,
\1
A SEMIANALYTICAL METHOD FOR STEADYSTATE
'SLUTION IN HVDC ANALYSIS
i:
.',
,.
,~
\ 1
l
!j
by
Nickie Menemenlis,
B.~ng.
(McGill University)
A thesis submitted to the Faculty of Graduate
Studies and Resear~4 in partial fulfillment
of the requirements for the degree l of
Master of Engineering
1:,
\1
1\
Departro.ent of Electrical Engineering,
i~
McG.l University,
Montreal, Canada.
March, 1980.\
1
1
(1
),
A SEMIANALYTICAL METHOD FOR S.S.
SLTN IN HVDC ANALYSIS
Nickie Menemenlis
B .Eng.
4'
i
i.
i.
f
...
'~
I~
..
~"..,..,.,j~"'
__
r.l""\"P'#~~~_"_""""'_"'
_ _ """'
: '
,~
i
l
1)."4'
1"
A SEMIANALYTICAL' MET'HOD
/'
i
FOR STEADY  STAT S6)LUTION
t
lN HVDe ANALYSlS
"
..
\'
,l>
"
i'
l
l
,
)
"
\
!
/
Nickie Menemenlis
B. Eng. (McGiII University)
1.
i
"
1
l
~
/
l,
0'
,1
ABSTRACT
"
A novel method for finding the steadystate
solution in HVDC analysis is developed. The rnethod is fast,
economical and accurat because i texploi ts fully the properties of periodicityinT, transposed periodicityinT/6,
.....
, balanced 3phase syrnrnetry in the ac
ci~cuit$
and the pro
perties,of linear circuit theory.
The correctness of this ,"semianalytical" method
is verified against the results of a
nurne~ical
integration
method.
The method
, contributes towards making the digital
computer a working tool for HVDC analysis.
,,
j'
/
~
_4"' .....~_ .. _~
\\
,, '
McGILL.UNlVERSITY
FACULTY OF GRADUATE STUDIES AND RESEARCH
)
'Date ..
,
2J. /. .?;.j. p.9. ......... .
l'h~'f.. ~ ~ ....~. ~ .'0:e:.~ ~.~\\? ...C.1YJ.~.l ...... .'.'~ ...... ~ .... ~'.:....... '
DEPARTMENT' EL'.t.1 L,RJ. .................... DEGREE SOUGHT ~.\~.t;'? 'f........... ..
AUTHOR 's' RAME
TITLE OF
~SIS
..
"
'
l\ .... 'S..~L:.f\J.N..~.\.'{~\...~.h ..~..r.~.Q.".....F.RB ................ .
. .~ J. ~.A..Q~.'::~ J:B.l"~ ... ~Q.l...\) .1.\.9N ... .\.N...... \;\:< l\... ~~.~.\;r?\.~ .. .' ..
',a
Il
_,'
Avthorization ia hereby given to McGil~ University to make this thesis available to
readers in a McGill University Library or other Library, either in its present forro
or in reproductjon. The author reserves other publication rights, and neither the
thesis nor extensive extracts from it may ~e printed'or otherwfae be reproduced
wi~out the author's written permission.
2.
The authorization ~a to have effect on the date given above uniess the Executive
Committee of Council shaH have voted to defer, the date on wh,ich it la to have
effect ... If so; the deferred date 18 : ......... ~ ........... ~ .. _ ........ "' ...../ ...... .
'.
Signture of Author
\
PermB.gen~
Address:,.
'
105 ' Ni \b V'), Ap\ '10 G
Mao\'jea.Q  Gu...
l
'\
,.
...
Signature of Dean required if 'date la
inserted in paragraph 2.
(Franais au, verso)
,f
JI,
ni
1\
ii
RESUME
J' ai
d~velopp
ici une nouvelle mth de pour trou.ver
[, '
la solution au rgime permanent
~s
l'analyse de haut tensjon
,
courant contlnu. i; Cette m~hode 'est rapid~. Coromique et
prcise, car elle ~mploie entirement les propri~ts de priodiciten T. de pri odicit transposen T/6,
~'un
ci rcui t
c~
de'~ a
balanc et l'es propri ~ts
s vmmjtri e tri phas
lat ori e" des ci rcui ts
1inai res.
L'exactidude de cette mthode "mianalytique ll est
i
verifie"A l'encontre des rs"uJtats d'une mthode: d' intgrati on
,1
numrique .1"
1
!1
Elle dmontre que l' ~rdinateur numr,ique peut tre
!i
utilis ffectivement pour l'analyse d'haute tension courant
,
continu.
1,
11
1
,
,,
'i
\.
.,
!,
t
1
'0
,
1 \
'1
"
III
iii
(j
ACKNOWLEDGEMENTS
il'
The au~hor is greatIy indebted to Dr. B.T. Ooi
,
for his constant supervision and numerous invaluable advice
during the course of this study. Without his guidance with
a
and friendly attitude this work wQuld not
wi~e
~ave
been
accomplished.
Th~
author wishes also to express his sincere
appreciation to Dr. H.L. Nakra for his supervision and
C
helpful,suggestions
during the absence of Dr. B.T. Ooi.
,
(
l ,
Last, bt not Ieast, the author is very grateful
\
to Mr. J. Lazar for his useful discuss~~ns and encouragement
l.
during the preparation of this work and Mr. S.L. Low for
.\, corrcting this manuscript .
\1
,/
/
"
,
\
'il
1
.9.
/'
    _  _._J:_
1.
CHAPTER
2.4.2
Noncommutative Subinterval
K=2
14
2.5
Magnetic Flux Continuity
Constraint
15
III
PROPERTIES OF STEADY~STATE
SOLUTONS IN HVDC
17
3.1
!Denition of Stead~Ee
Solutions in HVDC
1.7
Properties of the St~adyState
Condition in HVDC Analysis
3.2
3.2'.1
/Symmetrical Threephase Circuits
and S~etrical Triggering
19
19
3.2.2
Linear Circuit Theory
25
3.2.3
Magnetic Flux Continui ty.
