Professional Documents
Culture Documents
A thesis
presented for the degree of
Doctor of Philosophy in Electrical Engineering
in the
University of Canterbury,
Christchurch, New Zealand
by
1980
i
CONTENTS Page
CHAPTER 1 INTRODUCTION 1
2.0 INTRODUCTION 5
2.1 STEADY STATE STABILITY 6
2.2 TRANSIENT STABILITY 8
2.3 MULTI-MACHINE TRANSIENT STABILITY 10
2.4 MODELLING FOR TRANSIENT STABILITY STUDIES 11
2.4.1 Network Representation 12
2.4.2 Synchronous Machine Model 13
2.4.2.1 Algebraic equations 13
2.4.2.2 Differential equations 15
2.4.3 Speed Governor 16
2.4.4 Automatic Voltage Regulator 16
2.4.5 Loads 17
2.5 COMPUTATIONAL CONSIDERATIONS 18
2.6 SUMMARY 20
Page
3.3 ABNORMAL MODES OF OPERATION 37
3.3.1 Mode Classification 37
3.3.2 Equations for Abnormal Operation 38
3.4 UNIFIED ALGORITHM 39
3.4.1 Algorithm Proposal 39
3.4.2 Formulation of Equations for
Normal Operation 41
3.4.3 Formulation of Equations for
Abnormal Operation 43
3.4.4 Programme Implementation 44
3.4.4.1 Calculation of initial conditions 44
3.4.4.2 Choice of operating mode 46
3.4.4.3 Control specification 47
3.5 COMPARISON OF SEQUENTIAL AND UNIFIED
ALGORITHMS 47
3.6 RESULTS 50
3.6.1 System Studied 50
3.6.2 Discussion of Results 52
3.6.3 Rectifier Performance 55
3.7 CONCLUSIONS 56
Page
4.3.3.6 Significance 9f spectral
leakage in TCS waveforms 79
4.4 ALTERNATIVE TO SPECTRAL ANALYSIS 80
4.4.1 RMS Approximation for Voltage 81
4.4.2 RMS Approximation for Power 82
4.5 CONCLUSIONS 84
Page
CHAPTER 6 STUDIES WITH COMBINED. TCS AND QSS
DC LINK MODELS 114
6.0 INTRODUCTION 114
6.0.1 System Studied 114
6.0.2 Disturbances Examined 116
6.1 RECTIFIER AC SYSTEM FAULT 118
6.1~1 The Effect of Time Variant
Equivalents 118
6.1.2 Comparison of Results 120
6.1.2.1 Differences in voltage and 120
reactive power
6.1.2.2 Differences in real power 122
6.1.2.3 Coincidence at TCS end points 125
6.1.3 Approximations for TCS Equivalents 127
6.2 INVERTER AC SYSTEM FAULT 127
6.2.1 Fault Representation Using
TCS Equivalents 130
6.2.2 Low Voltage Mismatch 132
6.2.2.1 Second iteration 132
6.2.3 Comparison of Results 134
6.2.3.1 Differences in voltage and
reactive power 134
6.2.3.2 Differences in real power 136
6.2.3.3 Coincidence at TCS end points 138
6.2.4 Approximations for TCS Equivalents 138
6.3 DC FAULT STUDY 140
6.3.1 Link Performance During Fault 141
6.3.2 Matching QSS Restart with TCS 142
6.3.3 Termination of TCS 143
6.3.4 Comparison of Results 144
6.3.4.1 Recifier reactive power response 144
6.3.4.2 Differences in real power 145
6.3.5 Approximations for TCS Equivalents 147
6.4 CONCLUSION 147
Page
7.1.2 Full Damping Coritrol 152
7.1.3 Short Term DC Current Overload 152
7.2 POSSIBILITIES FOR TRANSIENT STABILITY
IMPROVEMENT 153
7.2.1 Classification of Systems 153
7.2.2 System for Study 155
7.3 THE EFFECT OF REALISTIC MACHINE MODELS 156
7.3.1 Summary of Results 156
7 3.2 Discussion 158
7.3.3 Effect on P-6 Curve 160
7.3.4 The Influence of Fault position 162
7.4 FACTORS AFFECTING TRANSIENT STABILITY
IMPROVEMENT 162
7.4.1 Magnitude of DC Current Increase 164
7.4.2 Period of DC Current Increase 167
7.4.3 DC Link position in the Network 168
7.4.4 Fault Resistance 169
7 4.5 Rate of DC Current Increase 170
7.4.6 Summary of Results 171
7.5 RESULTS WITH REALISTIC SYSTEMS 172
7.5.1 Two Generator SI System Equivalent 172
7.5.2 Full SI System 174
7.5.3 Full NZ System 174
7.6 CONSIDERATION OF FULL DAMPING CONTROL 177
7.6.1 Discussion of Controller 177
7.6.2 Results 178
7.7 ALTERNATIVE TS IMPROVEMENT SYSTEMS 180
7.7.1 Inverter End TS Improvement 180
7.7.2 DC Link Power Reversal 180
7.8 TRANSIENT CONVERTER SIMULATION RESULTS 182
7.8.1 Rectifier End Fault Case 184
7.8.2 Inverter End Fault Case 187
7.8.2.1 First Simulation 189
7.8.2.2 Second simulation 189
7.8.3 DC Link Reversal 191
7.9 CONCLUSION 194
REFERENCES 198
vi
Page
APPENDIX A1 RELATIONSHIP BETWEEN AC AND DC
CURRENTS IN ABNORMAL MODES 205
THE AUTHOR
Symbols
I Current
V Voltage
Synchronous Machine Field Voltage
Admittance
R Resistance
x Reactance
Commutation Reactance
Synchronous Machine Transient Reactance
X" Synchronous Machine Subtransient Reactance
L Inductance
C Capacitance
S Complex Power
P Real Power
Q Reactive Power
H Inertia Constant of Rotating Machine
T'
qo' T'do Synchronous Machine Transient Open C~rcuit Time
Constant
Til T" Synchronous Machine Subtransient Open Circuit
qo' do
Time Constant
Vd DC Voltage
Id DC Current
VLlL Thevenin Source Voltage
E~ Converter Fundamental AC Terminal Voltage
I~ Converter Fundamental AC Terminal Current
Zth L1L : Thevenin Delay Angle
a Converter Delay Angle
u Converter Commutation Angle
o a + u (converters)
o Generator Rotor Angle (for synchronous machines)
y Converter Extinction Angle
~ Power Factor
7T 3.14159
t Time
h Integration Step Length
T Period of Power System Frequency
W Angular Frequency
x SP Specified Value of x
x Derivative of x with t to Time
cos x + j sin x
j Complex Operator
f (x) Function of x
x Complex Variable x
x* Complex Conjugate of x
[ Matrix or Vector
.; Square Root
Integration
fE Summation
a>b a greater than b
a <b a less than b
++ Fourier Transform Pair
x Convolution
Subscripts
Abbreviations
N. Z. New Zealand
S.1. South Island of N.Z.
N.1. North Island of N.Z.
p.u Per unit
x
ABSTRACT
ACKNOWLEDGEMENTS
Proverbs 31.10.
1
CHAPTER 1
INTRODUCTION
CHAPTER 2
2.0 INTRODUCTION
Machine 1 Machine 2
Power
E
f
Increasing
steady tate
Stabili y
Limit
P
2
PI
increased, as
increases to 02' by the increase in field
vol tage, E , modifying the P-cS curve. In normal system
f
operation generators are generally operating well below their
steady state stability limit with angles ranging up to 40.
Angle (0)
Angle
Fig. 2.2 b
Time
Power
Powe
Prefault
curve
curve
Fault curve
P
2 - --
Angle (0)
around the network for the next point of the swing curve.
with the advent of the digital computGr the alternating
solution of the differential and network equations was
implemented entirely on the computer, eliminating the need
for any hand calculations. The complexity of the machine
model could be increased and the system size was only limited
by the amount and speed of the computing time available.
(X,Y) = 0 ( 2.2)
Y == f(X,Y,t) ( 2.3)
-
Y :::::
1
(2 .4)
ab R + jx
Y
aa = Y
bb
::::: -jB (2.5)
/2
a = 1 + O.OlT ( 2 .6)
and Y
ab = aYab ( 2 .7)
t 2
Y = (a - a)Y ( 2 .8)
aa ab
t
Y
bb = (1 - a)Y
ab
( 2 . 9)
t
R
a
(2.11)
x d E
q
- V
q
14
xq ~ Xi
q
~ X"
q
(2.12)
E ~ ED ~ E"
-j (~ - 6)
(I
r
+ j I
m
) = (I
d
+ j I )e
q
2 (2.13)
I = Yfict (E - \7) + I
sal
(2.14)
2
where Y
fict =
(R
a - j(x
d + x q )/2)/(R a + XdXq) (2.15)
I = Y (E - *e j26
V) (2.16)
sal sal
Y
sal = j(x d - x q )/[2(Ra 2 + XdXq)] (2.17)
Terminal
Busbar
w = (2.18)
o = (2.19)
E' = [E - (x - x') I
d d
- E I
q
] I T I
do
(2.21)
q f d
E"
d
= [E I
d
+ (x v
q
- x")
q
I
q
- E '~ ] IT"
d qo
(2.22)
E"
q = [E'
q
(x v
d
- Xli
d) Id - E"q J IT"00 (2.23)
16
2.4.5 Loads
Constant Power
Constant
Impedance
Nominal V
Voltage
[V] = [I J (2.24)
] I
I
I
Pe I
1- _ _ _ - 'i
Fig. 2.8 Functional Relationships of a T.S. Model (in the Transient State)
i-'
1,0
'.":
20
and [I]
- :::: [L] [D] [U]
-
[Vj (2.27)
In this form [L] = [U]T and hence only one half of the
admittance matrix need be stored.
The modified vector, [I]" f can be obtained by forward
substitution on [Lj and many solutions for [V] can be
obtained without re-triangularisation provided [Yj remaj.ns
unchanged. Re-triangularisation is therefore only necessary
when the system is subjected to a topological change.
2.6 SUMMARY
CHAPTER 3
3.0 INTRODUCTION
Stoke
Islington
Kikaw'a
Tekapo B
Ohau C
Twizel 212 MW
Ohau A
Ohau B
W
264 MW
itaki
Livingston
Half Way
320 MW Bush
tv
South
)l~~========~R~O~X~b~U~r~g~h~==========i===n
Manap0l:;:ur i
Dunedin
~t===================================~~~~~Tiwai300 MW
Smelter
Co
~
-B I
m
~
Control a u
Current
33kv/0.76kv
E Ia
r Ia 2
l 1
I
I x
Rectifier i r4 E
L ____ - _ I
'---'1---'----'---.1----
--,
Unit
21
Fig. 3.3 Single Line Connection Diagram for one Potline
at Tiwai Smelter.
= Xs + + (3.1 )
25
= (3.2)
E
rl = E r2 = E
r3 = E
r4 = Er ( 3 .3)
xc = ( 3 .6)
I LlJ!.l:a
r
ELQ
r V
cell
= kX
s
+ Xr (3 .7)
= ( 3.8 )
In terms of DC current:
aE
r
= ~ (cosa - coso) (3.9)
c
312
3X
c
= 1T
aE
r
cosa - - -
1T
(3.10)
(1 /~~2a - j2u 1
= 16
I - 20 -
I
r 1T l 4 (cosa - coso) J Id
(3.11)
I := ak 16 I (3.12)
r 1 'IT d
:= 13 Er I r coscP (3.13)
where cP = e- l/J
= + Vcell (3.15)
MVAb
AC
(3.16)
3/2 3 X I
V = aE coset.
d(pu)DC 7T r(pu)DC 7T c(pu)DC d(pu)DC
(3.18)
3.1.5 Sequential Algorithm Formulation
[VJ i + l := (3.19)
G
o = P /
0
IE r 12 (3.20)
B
o
= - Q0 / IE r 12 (3.21)
s0 = P
0
+ j Qo
(3.22)
Y
0
= G
0
+ j B0
I
r..
= (Y 0 Y) E (3.23)
r.
ln J 1
where Y (3.24)
Calculate Yo and
Include it in (yJ
Begin TS Run
Solve Rectifier to
obtain '9
-
Calcu\Qte I f'inj
Iter -= Iter t I
Update Vector of
Nodal Voltages
( Exit
= (3.25)
"--- --.,.
-- --- - - vd = k
2
ECOSC/,.
rnln
- k3 X Id
c
Slope given by
Regulator Gain (A)
I
d set
= (3.26)
S (3.27)
33
3/2
= (---
n
, 't - Vce 11)/(R + 3X c In)
aE r cosa l lml
d
(3.28)
where aI'lml' t = a mln
. or a
max
= + V
cell + (3.29)
Id = f (t) + C (3.30)
i. e. Id (t) = AE
r
(t)cosC/,(t) - B Id (t) + C (3.31)
1 3/2
where A = TRd
-- a
'IT
(3.32)
3X )
B = 1 [ c +
T 'lTRd IJ (3.33)
C = Vcell/TRd (3.34)
(3.38)
I
r
Impedance
Rectifier (C.C.C)
0.5 1.0 E
r
I
F.L
E
r
1 0 ~ 90 0 < u ~ 60
'" a
2 0 ~ a ~ 30 60 0
4 30 0 ~ a ~ 90 0 120 0
(3.39)
aE
r
= l6X (cosa' - coso') (3.40)
c
where a' = ~ - 30 0
(3.41)
0' = 0 + 30 0 (3.42)
v = 3I 6a E cosa I _ 9X c I (3.43)
d TI r TI d
15
-5 a == 0,15
a =60
-15
a =75
-25
Error (%)
[ Z] = [V i ]
= [YJ- l [I. .J (3.45)
lnJ
where 0
0
[I.
lnJ
.J = II
r
and I'
r = 1 + jo (3.46)
0
Dynamic
Load
Element
V
cell
where
-0
[Vo] = [y]-l [I. .] ( 3 . 48)
In]
-0
and [I. .] are the injected currents due to all other
In]
generation and loads in the system.
(3.49 )
3/2 3 (3.51 )
r coso: + -1T c d
Vd -- a E X I = 0
1T
(3.53)
= 0 (3.54)
a - a.
mln
= 0 (3.55)
k E cosa - kb
a r
= 0 (3.56)
where u is in radians.
3X
12
- a E E cosa l
+
c
= 0 (3.59)
7T r r 7T
k I a E
arb
cosa' - k' = 0 (3.60)
a - a .
mln = 0 (3.62)
vd =0
Dynamic
Load
Element
V
cell
where u is in radians.
