,
', I,
',\/ i
, , '
'.'\i

.
t, f.'
".,
'I \1

'.
, ,
, ,
0
..-' V ;
.,
\l
.,
-lp_l".c".o" ,,,,,-.,,,,
"-(0)
(b)

. ,
'. \. 'i
\ '. I ,
I '{ !
\ ,. !
, ,!
"

.
,

'-".l
, '.'
'.,
-"M"-"." -.,.' ", - ".coo."" ",- ."".",.,"
lim< (0)
(d)
Figure 7.19 Converter t ransformer winding currents, fault recovery (left hand side=thrcc-phasc
bank, ri ght hand side=three-iimb three-phase); SoLid= rcd- phase, Dash-dash=yellow-phaw, Dash-
dot=bluo-phase: (a) and (b) YyO primary current , (c) and (d) Ydll primary current.
Chapter 8
FUTURE WORK AND CONCLUSIONS
In this thesis magnetic circuits have been used to model single-phase and three-limb
three-phase transformers. The pl'oposed magnetic equivalent circuit (MEC) concept
has been implemented in all electromagnetic transient program as a Norton equiva-
lent. Since the transformer model applica.tions presented in this thesis arc specific to
HVde improvements, the PSCAD-EMTDC package has been used to incorpora.te the
proposed models. Moreover the effort spent on the PSCAD-EMTDC program trans-
former models, relative to other HYde components, bas been small.
The MEC model has been used to intervalidate the Harmonic Domain Analysis
(HDA) and PSCAD-EMTDC programs. Intervalidations ha.ve been performed for ovec-
voltage excitation of single-phase and three-limb transfonners with the test system
supply impedance being purely resistive, and series resonant respectively. Accepta.ble
matches in Time Domain wa.veforms and Harmonic Domain spectra. were achieved in
each case.
The interphase lI1agnetic coupling representation of the t hree limb three-phase
MEC model was verified with experimenta.l da.ta.. Currently PSCADEMTDC uses
magnetically independent phase models. Comparisons between equivalent PSCAD-
EMTDC three-phase bank and MEC three-limb three-phase models under steadysta.te
energisa.tion, flux decay and in rush opera.ting conditions showed considera.ble differ-
ences in the beha.viour of each model.
The present PSCAD-EMTDC transformer model core dux is calcula.ted from
the voltage of only one winding, a uniform flux density is assumed throughout the core
legs and yokes. The non-linea.r magnetisillg current is injected across the terminals of
one transformer winding. The unified magnetic equiva.lent circuit (UMEe) transformer
model does not apply any of these a.pproxima.tions.
LaboratOl'y results were used to show that the UMEC reprcsenta.tion provides
an accurate division of magnetising current between transformer windings. To demon-
stra.te the effccts of the PSCADEMTDC transformCl' approxima.tions in a practical test
systcm, the New Zealand HVdc system was used. For the case of load rejection, the
results showed tha.t the PSCAD-EMTDC modcllead to significant errors in predicting
116 CHAPTER 8 FUTURE WORI{ AND CONCLUSIONS
the magnetising current harmonics in the Benmore interconnecting bank windings.
The first set of group connected generator HVde converter scheme harmonic
measurements are presented. The field data provides evidence to support the feasi-
bility of the group connection, and is useful for verification of t he UMEC transformer
model. Benmore three-phase bank convert.er transformer UMEC models were derived
and placed in a. PSCAD-EMTDC model of the Benmore group connection. The simu-
lated generator voltage and current provided an acceptable match to the field data.
A new mea.us to reduce cost and increase the operational flexibility of the HV de
terminal was investigated. It consists of the smoothing transformer and de blocking
capacitor as a replacement for the HVde :;moothing reactor and conventional filter-
ing arrangement. The PSCAD-EMTDC model of the Berunore grOIlP connection, and
single-phase UMEC transformer model have also been used to investiga.te t he smoothing
transformer concept. Dc side hannonic cancella.tion was a.chieved with the smoothing
transformer a.t characteristic and non-cha.racteristic harmonic frequencies. Satisfactory
de side harmonic cancellation was verified for a range of synchronous generator operat-
ing speeds. Moreover, during a dc line fa.ult the transient behaviour of the smoothing
tra.nsformer was shown to be similar to tha.t of t he conventional filter arrangement.
The Single-phase UMEO was extended to represent three-limb thrce.phase trans-
formers and a model of the Manapouri 105MVA, 13.8/220kV transforruCl's was vali-
dated with factory test data. Three-phase bank and three-limb three-phase HVdc
converter transformers comparisons were also made. UMEC three-limb three-phase
models, equivalent to the single-phase Benmore converter transformers, were devel-
oped and model equivalence in the steady-sta.te was verified with the field data. In
the conventional star-grounded/star, star-grounded/delta winding configuration, the
response of the Single- phase bank and three-limb three-phase converter transformers
to a primary phase-to-ground fault are dissimilar. Significant fault currents flow in all
phases of the three-limb three-phase star-grounded/delta transformer while no fault
current flows in t he corresponding three-phase ba.nle.
The tra.nsformer models developed in this thesis will be incorporated in version 3
ofPSOAD-EMTDC (for release to all users in 1996). To place the UMEO model in this
program further work is required to fix the Norton equivalent adntittallce matrix at each
time step, or reIl10ve it completely. When the time step is small (50j,ts), the UMEC --
admittance matrix has little effect on the winding current as the Norton equivalent
current source dominates. Updating the tra;usformer adlnittance network significa.ntly
slows the PSCAD-EMTDC main program loop. Alternatively, the MEC models could
be applied to calculate transformer maguetising curreut. The MEC Norton equivalent
current source would then be injected across a six winding inductance transformer
representation.
The models proposed in this thp.sis should be extended to include hysteresis.
117
Branch penueancc should be calculated using hysteresis loops. The addition of hys-
teresis representation would be useful to verify models with small VA rating (laboratory
transformers). Winding resistance and core loss representation could also be added.
The addition of tank and winding capacitances would extend the capacitor
electromagnetic representation to a Norton equivalent electric field formulation. The
UMBO magnetic and electric field Norton equivalents could then be superimposed to
give a. transformer more suitable for simulating fast transients.
It would be beneficial if morc widespread use of accurate models could prompt
manufacturers to release improved transformer I:lpecifications. Core design, winding
turns numbers, and core steel characteristics are essential parameters of detailed trans-
formel' models. These parameters are ra.rely all listed in individual transformer techni-
cal doeumentation.
Only simple single-phase and three-limb three-phase transformer magnetic equiv-
alent circuits have been proposed in this thesis. The Norton equivalent formulation may
be used to interface more complex magnetic circuits in PSCAD-EMTDC. Individual
transformer windings can be represented with multiple magnetic circuit branches, such
a representation would be useful for studying turn to tum short-circuits. Zero-sequence
or winding-leakage flux paths may be broken into multiple magnetic branches. Small
magnetic networks may be applied to represent the core clamps, bolts, frames, or the
transformer tank; this detail would be useful for examining loca.lised saturation. Ex-
tensive three-dimensional magnetic equivalent circuits could be applied.
Core configurations other than single-phase and three-limb three-phase types
are yet to be examlncd. Four and five-limb, single and multi-phase transformers can
be modelled with magnetic equivalent circuits. Saturable reactor and dc measuring
transformer models would also benefit from accurate core and winding representation,
these devices have not been addressed in this thesis.
A magnetic equivalent circuit similar to the UMEC transformer representation
can be proposed for the salient pole synchronous generator. Figure 8.1 displays a
single-pha..<;e example. The permeance of the rotor path 'P2 and the field unnf N2i2
would be time varying accOl"ding to pole saliency and rotor speed. The superscript #
is used in the figure to indicate variables scaled according to the field windings rotating
frame ofreference. Leakage paths P4 and Pr- are associated with the armature and field
windings respectively, the flux retum path is represented with 'P3. It would be useful
to create a UMEC model for the Benmore synchronous generators using the practical
magnetic circuit design. Such a model could then be placed in the Benmore group
connected test system to try and improve the cbapter 5 voltage harmonic match.
The UMEC transformer model should be implemented in the Harmonic Domain
Analysis (!IDA) program to provide an accurate prediction of magnetising current
hat'monic distribution between transformer windings. To make the UMEC model more
118 CHAPTER 8 FUTURE WORK AND CONCLUSIONS
Magnetic circuits
Single-phase generator
(,)
II,;,{I/
- ~ ( I )
mR
~ ~ ) J
Unified magnetic equivalent circuit
(b)
Figure 8.1 Synchronous generator magnetic equivalent circuit.
generally availa.ble, it should also be placed in the EMTP program and comparisons
made with the existing three-limb three-phase duality based EMTP representation.
Appendix A
NORTON EQUIVALENT ADMITTANCE MATRIX
SYMMETRY
To prove the symmetry of the Norton admittance matrix [Yul, it must be shown that
the product [M1[P] (see equa.tion 2.25) is symmetric.
(A. I )
Now
Since the pe<meance matrix IP) is diagonal IPjT = [P), and therefore IAjTIPlIA)
is obviously symmetric. The inverse of a symmetric matrix is also symmetric, thus
(IAjTIPlIA)) - T = ( IA)TIPlIA)) - ' . Therefore
(IM)[P)) T = IP) - IP)IA)(IAfIPlIA)) - 'IAfIP) (A.3)
and the product IMJIP) = (IMlIP)) T.
JI (M]{P1 is symmetric the upper left-hand partition [M",,][P$J is also symmetric.
Now
( )
- ' ill
IY,,) = IM .. )I"P.lIN,) 2"IN.)- ' (A.4)
and Lhe scaler ~ does Dot affect [Y$lJ matrix symmetry. Using [AI -
1
[Bj - l = ([B][AJ) - l
leads to
(IM"lI"P.lIN,)) -' IN,r' = (IN.lIM lI"P,lIN,)) -' (A.5)
120 APPENDIX A NORTON EQUIVALENT ADMrTTANCE MATRJX SYMMETRY
Also
But [N,] is diagonal and the pwduct [M,,][P,] = ( [M"IIP,]) T Therefore
[N,][M"IIP,][N,] = (]N,][M,,][P,][N, ]) T
and
(A.6)
(A.7)
([N,II
Mu
IlP,IIN.,]) - I = (]N,IIM"IIP,IIN,]) - T (A.B)
Now
[y,,] = ( rAt
[N,IIM"IIP,IIN,] 2
-
( (At
[N,IIMu][P,][N,] 2
-
[Y,,]T
and the symmetry of [yssl lS shown.
(A.9)
(kIO)
(A.ll)
Appendix B
PER-UNIT DYNAMIC SIMULATION
With all varia.bles in perunit form, generalised equa.tion 2. 10 becomes B.l and equation
2.29 becomes B.2, where Wo is the base angula.r frequency. In the per-unit examples of
this thesis angular frequency is Dot normalised.
- - 6t [ 1
,(t) ~ ,(t - 6t) + TWo .,(t) + .,(t - 6t)
In the per-unit system the clements of equation 2.30 become
aud
[Y,,) ~ ([M,,)[P,)) - I ~
in. = ([M,,)[P.)) - I ~ ; ; , t _ 6t) + 4>.(t - 6t)
+lM,,)4>n. + lM,,)4>.,)
(B. l )
(B.2)
Appendix C
LEAKAGE PERMEANCE CALCULATION
Theory
Consider two mutually coupled windings having primary and secondary self in-
ductances L11 and ~ respectively. The mutual inductance M12 between the two
windings can be defined by the relationship
(C. 1)
where k is the coil coupling factor.
Mutually coupled coils can be represented with the electrical equivalent circuit of figure
C.l.
Lil =(l-kJL IJ