27
3.2.4
Periodicity in T
28
i'
CHAPTER
3.2.5
The Commutation Time
IV
DERIVATION OF THE SEMIANALYTICAL 30
/
fi.
29
MET~OD
Introduction
30
4.2
StateSpace Solutions in the
Subintervals\ k= l, 2
31
4.2.1
Subinterval
k=l
32
4.2.2
Sl:lbinterval
k:2
33
Matching of Initial and Final '
Conditions
34
Matching the Terminal
Conditions at t: fi.
34
4.1
1
4.3.1
,
J~
,
vi
(J
Page
4.3.2
Matching the Terminal
~,
Candi tians a t
1
1
4.4
35
,,
38
5~
4.4.1
Algebr~c
4.4.2
Algebraic Expression of
4.5
Newton'type Algorithm
Expression of ~l (0)
"
F(~)
39
41
41
)
5.1
/1
tO
Determina tion of f.1
CHAPTER '," V
'
RESULTS
43
Convergence of the Newtontype
44

Algorithrn
Cz
==
==....::::::...:;;:..  
Numerical
VI
CONCLUSIONS
50
CIRCUIT CONFIGURATIONS
51
CHAPTER
corre1a~ions
5.2
47
APPENDIX
AND STATEVECTORS IN THE
12 SUBINTERVALS
"
A.1
1 /
1
Subinterva1
51
A.2
Subinterva1
k=2
51
A.3
Subinterval
k=3
52,
A.4
Subinterval
k=4
52
Subinterval
k:5
53
A.5
",,::
Subinterval
k=6
53
A.7
Subinterval
k=7
54
A.a
Subinterval
k=8
54
A.9
Subinterval
k=9
55
A.1O
Subinterval
k:lO
55
A.ll
Subinterval
k=l1
56
1
\
A.6
'"
)
1
k=1
ill
'......;
"
L
J
\.
0'
~'
____________
,.
"1
"
viii
:.
(
'
LIST OF ILLUSTRATIONS
Figure No.
Title
~,
21
,,22
HVDC
inverter system.
(a)/ Threephase Voltages vI ,v 2and v 3'
10'
(b) De Voltage, =2 n /3
23
'(a) Circi t topology for subinterval,' 13
, k=l, when three valves are '
conducting. "
(bl
, . ...
circuit topology for subinterval, k=2,0 when two valves are
cpnducting .
1
1
24
,
i
!
31
{
~
32
Continuity Requirement
Peripdicity. and Continuity
, requirement of the~teadY7tate
solution in HVDC~a1YSiS.
(a) circuit topo1ogy for subinterval, k=3, when three valves
are co.nducting.
(h) Circuit topology for subinterval, k=2, when two valves are '
conducting.
''',.
33
Valve currents'i , i b , and i in
a
c
the 12 subintervafS (ks1, ... l2} of
the fundamental wave period T.
l
'.tt
1
l~~'
\
1
16
18
22
1
~.
24
(,
ix
1.
F~gure
No.
41
Tit1e
Newtont,ype algori thm
42
!
51
(a) Convergence of F (ft) towards
1
,,45
zero aS a function of the
i teration mlInber,.i
,
Cb) 'Conlmutation time, ftl l, against
the iteration number ,1
52
Commutation angle
ft
.)
as a func
tien of swi fching angle,
'1 ,
46
49
ct
Line: ",semianalytical" method.
Crossed ,points: Numericall inte,
gration
Al
Circuit configuration during
subinterval
51
kl
'"
A2
,Circuit, configuration dring
stlbinterval
A3
51
k=2
Circuit configuration during
52
subinterval
,
A4
'1
k=3
Circuit configuration during
subinterval
A5
52
k:4
Circuit configuration during
subinterval
A6
Circuit
k=5
config~ration
subinterval
53
1
during
53
~=6
_
..
_..
~._
"'
'
 ....
~~
... ___
__
 _.
'
 
1
1
,/
(,
'i
Fig:ur No.
!illi.
..
A7
Page
Circuit configuration during
subinterval k.7
54
1
1
,f
1
1
1
Aa
Circuit configuration during
subinterval k=S
54
A9
Circuit"", configuration during

55
\ sUbinte~al
k=9
"
AlO
Circuit configuration during
1
All
k ... 1O
Circuit configuration during
subinterval k=ll
Al2
1f
subint~rval
55
Circuit configura ti on
subinterval  k=12
<J.
dur~ng
56
56
!1
'1.,.
1
,
LIST OF TABLES
Table No.
Title
21
pa<fe
Valve switching sequence
during one period T.
(
.)
1.
12 l"
if
II
~'~~~~~~~~~~~~~""""""
'
,1
xi
:'
,t
NOMENCLATURE
Matrices
[A]
the~
Constant coefficient matrix of
system of
linear differential equations (equation 33).
" ,
[B ]
Constant mat ri x of the, system of linear
diffe~,~
ential equations (equation(33.
[C ij]
Partition of the state transition matrixfl (~~
,defined in equation (422).
[ CP]
Dimensional compatibility matrix defined in
equation (48).
[Cs]
Rearrahgement rnatrix defined in equation (413).
[~J
Constant matrix constituting of
1,0 and 1
(equation (AI .
[IJ
ldentity.
[L]
Inductance matrix in henry.1
[M]
Modal matrix of [A ] (equation (C2.
R
[RJ
General purpose vector.
Resistance matrix in ohm.
".
Forc~ng
function of the system of linear differ
entia1 equations (equation(33.
o
l
 '!J
1.\\
xii
Vector of the statevariable of the system of
'1:
diffrential equations:
~,
~ks
Particu!ar integral.
Diagonal matrix of the eigenvalues of rnatrix
[A]
(equation (C2)).
Constant of integration.
Statetransition rnatrix defined in
Transpose.
Inverse.
(
,
"
equa~ion
(35).
,1
xiii
Subscripts
,
a,b, c
Phases in the ac side.
d
j
. .
rDc quanti ty .

.\
Cycle number (Fig. 31).
Subinterval number.
Iteration number.
.s
particular part of th,e solution (equation (34) ) ._
Quantity of synchronous machine.