Er/8 (3.58 )
-
3X
12
- a E E cosa' +
c
= 0 (3.59)
1T r r 'IT
k' a E
arb
cosa' - k I = 0 (3.60)
(3.61)
a - a mln
. = 0 (3.62)
v =0
d
Dynamic
Load
Element
V
cell
Yes
I Convergence
I , No
I Calculate J"ac.obian for
~
I given Mode of Opera-Hon
I ~
I
I
Evaluate Updates Test Id )0
~o
Il ~
Iter Iter t' J ,
for Variables Yes
5hutdown
Update Variobles
I
I
, Rectifier
I
< convergence?) i I
(3.63)
(3.64)
K
m = aE
(3.65)
r
m
Mode 2 cos 60- ex <K m <0.8660
== o (3.67)
Iterat~ons Reconv~rge
I
I
J to Total Iterations
for 1 Second
Case Fault Rectifier Algorithm I Fault : Fault i Auto Study
Position Model On ! Off Reclose
--._"
i
Sequential 5 5 !, 4 324 I
1 Manapouri Basic
,,
I Unified I 2 4 , 3 332
,
Sequential 27 6 I 6 346
j
2 Roxburgh Dynamic i
Unified 3 ,
5 3 364
Sequential Failed - - - I
I
i
3 Manpouri Dynamic !
! I
fault X=0.OO60 Unified 3 3 3 j 302
!
I Sequential 8 Failed at first Time
i
4 Manapouri Dynamic I Step after Fault Application
fault X=O.OO65 I Unified 3 3 3 298
I
I
~
1.0
50
3.6 RESULTS
These are subsystems from Fig. 3.1 with high and low
short circuit ratios at the smelter terminals. The Roxburgh
51
Roxburgh
~ Halfway Bush
Fig.3,14a
South Dunedin
Invercargill
Manapouri
Tiwai
Halfway Bush
Roxburgh
Fig.3.l4b
South Dunedin
N Invercargill
HaniJ.pour i
Tiwai
vpu V
.9
p
.8 pu
2.8
.7 ,... .. -.--..,
,'t--.. . '-. .......... 2.4
.6
Impedance . . . ......
... , ' ....... ' ___ ..... - J
V
2.0
......
5
.4 Dynamic -- -
1.6
1.2
.3
.2
\ .8
.4
.1
,
Manapouri Impedance Fault
First Swing Maximum - Manapouri J Difference
Network , Case Generation (MW) Load (MW) Position ImpedanceB " aS1C "
Dynamlc II between
d
Model Model Model i m~eDance"
an ynamlc
Model
A
2 450 260 Invercargill ! 25.1 25.0
(1)
24.6
(3)
2.0%
I
3 570 340 I Manapouri 75.2 73.1 73.2 2.7%
i (1) ( 4)
4 570 340 i Invercargill 53.4 0 53.0 0 57.4 3.7%
I (1) ( 3)
1 450 260 ; Manapouri ! 66.0 0 60.10 62.6 5 . .1%
, I ( 2) (4 )
i
2 450 260 I! Invercargl"II 1
,I
37 . 20 36.40 3L 7 0 I 14.8
B I
I (1) (4) I
,,
3 570 340 i Invercargill i
I
92.0 0 82.0 0 72.0 0 I 21.7%
(1) ( 3) I
I i!
! 4 I
!
570 i 340 Invercargill Unstable 98.0 0 84.0 0 I -
( 5) (1) (3)
(1) Rectifier Shutdown during fault (4) No Short Circuit at Rectifier Terminals
(2) No Rectifier Shutdown during fault after Fault Application
(3) Short Circuit at Rectifier Terminals (5) Fault Period Increased by One Cycle VI
after Fault Application W
54
with the dynamic rectifier model the power demand during and
immediately after the fault is seen to be substantially
reduced and voltage recovery throughout the system is
improved. Although the governor action was adequately
represented, the mechanical power input for the three cases
is almost constant during the period up to the first swing.
swing curves for case 3B are illustrated in Fig. 3.16.
Swinq A gle (Degrees)
Impedance
120
Simple Rectifier
100
~b-+--- Dynamic Rectifier
80
60
40
20
.1 .2 .3 .4 .5 .6 .7 .8 Time (secs)
100
80
60
40
20
--
3.7 CONCLUSIONS
CHAPTER 4
4.0 INTRODUCTION
I
r
+ I
c = o ( 4. 1)
Fig. 4.1
" e 'O~()-~
Va t El Ll Rl t Vb
Fig. 4.2
Va + E
r
R I
r r - Vb = 0 ( 4 3)
e M/'
Va I E
r
R
r f Vb
Fig. 4.3
C
a
Fig. 4.4
61
l/J l = ( 4. 5)
= (4 . 6)
= C
a
V
a
(4.7)
- t
Kt V + KIS Kt V
l/J I = EI RIll la a VS + ly y
(4.8)
t t
l/J k = -RkI k + Kt V
ka a
+ KkS Vs + KkoVcS (4.9)
Q = -KalIl -K I - K I (4.10)
a ak k ar r
0.50
0.00
-0.50
-i.OO
'lI
-1~00~-71.~20;;-----:2~.4;';0~--;;3~.60~-4::-.:I;;eo:---:6:-;.00::--=-7.+.2O:----:8:-.4O~--8..J.eo--IO-.eo.L--1-2.J.00
TII' CYCt..S
20.00
0.00
-10.00
-2O.00~-~~--:~:--~~--~~_~=-__~:--__~:--~~____L -__- J
0.00 1.20 2.40 3.60 4.80 8.00 7.20 8.40 8.80 10.00 12.00
TIN: CYCLES
RECTIFIER VOlTRGES-1321R
-1.00
RECTIFIER CURRENTS-1321R
20.00
INVERTER VOLTAGES-2313R
0.50
0.00
-0.50
-'.00
-1.50 ~ _ _.b-_ _.l--_ _.l--_ _..I..-_ _..I.-_ _..l-_ _..J-_ _...l-_ _...l-_ _.-I
10.00 11.00 IZ.OO 13.00 14.00 15.00 16.00 17.00 IB.OO 19.00 ZO.OO
TlI' CVCU::S
INVERTER CURRENTS-2313R
-~i~.OO~--II-.00~--'2-.00L---'3-.00~--'4-.00..l..---'5-.LOO--'-6.~OO--'-7.~OO--'-8~.OO~--'-9~.OO--ZO-.-l.oo
T11' cvcu::s
Time
Sequence of
Sample Points
Aperture fa
7:~
i~
Sample Points Fig. 4.8b
INVERTER VOLTAGES-2311
-1.00
-I W,oo!;;:----:-l-=.2O~~Z:::-.::40:--=~3-:.60::--:-4.-=60:----::6-:.00L:--:':'"i'.~2O:--~e::-.40.L:----:-9...l.60--10-.L60- -1Z....J.00
TItE cvcu::s
RECTIFIER VOLTRGES-I00B
1.20
0.60
0.00
-0.60
O.dO
0.00
T
g (t) = 1 for It I <
2" (4.11)
T
g(t) = o for It I >
2"
(4.12)
(4.13)
(4.14)
where F (w)
g
= F(w) Q9 G(w) (4.15)
F(W)
FT for positive a
<I ~Frequency
F (w) G(w)]
at
FT for positive
Frequency
0<1 $
Fig.4.12c t1me
/w
Non periodic component
affectinq (jJ
l
Fig. 4.12 Convolution and the Effect of Spectral
Leakage on the Component at w.
74
-N/2
111111111111111111111111111111111
0 N/2
q '" r ~ ~ ~!~ -60
RECTANGULAR
- 0 clB
-10
-20
-30
-40
-50
-N/2
.,,1 III 11111111111111111
0
lilli, ,
N/2 -"-"-"'-"-"-''-'--'-"'--''-''-''~
-60
TRIANGULAR
- 0 dB
-10
-20
~
-30
-40
-50
-N/2 0 N/2 -60
COSINE SQUARED
- 0 dB
-10
-20
'!l1 1 1 1 ~1 1~1 1 1 1 1
-30
-40
-50
[,'
-N/2 0 -60
N/2
HAMMING
- 0 dB
-10
-20
-30
-40
,," 1111111111111111111111111""
-N/2 0
IYi\\~W#l -50
-60
N/2
G~"lUSSIAN
- 0 dB
-10
-20
~
-30
-40
-50
-N/2 o N/2 . -60
DOLPH-CHEBYSHEV
6 = cosh [1:
N
cosh (lOa)] (4.17)
-1 . TI 2
cos h x :=
2" - tan-l (x/II x ) for Ixl < 1. 0 (4.18 )
= Ln[x + Ix 2 ~
N/2 N/2-1
+ L wn Cn + l + k w 1 C (4.20)
n=3 n=4 -n+ -n
C =C
n -n - complex periodic Fourier components (sub-
script refers to harmonic order) .
(4.21)
7
+ ~ c cos (wIt + ) (4.22)
n=2 n n
where c
l
= 1.0
= 0.2 for n = 2 to 7.
V
rms -/ J
o
T v(t) 2 dt (4.23)
V
rms
= /1- N
E
N n=l
(4.24)
RMS values for discrete data are easily and quickly evaluated
in relation to spectral analysis.
I,()()
0,50
0.25
P
rms = 1
T
JT v(t) i(t) dt (4.26)
o
This is readily evaluated from discret,e signals using:
83
N
P
rms
= lEV i (4.27)
N n=l n n
9.00
4.00
0.00
-4.00
<1.00
2.00
0.00
-2.00
4.5 CONCLUSION
CHAPTER 5
5.0 INTRODUCTION
if required.
X = X ( 5.1)
pU(DC base) pu (AC base)
= ( 5 2)
89
and since V
b = V ( 5 .4 )
DC bAC
I
b = 13 I ( 5 . 5)
DC bAC
3/2
= 1T aId (pu) (5 6)
C/,
1
3/2 3
TI a.E. cosC/,J' - TI Xc.Id (5.7)
J J J
E. I. cos(8. - w.) = 0 ( 5 9)
J J J J
1
a.E. cosC/,. - - X Id - a.E. cos (8. - w.) = 0
J J J 12 Cj J J J J
(5.10)
(5.11)
91
3/2 3
- - (aIEl casal + a E cosa )
7T 2 2 2 - 7T
(XcI + Xc2
+ R ) Id
d = 0 (5.12)
(E.,
J
e.,
J
1jJ., a.).
J JJ=,
1 2 (5.13)
= 0 ( 5. 14)
= 0 (5.15)
(5.16)
or (5.17)
(Kimbark 1971) .
Eliminating Vd from equations 5.16 and 5.17 by subtraction
and substituting
8 = 1T y (5.19)
sp
cosy = o (5.20)
vd = NBG. V (5.21)
Link dBridge
1 a
ZLl!.. ILJIL Ic~c
TI
I
f
,j8 +-
2
E~
(5.22)
where If = E. B
f
and B
f
is the filter
admittance.
- E. L..
J
= 0 (5.23)
J
3/2
and K = --- a
Tr
1 Inverter
I
I
Inverter Current! Rectifier Current
setting I Setting
Margin ~~~~---Current
(5.24)
= 0 (5.25)
CX
l - cx
l = 0 (5.26)
min
Y2 - Y2 = 0 (5.27 )
min
F-- _ _
---- -~.
xcr < x .
c~
CI
Fig. 5.5 Modified Control Characteristic.
TI
< a. < TI
2 1
0 < E (5.28)
r,i
0 < (8 ~
~)r
(8 ~
~) .
1
< TI
" I
100
l-
I DC link Input
I
Dynamic Simulation Input
I
F "'"
I Calculate initial
From I Conditions
TS. I
Program
I
Obtoln Thevenin Equivalent
I of NetVJori<..
I
I ~
I QS.S. Model or Dynamic.
I Dynamic
t 0.55.
Under- Voltage
i No
Yes
Reset D.c Variables
--
(hange Current Yes Calculate new
Setting CUrrent Seiting
'No
Disrontinuity
Nof_
(- Calculate.
Thevenin
new
Impedance
r--
Calculate. new
,
Calculate Mismatches
Yes
Ini"liol Condition,:;
I
Iter- =Iter I
,
t Convergenc.e.
L I
Return
I \
p
dyn
Ed ynLiL
Using S = E . 1* (5.29)
+ (EdynCOS]l)/V] (5.33)
T\[i T!o
-
Tf\IYc I Restart
QSS
Model
I I
TS Time
A .. I ~ /i,
I I I \ I
A
I
/i,
I
A
I
I
A
I
&.
I
A ?: \ t
: I I \ I
A
I
Ii.
I
I
A
I
I
I I I \ I
I I I I
I
I
I
I
I
I
I I
I I : I \:
I
I I
II I \ I I I I I II I \ I I I
II I I I I, I I
I:
I \1 I !
I
I
I
I II
I
II
II
\1 ! !
i TCS
start
TCS
End
I TCS Time
~dormon(e
Translent Stabinty
f-'
o
U1
Fig. 5.9 Coordination Between Programmes.
106
L stage 1 1
!
+
1 Stage i-I
If observatio n of TCS
indicates mis match
with QSS mode 1 in TS,
adjust QSS pe rformance
accordingly. New TS
will then pro vide more
accurate Theve nin
Equivalent se quences
L Stage i I
I
j
I Stage n
End of TCS
I
Process TCS information
to enable its use in TS study
1
TS study
using processed TCS
information yields new
Thevenin Equivalents
Yes
!
Check difference between
consecutive equivalents -
more iterations necessary?
1 No
l STOP
I
(5.34)
vm = V
m
+ Z I
mm m
+ Z I
mn n
- Z I
mp f
(5.35)
f
and similarly for the other link and fault nodes, where
I and I are injected currents due to the link at
m n
terminals m and n and If is the fault current at fault
node po
vp
is obtained from:
= Vp / (Z
pp + Zf) (5.36)
SDlutton of Machines
and Networl~
1
Superimpose injeded Current
due to DC. link obtained
from previous iteratiDn.
[Vp]
Estimate of Thevenin Volluges
at fauH Bus
Superimpose If on Ne.twork
Voltages [VI]
[Vm,nJ
New Estimate of Thevenin
Voltages for D.C. link..
Return to 0
Fig. 5,12 Simultaneous Solution of Fault and DC Link.
112
!
[VI]
!
Solve for fault Current If
Using [VI]
1
Superimpose If on
[VtJ
!
[Vm,n]
1
Solve D.C Link and
Superimpose on [Vm,n]
Fig. 5.13
cb
Starting Process After Discontinuity.