Figure C.I Mutual ly coupLed coil equivalent circuit.
Calculation of transformer coupling factor k, and self inductances Ll1 and L22 gives
the primary and secondary winding leakage reactance Lll and LI2 respectively.
Measurement
Energise the transformer primary with the secondary open-circuit. Measure the
primary voltage VI, primary current iI, primary power Wl, and secondary voltage V2
Calculate L
u
) M21 using
(C.2)
124 APPENDIX C LEAKAGE PERMEANCE CALCULATION
'"
L IJ
MOl -
where Wo is the supply angular frequency.
../' .,
tl - fro
"1
t::z;oWn
v,

(0.3)
(0.4)
(0.5)
Now energise t he transformer secondary with the primar y open-cir cuit. Measure the
secondary voltage V'l, secondary current i2. secondary power W
2
an d primary voltage
VI _ Calculate 22. M12 using
"0
-
i:l;O -
L22
M12 -
w,
V,
../' .,
12 - tro
V,
!$QW
Q
VI
t::z;oWo
(0.6)
(0.7)
(0.8)
(0. 9)
Check that M12 ::= M
21
, and if they agree calculate the coupling factor using equation
C.1. The primary and secondary winding leakage inductances La and Ln are then
calculat ed with
L" _ (1 - kiLn
L" - (1 - k)L"
Single-phase UMEO leakage path perrucanees PI\. and Ps arc obtained from
L"
1'5 = -,
N,
(0.10)
(0. 11)
(0.12)
(0.13)
where NI and N2 are t.he primary and secondary winding turns number respectively.
If only the total transformer leakage react:mce is given, for example X/ =O.10p.u.,
and the above tests can not be perfoInJed, then the lllodeller must approximate. One
approximation is to halve the given leakage inductance between the t ransformer
mary and secondary windings, and then apply equations C. 12 and C.l3.
Appendix D
TRANSFORMER RELUCTANCE/ PERMEANCE
PATHS
D.l PRIMARY OPEN-c m CUIT RELUCTANCE
If a single-phase transformer primary winding is energised and the secondary is open-
circuit, the effective flux path reluctance is derived from the magnetic equivalent
circuit of figure 4.3(a).
1?j + 1?:/ /(1?; + 7<:,1/1?;)
_ 1?' 1?'//(1?' 7<:,1?!)
, + 4 '+ 7<:,+7<;
R:;R3(n; + n;) + n,F?;R;
R1 + R
4
(R; + R;) + R;(R; + R;) + RiRs
(RjR: + nrRj + Ra:R4)(Ri + Rs) + (Rj + R
4
)Rin;
(1?; + 1?')(7<:, + 1?;) + 1?i1?;
D.2 SECONDARY OPEN-CmCUIT RELUCTANCE
(D.l)
(D.2)
(D.3)
(D.4)
Similarly, if the transformer secondary winding is energised and the primary is open-
circuit the effective fl ux path reluctance is derived fI.'om
n*8ce
- 7<:, + 1?;/ /(1?j + 1?;/ /1?,) (D.S)
oc

'//(' 1?j1?,)
R2 +Rs Ra + 'R.j +n:
(D.6)

1?" n;nacRj + R:) +
2 + R;(Rj + 1<.4) + R;(Ri + 1<.:) + Ri1<-4
(D.7)
(n.W + WW + W7<;)(W + W) + (1?' + 1?')1?'1?'
2523314 li25
(D.S) -
(1?; + 1?')(1?j + 1?;J + 1?j1?'
126 APPENDlX D TRANSFORMER RELUCTANCE/ PERMEANCE PATHS
D.3 PRIMARY SHORT- CIRCUIT RELUCTANCE
If a single-phase transformer primary is energised and the secondary winding shorted,
the effective flux path reluctance is derived from the magnetic equivalent circuit
of figure 4.3(b).
D.4 SECONDARY SHORT-CIRCUIT RELUCTANCE
(D.9)
(D.lO)
(D. ll )
H the transformer secondary winding is energised and the primary is shorted, the
effective flux path reluctance is derived from
R"St:.t:.
- R; + R;II(R; + R,) (D.12)
"
-
R. + R,s(Rj + R.4)
2 Rj+R:i+'R;
(D.13)
Ri(Rj + R.i + Rs) + R5(R; + R
4
)
(D.14) -
R.j+R4+1t.
S
D.5 MUTUAL INDUCTANCE PATH PERMEANCE
The equivalent open-circuit reluctance network of figure 4.3(a) is redrawn as the equiv-
aJent permeance network shown in figure D. 1(a), where
. Nti t P; (D. 15)
P; - Pi + P1 (D.16)
Po = Pi
(D.17)
P' , = P; (D.18)
p'
- Pi (D.19)
The permeance network of figure D.1(a) can be simplified to the reluctance network of
figure D.l (b), where
(D.20)
D.5 MUTUAL INDUCTANCE PATH PERMEANCE
p'

1'.'

(a)
(0)
Figure D.l
n;
8,
Mutua! inductance equivalent circuits.
n'
1 1
=
'P. + 'P. + p .
a
1 4 J
n,
1
-
'1';
n'
1
= ,
p'
,
127
(b)
(d)
(D.21)
(D.22)
(D.23)
The reluctance network can be converted back to the permeance form of figure D.l (e),
where
.
N
1
i
1
PiP3(Pj + P4)
(D.24)
=
(Pi + P4l{P; + Pi + P.j)
'1': =
P" 'P;(Pi + 'P4)
s+Pi +'Pi+'P;
(D.25)
p'
/
= '1'; (D.26)
Finally the permeance netwOlk of figure D. l(c) can be placed in the thevenin form of
figure D.l(d), where
e, -
p,(pi + P,iJ + p,(Pj + Pi + '1',)
(p; +1",)('1'; + Pi + p,t) + PHPi + '1',)
'P
2
(P;(Pi + P:) + P; (P
3
+ Pi + Pt))
Solving for <Pc(t) in figure D.l(d) gives
t _ Nt it PiP;;p:r
,( ) - P,(pj + Pi + '1',) + 'I'j('I'i + '1',) + P,(P;i + Pi + '1',)
(D.27)
(D.28)
(D.29)
128 APPENDIX D TRANSFORMER RELUCTANCE/PERMEANCE PATHS
Now ,(t) = </>2(t) and
N d,(t)
V2 - ~
N,N
2
'PiPiP;' di l
1';(7'; + 1'i + 1'4) + 1';(1'; + 1';) + 1';(1'3 + 1'i + 'I'll dt
Which gjves the mutual inductance
where
.. 1'i1';'P;
1'21 = (P2 + P;)(Pj + P; + 1'4) + Pi (pr + P,i)
(D.30)
(D.31)
(D. 32)
(D.33)
If the equivalent open-ci rcuit reluctance network of figure 4.3(a.) is drawn with the
primary winding open-circult and the secondary winding energised 'Ph can be derived..
In this <:Me
. =
N
2
i
2
Pi (D.34)
p'