\ ,
Il.
. Quanti ty of' the trans former ""
'il
Constants
i
.~
l
,~
,
~
Amplitude of the source.
Id
Constant
J
f
dc' current.
'
IV
"
"
Inductance
(henry) .
Resistance
(ohm) .
fundamen tal wave period.
V
d
Constant
Firing angle of a valve.
peruni tized closure norm.
,."
'
dc
voltage.
Commutation time (sec)
'j.
/'

'"
_,41 _
_
9~
e,
xiv
(r~
Angular frequency of 60 H
,
"\
i(
Variables
i,e
Instantaneous current and voltage.
time.
In~tantaneous
voltage.
..
Ct
.j
1
,
!
1
j
(
"1
~~t
~ll \
~
,
,1
"
"
CHAPTElt
INTRODUCTION
y
..
Electrical pow,er may be transmitfed by me ans of
.. aerial Unes or
Although the
un~~rgFound
mai~
cables, in either ,ac' or dc forme
bulk of' electric power is transmitted using
ac, there are occasions where dc transmission finds its niche,
in:
/'
a) Very long transmission distances.
.)
"
b) Underwatr transmission.
l
c) Interconnection of systems at different frequencies.
d) Reinforcement of large
ac
systems.
,
Research in HVDC revolves around the
J~,\
of component~, the converter station
ers
[11,
developme~t
dc circuit break
[2J, Ithe effe'ct of harmonies [3], multiterrninal
topolog~es
etc [41~ . However, this thesis is concerned with~the predictions of the
syst~
Qehaviour, More precisely, it is
concerned,with a method of predicting the steadystate per
.,
,
formance of the HVDC system using the digital computer.'
, l
1~
1
c
\
'
).
l.~
""
."
Stte of Art
""
Cl
To date, the methods for predicting the perfor.rnance charpcteristics for design, research and operation, fall
under two major categories:
1.1.1
Ana10g sim~tion Methods
The simulation of the HVDC
~ystem
is performed
using; sorne. ana log representation of the system equations. The
converter station is u~ually modelled by a thyristo~ bridge.
~
This method is an excellent tool for system study,
and is the workhorse in rsearch laboratories [1113J
/
1.1.2
Digital simulation Methods
\
HVDC
The use of the digital computer as tool for
system analysis is still
i~
its infancy.
~evelopmental
wo;i
falls under two further sUQdivisionp:
a)
'1
Load!Flow Studi~s
Loadflow studies [1417J
t~~
needs of
th~
are oriented to
system dispatcher. In these studies
l,
a number of gross assumptions have been made to
,simplify th~ analysis. Typically the commutation
p~riod
i;
neg~d
and the' phase current takes a
rectangu1ar form.
Recently there 1 has been a spate of activities
1
(
in this area following
i~terest
in multiterm1nal
stations and acide system interactions [4
1/
10J
/
b)
Studies
The majority of papers in the area of HVDe
t'
analysis makesuse of numerical integration techniques [18  2SJ.' This is an allpowerfu,l method
which is'suitable for any conceivable situation.
Thus the system may be unbalanced, unsymmetrical
and faulted. One can use it for transient analysis'
and
~ady~stat
analysis. However, one pays
a cost for the generality of these 'allpurpose programs.
Bsically, they are slow in cornputing the
steadystate solutions which are the solutions
that
(}
the investigator is rnost interested in. This
is because one does not know what initiai conditions
to use in order to arrive at
As such, one makes
a guess
th~ steadystate solution.
of the initial conditions!
ti
tnd runs the prograrn for a sufficiently long time
\or ,the transients to subside. Usually, the time
constants assodiated with the transients of the
tt
sy~tem,
1
1.2
,are generally qui te long compared with the
ti,{y integration step,\
Staterneqt of the Problem
This thesis recognizes that in rnany applications
the investigator i5 interested rnainly in the steadystate
~.
'
solu~ion.
Furtherrnore, a large blass
o~
HVDC problerns falls
l~
1
j
under what may be described as symmetrical and balanced networks. As such, there is a 'need for a quick and fficient
"
methodifor the solution of this class of problern.
As has been stated
i~
the foregoing section, the
slowness of the numerical integration methods in arrivi~g at
the steadystate lies in the fact that transients are unavoidably excited
u~~ess
one is tortunate in guessing the
correct initial conditions. Thus, if on makes a correct guess'
Q
of the initial state
~(O)
such that, after the time period T,
the state vector !(T) is equal to
~(O),
then one is sure to
be at the steadystate. This method, of course, has ta ken
advantage of the periodicity of the steadystate solution.
The steadystate analysis of linear and nonlinear
,
1
circuits with periodic inputs have always treated the subject
as a tw6point boundary value prob1em [2630], in which one
,
<
l
,attempts to find a systematic method of converging quickly ta
J
the initial states which satisfy the periodicity requirl2ijlent.
l
!(O) eq'ual to !,(T). In fact, for linear systems,
~la
closed form
I
i.,
l
!1
of the initial state x(Q) is derivable.
J
However, as
app~ied
to HVDC analysis the problem
becomes more complexe This is because the six valves of the'
1 ,
1
converter
, bridge are triggered , sequentially once in a cycle.
,
This breaks the fundamental period
into 12 subintervals.
In
1 l
'
ech subinterval the system netwo~k
is
mode1led by a differ~
ent circuit topology. This changes the problem ta a 13point
(J
1
~

"
boundary 'value problern.
\
'
" The HVDC analysis is further complicated by the
fact that the durations of the commutation subintervals are
1
unknown, 'and are themselves solutions to the system equations.
Thus far, the statement of the 13point boundary
;/
enun~iated
valoe problem has been
in the most genetal way.
For the restricted cases in which the following assurnptions
can be made:
a) linear circuit representatioh of'system topologies,
b) symmetrical threephase circuits on the ac side, ,c) equally spaced triggering of the converter valves
at T/6,
it is possible to redu ce the problem to a threepoint bbund,
ary value problem. This reductio,n is amenable because:
a) The assumption
of
<
linearity allows one to write a
closedform analytical formula ,relting the termina~
state vector to the initial state vector of each
subinterval.
b) The ass~ption of symmetrical network topologies
and equally spaced triggerings allow one to consider
the
~solutions
of any two consecutive subintervals
to be the ;templates from which the solutions of the
entire period can be constructed.