5.3 CONCLUSION
CHAPTER 6
6.0 INTRODUCTION
Marsden
Tarukenga Edgecumbe
Otahuhu
Penrose
Atiamuri
Henderson
280 MW
Hamilton 84 MW
250 MW Ohakuri
Whakamaru
MW Whirinaki
r-------------+-~N
Wairakei 255 MW
Waipapa
'---t---!N
1.------1
155 HW
51
Maraetai
Tokaan Aratiatia
MW
Bunny thorp
Brunswick
New Plymouth
Stratford
600
Haywards
600 MW
N Synchronous Condensers
DC Lin]
Line
I
i I
head l Reactol{ System
I
i Line i
i
I
i
i : , i
i ,
I i
I i
i ".---- ----,
I System The-JijOmin
I I
I I I I. I
i i
i. I
I Inverter I '-.../ i
. J , I
I System Thevenj.n ~ -:;: "=:=' -:: ~ ' ....... ______ J
1...._-----,
Bus 1 Bus 2
5 7 9111 13 HP 5 7 19 11 13 HP
Filters Filters
'7
I--'
I--'
'-l
v (pu
1.1 2
~ TCS start
1 0
.9
.8
3
.7
(1) - QSS
(2) - TCS Fixed System Equivalent
.6
(3) - TCS Time Variant System
Equivalent
.5
.4
.3
.2
Fault On Fault Off
.1
.1 .2 .3 .4 .5 Time (seconds)
\ Q(pu)
4.0 TCS Start 3
3.0
2 &
I
/\/
r j
TCS
---~J-y~~-
2.0 1 End
1.0
-2.0
r
Fault On
- QSS
-3.0 - TCS Fixed System Equivalent
(3) - TCS Time Variant System
-4.0 Equivalent
-5.0
4.5 J/'
4.0 '~(3)
( 1)
3.5
(1) ~ QSS
3.0 (2) ~ TCS Fixed System Equivalent
(3) ~ TCS Time Variant System
2.5 Equivalent
2.0
1.5
1.0
Fault On Fault Off
0.5
4.5 (2 ) ~
..-----
..-,J
4.0
!-0(3)
3.5 7 "'(1)
,...
3.0 (1) ~ QSS
(2) - TCS Fixed System Equivalent
2.5
(3) ~ TCS Time Variant System
2.0 Equivalent
1.0
Fault On Fault Off
0.5
DC CURRENTS-2321
5.00
~. Rectifier
:1.00
-5.00
, -- I'"
Inverter \
-10.00
Fault Fault
-15.00
0.00 1.20
~
2.40
On
3.60 4.60 6.00
~
7.20
Off
0.40 9.60 10.60 IZ.OO
TJt1: cvcu:s
Benmore Haywards
TCS
Fixed Equivalent 14.7 0 - 5.9 0 :
i
TCS
variant Equivalent 16.4 0 - 7.10 I
1
I i
125
0.00
-0.50
-i.OO
0.50
0.00
-1.00
-
"
Power (pu) Voltage (pu) I
i
Rectifier Inverter Rectifier Inverter
--+-
I
! II
I
I
End
QSS 4.63 I-
I
4.42 I
i
i
0.964 :
0.965
I
I
TCS 4.51 ' - 4.28 i 0.955 0.963 I
I
I
I I
I
1.00
0.75
0.50
0.25
0.00
0.00 1.20 2.40 3.60 4.00 6.00 7.20 6.40 9.60 10.00 12.00
Tlr cvcu:s
1.00
0.75
0.50
0.25
0.00
0.00 1.20 2.40 3.60 4.00 11.00 7.20 8.40 9.60 10.00 12.00
T1r CYCU:S
6,00
0.00
-1.50
-3.00
-4.50
INVERTER CURRENTS-2311
40.00
0.00
-40.00
-00.00
0.00~--~I.~W~~2~.40~~3~.OO~--74.~OO~~6~.00~~7~.W=---~B.~40~~9~.OO~--I-O.~OO----12~.00
Tlt'E C'iCLS
8.00
DC
4.00
0.00 AC
-4.00
The fault power can be removed, leaving only the real power
flow to the converter using:
P
dc
= P
ac
- (E )2 /
2
R
fault
(6.1)
Integration Stepl 1 2 I 3 4 5 I 6 7 I 8
1
9 10
I
;
During Fault ! I
I
!
i
Ii ' I !
DC Power by TCS -0.752 0.244 ij O.002 0.043 0.024 0.015 '0.00910.006 !0.005 0.005 i:
(Pu)
I I I
i i j
I I
Integration Step During Fault. I
!
1
I
I
2 3
I
I
I-'
w
w
134
Table 6.4 shows that the results from the first and
second iterations of the two programmes are very close and
little benef is gained by performing the second iteration.
v (pu)
start TCS
1.1
La End TCS
0.9
0.8
0.7
(1) QSS
(2) TCS
0.6
0.5
0.4
0.3
0.2
0.1
Qde(pu)
3.0
Start TCS
~
rII
~
End TCS
(2) I
(1)
(1) QSS
(2) TCS
+--1
2.0
LO
1\
/,-_7
I
I A
0
O.l / V 0.2 0.3 0.4 0:5 Time
t (sees)
-1.0
4.0
+
~---~
L
(1)
End TCS
p
(pu) (tart TCS
5.0 -t-~--"":"'-
(,) ! End T:-_-_F_i_g_.6_.1_6_b_Reetificr
I
4.0 I
I
I I
1'1--7 I
3.0 I I
I
I //::: (2)
I
2.0 I I I (1) .. QSS
I I (2) .. TCS
I
1.0
I
I
o
0.3 0.4 0.5 Time
.. LO ( sees)
.. 2.0
.. 3.0
o t-.--~
~40 '\
~50
\ \
\
\
~60
-70
(1) ~ QSS
(2) - TCS
-80
-90
.1 .2 .3 .4 .5
Time (8)
.o'
0.00 ,'
, I
p
rms ,,
", 1
-<1.00
II
on
," ' ~F~Ul t off
-a~oo~-~:::--~~-"""",:-~_---1~_-I,.".--!..!.;...1..!-L_ _--'--_ _..L-_.....-l_ _.J
0.00 1.20 2.40 3.00 4.00 6.00 7.20 6.40 9.00 10.00 12.00
TI t1: CYCLES
RMS
1.00
- -
0.75
~~ Fundamental
0.50
0.25
0.00
0.00 I.W 2.40 3.60 4.60 6.00 7.20 9.40 9.60 10.00 12.00
TIf'E. CYCLES
RECTIFIER CURRENTS-1301
10.00
5.00
0.00
-5.00
2.00 4.00 6.00 0.00 10.00 12.00 14.00 16.00 18.00 20.00
Tlr-E CYCLES
9.00
AC
-9.00 =--::-::::-----:--:-::---:-=::---:--:---__
0.00 2.00 4.00 6.00 6.00
- -12.00
10.00
---- - - -16.00
14.00
- - -19.00
---1 20.00
TIME CYClS
p
(pu)
start TCS
5.0
4.0
3.0
(1) QSS - 1st run
(2) QSS - Delayed
2.0
(3) QSS - Matched
(4) TCS
LO
"
0.1 0.2 0.3 0.4 0.5 0.6 Time
( sees)
Fig. 6.22 Power Flow to Rectifier Terminal - Case 1301.
Q(pu)
3.0 (1)
LO I
-LO
1.0
(1) QSS - Matched
0.9 (2) TCS (2)
oSouth 0
North
50 -40
40 -50
.-~.-'
30 -60
I
I
I
I
/
20 I
-70
(1) - QSS)South ~ (4)
(3) I
(2) - TCS)Island
I
(3) - QSS)North I
I
10 -80 (4 ) - TCS Island I
I
I
.... - -"" "" "
.1 .2 ,3 .4 .5 Time (s)
,
i
South Island North Island
I
I
I
I
i
QSS i
24.1 45 I
(1st Run) i
!
:
QSS
(Delayed) 29.1 50.0
QSS
(Matched) 29.0 49.7
6.4 CONCLUSION
CHAPTER 7
7.0 INTRODUCTION
Direction of
Power Flow
_ - - I - . L -_ _ Infinite Bus
Fault
Rectifier
Terminal
p
elec
curve with
increased
after fault
"""""-\-- Prefaul t
P-o curve
P
me~~~~-r~~~-M~~~~----------------~~---
Fig. P-o
02 0403
Curve with Post Fault Id Increase.
The simple power angle curve of Fig. 7.2 demonstrates,
in principle, the effect of a sudden DC power increase. The
figure is a very simplified case of a much more complex
situation and applies only to a generator modelled as a
voltage behind reactance and connected to an infinite bus.
During a 3 phase terminal fault the generator experiences
accelerating energy proportional to area AI. Without any
change in DC current setting, after the fault, the machine
experiences a deceleration energy proportional to areas A2
plusA3 before peak swing angle 03 is reached.
7 1 2 Full Control
--------~--~---------
-0 Curve with
Decreased
p
mech
,, ,
\
\
\
\
2 3
*
Rect.
Mot.
~
3 Motor Rectifier Id Decrease
I or DC Link
sf Rect. Reversal
Mot.
4
C7 I Motor
II Inverter Id Increase
~ Inv. I
Bus 3
I = I
I
,1-
,,
_v_
"
,
I Alternative Link position
Bus 4
System 2
rv I
Fig. 7.4 System for Stability Improvement Investigations.
The data for the DC link and network of Fig. 7.4 was derived
from the N.Z. system and is given in Appendix A7.
! I
i
1
CASE IMMEDIATE POST FAULT SOLUTION
I Machine
I
Model
Id Setting 0
max P
dc (pu)
/:'P
gen Vterm(gen) Vterm (sys)
/:'P
dc
Generator
Simple 100%
System 1 Increase 89.2 4.37 0.55 0.55 0.8
Both Impedance
Detailed Model of Unstable II
DC Link II,
Note: t:,p / /:'P represents the proportion of extra DC power (due to the increase
gen dc I-'
in I ) which is being taken by the generator (i.e. stabilising U1
d .
power) -....J
158
7.3.2 Discussion
0
Degre s
70
60
50
40
30
20
10
.1 .2 .3 ,4 .5 .6 ,7 'T'ime
Fig. 7.5 Swing Curves for Different Machine Models. (s)
159
.9 (1)
.8
.7 ;
-- -- ---~4)
.6 '"
/' --_~3)
--
.5 (1) Id Constant )Sim le Model
(2) Id Increased) P
.4
(3) Id Constant )Detailed
Model
(4) Id Increased)
.3
.2
(s)
P
dc
(pU) 8
(2)
6
4
""- ~l)
3
2
""
..... -- --'
(1) Id Constant) .
~~s:-~:
Slmple Mode (3)
(41
(2) Id Increase d)
1
(3) Id Constant )Detailed Model
(4 )I Increased)
.1 .2 .3 .4 .5 .6 Time(s)
Fig. 7.7 DC Link Powers for Different Machine Models.
Qdc(pu)
4.0
3.0
2.5
2.0 (2)
L5
f--~-
L
~
LO
(1)
.5
.2 .3 .4 .5 .6 Time (sees)
Nominal Prefault
p-6 Curve
Intermediate
P Post Fault
mech--+-----~~~~~~~~------------~--~--~
P~6 Curve
6
1
Detailed
System
100%
Increase
55.4 6.42 0.74 I 0.75 0.84
I-'
0'1
W
,_..
164
P
(pu)
6 (2)
5
"
( 1)
4 (1) & (2)
---
3
2
I?------
()
(1) Id Constant )Mid Point Fault
(2) Id Increased)
.1 .2 .3 .4 .5 .6 Time (sees)
4) Fault resistance.
P
(pu)
Fault at Mid Point
8
9
_L___ .___
7
6
Fault Near Generator
5
2
1
60
------~.
-~
G:~
50
40 Fault Near
30
20
10
<5
max
(degr
100
90
I
80 (1) Fault N8Fr Generator
70
Poi1nt Fault
60 I
50
40
30
20
10
100
90
Fault near
Generator
80
70
60
Cycles of Id
1 2 3 4 5 6 7 8 9
Increase
Fig. 7.14 Maximum Swing Angle Against Period of Id
Increase.
(2) (3)
4
1~ (1)
3
(3) & (4) -- ""~ .....
------~----..
2
1
j------
(4)
.2 .3 .4 .5 .6 .7 Time (secs)
1.6
1.2
.8
50 100 150
% Increase in Id
-.4
-.8 6P
~
6P
dc
Fig. 7.16 Distribution of Increased DC Power for DC
Link at Mid Point.
169
Instant
start
Nominal
setting
'-'-;I--_L Ramped Start
I
I
I
I
I
I
I
I
5 12 15 18 Time (eye es)
1 0 108.40
Const.
2 2 Unstable
3 100% 0 89.4
Increase
4 3 95.4
System 2 r
Fig. 7.18 Simplified South Island System for
Stability Improvement.
15
max
(degre s)
llO
1---'-- 6
100
max
90
80
p
p / 6.P
(pu~ KPdc ( 4) ,I-'->---+-- gen 2
/ 6.Pdc
6 .6
5 .5
......... '-...-..-.. ./
'-"~_ _ _ _(~2) 1 I
4 .4 /'
" "
~ I
3 I
" (3) I
\
\
2 .2 \
\~~----~- 6.P 1
1 .1 \ $len
\ 6. P dc
\
7.5.3 Full NZ
Close to
Constant 37.6 3.34 0.75
Benmore
220 kv 100%
Bus
Increase 32.8 5.55 0.45 0.67
I-'
-...!
li1
176
60
Id Increased
50
40
30
20
10
Time
.2 .4 .6 .8 1.0 1.2 1.4
(sees)
-10
-30
-40
-50
-60
Id Constant Id Increased
-70
Id ' Nominal
Fig.7.21a 2.0
1.5
1.0
I .5
Id Nominal
Fig. 7.21b
2.0
1.5
.5
7 6.2 Results
<5
(degr es)
(1) Impedance Model of DC Link
(2) Link Model with Id Constant
(3) Link Model with I Modulated by Machine
120 Slip d
110
90 (1)
80
70
60
50
40
30
20
10
% Increase in Id
50 100 150 2do 250
~-----------+-----r----~--~~--~
-1 -0.1
-2 -0.2
-3
-0.4
-5 -0.5
-6 -0.6
P
de (pu) 6P~
-70 6P
max
-80
-90
-100
-110
0
max
(degre s)
P
dc
Inverting
time
Rectifying
Fault
t
Period I
I
I Txlansi tion to nominal
operAtion
Fig. 7.24 DC Power Flow with Link Reversal.