= P2+
P
S
(D.35)
1',
=
p'
3
(D.36)
p'
,
= Pi
(D.37)
'I'd = 1':
(D.38)
If the permeance network of figure D.1 (a) is again simplified to the reluctance network
of figure D.I (b)
o.
N2i 2Pi
-
'Pi +1'5
n'
1 1

-
p. +'P,. + p.
, 5 3
no
1
-
p'
I
n'
I
- ,
p.
The permeance network of figure D.l(c) becomes
(D.39)
(DAO)
(D.41)
(D. 42)
(DA3)
(D.44)
(D.45)
D.S MUTUAL INDUCTANCE PATH PERMEANCE
The figure D.l(d) thevenin form becomes
P;(Pi + Pt) + 'P:(P; + Pi + pn
'R.d = ('Pi +P.i)(1'j+1'i+'Pt) +'P3(P2'+1's)
Pi(P,(Pi + P;) + P4(P, + Pi + P;))
Solving for 1>e(t) now gives
Now <Po(t) = <P' (t) and
N
dJ(t)
VI - 1 dt
=
NIN2PjPi'P; di2
Pi(Pj + Pi + P; ) + P;(Pi + P;) + P4(Pj + Pi + P;) dt
Which gives the mutual inductance
where
* 1'j'P
2
1';
P
12
= (Pi + P4)(Pi + Pi + P;) + P,(Pi + P;)
129
(D.46)
(D.47)
(D.4B)
(D.49)
(D.50)
(D.5!)
(D.52)
Appendix E
PER-UNIT -I AND B-H SATURATION
REPRESENTATION
Although only a single B-H saturation characteristic is required to represent a trans-
former core, individual per-unit winding-limb and yoke saturation characteristics are
applied in this thesis. The relationships shown in figure E.!: F(Bd, between per-unit
branch flux {U and branch flux density Bk, and G(Hk), between per-unit branch mmf
i ~ u and branch magllchsing force H
k
, are described in this appendix.
!jf."

~
'.
q; cF( Hti
,------...
if' =G( H!)
"
Figure E.! Saturation chru-adetistic r.onvllntions.
For each branch k of the magnetic equivalent circuit
(E.l)
where: k is the branch flux, Bk is the branch flux density of branch k, and Ak is the
branch cross-sectional area. Now the transformer core flux base is defined as
b
/2Vlb
(E.2) -
NIWo
/2v"
(E.3) -
N2Wo
where: Vlb is the primary side base voltage, V
2b
is the secondary side base voltage, Nt
132 APPENDIX E PERt-UNIT </J-1 AND B-H SATURATION REPRESENTATION
is the primary winding turns number, N2 is the secondary winding turns number, and
Wo is the rated angular frequency. Per-unit flux is written
'"
k =
F(Bk) (E.4)
k
(E5)
=
.
AkN\wo B
J2
V
lb k
(E6)
AkN2
W
o B
(E7)
=
J2 k
2
V
2b
Also, for branch k
(E.8)
where: Ok is the branch romf, 11k is the branch magnetising force, and 1k is the branch
length. Now the tranformer branch mmf base is defined as
8, - J2NJlt.
- J2N212b
(E9)
(E.lO)
where: lIb is the primary side base current, and 1
2b
is the secondary side base current,
Per-unit mmf is written
plt =
' k
OP'
k
G(Hk)
Ok
0,
lk
J2 Hk
2N
1
l
1b
lk
Hk
J2N,l2b
(Ell)
(E.12)
(E.13)
(E14)
(E15)
The transformer winding-limb length and cross-sectional a.rea dimensions will not
usually equal those of yoke. Dissimilar length lk 8J1d cross-sectional area AI; dimensions
produce dissimilar per-unit winding-limb and yoke 1{u i ~ u saturation characteristics,
from the one core steel B - H curve.
Appendix F
LABORATORY AND SIMULATION PARAMETERS
Sine-wave generator
Rating
Voltage
F):equency
L
soorce
R Joltrce
Test transformer da.ta
Type
Rating
3.0 kVA
nov
50 Hz
115mH
O.Hl
Single-phase
2.0 kVA
Primary voltage
Secondary voltage
X,
210 V
210 V
0.3362 p.ll.
0.5 !l
R"rim
Rsee 0.5 n
U MEC test transformer model pa.rameters
Winding-limb length
Winding-limb area
Yoke length
Yoke area
Primary winding turn!! number
Secondacy winding tu rus number
L"
L"
m,
0.0580 m
0.0049 m'
0.2960 ill
0.0049 m
2
250 'IUrns
250 Turns
11.8 mIl
11.8 mIl
0.124
6.3e-4
0.0
0.77
1.85
134 APPENDlX F LABORATORY AND SIMULATION PARAMETERS
PSCAD-EMTDC test transformer model parameters
H.ating
Frequency
XL
No-load losses
Ideal transformer?
Tap changer winding?
Primary winding voltage
Primal'Y side magnetising current
Secondary winding voltage
Secondary side magnctising current
Saturation enabled?
Satura.tion placement
Air core reactance
In-rush decay time constant
Knee voltage
Time to release flux clipping
Series resoDant load
lllQad 0.364 n
LIO<1d 40 mH
C'"d 10 j.LF
2.0 kVA
50 Hz
0.3362 p.li.
0.0 p.n.
Yes
None
210 V
1.66%
210 V
1.66%
Yes
Primary
6.27 p.li.
le-3 sees
1.0 p.li . .
0.0 sees
Appendix G
BENMORE GENERATOR AND INTERCONNECTING
TRANSFORMER PARAMETERS
Generators GI-G3
iu.ting 112.5 MVA
Voltage 16.0 kV
F\'equency
x"
50 Hz
0.1820 p.ll.
0.0042 p.li.
Int erconnecting transformer data.
Type Three-phase ba.n.k; Delta-Star-Star
Rating 200 MVA
16 kV
220 kV
33 kV
Primary voltage
Secondary voltage
Tcltiary voltage
Xj(prim- ter)
Rprim
0.0392 p.n. (66.67 MVA b88e)
0.004 (l
t ~
0.016 (l
UMEC per-phase interconnecting transformer model parameters
Winding limb length LO m
Winding limb area 0.4536 m
2
Yoke length 9.836 m
Yoke area
Primary winding tUl'DS number
Secondary winding turns number
Tertiary winding turns number
Lit
L"
0.4536 m'
113 Turns
897 Thms
135 Turns
0.243 mH
0.345 mH
136APPENDIX G BENMORE GENERATOR AND lNTERCONNECTING TRANSFORMER PARAMETERS
ml 0.0145
m, 1.5ge-5
b
1
0.0
b, 1.72
E 1.90
The winding turns have been estimated assuming a peak operating flux density of 1. 4T
at rated voltage and frequency.
PSCAD-EMTDC per-phase interconnecting traJlsfol"lllel" model parameters
Rating
Frcquency
XL
No-load losses
Ideal transformer?
Tap changer winding'?
Prima.ry winding voltage
Primary side magnetising current
Secondary winding voltage
Secondary side magnetising current
SatlU'ation enabled?
Saturation placement
Air core reactance
n ~ r l l s h decay time constant
Knee voltage
Time to release flux elippjng
33kV Filters (Series RLC)
66.67 MVA
50 Hz
0.0392 p.u.
0.0 p.u.
Yes
None
16 kV
1.0%
19.0526 kV
1.0%
Yes
Primary
0.6667 p.n.
1.0 sees
1. 25 p.u.
0.0 sees
Harmonic number
5 7 9 11
Rfilt.o;r (0) 0.412 0.454 0.560 0.249
Lfilt<;r (mH) 13. 1 10.3 10.0 3.60
C !i"a (I'F)
30.9 20.0 12.1 23.2
, '
13
0.305
3.75
16.1
Appendix H
MANAPOURI TRANSFORMER TECHNICAL
DOCUMENTATION
Delta-star Dyll , connection matrix
1 0 0 0 - 1 0
0 1 0 0 0 0
- 1 0 1 0 0 0
IC] = 0 0 0 1 0 0
0 0 - 1 0 1 0
0 0 0 0 0 1
0 - 1 0 - 1 0 - 1
(H. 1)
138 APPENDIX H MANAPOURI TRANSFORMER TECHNICAL DOCUMENTATI ON
[SaVigliaiiO] PURCIlASEIt
lUI
OF NEI., ZFALAND
MINISTRY qr.
16 11\ ;
122-4-201 C
111386
TRANSFORMER H. 60)72 TYPE
TRA kV", 105.000
A 262 - 270 - 276 - 28) -
It. V.