The foregoing assumptions tantalize one to the
o
(

\ ,
(
\.
'
6
jo
prospect of an analytical solution.nUnfortunately, the 'com
mutation time, f.1, is basically unkno,,:,n. This has to be
.
s61ved from a .,transcendental equation involving the cornmutation current diminishing to zero. The solution of tl1~
transcendental equation is obtained numerically. For this
,
.,
wr
reason the method described in, this thesis .i:s called a
"semianalytical" method.
,
[
o
1.3
Organization of the Thesis
\
This thesis is organized in tne/following way:
Chapter II is' a brief review of the Graetz
1
bridge of' the converter station.
for an HVDC
[13J
~n
inverter station
is chosen as an Fllustrati ve ex
ample and is examined in detail. This illustrq..five
example is used throughout this thesis for the exp lanation
of the " semianalytical" method (Chapter lH)
,
and to glve quantitative results (C?apter V) .
\i
Chapter III presents in a detailed' fashion
the central ideas of the "semianalytical" meth~d.
lt
itemizes the properties of the steadystate solution
and t he
.
prop~rtl.es
f rom the Jund
'
.
erly~ng
assumptl.ons,
which are exploi ted in the "semianalytical" analysis.' r
i Chapter IV organizes aIl th~perties toge ther
and derives the pertipent equations.
".
Chapter V gives a quantitative check to the
,correct answer and effectiveness of the "semianalytical"
metllod.
,
CHAPTER
,<
II
REVIEW OF THE THEORY OF THE GRAETZ BRIDGE
,
2. 1
Introduction
The theory of an HVDC
conver~er
9tation is well
. This chapters devot~d to a quick review
known
of the theory of ft,s
opera~ion.
Rather t:han discussing the
subject in generar terms, a specifie examp1e is chosen as
,
,il1ustrated in Fig. 21.
ea+ Ls
Rs
Vfr
t
Id
(
V4
vs
~d
V2
Fig. 21
,f
HVDC system showing the Graetz bridge) inverter
connected to a,synchronous ~qchine and a balanced
load on the ac side and a constant current' source
,
.
Ij)
on the de side.
Fig. 21 represents an inverter station consisting of the following parts:
1"
:\
1
,..,8
1
"
1
1
al
An ideal/constant direct current source" Id' This
represents the curreht from an HVDC transmi'ssion
line.
b.)
A Graetz bridge, referred als,? as a threephase
six pulse bridge, consisting of six controlled
valves. The valves are nurnbered vl to V6 accordt~eir
ing to
c)
firing order.
A three phase transformer connecti'ng the ac side
1
of the Graetz bridge with the load, and the synchronous machine. Each phase of the transformer
o
is represented by its equivalent model consisting of an inductance LT ,and a resistance
d)
R.r'
,C'
'\
three phase load., Each phase co.nsists ,of an in
1,
1
duc tance oin parallel with a resistance, Land R
,
respectively.
....
1\
"
el
The last part of the station is a synchronous
machine. It is represented by its
simplestimo~el,
this being an ideal voltage source behind an
equivalent stator inductance
(subt~ansient),
LS
and a stator resistance, RS' The instantaneous
t
line to neutral EMFS of the machine are:
'11/2) 1
e C (t )
1
Em sin ( c,,) t .. 5'!: / 6 )
(>
.:;;.......~
..
 ..
~
..._.
(21)
where: Em
w
= Amplitude
of the source.
angular frequency of 60H
This synchronous ~achine is included in this
1
inverter station in order to provide natural
1
commutation t9 the bridge. In this inverter
system it also furnishes part of the inductive
power requirement on the ac side.
",,1
Bridge Operation
As shawn in Fig. 2Tl, the rGraetz bridge is cam
.posed of six controlled valves. As is well known, the ignition
of each controiled valve may be detained by the
"delay angle"
,01..
l3l]
It i6 well' known
.1 '
Il
~4
1
.J
that the bridge
operates as:.
1
ft
i)
a converter fbr
ii)
an inverter for
< < /2
"'/2 < <
cr
7r
cr
11"
and
,)
In or der 1;0 illustrate the meaning of the firing
1
!
,t..
angle cr,
the voltages vI' v 2 and v 3 (6e Fig. 21) are
shown in Fig. 22 (a). The DC voltage, Vd' for
cr
=2
;
11'
/3 :hs
shown in Fig 2 2 (b) .
j
i
!j
r
1
Although the trigger signal which fires a valve
is controlled by the value of a, the extinction of a valve
is uncontrolled. The method of swi tching a valve off is the reversal of the'current
'0
th~ough
that valve. This change of state
'\
,
.
1
 .. 
cr.
"
1\
!
1
10
(~l
"
J
~I
t)
. V (t)
2
Va(t)
1
;"
,
,t
..,
T
(a. )
'Id
fi
~a
111
Cl
1
l,
J
(b)
Fig. 22 (a)
(b)
Il.
\r\
3phase voltages VI' v 2 and v 3
..pC voltage Vd for' Cl = 271:/3.
\~
11
of a valve, from the conducting to the nonconducting position and
vice~versa,
is known as "commutation" of th.e ...
valve. Furtherrnore, the time required for the current of a
,valve to drop to zero is the "commutation" tirne and will be
denoted as "jJ." sec. throughout this work.
Du~ing
normal operation of the bridge, the valves
are triggered successively in the order shown in Fig. 21.
The system completes one cycle after each
val~~
has been
triggered once and has conducted for a specified period of
time, which is usually one third of a cycle,or
2~/3
radians.