Post Fault DC Constant 50% Reduction 6 Cycle 50% Setting 100% Setting
Current Change 6 Cycles Shutdown in Reverse in Reverse
for 6 Cycles for 6 Cycles
I-'
00
w
184
-1.00
:20.00
10.00
0.00
-10.00
13.50 14.50 15.50 18.50 17.50 19.50 19.50 ZO.50 21.50 22.50
TltE crcu:s
INVERTER CURRENTS-1321R
ZO.oo -
12.50 J3.50 14.50 J5.50 16.50 17.50 18.50 19.50 20.50 21.50 22.50
T I tE Cl'CLS
-1.50 !::::--~::-----::-:'::-----:::-::---:-:-~-=...J~--:~:-----L---...L.--.l--_......J
12.50 13.50 14.50 15.50 16.50 17.50 18.50 19.50 20.50 21.50 22.50
T11" cvcu:s
p
(pu)
5.5
\
\
5.0
/ I
4.5
Y
\
(3)
4.0
(1)
/,
Y
\
3.5
J
/,
3.0
Q \ I
!,/.1
start
L5 (2) QSS - 4 cycle ramped start
TCS (3) TCS
1 I
f
,/
.5
.2 .3 .5 Time (sees)
t
l
Fault On
t
Fault Off
Case QSS -
Instant
QSS - QSS -
4 Cycle
TCS
3 Cycle
s.tart S.t.ar.t s.tar.t
P
dc(pu
Fig. 7.27a Rectifier Power
8
QSS ~ TC = 0.03
7 First TCS
~~__S~econd TCS
6 ,
'\
/ ....
/ \
5 \\,j/ l
\ i\ (1) (3)
;'
4
'\ ~~-
/[-~~7~==
3 ;;-\
2
\/ 1\1
I " End
1 If TCS
\I
'v
r1
Fault On
r2
Fault Off
.3 .4 .5 Time (secs)
P
dc Fig. 7. 27b Inverter Power
(pu)
5.5 (1) QSS - TC = 0.03
",\
I \
I I ,
I
(2) First TCS
5.0 I
I I I (3) Second TCS
4.5
,I
I
I
\
. -t\
\ I
,
\
I
,
4.0 ,,
I
I
, \
J '\\
\
I (1)
3.5 I ~ I
: /, I
, I (3)
i
II rr--::::--------./
3.0 I \ '
2.5
: /'\ I
I, I \ ----- ""'--,
~
V I
\,
I
,\
\ .
2.0
I~ \ \
I
I
,II
,
;----
1.5
start
TCS
~~\ (2)
1.0
~ I End TCS
.5
.3 .4 .5 Time (sees)
_~.00~ __ ~b- ____b -____L -____ ~ ____ ~ ____J -____ ~ ____ ~ ____ ~ ____ ~
10.00 i 1.00 12.00 13.00 14.00 15.00 16.00 17.00 18.00 19.00 20.00
T I t1: CYCLS
INVERTER CURRENTS-2313R2
20.00
10.00
0.00
-10.00
~60
~70
-80 \
-90 \,
~100
(1) QSS
~
(2) TCS ~ 1st Run
~
~110
(3) ~ TCS - 2nd Run
(4) - QSS without current
-120 setting increase
-130
-140
~ 150
(3)
""
"
.1 .2 .3 .4 " " ,.5 .6
'\ Time (secs)
5.00
0.00
-5.00
-10.00
-15.00 =--:::-:::-----:~::---~~:---:--'---..L...---I...---I...---L...--..l-----I
12.00 13.20 14.40 15.60 16.00 IS.OO 19.20 20.40 21.60 22.00 24.00
TI tt: CYCLES
a.oo
Inverter
\
a.oo
0.00
-4.00
Rectifier
-8.00
12.00 13.20 14.40 15.60 16.80 18.00 19.20 20.40 21.60 22.80 24.00
T1tt: cvcu:s
Fig.7.30c DC VOLTAGES-2314
4.50
3.00
1.50
0.00
-1.50
-3.00
12.00 13.20 14.40 15.60 16.80 18.00 19.20 20.40 21.00 22.00 24.00
Tltt: CYCLES
'N/ 1r
Fig.7.31a ~--- -----
4.0 (1) First QSS
3.0
(2) Matched QSS
(3) TCS (1) /
1
End TCS
2.0
1.0
start
T1S
~/
I
I ,I1 '
Ii
I
.1 ' 3 I .5
t "<\. I
Time (sees)
-1.0
-2.0
Fault
On
I \,--~
~-"
Fault \ j
Off \ ,.~'-..J
-3.0 '--...-' (3)
5.0
4.0
Fig. 7.31b
(1) First QSS
(2) Matched QSS
(3 ) TCS
r-
(7-- lEnd ::"'.r- - ---~
TCS
-
3.0 I ,
I
I
2.0 (1)~1 [ (2)
start TCS I '
1.0 I !
I
~ I
I
I'~t~- ~I i
.1 .2 Time (sees)
-1. a
t
Falllt On
-2.0 Fault"
Off \,/'~ (3)
Fig. 7.31 Comparison of QSS and TCS Results ~ DC Link
Reversal for Stability Improvement.
194
7.9 CONCLUSION
CHAPTER 8
CONCLUSION
REFERENCES
Engineering)
ZOLENKOPF, K. "Bi-Factorisation - Basic Computational
Algorithm and Programming Techniques", Conference
on Large Sets of Sparse Linear Equations, Oxford,
1970 u pp.75-96.
205
APPENDIX Al
2
= -
7f
(AI)
where
8 : : : wt
and
i 1
: : : I [cosa' + 2 13
coso' - ~ cos(e+600)] (A2)
where
0' - 150 0( e < a' + 30
where
a'+300<e<0'-90
i = I[ !2 cosa I - '2
1 coSu~u (A4 )
where
0' - 90"8 <a'+ 90
i = I[cos(e-90o) - ~ cosa l
- ~ coso'] (AS)
where
1
aU + 90 <0 <0 - 30
i = I[/3 cos(6-600) - coso - ~ casal] (A6)
206
where
oI _ 30 0 ~ 8 < a U + 150
where
and where
2Id
I ::::
(cosa I - COSO I)
Using identities
j8 -j8
2cos8 :::: e + e (AS)
ja8 1 -ja8
Je- = j - e
a
(A9 )
TIlL (cosa l
- coso') :::: ~ j (/-30- /30) (j-2a I - /-20 I)
3
+ /-2a I - /-20 I - j "2 u . (AIO)
or
APPENDIX A2
(A2.1)
f (x) = 0 (A2. 2)
f (x. + 6x.)
1. 1.
= 0 (A2 . 3)
(A2 .4)
n
If the error 6x. is small then terms in (6x.) ,
1. 1.
where n > 1 and integer, can be neglected. Thus:
(A2.5)
or 6x.
1.
= - f(x.)
1.
_1_
f(x )
(A2 .6)
i
= x.1. + 6x.
1.
(A2 .7)
f (x . )
1.
= - J 6x.
1.
(A2.8)
= (A2 9)
by:
(A2.10)
APPENDIX A3
(A3 .1)
_00
(A3 .2)
_00
jnw t
00 l
fT (t) = r F(nW 1 ) e (A3 . 3 )
n=-oo
T
-jnw 1 t
F (nw )
l =
1
T C
-2"
fT (t) e (A3 .4)
21T
where WI = T
00
n cos (nwl t + n)
fT (t) = L: C (A3 .5)
n=o
T
J~
1
a
0
= T fT (t) dt (A3 .6)
-'2
210
T
a
n = 2
T J!
~2
fT (t) cos nUll t dt (A3. 7)
T
b
n = if2
J!
~2
fT (t) sin nUll t dt (A3. 8)
C (A3 .9)
n
-1
= - tan (b /a )
n n
(A3.10)
m
= r cn cos(nUllt + $ n ) (A3.11)
n=o
a
0
= 1:T r f(ql1T) I1T (A3 .12)
q
a
n = T~ r f(ql1T) cos [nUl (qI1T)] I1T (A3.13)
q
CONVOLUTION
++ (A3 .17)
(A3.18)
(t) ++ ( w) (A3.19)
(A3.20)
APPENDIX A4
T
2
= if2 coswi t coswlt dt (ALl)
fT
(w i - W ) T
2 1 1
= sin
2
:::: 314.16
Wi
1
= 310.28
Hence = 0.9935
APPENDIX AS
APPENDIX A6
OOOO~OOOOOOO~~~OO~OOOO~O
000000000000000000000000
(110
OOOOOOOOOOOOQOOOOOOOOOOO
0908 0
~ . .0
I-
U
00
x.ll. OOOOOOOOO~OOOOOOOOOOOOJO W M:..Il
<l:..U:l.. OGGOOOOOOGO.OOOOOOOOOGOO a:
X- O~~~~OOO~OOOOOO~~aOQOOOO Xf\J:7"
<tl
-.'I-'"
1.:>-"
~oo~oaoooo~oOOOOOOOooooo
o~~ooooooooooooo~oo~o~~~
o~~~ooooooo~oo~oooooooo~
eooooooOOOOOOOOOOOOO~OOO
OOOOOOOOOOOO~OOOOOOO~OOO
OO~OOOOOOOOOOO~OOOOOOOOO
>-
~
~
Woo
00
..
ON ....
1-1 IL- 1:1 1:1 ?-n~~ftD~~~~D~~n~~.~n D~ ~
'"
II II 1:1 I) 0 1:1 1:1 1:1 1:1 1:1 1:1 " 1:1 I) I) II I) It I) III
<to I ~~~~O~~~~OjO?~O~O~OOO~~O 1- ~~~~~~.Il~~~~~~-~~~~~~'-~~ ~
us I ::: 0080 OOOOOOGOOOOoooo.OOGO 0 III
I ,-~
12-
c~~~OOOOOO~O?Oooooo~a~~o
~~~~~m~N~~o~~~MMN~N~NMM
"-::: ......"\J:l"~~=>'O.Jla-Nf""')..(Y"')Jlf"""."""l31"O-:O:\jOfV')
-M ON
.z "':::>
uJi I- 6~~~~
.... .... ....
oo~..oo N~ W (f)-\,()oO
-.U.I'
2:1 I uJx
I -<I.
1..1'11
IL~
..ot'\.JMN.$lJ)LJ,I11,.....I.fICC"OON..tIJ)(\Jf\J_,...._~f"'llll
o~r"oooooo
000.0.9
~ooo~ooooooooocooooooooo
____ OOOo_OOf""lOO.,04
... 60000 0000
:::>
<.>
a:
o
,
U
WV'lLn!..n
a.;t. 0
UJONf\J
Ynn
""
::1
III
:::>.
1
'4
Ixm
-IK
eD4
~-
o~oooooo~oooooo~oooo~~~o
D~n~-~n~~~o~-~~~~~~~-N~~
~NMN~m~~~~~~ON~~NN_~~Mm
OO~OO~OOO____ OOQO_OOMOO~ -
<II
I!)
z
o
W
)I
0
~ 000000000000000000000000
000000000000000000000000
::
~
w
. z-
.....
J
01 :::>:l'
o~~
e III G (, G G .. It It a It . . . a It " 0 9 lit . . . III ..,
OOOOOOOOOO~OOOOOOOOO~OOO
o U
, 000000000000000000000000 a
218 oooooo~OOOOOOOOooooooooo
01 :l. ...J XJ< oooooooroooooooOOOOOOOOOM W
- ~ oooooooo~OOOOOOOO~o~OOOO ::: iii 0 0 > 0 0 > "Q0 0 0 0 0
1) 0 0 0 > 1).
'" II
I ~- ~~~----OOOOO~~OMN~~~OO~~ oooo..;tm~~ooooooooooooooo_
WI .J)"::> -< :J'o y.. 9)'1)_.0 D .Ilj'oO.,;f-.J)l',...lI.o::o.::t ""'V-nr-.n N~ O_~~DD~~~ft -JlO~OD~~D~ p
JI:'
>-1
uJO
J":! I
N-:'"l-..$4"..:t.:1.::tIJ)3':7'..;tFlN..:tN..of'1MUl"Jl..:tXl N~ ..:fN-- - - N_~_N_ -- 4
Z
Ill'
II I> 'Ii> <11. 0 0 " 9 0 .... 0 " a 0 (II 0 (10'" 0 0 0
ZO:J 00000000000000000000-000 0:
"'10
~L~
uJ-
oooooooo~ooooooooooooooo
~~./l--~~Ooo~O~.::tO~N~~~oon~ Uo
...
W
...J
~Z~ ~~M~-~~~~O..:t-~..:f~ro~ro~~ro~~
~~~----------~-~- ~
'0
W~. N-~- ~~~.nN.~~Mn~~.~ :i:Z
n::lf' 0 ... 0 0 0 0 0 0 1\1 0 0 0 0' 0 0' .0 0 0 lit 0
-~o~~ ______ -o __________ ~
C>
o oooooooooooooooooooo~oo Z
: ~"oooooooooooooooooooooooo
,
I~~
- ~~~..:tNN~~oom~NmMnmN~ONmm
~~~~~:7'~ro~~~~Nft~~~~~O..;t_~N
--N----------N------~_~
'INI II. IIIIIINIIIII" III ~
;:)
12<
'WI
oO~OOOOO-NN_OOOO_O~~_ON ~:~~11:4~:!~~~~I~lf:I~~~ u
.0 00(11.00 08004100 ~W~~'II~.~~~OO~ZX~~~4W<4 ,
a-o OOOOOOOOOOQOOOOOOOOOOOOO oa~o~wu~o~~~~a~w>a~z~~z1 1
I Jl XOZ~II.~~4-~J4W~J~~4~4~~
.~~ O~OOOOOOOOOOOOOOOOOOOOOO w~~~~~~~~~~~-J~~~~~~Q~OC~ IIl00
1.- ~~~~.NN~~OOft.~ft~D~N"'ONDn ~ZJZ<~4~~~4-~~XXk~~~_~4 uJoe:>
--
1.:1:)(. :y.""'r'"'IMO'O"C1'.r0'C7'..:rf')N~..oIJ""ccmo4_:.nN >w~4XXXOW4n~d4~4wI~04<XZ U'H!) 0
I~~ ~OMO~OOO~-~N_~OOO~O~~~ON ~D~~OOOX~~4~I~~XZOO~~B~E wo_
I I 000&00000.0111(110000000.000
-
..
1-1->
::nll",
~
",
OOO~~OOOoooo~oOOOOOO~OOO
~~No---oo~~oooooomoo~~oo
Mom4~NN~~_~~m~~o~~NN~~
a::r
w:o
III
ro~o~~..;t~~~~OO~N~~OO~~~~~
.aJ:n
~:> ...
-Ul<:
0000000000,,0000.0.0(1100(110
Dn~.D---ftn~~~n_~,.nD4D_~
..
....""''''
~':::ll
......