..2 kV

,
. A
2x18 2%.18
kV 231 225,5 .. :: ...... 220 ... - . 2C1j
Tt>,,,, 15( 718. . 682
II.V.
-: kY 13,S . Turn. 78
L.V.
<
"
Tum.
CONNECTIONS CROUP
Td1
. l'nASF.5. ) FItf.Q 50
1. , 'EN . 13:1 102,.3 .
1. - BaN 1.3!:1 ,"
102,6.
1 - CaN
102,6.
vg._
" we
RESISTANCES ...
..... B.V. Winding
T. ,....: h
",,<lItO.
Ph ....
Y' - !'.".-
I .
,
:
K
,
, 6.793 " - y .. 130,) 102,8 f ,627
,679) " BSR .. 1)0. i 102,6 ;0,6265
,676 ... " - C
6
N 1)0, 2 i 102,8 ;0,6265
.. fvg ; . )0, 6266
.7862 .. 7S' ; 7:142
......... ........ 1 "' !
2...1.aH .... .. 136 ... .:102,4 .. ,664 ... 5 .... "sIJ... .. 127," i 10),2 :0,617"
.. 2'""!1IsN .... ..1)6,2 .102,6 .. 166) ... 5 .. _.-gN. .. . 121.,4 .102.,6.
. 2.,..C
a
li . 1.36,S .102,a .. ,66-4. .127
... .&.v, .. .,.66)6 ............ _ .... _ .J.vg..... , 6180
,178 75C ,7245
.. .3 .. := ... ...... .1.)), .... O .. 6.4!I. 2.._._ ... _ .. _ ............ .
.. l..:::-... J,JeP. . .... .1.3) , ..... .0.,.6.46. .., ............... .
.. l .. ... .. _.1.)) ... .... L ..... _ ...... _ ._ ....... .
. r& . ... . 9,.6.406 J .... ........ .
.. . ... Q.1.6Q.
..... )O""3j /..l/t!J6!l ...................... l a. p<ecI h, .... . Mr.Dc)Var
D .... _ ... J .... ... _ ...... ,"p", .......... ..
U ' ... c. r . ..... . . -
c . "i\ . ....
L. v. Windinc
I
0,1
e)
8), 0 : 11 6 b,0036 1
I
: 116, 8 P,OO)5 5
i Avg. _ 3
. 0
139
- N O .L O A D RAT IO TES T
M . .. u d 1.. ,.
Top. C.I.al. 11td l.
eh .... e.I - "''',y,--,-::--i---:;------,-:;-----,----;;-Ai,--j
r II. L. V. K V. .. V. . K M
1'0"
kY " "kY" -, ' J
231 1) , 8 16,75
2 225,5
..
16, 35
3 220
..

" . .. " ...... ,.

21'(.5
..
15.55
....... . .......... ... ....
5 209
..
15,15
....... ' ........ ....
TEST
Feedl ... V. kV .... ... ............. __ F. 6,,".ncl .... 50
V, ..... " V .... II.I I I, I, I, 0, w, w,
... _ .. ... 5. .... ,' 1 .. 5 ....
.... 5 . . .. . :4.(?!L
I
II<.
"v ..... , ky", . l.ll

.(5
:1 , 215 ..
..
8,152
..
.. .(, 61. .. .. '.0.1,3 .. 13,)8 p ,-I J zq
:x10 xl0 xl0 xBOO xBoo xBOO
.. 8.7, 6 . 3., 18 .. ,, 2 , 6!1 .. .. . 2, 34 .. .73. ...... 1.( ,0, 1.(.022,.35 0'j,):I2
.. .... . .. ..... ... .. . J. _ ... .. ___ _ . ...... ,,14.41 )4,4 23 . 1.2
120 120 x20 :x1600 x 1600 :x1600
5.(, 3 - 'J, 'J 15. J 15. 12 52. !1J
... !'-.t, ..
.. ! . ...,."
.J .3&,. 13, 8 ... 8, 1.- , ..
U TO,. " ., . ............. .... ,
co",p
140 APPENDIX H MANAPOUlU TRANSFORMER TECHNICAL DOCUMENTATION
,
. " ... pe ,. ..... ..
T'poeb.". <.
po,l, Io ..
", I' '''
III' , ......
5".y lo. d
lo ..
I''''
.... ,
........... ............ ..... _ ... .. .. .... .......... _._ ............... _ .. .
............. ___ .... .. ........ .... _ ... .
..
c
Tn. , l
I .. .. " . ", .. , q>..
(' '''
....... .... .... .
2.(, ............. .......... .}>.).& ........... l'J ...
..:j ..
[.
......... .: ..... 2.4., ij" ....................... '" ......... ...... -_ ...... -_ .
... ......... .............. ........... 2.20 .... . .. :: ... ..... .
... _J _ .. _._..... _ .. _22 . 23...... . ...... 252,.15 ...... _ ..... .
.... _2J. ,,1.1... ......... g?.QS_. __ _ ... m ... 6._ .. _._ ...... _ ..... _ ... _ ....... _
............................. .. t! .. _. ___ ...... .. _._ ............... .
_. __ .. _ ....................... ........... ...... _. ..... .. .. _ ... 1
..... ..........;,. .....4....... .. .. .. .. ........
. -_.- --- .. -.... ... 2-3.1 ..... ..... ......... ..... ... ... : .. f.: ........................................ .
........ 5 .... _ ......... ...... _ . .. J. .... .... ......... _ ....... .
_ ........ ................... 1.1 J:t.!L .. . ... _?1. . 15. ...... ............... .
..... _ ....... .. ..... ..... _ .. H . 4.L ......... ......... ... .
.. ........... ........ .... .. _ ... _ ._ .. _J.16. .5. ........ ... . ,200 ... " , OllJ ..
...... ... .. .. ... %..
I
i
1
H IGH -V OLTAGE TESTS (oillemperolUre .. 3:r:.I.:? .' C)
\:L".d "," "v ...... ..... .. -.. -......... _ .....
" .
............ ... 1..Cl . . ........
,; ....
..c, .... . :4.Q .....
I"." (II
- . ..... ............. ................. _ .. ...... 11,
-
-
. .......... -... . ..... - - _ .. _ .. _ ... .... CJ
I,',!;
OI.tone.

, ,
,
.. ,"" I"" ." ..
.... , ... T"" oj
"
. .......
,
II.V.

J.
L'.
. .
.............
(0) ... H.V,
t enainal llll railJed
...
induoed vol ta.go to kV for 40" at
C.p.lI. with neutral terrninal earthed (B. S. 171
-
ClaJJ.Be 1810).
....
. ..
T;",.
.. ,
16
141
,'. i':i":" . ..:.. -:: [::.;"j' ..... I , I .. ,.:: :
L _. . . ":;. I ':' 1_, :;.- ., ' I
' ': i'" -- . ':' ! :.: ': ,' :. 1 .:: . ;;; 'i :J :1
i . . j, i , ': "':1 [: ' F :; i' i "' 1' i: .' :: ' .. ".:;,; ,; 1
L ,j u :i :: .'; , ;: :1 : . : :- ',' '', :
1: ! " ' j ; .:, I.!' :" '1 : .:! . :" ':,t i ':
' ' , ' ;i;' r "; " 1"" ...... ; ', : 12:1' 1 .; '
, i" " :":.' i:- '. 1:.1'.1:., . ... . : . . ' ':::" i ' .. h
: : ;. [ . , '" .. 1' ," 'I': : J .::' '/ I, :' ,: ": :: i ...1: . '
.. ', i
l
:! : I : I i ', ::,' .:::' ,,' .. .. .jmm, .' t.:: ;-; :;
I::; :: : .: <, .' 1 . ' , I':;' .. : .. ' .' , :'; . , ...' j .,' ., ,.
.. ..... : .. ... : ' IM' j'l':r J" I 'I :. :. i. "j : I, ' ! J., .' I '':' i: :1
. - - " , ' :,: ' . ", .'.'.: ::.; ' :.,: , I:" . .... .:' . ., . .... .. ' . . ' , . .. .... T
I:" : " ::' I,': j;::' :;':"' I :; '.',; 1:::: ',1:;: '.". . , I ; . ;; ,
: .' .... :. 'I : . ' ::: :.i 1'(: I':: i,' : ,' .. , ., i: ,':.; '1
. 1- " : .':' T: 'i ' . ,: ':1:: ,': ", '. I ' : .: 'j' .: ii ::
"
..... 1M ""C .. .. 1 :-_ " ', " . ".,. '1 ::""":: j . !. ' .. ". :i, .. ., L.,:.' ...
I [ - i ;';; : . -f ' ; I "; j . ' "1' . j:: " :, . -:' .I hi ,!, ; : 'j: .
' ,,. , .. ".,,'" '. : , :. , .. ,::,' .. ,:,y... . . ' :',., '"A:'''' ' " " '.
; .- , I: C
T
: I' . ; I i :,' :.: ... !' :::iJC; " "
, . i" ", '.,d , !.,' ... :, i ' .. ... i ( ':L: , ::
L " " . .. - . : ' 'I ' E..' . .. i" 1" 1 ill ' .!!. . I ' .:; : j' !'!,' :!
: . ' . ' : :, . . ' .i,. :-i,: ", . , ....... " .. ,::.... , . ". 'II i
i-"':" .. i i ' I. '.. -: .. :' : _ .. : (I: I : : I
! . -l :: I ' j i . :'.- ! ,,: .: - :: . :' :: : ;. . , . ': i ' : .,1 I
1-' " . : -", .. . , . , I: :'. 1 " I: .', i/" " :, I... ., ." 1
; : : : !"I ! : "; .. ! I 1'-; ' , , : I -. I . ':' :. -, ': . ..:
:. ' .. ,0 ; 1::' 1 ,. /-! . . . i . : : 1, . ' i:"-I ' ; i,' ,I i .. ;'j' i
. . . '.: -';::' -1 ' " I 1,: -.' . ' . " 1 . ;,'1:/ -, : _.J
: - : , ! ! , ' - ;.' ! ... i . _. - . .;:. J;.!.. . ... :t .i
: ."" .,.; .;. i ':"' :' 1:: : 1 '. i ; '. ' 1/, : : ' I' ;IF;. '.' .' i '.
I . 1._, , .. , I . " I': .' . "
:""' '1'" : . : :"" l ,1: 1 1-; , 'Y' .i'; ']7 ; , .. ,.' i' '
i,,;' , ! 1,i' l ie-j':H . ' :" 1): ' ; <>: :", , ii,:
/+ f.. : : . i l b: ;: 1 i iJJr '; < j_; I; :: .
! ,,, . : ; i li :1': I!h i v . ,<: :'! : : ,'
, " , ., : " ;" !!" I " , ' . ..
I ' i".i': ,.:. , ... , .. ' . . ........ , ... .. . . , ,: '::
j ' ii. T l vP'" ,'; " ,': ". ii
1, ';' . j li.,' j .; Y il '. I'll ' " .' ' .... I . i " : ".. fY j
; . '" '. , V : . I, " .. ", I .. ... . j: .. ;' ,
: ... i : I,: i -'
1
.. 1 .. L .' i ' : : ,;.: : .... , :i: .. "
1- 10 /' Vj'" I : : I ' .:. .. '0' : ... -:' 1 " i .... - j . ' ,!:::,
t, ' " I : ' ;! :.;" " r+ . . . : ': 1.'
l " L t': . _.. . __ . .. __ ._ .. __
112
u'
,.,
<h UR"
..
APPENDIX H MANAPOURl TRANSFORMER TECHNICAL DOCUMENTATION
SHORT - CIRCU IT TEST
H.V. 216 AllIpg (Po)
< .. 1' . .
I '.
'( , ', )
I
."
...
,
" If'
.. , ,.,
j
__ ... L .... 1.?f) ...... 4,H ...... .. .. ... ... 0,228 1 0,06
J .?.t. .... ...... ... "?J.:L ..... ... 9,)0