1<1
Thus under nonfaulty conditions there is always a pa th from
the dc ~ide of the system t? the ac side and back to the dc
side through tHe conducting valves.
t'
~.3
12 Switching Subintervals
During normal operation, at least two valves are
conducting simultaneously; one from the top of the bridge
(namely VI, V3 or VS)and a second one from the bottom part of
the bridge (narnely V2, V4 or V6).The valve from the top of
the bridge perrnits the dc current to
flow~towards
the ac "
side, and the valve from the bottom of the. bridge leads the
current out df the ac side\.
T~e
six triggerings of
break the prob1em in 12

~he
sbintervals~
valves in one period
'it0
12
Ir
(_1
SUBINTERVAL NUMBER
1
=1
V2
V3
V4
2 1
V3
V4
31
V3
V4
VS
V4
VS
7 1
1 6 1
VALVES CONDUCTING
V4
VS
V6
VS
V6
VS
V6
VI
V6
VI
V6 VI
VI V'2
V2
VI
V2
V3
V2
V3
Valve
Il
V2
V3
V4
.,
Table 21
8 1 9 1 10 1 III 12 1 13
switching sequence dUl:'ing one period T.
'Table 21 enumerates aIl possible combinations of
l
,j
the conduc,ting valves during one cycle in the sequence they
occur in the partieular example.
( ,J
i)
"
During the even subinterval, k
2n (n
= 1,2,3 ... ),
'1
only two valves are conducting.
ii)
Durin9rthe odd subinterval, k
= 2n
+ l(n
= o~
)
,1
l,
2, ..... ), three valves are conducting simultanr
eously. This is the commutation subinterval, during which the changeover of current, from one
valve to another valve, oceurs. This changeover
of current does not happen instantaneously. At
1
,
the beginning of such a subinterval a new valve
is fired while one that was already conducting
ceases \ at the end of the subinterval.
o
, '
13
2.4
Circuit Topology
As describe4 in section 2.3 the circuit topology .
,may exist in two main configurations with eit.her two or three
valves conducting. In the chosen example,' the circuit topo'\
logies of the HVDC inverter system for the subinterval numbers
k=l and'2 are shown in Fig. 23(a) and (b) +espectively.
(a)
1
\ld
V2'
(:'
"
(b)
V3
V4
.,
'"
'
Fig. 23(a)
Cb)
Circuit topology for subinterval,
k = l, when three valves are conducting.
Circuit topology for sUbinterval,
k
2, when two valves are conducti~g.
'
14
Commutation Subinterval k
2.4.1
It is worthwhile emphasizing that the reversaI
of current of the extinguishing valve (in this case V2) does
not occur
the
This is due to the inductances in
instanta~eously.
circ~it.
Thus, it is clear that the magnitude of the
depends on the size of these induc
commutatio~ time,~,
tances.
Commutation occurs during the odd subintervals,
r,
= 2n
at
=0
(n
O,1,2.~.).
The subinterval k = l
with the triggering
earlier) , and ceases at
= IJ.
o~
begins
valve V4 (which was off
sec. with the extinction of
valve\V2.
2.4.2
Noncommutative Subinterva1
=2
i'
The noncommutative subinterval is the period
<,
between two consecutive commutation subintervals. This is
taking place for
k  2n (n
= 1,2,3 ... )
and exactly, two
valves are conducting.,
The valves are considered as ideal switching
elements with the 'following characteristics:
a)
,f
TheYf.have zero resistance to positive anode CUrr;.:f..l
rentithat is to say, they are considered as short
circ~~ts
b)
when'they are conducting.
They have an infinite resistance to current
ir.f
15
the reverse direction.
Also they are open circuits
when they cease to conduct.
2.5
Magnetic Flux Continuity Constraint
,In each subinterval the
dynamic~
of the system "of
Fig.2l, may be described by rnqthematical equations in a
"
statespace variable forro:
~k
= ~k (xk, t)
= 1,2,3 ..
where
( 22)
'"
is the subinterval number.
,,
From a circuit point of view, Fig. 21 contains
a three phase voltage source, a dc urrent source and linear
l
l
circui t elements'.~ Thus the system differentia1 equations
describing the network in each particular subinterval, are
linear. Therefore, J it has a closed forrn solution. These solu\
,1
1
tions may be expressed analytically except of the cdnstants of
~
integration. The constants of integration
!
1
ing
~e
obtained by invok
the cong~raints of periodicity (to be iscussed in sec
tion 3.2) and of magnetic flux continuity.
1
1
t
1
At the instant of time when any valve begins or
ceases conduction, the circuit changes its configuration.
The
sqlut~ons
of, twc
pieced together by
contiguou~
satisfyin~
systems of equations are
the conditions of the magnet
ic flux continuity constraint which in this case is the currents in the inductance elements
L , Ls and L.
T
.~
"
.
IG
(~
..
SUBINT~~ ~UMBER
):
,1,
'k
"'k+l
 \
1tK+I
\
Fig. 24
1
1
Continuiy requirement.
Cf
,More
\
kth subinterval
pr~cisely,
evalu~ted
the terminal state
1
t~rmination
..
tirne t
of the
is used
k
to fom the initial sta te ~k'+ 1 (tk ) of the next subinterval.
at the
~k(tk)
/
1
1
,r
j
t
17
lI
!,
(:
1HAPTER
III
>
PROPERTIES OF STEADYSTATE SOLUTIONS IN HVDC
\
\
3.1
Definition of SteadyState Solutions in HVDC
For the purpose of this analysis, it is necessary
to assume that a steadystate in .the HVDC system existlil. An
HVDC system is said to have arrived at its steadystate when
\
aIl the transients have disappeared and the system has resumed a cyclic operation of period T. This definition correto the sarne notion of steadystate in
"
sponds
Thus the system in the steadystate fulfils the following
analysis.
conditions:
a)
a~
AlI transients from faults, switchings and reclosures have died out.
b)
AlI transients due to a transition of the sY,stem
from one steadystate condition to another ~eadyl
state condition due to a change in the load,have
disappeared.
c) . Each cycle of operation is identical to every
other one.
,
Therefore the steadystate is defined as the
\
periodic; solution of the statespace variable, that i5:
1
, .
\
~)
where:
~(t)
= ~(t
T)
(31)
is the stateyector and
T is the pe'riod of the fundamental wave.