- N - --I\J .....,
t#llC""fJ;)
oOOOOOOOO~~OOOOOOOOOOooO I-
~~O~Oj~~~~~~~~O~o~~~o~o~
000000(/100"0001110000900090 :Ii:
O~OOOOO~JOOOOOOOOOOOOOOO
~~?~~~jJ~~~J~~~~~~~~~~~~ J
~~~~~--~~~-~-~~~~~~~~~~~
uJ 00
J 00
t!h< " 0
Z <{ClL:!J
:;[0-0"
_n03 _ _ _ _ _ _ _ _ ~ _ _ _ _ _ ~ _ _ ~
--N----------N----------
I I N I I ' 1111111"'111111 BIll
~::;~: I: ~b::!~.~}'!:'=: Ill~~
~~~, ~'~nO~~z1_~~~~~c
::J~~a.~u~~~.~I-~I-~~nLZ~l<:ZX
~O~~I I .~~4-~~~wOJ~J~44-~
uJ:T ..... .., -:'t ... ..., O<H-!=-"4_VI <I.Il.)...'tflt:<Q:(\;1 .lC.~
-.::_, .': I. "",\2-d.!-' )-.:k:" Xw.("ot".l'l("-_ _ <tt
~dn~lL~~~~~I-~~ ~L~O.cLE
~~~X~v~x~$~q~~XXZO~B~8~
218
SHUNT LOAuS'
APPENDIX A7
M/C PARAMETERS
BU~BAR
N Mt:.
MACHINE
j>.jU MODEL.
UAI A. INERt.
MV CUNS.
1-------------
TRANSIENT
REACTAN~ES IP.U.) -------------) ARM. (----- OIC Tes I~EC~)
SYNCHROhOUS SU~-TRANSIENT POT- RESIST TRANS1~NT SU -T ANSIEN
~AtN.
A T.
----t RlR.
TYPE
\jASE. KWSKVA D-AXIS Q-AXIS D-AXIS Q~AX!S D-AXIS Q-AXIS IER (P.U.) D-AXIS Y-AXIS D-AXIS Q-AXIS
g~N~ORE---~o J. ~ J,uO.uo 2~.000 0.0150 a.ol50 0.01~0 0.0150 0.0100 0.0100 0.000 o.ooo~ 6.00 v.luO 0. 00 0.1000 0.000 SAL
S EM---2 8 J,
HAY"'ARDS-22 J.
:;
!
Aoo8a99~.900 0.0 00 0.0 00 8"8200 o.o~oo 0.oo~8 0.005u u.ooo 8.08 6.00 U. uo
Ava. 099~.900 0.0100 0.0100 100 0.0100 0.010 0.010U o.OUO .00 0 99.99 9.9~9 000 .000800 8.0o o
.0000 0 8" 1 8" g SAL
M/C LOADII~(jS
DC LINK P~RAMETE.RS
BU$ NAME ANGLE LIMITS SERIES VOLTAGE----CU~RtNT---- LINE RESIS REACTANCE (PU) TAP
MAX MI>~ BRWGES BASE{KVI MA~GIN(AI (QHI"S) COM TRAN fiLTER %
BENMORE---DC 90.00 3.1,10 1.00 420.00 0.00 2~.560 0.02~43 0.02243 0.99200 0.000 RECTIfIER
HAYWARUS--DC 90.00 Itl.1.I0 1.00 420.00 12U.00 2!:>.56U 0.01905 0.01905 I.U32!:>0 0.000 INVERTER
RESTART T.C.= 0.0600
10Y.00~AGE INCREASE IN CURRENT SETTING ON AT 0.240 SECS AND uff AT 0.360 S~CS
I\)
I\)
o
*~**~***o***~********~*~****************~**********************************~********************************
* NUTE - S"Y NON-~~SENTIAL PARAMETERS MISSING fROM THE INPUT DA1A HAV~ ~EEN ASSIGNED APPRuPRIATE VALUES *
IN THE CAL~ULATIO~ PRUC~SS, AS SHUWN ABevt. A VALUE IND1CATEC AS A SERIES Of 9~ IS ACIU,LLY VE.RY LAKGE.
****~*******il********~**************************w**** *****{r*********************************************** ***
221
APPENDIX AS
M/C PARAMETERS
tlUSBAR
NAf'A1:.
MACHINE
ilJU MOOEL
UATA
MVA
INERT. (------------- REACTAN~E~ IP.U.) -------------)
CUNST. TRANSIENT SYNCbRONOUS SU~-TRANSIENT POT-
ARM.
RESIST
(----- OI.C rcs <SEeS)
TRANSIENT
----I
SATN.
SUB-TRANSIENT fACT.
RTR.
TYPE
bASE KwSKVA D-AxIS O-AXIS D-AXIS Q-AXIS a-AXIS Q-AXI~ lER (P.U.) a-AXIS Q-AXlS D-AXIS Q-AXIS
BENMORE---lo .!. 5 !OO.OO 20.000 O.OlSO 0.0150 0.07S0 0.0750 0.0100 O.OlOv 0.000 0.0003 6.00 o.IUO 0.1000 0.1000 0.000 SAL
SYSTEM---220 A S !VO.0099~.900 0.0100 0.0100 0.0200 0.0200 o.OO~O 0.0050 0.000 0.0003 6.00 0.100 0.1000 0.1000 0.000 SAL
HAYWARD!>-220 A 1 !OO.00~9~.900 0.0100 0.0100 0.0100 0.0100 0.0100 0.0100 0.000 0.0000 99.99 ~.9~~ 0.0000 0.0000 0.000
M/C LUAaII~GS
BUS BAR M/(; MIC PUWER OUTPUT DAMPING
NAME NO. Mill MVAR fACTOR
BENt'ORE---16 1 -1000.00000 238.77000 0.00
SYSTEM---220 1 529.9':1000 30.8~000 0.00
HAYWARDS-220 1 ~OO.OOOOO 12':1.0JOOO 0.00
DC LINK PARAMETERS
BUS NAME ANGLE LIMITS SERIES
------------------
VOLTAGE CUh(Rt:.NT LINE RESIS REACTANCE(PU) TAP
MAX MIN BRIDGES BASEIKV) MAh(G!NIA) IOHt'S) COM TRAN flLTER %
HAYWARDS--DC 9().OO 3.i!O 1.00 420.00 0.00 25.560 0.01905 0.01905 1.03250 0.000 RECTIFIER
BENMORI:.---DC 90.00 11:l.1J0 1.00 420.00 120.00 2S.56U 0.02243 0.02243 0.':19200 0.000 INVEf<TER
APPENDIX A9
I'l/C PARAMETERS
eUSBAR
NAMI:.
MACHINE
NU MODEL
UATA
MVA INERT. (------------- REACTAN~ES (P.U.) -------------)
CUNST. TRANSIENT SYNChROI~OUS SUd-TRANSIENT POT-
~RM.
R SIST
<-----
O/C TCS <SECS)
TRANSIENT
----I
SU~-THANSIENT
?ATNe
ACT.
RTRE
TYP
IjASE D-AXIS Q-AXIS D-AXIS Q-AXIS O-AXlS Q-AXIS IER
K~~KVA iP.U.> D-AXIS Y-AXIS D-AXIS Q-AXIS
I:3EN~ORE---16 ! 5 100.00 20.000 0.0150 0.0150 0.0150 0.0750 0.0100 0.0100 0.000 0.0003 6.00 0.100 0.1000 0.1000 0.000 SAL
5'1'5EM---220 .1. 5 !UO.00~9~.900 0.0010 0.0010 0.0020 0.0020 0.0010 0.0010 0.000 0.0003 6.00 O.lUO 0.1000 0.1000 0.000 SAL
HA),WARDS-220 ! 1 100.0099~.900 0.0100 0.0100 O.OlUO 0.0100 0.0100 0.0100 0.000 0.0000 99.99 9.9~9 0.0000 0.0000 0.000
MIC LOADINGS
I:lUSBAR M/C MIC puwtR OUTPUT DAMPING
NAME NO. MW MVAR f~CTOR
DC LINK PARAMETERS
NAME
IdU$ ANGLE LIMITS SERIES
------------------
VOL TAGE CU~RENT LINE REsIS REACTANCE(PU) TAP
MAX MII'l BRIDGES BASEiKV) MAh<GIN(A) (OHr-Sl COM TRAN fILTER %
HAYWARDS--DC 90.00 3.00 1.00 420.00 U.OO 2~.56U U.01905 0.01905 1.0.J250 0.000 REcTIfIER
BENMURE---DC 90.00 H!.UO 1.00 420.00 120.00 25.560 0.02243 0.02243 o.'::Inoo 0.000 INvERTER
RESTART T.C.= 0.0400 IV
DC LINK ~EVERSAL WITH CUHR~NT SETTING Of 413.60 REVERSED AT 0.24 SECS AND ReSTORED AT 0.3b SECS IV
ol:::.
*******************~*t******************************** ****************************************************** *
~UTE - AI~Y NON-tSSENTIAL PARAMETERS MISSING fkOM THE INPUT DATA hAVE ~EEN ASSIGNED APPRUPRIATE VALUES *
1~ THt CA~~ULATIO~ PROCESS. AS SHOWN ABOVE. A VALUE INDICATEC AS A StkIES OF ~s IS ACTUALLt VERY LAHbE.
****~****************f***************************************i~***********************************************
225
APPENDIX AI0
C. P. Arnold, Member I.E.E.E. K. S. Turner, Student /lember LE.E.E. J. Arrillaga, Non Member
University of Canterbury New Zealand Electricity University of Canterbury
Christchurch Christchurch Christchurch
New Zealand New Zealand New Zealand
-~
the dynamics of the d.c. load,is used to investigate the
stability of a pOl:ler-system where the rectifier can rep-
resent a large percentage of load.
Kikiwil
INTRODUCTION
Is} ington
~-;:OUlh
reactive pm1er absorption. This allows overall system
stability to be evaluated and provides information on
the operating characteristics of the load during large DUI1'~d.in
disturbances.
This paper describes various rectifier load model~
discusses their effect on transient-stability and pro-
poses a unified algorithm of superior convergence pro-
perties, than the conventional sequential algorithm.
1 r
x
ally'examine the effects which rectifiers have on the
transient-stability of the system, this time constant
must be taken into account. This requires a more complex
model to account for extended overlap angles due to a
10>1 commutating voltage.
Modes of Operation
Fig. 2. simple Rectifier Load Equivalent. The full range of rectifier operation can be clas-
sified into four modes [4J.
converto~ transformer or the switching devices, Mode 1 - Normal operation.
rectangular current waveforms occur~at the rec~ Mode 2 - Enforced delay. This is imposed on the firing
tifier terminals, angle as the forward voltage across the incom-
there is no harmonic distortion. ing thyristor is negative until the previous
commutation is ocomplete or until the firing
With an assumed a.c. terminal voltage (E ). and angle exceeds 30 .
operating under constant current control, a simpl~ rect- Mode 3 - Periods of three-phase and d.c. short-circuit
ifier can be solved directly without iteration. In the which exist when 4 valves are conducting. (Two
case of aluminium smelters, effects such as cell counter commutations overlap) .
voltage (V 11) can be included in the d.c. load. The Node 4 - Continuous three-phase and d. c. short-circuit,
following g~ations may be used to describe the action caused by two commutations taking place contin-
of the rectifier [2J[3J. uously. The a.c. and d.c. current paths are
aE independent.
r
l2Xc (cos i< - cos 6)
I = (1) Table'l gives a brief summary of the four modes.
d
(9)
(~ - !.:::li -j 2u)
k (16)
/...,. (cosa - coso5)
Q P tan ~ (10) This expression is only valid for mode 1 and 2 operation.
2
228
Overlap (Deg.)
0.5 1.0 E
20 40 60 80 100 120 r
25
Fig. 4b
15
=30
5
--... ~- 0.=45
~
- 5 =0,15 a=60
~15
a=75 1.0
-25
0.5 E
r
Error (%)
o " - ".
m1n = 0 (32)
o
and equation (27) is replaced by an algebraic form of
circuit sho~m in figure
The equivalent can now the differential equation for ld' obtained using the
be applied to find the rectifier current (I,,) by using trapezoidal rule (see Appendix B):
the Newton-Raphson technique.
The effect of the rectifier on the rest of the Id - Ka Ercosa - kb = 0 (33)
system can be determined by superposition:
Equations for Abnormal Operation
(24)
In mode 3, account has to be taken of the effect
where of the periods of three-phase short-circuit on the a.c.
voltage at the rectifier terminals. These periods reduce
(25) the fundamental voltage at the rectifier to zero when
0
the overlap reaches 120 . In addition, the relationship
and [I~ .J are the injected currents due to all other between I and I , given by equation (17), can have a
d
generat~6n and loads in the system. significafit effect on the fundamental of a.c. current.
If the network remains constant, vector [Z'J is also To account for this, a linearised form of the curve for
0
constant and thus only needs re-evaluation on the occur- ,,=30 in figure 3 is included explicitly in the formul-
rence of a discontinuity such as line switching, fault ation:
application or removal, or motor s\"i tching.
Thus the advantages of the unified and sequential viS - Ej! -Z/~ kId (f (uL.l
l
= 0 (34)
methods are combined. That is, good convergence for a
difficult element in the system is achieved while the ~aErCOS(e-~) (f l (u) )-k6aErCOs(a-300)+~Xcld 0
programming for the rest of the systems remains simple
and storage requirements are kept low. (35)
Id - ka ErCOs(a-300)-~ = 0 (36)
Equations for Normal Operation
X I
o 0 k4 c d
cos(,,+u+30 )-cos(a-30 )+~ = 0 (37)
r
a - exm>n
. o
" (38)
Programme Aspects
Vd IdRd .. Vcell (30) Evaluation of other variables can be made using the
values of V, S, E and ld" The initial conditions so
k~ obtaine~ give co~vergence in 2 to 5 iterations for the
I Id (31)
r 11 first over~ll 1teration of the whole system and only 1
4
230
or 2 iterations on successive system iterations. Rcconv- CCIolPARISON OF SEQUENTIAL AND UNIFIED ALGORITHMS
ergence of the' a. c. system is unaltered by the rectifier.
Choice of operating mode after a discontinuity is The netvlork shown in figure 1 was used as a test
just as important as good initial conditions. The system to examine the convergence properties of the seq-
incorrect choice can lead to an attempt to solve the uential and unified algorithms. The disturbance applied
wrong set of equations for the prevailing conditions. to the network \1aS a three-phase line fault located at
This can lead to failure of the algorithm and study ter- various points in the system. Line svlitching and fault
mination. Mode choice is made using a factor k [4J clearance occurred 140 mS later, this being the normal
where: m
primary protection fault clearing time for the system.