. 11.(, 5 .. .... J#i&. .. Ol2);! 0,085:
, : I
30,4
2
1. 0, 226 ;-0, 072 I. I 13
,
... .......... 3 1:1, 5 1 I .. _ .. }C! , '. La I 07 -4 ' .' "
.......... .... .. . 86 )0,24 ... -o, on l , ,37
.. "
... ,
77,'5
5 ... .. .... ..... ... .. ,. 28,1) 0,228 -0,066 8,8) 11,15
.. Ug,:4: ... :hl(L
80
... .. ....... ... l>., ... , .. . .. .. . .. . . . .... ..
................
,08 8 22 7<
8 1] <,
2" 7< 8 28,)2 0,2)4 -0,089
"
67. "
) ... 126... ..... ... 2;0.,1.6 ... . ... 3Jh.6.li .... ... q,.,,,a .. 1.h5.:L.
..... 1
12.1..... .. . .(,)2. ....... 78.,.8._ ... 2.16. ...... .. J6,.e ... 3.Q.J6.Q .... j ... Q,.?9J j-:-OdOQ ... .. .. .
"1,5 )9,') 102
Y
,)1 . 7,75
135L .. .... 4,,5.1.5 . . 22.,23.. _22.5,.1 ..... !:I2,.! ._3.1.,.45 1...9 . 1n ... ...
...... 1.36 ......... 4,42 ... 88,4 .. .. 21.,7.1- .. 22.1.... , ... 66,.( ...... )1 .... 0, 9 , 01. 80, 55. ..
5[ ............... 1-:32 ...... . ,. ", .8.l,6 .. . 2.1.,.1.1.. .. 2.14-,2 .. 83,6 .. .. 31.22 .... 0,202. 18 15,99 .. .
If' . .............. .............. ............... ............... ................ [
. _ .. .5 ... ... 115S._ ... ,.eZ .......... . .. ??5 ........ JI2: ..... .
.... ' .. .. .. ..1.1 ..... .. . ... ..
108 8 " 2) 17 41 211,5
... -0., 1.ot .. ...
.... ... -0, 10$ ... . . ..
0,203 -0,1 0 67,3
..... ,L ....... . .. .. ... .. . .. :-0 ..':J! .. .
. ...... .... .. ...... .... 30,83 0,226 10,00 7
0
,05
tn"s ...... ... ..1L .... ... ....... B ...... ... 66,45
.. .. 1
"
:'
............ .... .. ... .. ... ....... .... .... __ .. 88 .... -0,0')1 ' ,97 19,25
.. . .... ....... .. ..... ... _ ..... ... , ..... -O,Dt!5 14,6
d!, '''''''''' """.,.,,,,, """'"",,,. """"""''' """'-,.. """,.,,,,,
, ... ............ .. ... .... 1.1r.J. ..... ... ....... 2.9,.:W 0,225 -0,1 05
........ __ . .... .. _ .. .. .... ... ...
T'ra.no(oMner Serial No. 60312
'.
T EST PER F 'O R X A NeE
( Referred to Exhibit G of the contraot)
Ref.
01.1
(.) Rulltanoea (75-C)
Tran_(or.er voltage percent
on IIled.iUl'll >tap.
High tel!lIion winding Ohmll per phaae
Low tendon windill8 ohmu per phue
(b) Impedano-e
Tranefonnor i1l!podanoo peroent I
on mediUIII tap .
Zero eequenoe percent
(0) Tran_former volta&o percent ,
on IIIf11f.'iWII
(d) EJ::citi ng ourrflllt (13,6 tv)
0.1. 2
(a) No load 10BBOB at rated
(b) Load lOBBBB
" "
"
"
" "
"
."
"
at 11 5 rated kVA
n '100"''' .. '
.. 15"''' Ii
50:1" "
(0) Regulation on medium tap
Regulation at 100 rated kVA and.
100 f. p,.F.-
" .'"
80" "
(d) EfficientlY
at 100 power factor
at 11 5 rated kV.l
at 100 n .. ..
at 15 " 'n n
at SO " " ..
Unit

ohm
"
"
"
Amp.
kW
"
, "
"
"
"
"
"
"
"
"
143
[iiiiiiijjliiiiiii]
0,36.(
0,760
0,006)0
11, 26
24, 2

506
}O2,l 13
215 , 0
:;0 5 , 5
: ..
'j':J ,.(%
'j';l,5.( 1

Guarll.l1teBd
0,333
0,7';1
0 ,006.(
12
11

20

350
1;6
87,5
1,05
6,08
7
,",,48
'9.53
9';1; 56
'i9 , 6
0
144 APPENDIX H MANAPOUIU TRANSFORMER TECHNICAL DOCUMENTATION
.

,
III- 0
.1

,
3 t.
Iii
,
,

8
0
,
,
-
-,
-
1
:i
<: 1
.. -
----;-+

.;:." . .
, "1::'

Di..l m dro (olol'l'\Q "1 850 III .. PUG! tot Sb,4 s
... .) : h' '2 I.u .. per 'J,'tt..at.,
.- /
.. '; d.ilr. '; : lille ...... nit" (=J S1t.ltj lin, ... t-,lttj (-.::)_
:. :.
,--
H ... t(..,_k: -'0 . 1 ... \1; Of ",t .. t: - liro MbX 'r,nuda 0;35 - !JOe

<.
.....
', ; "
-', :.(
Otlro ;\ I; 0 Ct I. ... <:c '
tt Boo' c..
1
1
5

T
e
r
n
i
-
M
6
T
3
5

M
a
g
n
e
t
i
z
z
a
z
i
o
n
e
e

p
e
r
m
e
a
b
i
l
i
t
a

i
n

c
.
c
.

,

,

,

"

u

;

,

,

,

i

!

.
,

h

.
,

.
-
"

H

"
.

'
.

j
'

,
.

'
.

:
!
.
:
:
:

i
.
;

;
.
1
.

.
.
.

,

.
,

.
.
;

.

!

,
.

.
,

"

.
'

:
i
"

!
;

j
"

I
i

,

.
,

"

"

,

0

.

.

,

,

"

,

.
.

'
"

,
,
,
O
l
l
n
o
.
.

,

1
4
6

"

,

,

,

~

,

,

"

;

A
P
P
E
N
D
I
X

H

M
A
N
A
P
O
U
I
U

T
R
A
N
S
F
O
R
M
E
R

T
E
C
H
N
I
C
A
L

D
O
C
U
M
E
N
T
A
T
I
O
N

F
o
r
z
a

d
i

e
c
c
i
t
a
z
i
o
n
e

T
e
r
n
i
-
M
6
T
3
5

t
J
l

"
N
O
l
z
n
O
N
I

,

,

"

<

,

"

<

"

1

t
'

>
;
,

:
.
!

.
.

r
.
i

'
t
'

.
'

.
.
.

!
.
'
,
.

P
e
r
d
i
t
a

T
e
r
n
i
-
M
6
T
3
5

1
4
7

t

,

,

,

,

,

,

!

"

.

"

,

,

,

"

'.
"

"

,

i
,
;

1
4
8

A
P
P
E
N
D
I
X

H

M
A
N
A
P
O
U
R
I

T
R
A
N
S
F
O
R
M
E
R

T
E
C
H
N
I
C
A
L

D
O
C
U
M
E
N
T
A
T
I
O
N

T
e
r
n
i
-
M
6
T
3
5

P
o
t
e
n
z
a

d
i

e
c
c
i
t
a
z
i
o
n
e

.