\
..
".
,'
l'
18
f'
Il
1
11 1
The periodicity of eguation (31) together with
i1lustra~
the continuity requirements of equation (22) are
ted in Fig. 3,1. From this the following can be inferred:
a) The final conditions of each cycle
are equal
to the initial conditions of the next cycle
j+l.
b) The intermediate events within a cycle are repeated with a period
T.
"
CYCLE
~~ER
"
,\\
t.T
Fig. 31
,,
""'4'..~lfooI.t T "+1
Periodicity and continuity reguirement of the
steadystate solution in HVDC analysis.
\
,In
\ ,
'~
ac analysis the "s1readystate'~ution is
the "particular integral" alone and the
"transi~nt"
solution
is the "complementary'function" of the system equations. In
HVDC it is
worthwh~le
to stress that in contrast to
. "steadystate" solutiqn is composed of the
ular
SUIn
ac" the
of the "partic
integrals" and the "complementary functions" of the
system differential equations in the subintervals.
"
$
,
"
,.
i1
19
3.2
Properties of the SteadyState .Condition in HVDC Analysis
"'
The semianalytical method of finding the steadystate solution in HVDC recognizes and exploits the following
properties:
3.2.1
Symmetrical Threephase Circuits and symetrical
Triggerin<]
1.
The
ac side of the system is assumed,to fulfil
the requirements of:
a)
thr~ephase
balanced and syrnrnetrical circuit.
i1
This means that:
J.
i) the impedances of
er' are
~ach
Phase\of the
transform~'
~),
'~
equa~ 1
, '4
lJ
ii)
the load is balanced"
1i
iii) the "subtransient" inductance and the stator
.
resistanc
1
hav~
\
1
the same value for each phase,
Cl
1
1
1
iv) the phase voltags of the synchronous machine
haye equal amp'litude and have a pha,se ang:le of
~ radians between them.
b)
\.
d
:~
';1
J
J>
Symmetrical triggering. This means that the valves
of the Graetz ?ridge receive the triggering pulse
at equally spaced intervals of T/6.
,
As a, oonsequence of the symmetry and equally spaced
triggering, there exists only two basic circuit topologies
fo~
,
l
~
20
the 12 subintervals in the fundamental period T. These basic
circuit topologies are as shown in Fig. 23(a) and {b). The
topologies of
th~
other ten subintervals are repeated forms
of Fig. 32(a)an(b), except that ,the currents have to be
relabel1ed because of the valve swi tching se,quences as
Q.
shawn in Table 21. This property is defined as transposed
symmetry.
To illustrate the
the subintervals k
= 1,2,3
transpo~ed
symmetry property,
and 4 will be examined. Their
circui ts topologies are shown in Fig, 23~}', (b) and "Fig.
32 (a)
(b),
The odd subintervals ,are essentially the commutation subintervals in which three valves are conducting
(~ }
._
1/
simultaneously"
During subinterval
!
.i
1
= l,
V4 is
starts conducting. The current through V2 tends
V3 carries the entire
current Id' At the beginning of subinterval
l
!
1
1
= l,
";
due ta natural commutation. The end of'this
~
~t
"
subinterval 's marked by ,the cessation of conduction of V2
"
2
and the complete transfer of current I~ t~ the b~an~h of V
The sarne effects are observed during thesubinterval
= 3,
The
differenc~
1s that the valves involved
in this commutation subinterval have changed and the branch
currents must be relabelled. The raIe of V3 is now played
'{
')

=~~~
\,
"'
',
21
,1
.[!' (',
by V4. 'The valve te be fired at the beginning of this period
p'
<'
vs whereas the valve 'which is going to cease
i5
con,~uction
at the end of this subinterval is V3. By ind~btioh, it can
be concluded that aIl the odd numbered supintervals are
transposed forms of the subinterval
=,1.
The ev en subintervls are essentially the noncommutative subintervals in which two valves are conducting
sirnultaneously.
.j
'\
The transposed symmetry is also observed in the'
ev en subintervjls. 'During subinterval
=2
the valve V3
is conducting on the top of the bridge and V4 is onducting
~
on the bottom of the bridge. During subinterval ,k
=4
.1
the
valve VS is conducting on the top of the bridge and the
J1
~
valve V4 on the bottom of the bridge'.
\'
"
!f
"".,.
~
l
.j
~"
1
"
1
\,
"
(
&
\.
. 1
1
1
.22
1
V3 \15
'1
c,
IJ
(
"
i!
1
!
,
'1
Fig. 32{a)
(b)
.,
Circuit topology for subinterval, k = 3,
<
when three valves are conductirig.
Circuit tOP9logy for subi"nterval, k = 4,
when two valves are conducting.
(
\
l '
  
ij
.1
23
The ab'ove paragraphs explained the transposition
'1
symmetry in the Graetz bridge. Fg. 33 shows the current's
i , ib and i on the transformer branches as a function of
a
c
'time. The transposition property is clearly observed. The
current 'waveforms in the subinterval
=1
are repeatedly
transposed in all the odd subintervals. Sirnilarly, the
cur~
1
rent waveforrns olf the even subintervals are repeatedly the
transposed forms of' those in
~,
This transposed (T/6)
= 2.
~
.,..
periodicity means" that
the behaviour in a basic subinterval spanning a brne of T/6
suffices in describing the behaviour of the complet period
of tirne T. Any two adjacent subintervals such as the one for
k ;
ft
,
~
1 and 2 constitute this subperiod of TV6.
d~
,~
~)
~l
This transposed T/6 periodicity is illustrated
in Append~x~A. Appendix A shows the' circuit configurations
.;
for the 12 subintervals for the l in,verter ~xarnple of Fig. 2~'l.
.,
In this exarnple, the statevectors of the commutation subI[
intervals
.1
"
"
(k odd) ar 7 all sixtuple vectors. The current
elements which consti tute
!.1'!.3'
... !.ll
'1
'..1
are listed in
!.
this Appendix A. The statevectors of the noncornrnu tative
.,
~ubintervals
(k even) contain five currentl elements. The
1
current elements wh~ch consti tut 7 !.2' ~4' . ~12 are alsOi
J .