For longer studies auto-reclosure of lines took place
12X I 360 mS after line opening. In all studies the rectifier
c d
k (40) steady-state load was constant at 288 H\"I and 100MVAR and
m aE
r generators were modelled to include transient and sub-
tr~nsient effects and automatic voltage regulators and
The range of k for each mode is as follows:- speed governors were included. All other loads ,-"ere of
m
the impedance type.
~lode km<cos(60-a) The most difficult computational problem occurred
Mode cos(60-a)<k <0.8660 when re-converging after a discontinuity. Once recoverg-
m
Node 0.8660 <k <1.1547 ence had been achieved and the first integration step
Mode m
1.1547 <k following this had been successfully Solved, the solu-
m
tion was normally numerically well behaved.
Immediately after a discontinuity, the choice of mode is Table 2 presents a comparison of the convergence
influenced by the estimate made for E. If the value of properties of the two algorithms. In cases 1, 2 and 4,
k indicates mode 4, iterations are m~de in mode 3, to the number of iterations required to reconverge after
cWeck that the overlap angle will in fact reach 1200 three different discontinuities during each study was
This test is repeated at the beginning of each time step used as the basis for comparison. Cases I, 3 & 4 were
so that reversion to mode 3 can be determined correctly. more severe faults than, case when observed from the
This can be justified as, in fact, mode 4 is merely rectifier terminals.
a boundary condition of mode 3. In case 1, the unified algorithm showed a slight
A particular advantage of the Newton-Raphson form- advantage when reconverging following a discontinuity.
ulation is the flexibility it affords to the setting of However the number of iterations required during the in-
control specifications. The normal specifications are tegration procedure was almost identical. The slight
either constant current or constant delay angle. During increase in the time-step iterations using the unified
the transition through mode 2,the overlap angle is fixed method was due to a tighter tolerance being set for con-
0
at 60 and the enforced delay angle is calculated. This vergence of the rectifier variables of the Newton-
can be.easily accomplished by replacing equation (29) Raphson process in all the studies.
by: Considering case 2, it can be seen that the rate
of reconvergence after a discontinuity using the sequen-
-1 _ hXCId] _ (t _ U
tial method was slow. This was so, even though the rec-
cos
aE sp
o (41) tifier model operated in mode 1 throughout the study,
r with a maximum overlap angle of 35. However once re-
convergence had been achieved then the number of time-
where u = 600 step iterations compared favourably with the unified
sp
method.
Equation (33), the algebraic form of the defferen- The unified algorithm was clearly superior in case
tial equation for I , also becomes a control specifica- 3 as divergence occurred when the sequential method was
d
tion when reconverging after a discontinuity. This used. Case 4 demonstrated that convergence problems al-
equation reduces to: so occur using the sequential method during the time-step
iterations, but these difficult cases do not cause prob-
lems to the unified method. The unified method consist-
ently gave reconvergence after a discontinuity within 5
for zero step length, holding Id at its previous value. iterations and the number of time-step iterations was
Fault
Iterations to Reconverge Total Iterations
Rectifier
Case Algorithm For 1 second
position Hodel Fault Fault Auto
Study
On Off Reclose
Sequential 5 5 4 324
1 Manapouri Basic
Unified 2 4 3 332
Sequential 27 6 6 346
2 Roxburgh Dynamic
Unified 3 5 3 364
Hanapouri
Sequential Failed - - -
3 fault X=0.0060 Dynamic
Unified 3 3 3 302
Failed at first Time
Sequential 8
Step after Fault I\pplication
Hanapouri
4
fault X=0.0065 Dynamic
Unified 3 3 3 298
5
231
also small provided that the step length 'laS suitable for A comparison between cases Bl and B2 illustrates
the system as a whole. the importance of accurate rectifier r~presentation. As
the fault gets closer, the rectifier is forced into inc-
EFFECTS OF RECTIFIER LOADS ON TRANSIENT-STABILITY reasingly non-linear operating modes.
Fig. 7 illustrates the differences in power and
In order that the effects of rectifier loads on voltage behaviour for case B3 under the three alternative
transient-stability could be examined the two systems of rectifier models. ~Iith the dynamic rectifier model the
figure 6 were used. These are sub-systems of figure It power demand during and immediately after the fault is
with high and low short-circuit ratios. The disturbance seen to be substantially reduced and the voltage recovery
to the network was a three-phase line fault applied at throughout the system is improved. Although the governor
either l1anapouri or Invercargill. This permitted speci- action \<las adequately represented, the mechanical power
fic rectifier effects to be observed. The studies \olere input for the three cases is almost constant during the
continued only long enough to record first swing stab- period up to the first swing.
ility and the results are presented in table 3. In case B4, fault clearance was delayed by one
Roxburgh cycle and the impedance representation led to instabil-
ity, while under the Basic and Dynamic rectifier models
marginal stability was observed.
Fig. 6a vpu
.9
.8 P
pu
.7 2.8
south
I'
Dunedin .6 2.4
.5 2.0
.4 1.6
.3 1.2
.2 }p .8
.1 .4
6
232
60
ACKNOWLEDGEMENTS
LIST OF SYMBOLS
0.1 0.2 0.3
Time (secs) Id - direct current
Fault Period - direct voltage
Vd
a - firing angle
Fig. 8. Rectifier OVerlap Angle vs. time for case B3. u - overlap angle
- a+u
Xc - commutation reactance
Rd - d.c. resistance
Run Time MW/MVAR P.F. - d.c. residual voltage at no load
~cell - power factor at rectifier terminals
Pre Fault 3.0 0.95
Ir or 1ft ~ fundamental line current at rectifier
terminals
During Fault 0.0-1. 5 0.0 -0.83
Er or E,L! - fundamental a.c. phase to phase voltage
at rectifier transformer terminals
Early Post Fault 6.5-4.5 0.98-0.97 ~ constant current controller gain
A
P - rectifier real power
At Swing Maximum 3.9 0.97 - rectifier reactive power
Q
I .. - complex nodal injected current
Table 4. Ratio of HW/MVAR of Rectifier for Case B3. ylnJ - complex nodal shunt admittance
S - complex power
CONCLUSIONS 'Z - driving point or transfer impedance
'Zth or Zth/~ ~ system equivalent Thevenin impedance as
Three different methods of representing a ,ectif- seen from the rectifier
ier load have been investigated. It has been demonstra- a rectifier transformer tap
ted that a detailed rectifier model, which includes the V or v/B system equivalent 1.'hevenin vol tage as
dynamic effect of the d.c. loads, and abnormal modes of seen from the rectifier
operation, is necessary to correctly assess the influence - complex conjugate
of the rectifier in transient-stability studies.
7
233
1 13 (B4)
i = I [cosa' t 2cos6' 2" cos(8+60oJ
) (A2)
where
6'-l500<8<a'+300
where
APPENDIX All
TECHNICAL NOTE
Prof. J. Arrillaga, M.Sc.Tech., Ph.D., C.Eng., F.I.E.E., B.J. Harker and K.S. Turner
Indexing terms: Power cOllversion, Load alld voltage regulation
Abstract: The note presents a fonnulation for a.c.-d.c. load flow that is far simpler in usc than former.
methods described in the literature. Voltage ambiguity which is present in the authors' and other earlier
papers, lS femoved.
total set of nine convertor variables,i.e.E,.p, ",,lp,ls,o:,a, R (I) Vd - K.aV cos '"
Vd,!d' 3
In theory, two independent variables are sufficient to R(2) Vd - KI a V cos 0: + -/ d Xc
11
model a d .c. convertor, operating under balanced con
ditions, from a known terminal voltage source. However,
the control requirements of h.v.d.c. convertors are such
that a range of variables, or functions of them (e.g. constant R( 4) '" control equation
power), are the specified conditions. If the minimum
number of variables are used then the control specifications R(S) = control equation
must be translated into equations in terms of these two
variables. These eq uations will often contain complex and
nonlinearities, and present difficulties in their derivation
and program implementation. In addition, the expressions
used for the active and reactive power at the convertor
nodes may be rather complex and this will make the V can either be a specified quantity or an a.c. system
programming of a unified solution more difficult. variable. The equations for Pdc and Qdc may now be written
For these reasons a nonminimal set of variables is as
recommended, i.e. all variables which are influenced by
control action are retained in the model. This'is in contrast ~c = VIp sin '"
to a.c. load flows where, because of the restricted nature of (12)
control specifications, the minimum set is normally used. = VK.aI d sin '"
Variables Ip ' Is, E and .p can be eliminated as they play
no part in defming control specifications. VIp cos '"
Thus eqns. 3,4,7 and 8 can be combined into
(13)
(10
or Pdc = Vd ld (14)
3"';2
wherek 1 = - - 3 References
11
ARRILLAGA, J., and BODGER, P.: 'Integration of h.v.d.c.
links with fasl-decoupled loadilow solutions', Proc. lEE, 1977,
The final two independent equations required are derived 124, (5), pp.463-468
from the specified control mode. 2 ARRILLAGA, J .. and BODGER, 1'.: 'A.e.-D.e. load 110ll's with
realistic representation of the convertor plant', ibid., 1978,
:2 D.C. model summary 125, (I), ppAI-46
3 ELMARSAFAWY. M.M. and MATHUR, R.M.: 'A new fast
The d.c. model may thus be summarised as follows: technique for load-noll' solutions of integrated mulli.erminal
d.c./a.c. systems', TrailS. IEEE, 1979, Winter meeting, Paper
F79 174-4
R(x, V)/l = 0 (J I) 4 ARRILLAGA. J., BARKER. llJ .. and BODGER. P.: 'Unified
and sequential load flo\\'s for a.c. systems containing static
where convertqr,', PSCC, 1978, {uarmstadt), pp.719729
APPENDIX A12
l\bstract - In the presence of h.v.d.c. links the source and use detailed transient simulation for the
convent.ional quasi ~ steady state simulation of power areas of changing topology, i.e. the convertors. Ho\v-
system disturbances is not justifiable. This paper ever the use of a fixed a.c. system Thevcnin equivalent
d.escribes a realistic solution \<:hich combines the effi- at the convertor terminals, even for a brief disturhan-
ciency of single-phase multimachine transient stilbility ce, will provide wlrealistic levels of current and vol-
analysis and the accuracy of detailed transient conver- tage.
tor simulation. Versatility and moderate computing Some compromise is therefore required between the
requirements establish the proposed solution as a via- accurate, but computationally prohibitive, transient
ble alternative to the use of analogue simulators. convertor simulation an1 the computationally efficient,
but less detailed, transient stability study. II com-
promise which exploits the advantages of both these
methods is achieved by simultaneously perfonning trans-
Tradi tionally the analys is of a.c. po.,er system ient stability and transient convertor sirr.ulations
disturbances involves the use of two separate program- .,ith periodic coordination of the results. In this .,ay
mes. IIquasi-steady state, linearized, fault study is the detailed convertor model is provided with a time-
carried out to obtain information of the fault current variant Thevenin equivalent for correct a.c. system
nnd voltage levels at a discrete instant. II dynamic representation [2J. Similarly, from the transient
study is also carried out to assess the stability of convertor simulation, accurate power frequency informa-
the system follo.,in9 the disturbance. tion can be derived to represent the d.c. system be-
lIo.,ever the highly non-linear behaviour of h.v.d.c. haviour at the convertor terminals. The computational
transmission systems can only be predicted wi th reason~ requirements are minimized by restricting the use of
able confidence under clearly defined steady state op- the transient convertor simulation to the most dis-
erating conditions. During disturbances there is no turbed part of the study. as the d.c. link regains
guarantee that normal convertor operation will remain normal controllability and attains predictable behav-
within the region of applicability of the steady state iour, this detailed simulation is replaced by a quasi-
equations, or that the convertors will retain controll- steady state (gsS) model.
ability. Therefore in systems containing relatively
large h.v.d.c. transmission schemes and with disturban- COMPUTATION OF CONVERTOR TRANSIENTS
CeS at or close by the convertor terminals, only a ful-
ly detailed elemental transient analysis can provide Simulation of the transient behaviour of an a.c.
accurate simulation of the convertor behaviour [lJ. -d.c. interconnection requires the simultaneous solution
The dynamic behaviour of the generators can be ac- of a set of differential equations Hhich vary with the
curately predicted by means of fundamental frequency convertor's topology. Any a.c. system components of
studies. This form of investigation is commonly kno>ln immediate influence on the convertor plant ~ust be exp-
as Transient Stability (TS) , and is primarily concerned liei tly represented in detail, e.9. convertor transfor-
with evaluating the ability of the system to remain in mer, a.c. filters, synchronous machines directly conn-
synchronism. The differential equations describing ected to the convertor terminals, etc.
the system can be solved with sufficient accuracy using Convertor transformers are represented by the
integration steps in the order of 10 to 50 mS. coupled circuit model. In purely a.c. systems the rec-
. ~I the other hand the continuously vary~ng topo- overy voltage is often strongly influenced by the tr~n
logy of convertor plant, .,ith commutation discontinuit- sformer magnetizing current, particularly when switch-
ies occurring every few milliseconds, requires an elem- ing is involved. Convertor saturation is also a feature
ental representation of the system components, and a of a.c.-d.c o interconnection~ following disturbances,
much shorter integration step. This type of study, and if switching of theconv~or transformers is inv-
defined here as Transient Convertor Simulation (TCS) , olved, the coupled circuit model permits representat-
provides very detailed results in the form of 3-phase ion of the non-linear magnetizing currents.
waveshapes, but requires integration step lengths in Machines are represented by their E~W behind ~n
the order of 50-100 ~s. appropriate reactance based on a Thevenin equivalent,
There is no doubt that a transient analysis of the or if necessary by their full transient equations.
power system including a detailed representation of the The combined effects of other a.c. system compon-
generators' behaviour and following the topological ents may be represented by an equivalent Thevenin circ-
changes of the d.c. link can provide realistic informa- uit which reflects the generators' and system's behav-
tion of the disturbance. However the demands in com- iour, and offers a realistic impedance to the lO~'er
puter storage and time are prohibitive. power-frequency related harmonics. The model for Thev-
An alternative to full system representation is to enin equivalent circuits permits variation of the reac-
reduce the a.c. system to an equivalent Thevenin tance (or EHF) , starting .,ith the subtransient and
changing into the transient as the study progresses.
The development of a time-variant Thevenin source is
described later (Interactive Coordination of Transient
Equivalents).
The time-variant Thevenin equivalent of the a.c.
$ystem (or part of it) must be accurate for all harmon-
ics of the power frequency, in order to correctly simu-
late possible resonances between the a.c. system and
the convertor equipment. By rearranging the Thevenin
impedance into a series-parallel configuration, an equ-
ivalent which maintains a constant impedance angle for
the major pO'ler frequency harmonics, and provides real-
239
: ' - - - -, - -- - - --1
:
I E
From TS mod~l :
Z I L Input
~
Data and ESlablish Equations I
I eTllt;v TIIEV I
I ~ I
I I J
\..----'----- _____ J L Input Fault Data I
~
I Monitor and Control convertor Operation: 1
any s\.litching?