.

,

c

,
.

:::

"

,

c
.
,

,

E

,

t

,

.
\

"

"

,

-
"

.
-

"

"

.
'

,

H

,

,

,

.
,

l

i

,

,

I
s
t
e
r
e
s
i

I
n

G
.
G
.

"

"
.

'
.

0
1

T
e
r
n
i
-
M
6
T
3
5

1
4
D

-

"

<

>

Appendix I
BENMORE POLE lA PARAMETERS
I.l UMEC SINGLE-PHASE VALIDATION DATA
chapter 5
Generators 0 1-G3 data
Rating 112.5 MVA
Voltage 16.0 kV
Xd
1.168 p.ll.
x'
d
0.264 p.ll.
~
0.174 p.ll.
Xq 0.672 p.ll.
x"
q
0. 19 p.u.
T';" 8.7 sees
T"
'"
0.087 sees
T"
,.
0. 132 sees
X, 0. 15 p.u.
R.
0.0042 p.ll.
Converter transformer data
Type
Rating
Three-phase banks; Delta-Star, and Delta-Delta
187.5 MVA
16 kV Prima.ry voltage
Secondary voltage
X1(prim-t.er)
110 kV
0.113 p.ll. (62.5 MVA base)
PSCAD-EMTDC generator model parameters
Lnertia. constant 3.42 MW /MVA
Base angular frequency
Mechanical damping
Rated nns phase voltage
Rated [InS phase current
314.15926 rad/sec
le-4 p.lI .
9. 238 kV
4.0595 kA
152 APPENDIX I BENMORE POLE lA PARAMETERS
Terminal voltage magnitude a.t t-
Terminal voltage phase at ~
Real power at t-
Reactive power at t-
Number of coderant machines
Potier reactance
Direct-axis reactance
Direct-axis transient reactance
Direct-axis sub-transient reactance
Damper-field mutual reactance
Quad-axis reactance
Quad-axis sub-transient reactance
Armature resistance
Armature time constant
Direct-a.:x.is transient time constant
Direct-axis subtransient time constant
Air-gap factor
X,
, '
Y
,
X,
Y,
X,
Y3
X,
Y,
X,
X,
Yo
X,
Y,
X,
Y,
X,
Y,
X,O
Y,O
o p.o.
Orad
OMW
o MVA,
3.0
0.14587 p.u.
1.16775 p.u.
0.30814 p.u.
0.19913 p.u.
o p.ll.
0.67163
0.19913 p.u.
0.00042 p.u.
0.37127 sees
8.7 sees
0.087sec
0.8
0.0
0.0
0.5
0.5
0.8
0.79
0.1
0.947
1.2
1.076
1.5
1.2
1.8
1.26
2.2
1.32
3.2
1.53
4.2
1.74
I.l UMEC SINGLE-PHASE VALIDATION DATA
UMEC per-phase converter transformer model parameters
Winding limb length
Winding limb area
Yoke length
Yoke area.
Primary winding turns number (Delta-Star and DeltarDelta)
Secondary winding turns number (Delta-Star)
Secondary winding turns number (Delta-Delta)
Ltl (Delta-Star and Delta-Delta)
L" (Delta-Star)
LI2 (Delta-Delt.)
Inj
In,
bj
b,
E
PSCAD-EMTDC converter model pru:ameters
Rated frequency
Thyristor configuration variable (Delta-star)
Thyristor configuration variable (Delta-Delta)
Firing pulses
PLO proportional gain
PLO integral gain
PLO input reference varaible
Snubber resistance
Snubber capacitance
Thyristor on resistance
Thyristor off resistance
Forward voltage drop
Forward break-over voltage
Minimum extinction time
50 Hz
o
-1
PLO
10.0
100.0
o
240051
0.29 }iF
0.01 51
1.0e6 51
0.001 kV
1.0c5 kV
0.0 ~ s e s
3.59 m
0.4536 m
2
2.656 m
0.4536 m'
113 Thrns
450 Thms
780 Turns
0.768 mH
11.8 mll
35.6 mH
0.021
3.0e-5
0.0
1.63
1.90
153
154 APPENDIX I BENMORE POLE IA PARAMETERS
1.2 UMEC CONVERTER TRANSFORMER COMPARlSON
DATA
chapter 7
1. 2.1 Delta-primary, llOkV fa ult
UMEC three-limb three-phase converter transformer DdO
Winding-limb length 3.59 ill
Wirlding-limb area
Yoke length
Yoke ru:ea
Primary winding turns number
Secondary winding turns number
'P7,9 ,ll
'Pa,lo,12
'P15,16,17
ml
m,
b
l
b,
&
Connection matrix
1
0
[OJ
- I
0
0
0
0
1
0
0.4536 m'
2.656 m
0.4536 m
2
113 Turns
780 Thrns
5.85e-8
5.85e-8
5.850-8
0.021
3.0e-5
0.0
1. 63
1. 90
0 0
0 0
1 0
- I 0 1
0 - I 0
0 0 -I
- I
0
0
0
1
0
0
- I
0
0
0
1
(1. 1)
1.2 UMEO CONVERTER TRANSFORMER COMPARlSON DATA
UMEC three-limb converter tra..nsformer Dyll
Winding limb length 3.59 m
Winding limb area 0.4536 m'
Yoke length 2.656 ill
Yoke area 0.4536 m'
Primary winding turns number 113 Turns
Secondary windillg turns nwnber 450 Turns
'P7,9,11
5.85e-8
'PS, l O,12
5.85e8
'PUi,lG,17
5.85e-8
Tn, 0.021
Tn, 3.0e-5
b, 0.0
b, 1.63
: 1.90
Connection matrix
1 0 0 0 -1
0 1 0 0 0
- 1 0 1 0 0
IC)= 0 0 0 1 0
0 0 -1 0 1
0 0 0 0 0
0 - 1 0 - 1 0
circuit-breaker model
Init.ial state
Use pre-insertion resistances?
Breaker open resistance
Breaker closed resistance
Open possible if current flowing?
Time to open breaker
Closed
No
1. 0e6 n
0.005 n
No
1.05 sees
1.08 sees
No
Time to close breaker
.Repea.t
PSCAD-EMTDC bypass valve model
Snubber resista.nce
Snubber capacitance
Thyristor OIl resistance
Thyristor off resistance
Forward voltage drop
Forward break-over voltage
Minimum extinction time
2400 fl
0.01 J.'F
0.005 fl
1.0e6 n
0.001 kV
1.0c5 kV
0.0 lJ.secs
0
0
0
0
0
1
- 1
155
(1.2)
156 APPENDIX I BENMORE POLE lA PARAMETERS
PSCAD-EMTDC fault parameters
Fault on resistance
Fault off resistance
Is phase A in fault?
Is phase B in fault?
Is phase C in fault?
Is fault to ground?
Clear possible if current flowing
Time to apply fault
Duration of fault
Repeat
1.2.2 Star-primary, 16kV fault
30 n
1.0,6 n
Yes
No
No
Yes
No
1.00 sees
0.08 sees
No
UMEO three-limb three-phase converter transformer Yl:0
All parameters the same as DdO except:
Primary winding turns number 65 'DIms
Secondary winding turns munber 450 Thrns
Connection matrix
1 0 0 0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
ICI = 0 0 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1
0 - 1 0 - 1 0 - 1
UMEO three-limb three-phase converter transformer Ydll
All parameters the same as Dyll except:
Primary winding turns number 65 Thrns
Secondary winding turns number 780 Thrns
Connection matrix
1 0 0 0 0 0
0 1 0 - 1 0 0
0 0 1 0 0 0
ICI =
0 0 0 1 0 - I
0 0 0 0 1 0
0 - I 0 0 0 1
(1. 3)
(1.4)
PSCAD-EMTDC circuit-breaker, bypa.<;s valve and fa.ult paramtel's for the 16k V fa.ult
are identical to the llOk V fault except the fault resistance is changed 0.1 o.
Appendix J
SMOOTHING TRANSFORMER TEST SYSTEM
PARAMETERS
PSCAD-EMTDC three-phase source model parameters
Type resistive
Extemal voltage control?
External frequency control'r
Initial pbase
Impedance
Voltage magnitude (ac, L-L, rrns)
Ramp up time
No
No
0.0 deg
1.0 n
0.707 kV
0.2 sec
Frequency 50 Hz
PSCAD-EMTDC converter model parameters
Rated frequency 50 Hz
Thyristor configuration variable 0
Firing pulses PLO
PLO proportional gain 10.0
PLO integral gain 100.0
PLO input reference varaible 0
Snubber resistance 5000 n
Snubber capacitance 0.051'F
Thyristor 011 resistance 0.01 n
Thyristor off resistance 1.0e6 n
Forwa.rd voltage drop 0.001 kV
}i'orward break-over voltage l.(le5 kV
Minimum extinction time 0.0 J,LSecs
158 APPENDIX J SMOOTHING TRANSFORMER TEST SYSTEM PARAMETERS
UMEC smoothing transformer model parameters
(a) Linea.r model.
Branch
m l (1ine4r}
Winding limb 292.38
Yoke 292.38
leak-path 4.17e-2
Table J.1 Pet-1Init smoothing transformer iron-core characteristi cs.
Base frequency
Dc blocking capacitor
(a) Non-linear model.
Branch
Winding limb
Yoke
leak-path
50 Hz
10 mF
ml linear}
292.38
292.38
4.17e-2
m2{64tl b. b,
0.67 0.0 1.03 0.85
1.67 0.0 1.03 0.0
Table J .2 Per-unit smoothing transformer iron-core saturation characteristics.
Base frequency
Dc blocking capacitor
Dc load parameters
50 Hz
lOmF
Linear smoothing transformer Ide
Non-linear smoothing transformer (0'=15) ide
Non-linear smoothing transformer (0=60) Ilk
0.62 p.lI.
0.75 p.n.
0.18 p.u.
Appendix K
CONVERTER TRANSFORMER MODEL
PARAMETERS AND DC LINE DATA
PSCAO-EMTDC deltarstar converter transformer model para.meters
'fransformer name
Three-phase transformer MVA
Base opera.tion frequency
Winding #1 Type
Winding #2 Type
Delta leads of lags star?
Positive sequence leakage reactance
Ideal transformer model ?
No-load losses
Tap-changer winding
Winding 1 line to line voltage
Magnctising current
Magnetislng current
Saturation enabled?
Delta...-Star
187.5 MVA
50 Hz
Delta
St.,.
Lags
0.113 p.u.
No
0.0 p.n.
#2
16 kV
1.0 %
110 kV
1.0 %
No
160APPBNDIX K CONVERTER TRANSFORMER MODEL PARAMETERS AND DC LINE DATA
PSCAD-EMTDC delta-delta converter transformer model parameters
Transformer name
Three-phase transformer M VA
Base operation frequency
Winding #1 Type
Winding #2 Type
Positive sequence leakage reactance
Ideal transformer model ?
No-Ioa.d losses
Tap-changer winding
Winding l1ine to line voltage
Magnctising current
Magnetising current
Saturation enabled?
'IIansrnission line data
Delta-Delta.
187.5 MVA
50 Hz
Delta.
Delta.
0.113 p.u.
No
0.0 p.u.
#2
16 kV
1.0 %
110 kV
1.0 %
No
The data for the NZ HVdc transmission line was obtained from New Zealand
Electricity. The per pole line resistance is made up of: Overhead line lO.9S!l per pole
and submarine cable resistance l. SHl. The submarine cable capacitance is O.333/-,F jkm;
a total capacitance of 13.3331J.'F. The series inductance for the overhead line is 1.33mHjkm,
and the shunt capacitance O.0121JF jkm. The submarine cable divides the overhead line
into two segments; the South Isla.nd and North Island partitions are 530km and 40km
respectively.
Dc filter data