,l~sted
in Appendix A.
.

_.
~~
: ...... _
......
""'~,b~
...,r...~ ...~~.>k'"l", ....M:>!""~~~.ti(,..4~.y~ ..I."
'
'
24
,
1
With
the definitions of the statevectors
!k (k  1,2,3 ... ) as gi ven in Appendix A, the transposed"\
b~
(T/6) periodicity can
expressed mathemat1cally as:
r'
(t + nT/6)
~k+2n
= ~k(t)
(32)
,
n = 1,2,3'0'
Il
where
k+2n
1,2
is the subinterval number.
sus :INTERVAL
K:i Il 1 2 1 3 1 4 1 ~ 1 6
NUMBER.
'1 7 1 8 1
9,1 10 1111121131141
J
VALVES ' CONDUCTING
(
i
ha
:r,
!(
V2 V3 V3 V4 V4 V5 V5 Vs VS VI \fI V2 V2 V3
V2 V2 V3 v3 V4
VI
V6
V3 V4 V4 V5
V4
~ ,.~f .~~ y'3
V4
V5
7'
!'iiiI::
ib
~f
,
1
a..
"
T
1"
ro.;,:'
a..
t
t
.1
tl
Fig. 33
1
1
Valve
currents i a , ib and i c in the 12 subintervals
(k =~l, 0.012) of the'fundarnenta1 wave period, T.
J'
25
3.2.2
Linear Circuit Theory
As has already been discussed in section 2.4,
the system equations describing ea,ch subinterval are linear.
Equation (22) in its linear form becomes:
"
1
( 33)
= 1,2,3 ..
= subinterval
~k
=a
statevector of dimension n .
k
=,forcing
~k(t)
number
~)
function vector of dimension m conk
taining elements of the threephase
source and the dc current source.
\.
nk
[B~J
x,~
= nk x mk
.~
~"
constant matrix
~
l
constant matrix
1
1
From linear' circui t 1;heory
.1
Il
equation (33) ,
1~
= ~ks (t)
+ [cI>k (t)]
lk
~
~
has a closed forro solution as f6l1ows:
x k (t)
[32 ]
.J
1~
(34 )
~
~
where
,1
X~s(t)
= particular
integral (dimension n )
k
(35)
= statetransition
fk
matrix
 constant of integration
.
26
T~e
forcing function
~k
(t) ,which is known ex
plicitly,together with the constant matrices
[Ak] and [Bk]
determine uniqu~ly the solutions o.f the particular integral,
,.,:P
x k (t). Themethod for solving for
 s
 ks
(t) is shown in
Appendix B.
The statetransition matrix
[!l>k (t)]
ls a function
alone. The method for evaluating [cJI k (t)]
is given in Appendix C.
of the matrix [Ak]
The solutiors of equation (34) are known except
for the unknown constants of integration
,~,
= 1,2.
These
unknown constants are obtained from the initiallconditions
of each of the subintervals. In the case where subintervals
k'::
1,2 are chosen as the representatives of the complete
cycle, the initial conditions needed are at t
As will be
sho~n,
= 0 and
.~.
the magnetic flux continuity
requirements will impose the conditions that the states at

>
the end 'of one subinterval become the states of the initial
\
conditions of the subsequent subinterval. This requirement
1
.
,;
combined_ with the periodicity requirement will make it possi,
ble
te obtain two sets of linear simultaneous algebraic
equations, from which the unknown constants, !k (k
= 1,2
will be solved.
()
27
(
3.2.3
Magnetic Flux
Contin~ity
The "magnetic flux continuity" condition requires
that at the jqint of two subinterva1s there is a continuity
;
of currents
ih
the inductance.
conditions of each subinterval
'
In consequence, the initial
are obtained from the final
.
\ su b'~nterva 1 .
con d ~'t'~ons 0 f t h e prev10us
In the case where interva1s
= 1,2
are chosen
as the representatives ?f a complete cycle, their joint is
:,
'"'4
at
t=~.
At this point of
of subinterval
tions of
~~(~)
=l
t~e,
the final conditions of
~1 (~)
are'obtained from the initial condi
of subinterval
= 2.
This property is
used in solving for the integration constants lk
in equa
tion (34).
Mathematically this is expressed as:
1
~2 (~)
[cp J ~l (~)
(3 6)
,
where
~~
~he
This
li
[cp]
i5 an (n ' x nI)' dimension compatibility matrix.
2
compatibility matrix consists of O's and l's as elements.
[cp]
matrix is required because !.2 (t) and
~l(t)
do not
have the sarne dimensions.
!}
As will be shown in the next
the
?eriod~
i
\
icity in
s~ction
requirernent together with the (T/6)  period
icity requirement will provide the second necessary initial
\
condition at t = O. This is needed in order to solve for
the unknown integration
constants
l,
J
5,..
,...
r
j
28
(:
3.2.4
Periodici ty in
T_
"\
The periodici{y in
T requirernent demands that
the initial conditions of each cycle are equal to the initial
conditions of any other cycle. Simi+arly, the final conditions
of each cycle are equal ,to the final conditions of any other
cycle.
Mathematically, the periodicity in
T requirement
of the ini,tial conditions of the first cycle is expressed
as:
~l (0)
= ~13 (Tl
(37 )
Equations (37') is needed in order to solve for
the initial conditions of the system at
= O.
Equati9ns (36) and (37) are the two terminal
conditions by which the constants of integration
ik
: (k=1,2)
can he solved. Of course, equation (37) inyo1ves the subinterval
= 13.
However, by exploiting the
T/6 transposed
periodicity property of equation (32), one does not need to
solve beyond the intervals
= 1,2.
stress that the commutation tirne,
It is important to
, itself is an algebra
ic unknown. In consequence, the solutions of
lk
are functions Qf ~ . The next section shows hO;
(k
= 1,2)
may be
obtained.
(~)
",
29
3.2.5
The Conunuta tion Time
JJ.