AoC.
Harmonic Yes +
Filters 12 pulse No 1 Modify ~quations, 1
convertor unit produce output
!
Fig. 1. Representation of each convertor and associated
a.c. system. I Time variant Inductances? Le. Ll=f(t) e.g'l
Saturation or Synchronous Machine r.lodels
~ Yes
istic dampingharmonic voltages, is obtained [3J.
of
This representation assumes there are no aec. network
No
~ Calculate new values for
relevant blocks in L '
I
resonances, yet gives an accurate treatment to possible Recalculate L]-l 1
resonances between the a.c. network and the harmonic
filters at each convertor terminal. The representation t
for each convertor and its associated a.c. system is
I Is Ly time variant I
depicted in Figure 1, where Z represents the series t Yes
parallel configuration. THEV
lNO
l Form Ly I
To determine in each case the appropriate equiva- .j.
lent circuit, the areas, of specific interest to the [Any time variant NeblOrk Equivalents ,
particular disturbance must
fault studies the major areas of interest are
be
No
I .j. Yes
Update Equivalents using
Data produced by Transient
5 tabi l i ty Study
I
doc. system being simulated.
For efficient computation it is convenient to seg-
regate the system nodes according to the type of bran- I Solve Network Equations for One step
+ I
ches interconnected by them. This segregation defines
a nodes (which have at least one capacitive connection), i
I Advance time by h I
a nodes (which have at least one resistive connection
but no capacitance), y nodes (with only inductive comp- .j.
onents) and 6 nodes (convertor nodes, involving only I Time for Output? I
inductive components, i.e. convertor transformer and
~ Yes
smoothing reactor). Thus the' topological changes
caused by valve switchings can be implemented by alter-
ing only the matrices involving convertor (k) branches
No I Produce output and
write output to mass
I
storage medium
and 6 nodes.
A partioned state variable formulation is then
developed which involves the following set of equations
..
I Time for Disturbance? J
in terms of the state variables ~t' ~k and Q , i.e.
a
(1) No I ,Yes
Modify equations according
to fault data. Input any
further fault data
I
(2)
.j.
(3) I
End of Study? I
No
Yes t
(4)
Store all variables
and I
(5)
(6)
I state information on mass
storage medium
+-
I Exit I
(7)
Hl.
240
i-.:xITnC~ t l on
detected? J ~nary
In the first place the three-phase real and imag-
terms of the branch impedances of the steady
state equations describing the system are transferred
~ Yes to the topological matrices R and L respectively.
No t t
Interpolate all state Conducting valves in each brldge are determined
variables back to extin- from a knowledge of the phase ' a' vol tage phase angle,
ction instant. Calculate measured with respect to the steady-state load flow re-
dependent variables ference, and taking into consideration the inherent
phase-shift in the bridge transformer (i.e. 0 or 30)
J and the delay angle.
Calculate valve voltages Having established the pair of conducting valves
per bridge, the currents in the transformer windings
~ must be ~stablished on both sides. In the case of a
For each bridge: delta-connected convertor transformer initial currents
check for cross-overs, must be established in the three delta windings, even
and activate control though only two valves are conducting.
system in relevant A steady state analysis such as a load flow, which
bridge i f they occur is based on the assumption of balanced sinusoidal a.c.
waveforms, cannot provide exact initial conditions for
i the transient analysis of a system which includes con-
for each bridge: i f vertor models, due to the distorted wave conditions in-
firing due issue grid herent in convertor operation. It is therefore neces-
pulses and advance ring sary to perform a preliminary transient convertor simu-
counter lation (equations (1) to (10, using the load flow
I conditions as initial values, in order to obtain the
I any switching operation above? desired steady state operating point dynamically.
To facilitate the application of a particular
No yesl fault, the fault branch is previously established in
Produce output, the branch list. This branch impedance may then be
switched in/out by modification of an element in the
modify Lo' Kok ' appropriate connection matrix. In some configurations
set next
output time this switching may contravene topology constraints, in
which case the numerical value of the branch impedance
firing
I
next
is changed at the time for fault application from a
If time to high pre-fault value (so that the pre-fault branch cur-
less than integration. rent is effectively zero) to the value required to pro-
step, reduce integration vide the appropriate faulted network.
step
A.C.-D.C. QUASI-STEADY STATE ANALYSIS
I
Zla.l Xc1L!lel d2
V '1
vd1
8
1
1,~'
Fig. 4. A.C. and D.C. system for unified algorithm.
241
I) Equating voltages at the Thevenin source, studies. It is essential that, during the study period,
the d.c. link solution begins with good estimates for
the d.c. variables. In contrast, when the Newton-Raph-
son algorithm is used in a load flow study [SJ it is
only necessary to make one estimate for the initial
conditions.
For Transient stability, ne", initial conditions
The fundamental a.c. current and the d.c. current are calculated by a separate estimating process at the
are related (in per unit) [5J by the expression beginning of a study and immediately after a discontin-
uity (i.e. fault application or switching operation).
312 The calculation of initial conditions is based on the
Ie) ~ -w- a Id (l2)
j new Thevenin voltages obtained for the convertors at
H) Equating powers across the convertor, the beginning of each step. Again referring to Figure
4 the following initial estimates are made for the var-
iables involved.
Vd~ Id - 13 Vj Icj Cos(6 j - nij ) = 0 (13)
I'll
242
'tIl
243
I Return to A I tors.
Figures 9 (a) and (b) indicate the effect of the
fault on the Thevenin source voltage at the two conver-
The invertor source exhibits a considerable
decay in voltage during the fault period >!hich modifies
Fig. 7. Simultaneous solution at fault and d.c. link.
- +
Tfc Tfc Tfc Restart
in obtaining the correct solution for the d.c. link.
1
\!/
The solutions of both link and fault were thus combined QSS
fI'Iodcl
in the algorithm of Figure 7.
The algorithm requires a good estimate of the d.c.
injected currents at the first iteration of each step
for the speed of convergence of the Transient Stability
III
TS ti~e
, 'II'
f: \ t
programme to be maintained. The d.c. link solution
I'
I
i
accurate except for the first iteration after fault I \1 I
REFERENCES
Fig. 9(E)
312 INVERTOR f\C VOLTAX5
[1J Arrillaga, J., Al-Khashali, H.J. and Campos Barros
J.G.. uGeneral Formulation for Dynamic studies in
Power Systems including static convertors II, ~
eedings lEE, Vol. 124, No. II, November 1977, pp.
1047-1052.
[2J Heffernan, H.D., Arrillaga, J., Turner, K.S. and
Arnold, C.P. "Recovery from Temporary h.v.d.c.
"1.$
.. 4.0 5.0 6,0 7.0 6.0 9.0 10.0 11.0 ll.o
Line Faults". Paper 80 SM 675-9 presented at the
IEEE PES Summer fleeting, Minneapolis, July 13-18,
1900.
[3J Bowles, J.P. "A.C. System and Transformer Repres-
Fig. 9. Results. for a 3-phase fault at the invertor entation for HVDe Transmission Studies", Trans.
terminal..
245
.-'''1''
246
APPENDIX Al3
Abstract - The aperiodic and distorted waveforms from sinusoidal components of voltage and current of
produced by, h.v.d.c. convertors following system dist- the fundamental frequency. This fWldamental frequency
urbances need to be procBssed in order to derive appro~ reactive power can only be derived from spectral analy-
priate transient equivalents compatible \-lith transient sis of the voltage and current waveforms and since both
stability studies. This paper describes various tech- these variables can be used on their own, reactive pow-
niques for the derivation of these equivalents and con- er becomes redundant. The above problems do not exist
cludes that, while rms values provide realistic app- in the definition of real power and further justificat-
roximations for power, spectral analysis is required ion for its use as a variable is given later in the
for the determination of voltage magnitudes. paper.
The choice of the second variable is between volt-
INTRODUCTION age and current magnitude. Since the convertor acts as
a current harmonic source, the current waveform, part-
A comprehensive programme has been described in a icularly during and immediately after a disturbance,
companion paper (Part I) which combines two separate exhibits greater distortion and modulation effects than
and distinct methods of transient analysis, i.e. Trans- the voltage waveform. This has been tested in many TCS
ient Stability (TS) and Transient Convertor Simulation studies, two examples of which are illustrated in Fig-
(TCS) ures 1 and 2. The spectral analysis of the vol~age
Transient Stability is normally a fundamental fre- waveform is, therefore, less prone to error and the
quency single-phase rms study and is therefore based voltage magnitude is a better choice of variable.
on the assumptions of balanced and sinusoidal waveforms. The specification of voltage and real power is
~n the presence of h.v.d.c. links, such assumptions are required at each integration step of the TS study when
extended to the voltage and current wave forms at the the Quasi-Steady State d.c. link model is replaced by
convertor terminals. On the other hand, Transient Con- the results of a TCS involving the d.c. link.
vertor Simulation is waveform orientated and exhibits To obtain fundamental components of vol tage and
the distortions and non-linearities associated with power, Fourier processing of the TCS waveforms is req-
convertor operationo uired and this must be performed over at least one fun-
The interactive coordination of the two parts of damental cycle of the waveforms~component of voltage
the programme requires periodic exchanges of informat- and power can be obtained at any sample point using a
ion between them in order to achieve a realistic simu- complete cycle (period T) spanning T/2 both sides of
lation of the behaviour of the a.c. system and the con- the sample point as illustrated in Figure 3(a). The
vertors. Transferring data from the Transient Conver- sampling points must be synchronised with the TS study
tor Simulation to the Transient Stability study is not
straightforward, however, as during disturbances the RECTIFIER VOLTRGES-132IR
convertors produce aperiodic and very distorted voltage .. or---~--~---r--~--~r---~--~--~--~---,
<>20.0U .$ U.5 14.$ lS'.s 16,5 n.s 18.5 19.5 ~o.s u.s n.s
Cycles
I
I
i
I
Sample Point
~~.Ol'~.'.,-.,,~.-.1~'.~1~'~.O.-'1~'~.-.,T,..-176.O~1~7~.~~10~.'-~~1'~.'-~~O.O for TS
Cycles
Sequence of
t t s,amPle...points
I I
I I :
i1~i2ii ~II
I I I
Aperture for :~,
Sample Points I
Figures 6(a) and 6(b) illustrate single cycles of Fig. 6. Single cycles of one phase obtained from TCS
one phase of the invertor and rectifier voltage wave- waveforms of Figures 4 and 5.
forms taken from distorted periods of the results pres-
ented in Figures 4 and 5. These individual cycles show T
that considerable levels of non periodic frequencies get) 1 for JtJ < 2 (1)
exist in the >!aveforms due to the nonlinearities assoc-
iated with control action, filter response and convert-
or behaviour. If these waveforms are assumed to be get) =0 for It I > ~ (2)
periodic and are subjected to Fourier transformation to
obtain the spectral components of the 'signals (in par- Figure 7(a) sho>!s the gate function and its spec-
ticular the fundamental component), a phenomenon known trum, G(w). In the frequency domain this multiplication
as spectral leakage [2J occurs. Spectral leakage res- is equivalent to the convolution of F(w) >lith G(w), i.e.
ults in the non periodic noise contributing to each of
the periodic spectral components present, introducing fT(t) _ - - _ , F(w) (3)
uncertainty in their identification by the Fourier
transfotmation. fT(t) 9(t) -<---"" Fg(W) (4)
The phenomenon of spectral leakage is best expl-
ained with the use of the convolution theorem [1J. where F (wI F(w) 0 G(w) (5)
In order to make the Fourier Transform of a 9
periodic signal fT(t) finite, it is necessary to multi- and < denotes a Fourier Transform pair.
ply the periodic signal, in the time domain, by a unit
qate function, g(t), >!here, When f (t) contains only the fundamental frequency,
WI' as sho$n in Figure 7(b), the convolution of F(w)
250
Fig. ?(b)
F(w)
F'T for Positive a
Frequency
---1I---'--"'"w
Fig. 7(c)
F(w) G(w)
F'T'for Positive
~ Frequency
Dolph Chebyshev Windo>l
2.5~~
~~~~-r-4~~~~~~~
w
non periodic
component
,,,'IIIIIIIII!IIIIIII,,,
I111111I a =
I\CKNONLEDGEHENTS
The real power evaluated in this way is the sum of fun- TS - Transient Stability
damental and harmonic real powers. The results from a TCS - Transient Convertor Simulation
TCS have shown that, in general, the harmonic real pow- FFT - Fast Fourier Transform
er is small and that the rms approximation for real db - Decibels
fundamental power is a good one. Figures LOla) and W - Angular Frequency
10 (b) shml blO representative cases. In the case of T - Period of fundamental frequency
Figure 10(a). although the maximum error is 4%, the cum- k - Sample index
mulativc error over the simulation period is small. vet) ~ Instantaneous voltage
The maximum error occurs for a short period when highly i(t) - Instantaneous current
distorted signals exist. set) - Instantaneous apparent power
The harmonic power flow can only result from 1n-
phase components of harmonic currents and voltages. APPENDIX
Considering the relatively low resistive component of
the convertor transformer and the large X!R ratio of The Fourier transform pair of a periodic function
the a.c. system, the in-phase components of the harm- fT(t) of period T can be written as:
onic currents and voltages are very small. Therefore
the use of rms power is a good approximation and spec-
t
tral analysis is not needed to derive the power vari- fT(t) L f(nw ) ejnwl (Al)
able. ne:-OG
1
CONCLUSIONS T
:2
!
An efficient technique has been developed for pro-
cessing the aperiodic and distorted waveforms obtained
F(nw )
1
~
T f-2 T
fT(t) e-jnWl tdt (1\2)
'.I 11
253
(w'-w IT
T . 1 1
sw
J-2!
2 2
T
For a fundamental frequency of 50 Hz (i.e. ~20 ms)
c _2 :2
b
n ~.
f-2
T
fT(t) sin (nw l t) dt lAG) WI c 314.16
-2
254
APPENDIX A14
!I ,
255
J\bstract - Quasi-steady state and transient conv- olved in first swing transient stability (TS). its acc-
ertor simulation models of h.v.d.c. transmission are urate determination requires a correspondingly accura te
compared in multimachine transient stability studies. representation of the d.c. link behaviour during and
The paper analyses their respeclive aoc. and d.c. syst- immediately after the disturbance.
em responses to typical disturbances, discusses the The Use of a small step transient simulation
limitations and extent of applicability of the quasi- throughout the stability study has been proposed in a
steady state model and justifies the need for temporary recent publication [5J. flowever the computational
transient convertor simulation. It is concluded that requirements appear to be prohibitive and the size of
after preliminary transient convertor simulation stud- a. c. system representation restricted. It is t.hus a!:Jp-
ies in a particular system, a suitably modified quasi- arent that a QSS model must be used for the d.c. link
steady state model can be used in the assessment of response if it can be justified.
transient stability. It can not be established in advance whether a QSS
model will provide realistic results. Only be attempt-
INTRODUCTION ing several worst case detailed convertor simulations
can the validity of a QSS model be assessed.