1400Q
0.037H 0.019H

FigUl'e K.l Benmore P1A de line filter design.
, ,
REFERENCES
ABU-NASSER, A.I. (1981), 'Transformer Inrush in 'fraction and HVde Schemes', PhD
thesis, Victoria University of Manchester, October, pp.135- 168.
ACHA, E. (1988), Modelling of P01lJCr' System Transformers in the Complex Conju-
gale Harmonic Space, PhD thesis, University of Canterbury, Christchurch, New
Zealand.
ACHA, E., ARRILLAGA, J. , MEDINA, A. AND SEMLYEN, A. (1989), ' General fi-rune of
refereuce for analysis of ha.rmonic distortion in systems with multiple transformer
llon-linearities', lEE Proceedin9s Pt. C, Vol. 136, No.5, September, pp. 271- 278.
ANDERSON, G. W.J, (1995), Hybrid simulation of ac-de pOllJcr systems, PhD thesis,
University of Canterbury, Christchurch, New Zealand, ApriL
ANDO, M., SCHWEITZER, E.O. AND BAKER, RA. (1985), 'Development and field-
dat a evaluation of singlc-Imd fault locator for two-terminal hvdc tra.nsmission
lines: Part 1- data collection system arid field data', IEEE 1Tansactions on P(J1IJer
Apparatus and Systems, VoL P AS-I04, No. 12, December, pp.3525- 3537.
ARRILLACA, J. (1983), High Voltage Direct Current Transmission, Peter PerCf:,'Tinus,
London.
AR.R.fi,LACA, J. AND ARNOLD, C.P. (1990), Computer Analysis oj Power Systems,
John Wiley and Sons, Ba.ffins Lane, Chichester, West Sussex P019 IUD England.
ARRILLAGA, J., SANKAR, S . , WATSON, N.R. AND ARNOLD, C.P. (1991) , ' Operar
tional Capa.bility of generator RVDC Convertor Units', IEEE Ihmsactions on
Power Delive1"1}, Vol. 6, No.3, July, pp.1171- 1l76.
ARRILLAGA, J ., SANKAR, S., ARNOLD, C.P. AND WATSON, N.R. (1992), 'Charac-
teristics of Unit- Connected HVDC Cenerator Convertors Operating at Variable
Speeds', Proceedings oj the lEE, Part C, VoL 139, No.3, May, pp.295- 299.
ARRILLAGA, J., BOWLES, J. , BARROS, J.C., K ANNGIESSER, KW. AND INGRAM, L.
( 1993a), ' Direct connection of generators to HVdc converters: ma.in characteris-
tics and comparitive advantages', ELECTRA, Vol. 149, August, pp.19-39.
162 REFERENCES
ARRll..LAGA, J., MACDONALD, 8.J., WATSON, N.R. AND WATSON, S. (1993b), 'Direct
connection of series generators and hyde converters', iEEE Transac-
tions on Power Delivery, Vol. 8, No.4, October, pp. 1860- 1866.
ARlULLAGA, J., CAMACHO, J.R., MACDONALD, S.J. AND ARNOLD, C.P, (1993c),
'Operating Charactedstics of unit and group connected generator HVDe con-
verter schemes', Proceedings of the lEE, Part 0, Vol. 140, No.6, November,
pp.503- 508.
ARRILLACA, J., MEDINA, A., LrsBoA, M.L.V., CAVIA, M.A. AND SANCHEZ, P.
(1995), 'The Harmonic Domain. A fra:me of reference for power system Har-
monic Analysis', IEEE 7ransactions on Power Systems, Vol. 10, No. 1, February,
pp.433- 440.
ARTURl, C.M. (1991), 'Tr ansient Simulation a.nd Analysis of a Thl'ce-Phase Five-
Limb Step- Up Transformer Following Out- Of-Phase Synchronisation', IEEE
l1'ansactio1l3 on Power Delivery, VoL 6, No.1, January, pp.19fi- 207.
ASPLUND, G., ASTROM, U., CANALHAS, A., ABERG, M., PUR-RA, E. AND HEYMAN,
O. 'A novel approach on UHVdc 800kV station and equipment design'. Pa.per
presented at the CIGRE Study Committee 14 Meeting, Wellington, New Zealand,
29 September-1 October, 1993.
ASTROM, G., ASPLUND, G., CARLSSON, L., JONSSON, U. AND LOaF, O. 'Outdoor
thyristor valve-A test installation in the Konti-Skan dc link at Stenkullen con-
verter sta,tion'. Pa.per presented at t he CIGRE Study Committee 14 Meeting,
Wellington, New Zealand, 29 September-l October, 1993.
BICKFORD, J.P. AND HEATON, A.G. (1986), 'Transient overvoltages on power sys-
tems', lEE Proceedings, VoLPt C, No.4, May, pp. 201-225.
CALVERLEY, T .E., O'l"rAWAY, C.H. AND TUFNELL, D,H.A. (1973), 'Concept of a.
unit generator converter tra.nsmission system', lEE Conference on High Voltage
de and ae Transmission, London.
CHEN, M.S. AND DILLON, W.E. (1974), 'Power Systems Modeling', Proceedings IEEE,
Vol. 62, No.7, J uly, pp. 901-915.
CHEN, X .S. AND NEUDORFER, P . (1993), 'The Development of a Truee-Phase Multi-
Legged Tra.nsformer MoclpJ for Use With EMTP', Department of Energy Contract
#DE-A C79-92BP26702, March.
CUEUKSUN, W., MOl-IAN, N., WRIGHT, S.E. AND MORTENSEN, K.N. (1989), 'Fea-
sibility study of AC- and DC-side active filters for HVdc converter tenninru!;',
IEEE Transactions on P01uer Delivery, Vo1. 4, No.4, October, pp. 2067-2075,
163
CIGRE JOINT WORKING GROUP 11/14-09 (1993), 'Hvde unit connected genera.tors',
ELECTRA, VoL 149, August.
DICK, E.P. AND WATSON, W. (1981)' 'Transformer Models for Transient Studies
based on Field Measurements', IEEE Transactions on Power Apparatus and y ~
terns, Vol. PAS- I OO, No. 1, January, pp.409-419.
DIJK, H.E. (1986L 'Voltage jump a.t transformer terminals due to non-simultaneous
switching" i EEE 1hmsactions on Power Deli1Jery, Vol. PWRD-l, No.3, July,
pp. 147- 155.
DOMMEL, H.W. (1969), 'Digital Computer Simulation of Electromabrnetic Transients
in Single- and Multipbase Networks', IEEE 7ransadions on Power Apparatus
and Systems, Vol.PAS-88, No.4, April, pp.388-399.
DOMMEL, H.W. (1975), 'Tr ansformer Models in t he Simula.tion of Electromagnetic
Tra.nsients" 5th Power Systems Computation Conference, Cambridgc, England,
Vol. 3.1/4, pp. I- 16.
DOMMELEN, D.V. (1987), Alternative 7rcmsients Program Rule Book, Leuven EMTP
Center, July.
EMTDC MODELS (1994), EMTDC V9 Custom models, Manitoba. HVdc Research Cen-
t re, 400-1619 Pembina. Highwa.y, Winnipeg, Manitoba, Canada R3T 2G5.
EMTDC USER (1994), EMTDC V9 User 's manual, Manitoba. HVdc Research Centre,
400-1619 Pembina. Highwa.y, Winnipeg, Manitoba, Canada R3T 2G5.
FLUX, R.W. (1971), 'The sealed Windings site-assembled t ransformer-A cas study',
IEEE transactions of Power Apparotw;: and Systems, Vol. PAS-90, No.5, October,
pp.2002- 2009.
GENERAL ELEC'l'ruC COMPANY (1987), Protective relays application guide, GEC mea-
surements, St. Leonards Works, St afford ST174LX, England, t hird edition cd.,
June.
GERMAN, D.M. AND DAVIES, A.E. (1981), 'The simulation of transformer feeders fol-
low:ing switching operations', iEEE 7ransactions on Power Apparatus and Sys-
tems, VoL PAS-lOO, No. 11, November, pp. 4510-4514.
GIESNER, D.S. (1971), The dynamic behamour of ac/dc power systems under ac fault
condition.1, PhD t.hesis , Unjversity of Canterbury, Christchurch, New Zeala.nd.