It is generally accepted that some form of Transi- This paper investigates the effect of different
ent Convertor Simulation (TCS) , as discussed in a comp- convertor models on the power behaviour of a d.c. link
anion paper [lJ, is essential when assessing fault cur- during disturbances and its relevance to the assessment
rent levels, d.c. link response, recovery voltages [2J, of transient stability. The models are then used to
etc. On the other hand, in stability studies the investigate a worst case study, with a temporary d.c.
h.v.d.c. Hnk is normally rE'presented by a Quasi-Steady current setting change, to explore the possibility of
State (QSS) behaviour [3J[4J. This is justified by the first swing stability improvement.
fast controllability of the convertor plant as compared
with the a.c. generator response. COMPARATIVE BEHAVIOUR OF THE QSS AND TCS HODELS
However consi.dering the relatively small time inv-
The differences between the two types of convertor
South Island a.c. system
r - - - - - ..... North
Faul~lt Bus Island a.c. system
llV lIE:
i
I
I
I
I
:
I
I
;
I
I I : P"----~-- ...
I : System "
r r 1 J1
I I : Thevenin r
I I II 12 I I
:. @:-c:? I
'
IV :
System Thevenin'/ ________ JI
'---------"'"
S 7 5 7 11 13 H .P.
Filters Filters
V(pu)
start 1'CS End 1'CS
1.1
~
1.0
.9
.6
1 - QSS
;7 2 TCS
.6
~l.:l.'--'lc".:-,--',:'.:-,--;1.,;,---,;-,.';;",-76.~~-',-.';"2--;;.~.':--"~.6'-1;;;.;:;.'-'':2.. 0.1 0.2 0.3 0.4
",elLS
Q(pu)
4.0
"'].0
"l,S
... 1.'
1, .. J.'
I.1.1
B.' 10.0
CTCLU
12.0
0.1
t
Fault ON
0.2
I
Fault OFF
0.3 0.4 0.5
Time (s)
2.
257
Start TCS
P(pu) J End TCS
~~-~---
Fig. Sea) ReCTIFIER Cl~[NTS-I301
5.o-1----. ., . r-'::"'::~-':""~~---'--~--'----~-~--~--'
4.5
4.0
'}/ 10,0
3.5
3.0
2.5
'X~
~ f
"':_1
1 - QSS
2 - TCS
2.0L-----~O~.b-----~O+.2~----~0~.3-------0~.4-------
Time (s)
Fig. 4 (a) Rectifier pO>ler.
Fig. SIb) S-AC s-oc RECT-1301
P(pu) start TCS 12,0
5.0 ~ ...
J'<.~\1'I
4.5
4.0
I\lli "f:
3.5 ... "!(Jo~I~
3.0 .. 4.0
2.5
2.0
Fault ON Fault OFF ...
.. s.o
> 4 ... ... 10.0 U.O 14.0 16.0 18.0
CYClES
JO.O
1."'---7,--~;;---'-;;'f-;----;;-+;----+-
0.1 0.2 0.3 0.4 Fig. 5. D.C. line fault - TCS rectifier response.
Time (s)
Fig.4(b) Invertor power. Arc extinction and deionization times are greater for a
d.c. fault than for an equivalent a.c. line fault and
throughout this period no power can be transmitted.
:Fig". 4. Rectifier a.c. system fault - real pO>ler. The reactive power responses at the rectifier are
illustrated in Figure 6(a). The TCS response indicates
served, the nett energy transfer as calculated by the that at fault initiation the reactive power increases
TCS and QSS models are similar, thus causing relatively briefly, due to the half cycle of fault current. Once
small first swing maximum angle differences. For the convertor action clears the fault ~urrnt, the filters'
high inertia machines at the rectifier terminals the total reactive power is fed to the a.c. network until
difference is 3%, while for the 1m/ inertia synchronous d.c. line reenergization starts.
condensers at the invertor, the difference is 12\. At The real power responses are illustrated in Figure
both terminals the QSS model gives the larger maximum 6(b). There is little difference between the results
angles, and thus provides pessimistic but safe results. of the QSS and TCS models at fault initiation.
The TCS results from the fault period indicate
D.C. Line Fault that extinction and deionization time require 11 cycles
[7J. At the end of this period transmission recovery
A d.c. line fault is applied on the d.c. line,next begins and in the absence of prior knowledge of the TCS
to the rectifier smoothing reactor, which results in a response, the QSS model produces instantaneous power
complete loss of power transmission during the fault recovery at this time, as illustrated by curve 1 of
period. Figure 6(b). As the transient convertor simulation of
The QSS model has severe limitations in its ~il the recovery period progresses it becomes clear that,
ity to represent this type of disturbance. A d.c. due to the line recharging delay, a considerable mis-
fault, using this model, is represented by instantan- match exists between the two models. The difference,
eous shutdown of the link at the prescribed time foll- clearly demonstrated by curves 1 and 4 in Figure 6(b),
o>led by restarting at the end of the specified fault causes considerable error in the TS assessment as ill-
period. Moreover the QSS model can not represent the ustrated by the swing curves of Figure 7. The South
timing of valve firings in relation to the fault occur- Island machine shows a 16% error in the 1st swing peak
rence and therefore gives no information regarding of the QSS response.
fault currents or subsequent control action. By using interactive coordination between the mod-
For the TCS the d.c.fault is initiated immediately els, as described in a companion paper [IJ, the QSS
Bfter a valve firing. This provides a worst case study model start-up characteristic can be delayed to corres-
since the control action influence is delayed for al- pond with that of the TCS model. CUrve 2 of Figure
0
most 30 until the next firing. The TCS provides more 6(b) shows a more accurate characteristic for the QSS
detailed information of the d.c. link behaviour, as mode~. with this approach, the differences in first
shown in the >lave forms of Figure 5. swing maximum angles are reduced to less than 2%. This
The fault is applied at cycle 2 and Figure 5(a) is due, in part, to the close correspondence of "the two
shows that the a.c. fault current flows fpr half a models throughout the fault period and also to the fact
cycle after fault initiation. During this period the that the peak s>ling of the machines is very close to
rectifier delay angle is 6etarded into the inverting the transmission recovery time Consequently the di ff-
Q
0
region (between 120 and 130 ), to accelerate the d .c. erences observed over the power recovery period do ~.o.F.. _
line discharge. Figure SIb) shows that the fault is affect the first swing significantly. By ramping, as
cleared rapidly as a result of this control action. well as delaying the restart, as illustrated by curve 3
258
Q(pu)
TCS
2.0
SO -40
1.
40 -so
0.4 0.5 0.6
Time (s)
30 -60
, I
I
, I
20 -70
1 - QSSI South
~ I
,
I
5.0
! ._-----f': -
J ______ - ",
.1 .2 .3 .4 .5
4.0 ,( \/ <:I Time (s)
I 2
Fig. 7. D.C. line fault - swing curves.
3.0 1 - QSS 1st RunL
2 - QSS Delayed
2.0 3 - QSS Matched I 1
I INVERTER CU'lRENTS-23 I I
4 - TeS I
.o .,r----r--~~--r---,----r----r---,---~----~--_,
1.0 I
I "...., 40.0
Invertor a.c. System Fault Fig. O. Invertor a.c. system fault - TeS current (1 )
2
waveforms.
Faults at invertor terminals provide a severe test
for the d.c. link controls and often result in repeated paring the apparent power at the a.c. and d.c. termin-
commutation failures causing considerable disruption to als of the invertor system. The difference between the
the normal valve firing sequences at the invertor. A two represen ts the power flovl ing to the faul t.
three phase fault, of 5 cycle duration, is thus applied Immediately after the fault is applied, the inver-
at the invertor a.c. terminal busbar (Bus 2 of "Figure tor experiences repeated commutation failures vlhich
1). A 1 ohm fault impedance is used, causing an 80% persist during the fault period. The commutation fail-
voltage drop at the invertor terminals but providing ures produce oscillations in the d.c. power at the inv-
sufficient source voltage for commutations to continue. ertor terminal. At the rectifier terminal the oscilla-
For this fault location, the TCS variables E2 and 12, tions are smoothed by the d.c. line, as shown in Figure
used to derive voltage and power equivalents for the TS 9(b). These effects can not be represented by a QSS
study, include the effects of both the fault and the model.
aOcD link. Since the fault is severe, it masks the The use of the QSS model produces responses vlhich
convertor response, as illustrated in Figure B. differ significantly from those obtained by TCS. The
In this case it is not necessary to obtain a sep- results .. from the two models are compared in Figure 10.
arate fault Thevenin equivalent for the TCS since the In order to compare directly the rectifier and
system and fault equivalents are derived from the same invertor responses, the results plotted in figure lOla)
busa Horeover when TCS equivalents are used to rep- include the invertor power without the fault contribut-
resent the convertor behaviour, the variables obtained ion. This is obtained by removing the fault power con-
from the TCS can be used to specify the fault in the tent from the TCS equivalent at each step of the TS
a.c. network of the TS model. Thus it is not necessary study.
to represent the fault explicitly in the admittance ma- The differences between the TCS and QSS results
trix of the TS study. If explicit fault representation are caused by a number of factors:
is required in the TS study, it is necessary to remove a) Due to the d.c. line response, power continues to
the effect of the fault from the TeS variables so that be fed into the rectifier end of the line for one cycle
they can be used to represent the convertor only~ Fig~ after the fault occurs at the invertor bus.
ure 9(a) illustrates the effect of the fault by com- b) At the invertor terminal, immediately before fault
259
P(pu)
End TCS
start TCS
l>.o~:..F:..ig.:;.-._9_(:.,a,..:)_~S_-A:-T_S_-_OC~_J_tN-.--_2_3~11_ _,--_,__.,
5.0
l 1 1
e.o
4.0 ~
,!-r---
.
'.0
3.0 /" 1-2
0.0
2.0 1/
I
1:0
1/
I. 2
- QSS
- TCS
~8:L,.0;---:,-.,.,:--.,,~.C';--J,.t6-~.o'.
e-"t.on'-,.'2,--,~. ,-,,12.0.
-,,-:0.,:-,,",,:0. 0
~~PL------1I~.----~I~
0.3 0.4
cycus 0.1 0.2
Time (s)
-1.0
Fig. 9(b) S-A: s-oc RECT-2311
".o.--:..--~--,~--.--~----..,--~--'---, Fig. lOla) Invertor power
start TCS
)~
P(pu) .
I
5.0 +-----,C-.
4.0
I ,
3.0
2.0
//-2
I I 1 - QSS
1.0
\
o~________o~~*o~i=~~~--------~'--------r~1'"
0.3 0.4
2 - TCS
P(pu)
Start TCS End TCS
'-'---"
:(\l~(\ ~~ -
5.5
-40
\ 5.0
\../ f i I = 1st Run
-so
\ 4.5
4.0 ,i " '\l
3 .... TCS - 2nd Run
\ 3.5 ;~
il\.~
l' i 1
-60 3.0
\I
!J 11\ r'--:~::;--L::---
2.0 ,I \ ;-----
~J
-70
1 - QSS 1.5 I 2
2 - TCS 1.0
-80 IJ
.5
.1 .2 .3 .4, .5 .6
-90 , , I , I &m- Time (s)
.1 .2 .3 .4 .5 Real power at the invertor
Time (s) Fig. 13 (a)
P(pu)
V
I
.
'~",",
.1
2'
tV
t .1
Fault ON OFF
.4 .5
Time (s)
.6
... ACKNOt~LEDGEHENTS
"M.O REFERENCES
,.,L. --~~-:c:-:-~17J.-:o--:7:--:I-:".-:O--':-:6~.O~-,:-:,L.O---1O-.LO---,,~.-o---'.1().o
=-.
<YCLtS
Heffernan, M.D., furner, K.S., Arrillaga, J. and
Arnold, C.P. flComputation of a.c.-d.c. system
disturbances - Part I: Interactive coordination of
Fig. 14. Current setting control - TCS current wave-
generator and convertor transient models ll (Compan-
forms.
ion paper).
Arrillaga, J" Heffernan, N.D., Lake, C.B. and
Arnold, C.P. "Fault studies in a.c. systems int-
erconnected by h.v.d.c. links ll , Proe. lEE, Part C,
.- .......... vol. 127, No.1, Jan. 1980, pp. 14-19
-60
''\ [3J Arril1aga, J. and El Amin, I.M. "Transient stab-
ility performance of a 3-machine system including
-70
-80 \ [4J
an h.v.d.c. link", Proe. lEE, vol. 123, No. 11,
Nov. 1976.
Brameller, A' I Yacamini, R' I El Amin, I.M. and
-90 Lynch, C.A. "Transient stability of AC-HVDC
-100 systems using a direct solution", Paper A79074-6,
PES l'linter ~Iecting, N, Y., Feb. 1979.
-1l0 [5J Vovos, N.. A. and Galanos, G.D. "Transient stabil-
ity of a.c.-d.c. systems", IEEE Transactions, vol.
-120 1 _ QSS '\.. 1, /
PAS-98, No.4, July/Aug. 1979, pp. 1376-1383.
-130 2 - 'reS - 1st Run \~. .7' / [6J 'furner, K.S., Heffernan, M.D., Arnold, C.P. and
-140
-150
3 -. 'reS - 2nd Run ~- - - . - 4
4 - QSs '.lith:,ut current ", ~ _____
sett1ng 1ncrease
\
,
;>. / Arri1laga, J. "Computation of a.c.-d.c. system
disturbances - Part II: Derivation of power frequ-
ency variables from convertor transient response"
-160 2/'- , 3 [7J
(Companion paper),
Heffernan, H.D., Arrillaga, J., Turner, K.S. and
-17~0t-__________- 4______~____~__~'~______~_ " Arnold, C.P. IIRecovery from temporary H.V.D.C.
.1 .2 .3 .4 ',.5 .6 line faultsll, Paper 80sr-t 675-9, PES Summer Heetjno,
, Time (s) Ninneapolis, July 1980.
[8J Uhlmann, E. "A.C. network stabilization by D.C.
Fig. 15. current setting control - swing curves. links",~, paper 32-01, Paris, 1970.