HATZIAROYRlOU, N. D., PROUSALIDIS, J.M. AND PAPADIAS, B.C. (1993), ' Gen-
eralised Transformer Model Ba.<;cd on Analysis of its Magnetic Circuit', lEE
Proceedings- C, Vol. ]40, No.4, July, pp.269- 278,
164 REFERENCES
HEFFERN'AN, M.D., ARRILLAGA, J. AND TURNER, K.S. (1981), 'Recovery from t m ~
porary HVde line faults', IEEE 1ransactions on Power Apparatus and Systems,
Vol. PASlOO, No.4, April, pp.1864-1870.
HrnaORANI, N,C. (1961)' Control and P1'otection of high voltage de systems, PhD
thesis, The Victoria University of Manchester, January.
HINGORANI, N.C. (1970), 'Transient overvoltages on a hi-polar HVde overhead line
caused by de line fa.ults', IEEE 'Iransactions on Power Apparatus and Systems,
Vol. PAS-S9, No.4, April, pp. 592- 610.
INGRAM, L. (1988), 'A practical design for an integrated HYDe unit connected Hydro
Electric Power Statioll', IEEE 7ransactions on Power Delivery, Vol. 3, No.4,
October, pp.1615-1621.
.JOOSTEN, A.P.B., ARIULLACA, J., ARNOLD, C .P. AND WATSON, N.R. (1990), 'Sim-
ulation of hvdc system disturbances with reference to the magnetizing history of
the converter transformer', IEEE Transactions on Power Delivenj, Vol. 5, No.1,
January, pp. 330- 336.
LEON, P.D. AND SEMLYEN, A. (1994), 'Complete ']}ansformer Model for ElectroMag-
netic Transients', IEEE Transactions on Power Delivery, Vol. 9, No.1, pp.231-
239.
LIN, C.E., YEH, J.C., HUANG, C.L. AND CHENG, C.L. (1995), 'Transient model and
simulation in three-pbase tbree-limb transformers', IEEE TI'ansactions on Power
Delivery, VaLlO, No.2, April, pp. 896- 905.
LrSBOA, M.L.V., ENRIGHT, W. AND ARRll.LACA, J. 'Improved repcsentation of tho
magnetic non-linearities of multi-limb transformers for use in power system har-
monic studies' . Paper accepted for presenta.tion at the 12th PSCC conference,
Dresden, Germany, 19-23 August 1996.
LrSBOA, M.L.V., ENIUGIIT, W. AND ARRILLACA, J. (1995), 'Harmonic and time
domain simulation of transformer magnetisation non-linearities\ Proceeedings
IPENZ Annual Conference (Innovative Technology), Vol. 2, February, pp. 72- 77.
MACDONALD, S.J., ENRIGHT, W., ARRlLLAGA, J. AND O'BRIEN, M.T. (1995), 'Har-
monic measurements from a group connected genet'ator HVdc converter scheme',
iEEE Ihmsactions on Power Delivery, Vol. 10, No.4, October, pp.1937- 1943.
MCPHERSON, G. (1981), An Introduction to Electrical Machines and Transformers,
John Wiley and Sons.
MEDINA, A. (1992), Power Systems Modelling in the llarmonic Domain, PhD thesis,
University of Canterbury.
165
MEDlliA, A. AND AllRn. LAGA, J. (1992a), ISimulation of multi limb power t ransformers
jn t he harmonic domain" lEE Proceedings-C, Vol. 139, No.3, Ma.y, pp. 269-276.
MEDINA, A. AND ARRlLLAGA, J . (1992b), 'Genera.lised modelling of power trans-
formers in the Harmonic Domain', IEEE TI-ansactions on Power Delivery, Vol. 7,
No. 3, July, pp. 1458-1465.
MILLER, A.J.V. (1992), IMultichannel continuous harmonic analysis in real time' ,
i EEE TIunsactions on Power Delivery, Vol. 7, No.4, October, pp.1813- 1819.
NAIDU, M. AND MATHUfi, R.M. (1989), ' Evaluation of Unit Connected, Variable
Speed Hydropower Station for HVDe Power Transmission', IEEE 7ransactions
on Power Systems, Vol. 4, No.2, May, pp. 668-676.
PAI"'iC, C.X. AND TARNAWECKY, M.Z. (1995), ' Generator winding 12R losses a.nd
Harmonic Interference 1lllder varia.ble frequency operation of a. HVdc unit-
connected scheme', IEEE TI-ansactions on Energy Conversion, Vol. 10, No. 1,
March, pp. 133- 139.
PETERSON, H.A., PHADKE, A.G. AND REITAN, D.K. (1969) , l'IhUlsients in ERVne
power systems: Part I-Rectifier fault currents' , IEEE 7hmsactions on Power
Apparatus and Systems, Vol. PAs..88, No.7, July, pp.981-989.
SAY, M.G. (1983), Alternating current machines, Longman scientific and technical,
Longman house, Burnt Mill, Harlow, Essex GM20 2JE, Engla.nd, fifth ed.
SLEMON, G. (1966), Magnetoelectric Devices: 'I'ransducers, TI-ansJormers and Ma
chines, John Wiley and Sons, Inc.
SMI"l'l-I, n.R. , SWANSON, S.R. AND BORST, J.D. (1975) , !Overvolta.ges with remotely-
switched ca.ble-fed grounded wye-wye tra.nsformers', IEEE Transactions on Power
Apparotu.7 and Systems, Vol. PAS-94, No.5, September/October, pp.1843- 1851.
STEINME'l'Z, C.P. (1895), 'Theory of the general Alterna.ting Current transformer',
American i nstitute of Electrical Engineers, Vol.xij, .Tune, pp.245 - 256.
STUEHM, D. L. (1993), Three-phasc n'ans/01'mc7' Gore Modding, Technical Report,
North Dakota State University.
TAKASU, N. , OSHI, T. , MIYAWAKI, F., SAlTO, S. AND FUJIWARA, Y. (1994), IAn
experimental a.nalysis of de excita.tion of transformers by geomagnetically induced
currents' , IEEE 7hmsactions on Power Delivery, VoL 9, No.2, April, pp.1173-
1182.
TESTI, G. (1994), Dc side harmonics and filtering in HVdc transmission systems,
CI GRE Task Force 02, Working Group 14.03.
166 REFERENCES
THE ELECl'RICI'1'Y COUNCIL (1981), Power system protection, Vol. Principles and
components, Peter Peregrinus Ltd, Stcvenage UK and New York.
TURNER, S. (1977), Modelling the Magnetic Characteristics of a Power Transformer,
PhD thesis, The Victoria. University of Manchester.
TURNER, K.S. (1980), Transient stability analysis of inte.gmted ac Mid de power sys-
tems, PhD thesis, University of Canterbury.
UHLMAN, E. (1975), Power 1'ranBmission by Direct GUrTCnt, Springer- Verlag.
WOODFORD, O.A. (1985), (Validation of digital simulation of DC links', IEEE Irans-
actions on Power Apparatus and Systems, VoL PAS-I04, No.9, pp. 2588-2595.
WOODFORD, D.A., COLE, A.M. AND MENZIES, RW. (1983), 'Digital Simulat.ion of
DC Links', IEEE 'nunsactions on Power Apparatus and Systems, VoLPAS-102,
No.6, pp.1616- 1623.
ZAVAHlR, .1.M. (1992), Ele.ctromagnetic rransient Simulation with State Vanl1blc
Representation of HVdc Conversion, PhD thesis, Univcrslty of Canterbury,
Christchurch, New Zealand, June.

Thesis Fulltext

Uploaded by

Copyright:

Available Formats

Thesis Fulltext

Uploaded by

Document Information

Original Description:

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

Thesis Fulltext

Uploaded by

Copyright:

Available Formats

Transformer models for

electromagnetic transient studies

= [[M:'.) [M:,)] [[1';)

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.