In order to evaluate
~2(~)'
the commutation
"
time, p. , needs to be known explici tly. It has already been
\
discussed in section 2. ~,that t,he triggering of a valve is
controlled at will but the switching off is not controllable.
, This
commutation time
depends on the parameters of the
system. It is the point in
t~e
when the current in the corn
mutating valve tends to reverse in di:ction.
In order to find the commutation tirne p.
one must
f,
write an equation of the branch current of the commutating
valve and solve for
when the value of this current equals
zero.
(
The expression of the commutation current is a
nonlinear function of
~.
It is therefore necessary to solve
it numerically. This is the reason why the rnethod'for solving the steadystate in HVDC analysis is called
Il
semianaly.:
tical" . ,A recursi ve formula, akin to Newton' s rnethod, is
employed in order to solve the nonlinear
equatio~
~""~
in
~~~~~~~~
~=~====
\
1
CHAPTER
IV
DERIVATION OF THE SEMIANALYTICAL METHOD
.' ....
4:\. Introduction
Instead of deriving the semianalytical method
in the most geneFal way, this thesis is
devoted
ent~rely
to the Graetz inverter system as given in Fig. 21. This
\
Il
specifie example gives sufficient structure from which sorne
of the fine points will have to be solved. Among thes
r points
are:
a)
the system dimensions of the odd and the even
subintervals are
ap~iieation
incompatibl r
b)
the
of the T/6 transposed syrnmetry,
c)
the determination of the commutation time
p.,(
Essentially, the derivation consists of organizing
the information based on the properties described 'in
chapter III,
50
that the constants of integratton
"
and p can be solved explicitly.
(k=1,2)
'
Although this specifie ex
ample lacks generality, the ideas have applicability in
more eomplex situations.
fo
In the development of the derivation, the two
systems of linear differential
interva1s
e~uations
describing the sub
1!2 are so1ved exp1ieitly. The solutions of
\
these two systems of equations give rise to two algebrae
C)
unknowns
~l(O)
and
,
~2(P).
The matching of the terminal con
'\
\
  
.
~ 
~
 
31
ditions of the two subintervals resu1ts in an equation of
~l(O)
in terms of the single algebraic unknown
Finally by swdstituting
of subinterval
constant,~
~l (0) into the system of equations
l, one derives a nonlinear equation in
This equation is solved for
Il'
Il
by ernploying a recursive nuI
merical routine akin to Newton's rnethod.
4.2
StateSpace Solutions in the Subintervals,
k~1,2
The first step leading to the drivation df the
sernianalytical method is te1 obtain solutions of the system
,of equations for each\ subinterval,
k =1,2. It is :to be
pointed out that the statevariables in each subinterval have
different dimensions.
It should be stressed, that for the purpose
~f
this analysis, under symmetrical conditions, the coefficient
\
matrices [~],[BkJ, and the forcing function ~k(t) are
equal in aIl the odd subinterva1s.Similarly, [~],[BkJ and
1)
~k(t)
are equal in the even subintervals.'These are con
sequences of the T/6 transposed,syrnmetry. As a result, one
needs only to solve for the state equations in the subintervals
k'~
1,2 only. The solutions in the other subinter
vals can be derived by relabelling the current e1ernents
with respect to the elements of the solution vectors ~k(k=l,2).
The statevectors in all the 12 subintervals together with
()
the circuit configurations are shown in Appendix A'.
, ;
"
32
\
4,2.1
Subinterval
=1
The system of linear differential equations describittq the network in subinterval k
= l,
shown .in Fig, 2~) ,
is given by:
(41 )
1
where the constant matrics
[Al] and [Bl]
are of dimension
x n ) and (n x ml) respectively, and the forcing funcl
l
tion vector, ~l (t) is of dimension ml' These are listed in
(n
Appendix D. l,
The statevariables are of dimension
~re
=6
and
choseri to be:
1
1
(42 )
t
t
The total solution of equatfon (41), aqcording
\
to the general solution, equation (34l, is as follows:
1
1
!.l(t)
= ~ls(t) +
[4>I(t)]
1~1(O)
~lS(O)}
(43 )
\.
where
!ls (t) a~~ [CPI Ct)]
T,he expression
1~1 (0)
are listed in Appendices B and \ '
 ~ls (O)} on the right hand side is
the unknown integration constant
Il
accord~ng to equation
(34). This i.s obtained by substi tuting
and
(34),
~ls
(0) may be evaluated
=0
in equation_
~xp1icitly
as in
7
1
33
Appendix
while Xl (0) is an algebraic' unknown of the initial
condi tions at
4.2.2
= o.
Subinterval
k 
The system of linear differential equations
des;,
cribing the network in subinterval k=2, shown in Fig. 23 (b) ,
is given by:
~2
= [A 2 ]
~2
[B2J~2(,t)
where the constant matrices [A2J
(n
x n ) and (n
2
x m2 )
u (t) is of: dimension
2
\
and
(44)
[B2]
are of dimension
respectively and the forcing vector
ID
These are listed in Appendix D.2.
The statevariables are of dimension
n  5 and
2
are chosen to be :,'
K2 Ct) T
[i2 (t), i3 (t.) , i5 (t), i7 (t), i9 (t)]
( 45)
equ~tion
The total solution of
(441 is similar
to equation (43). and is given by:
~2
(t)
= !.2s (t)
+'
[4> 2 (t.u>]
{ !2 (p)
~~ ~,u)}
(46)
where
!.2s (t)
and
~~ (tp~
are
~i.sted
The expression {!2 (,u)  ~2S (,Il)}
the unknown in tegra tion constant
~ ,
in
!.2'
'and C.
on the right hand side i6
1
i2
according to 1:.he equa tion
(34). This i6 obtained by substituting
(34) and solving for
APpendi~e6
=p.. in
equa'tion
As in equation (43), x
2s
(,Il)
is
34
evaluated explid:tly in Appendix
. Unfortunately,
(~) is an algebraic unknown of the initial conditions
2
of the subinterval k = 2, at t = p.
The algebraic unknown as posed in the solUtions
in equatins (43) and (46) are
~1(O)
andx (p.). They are
2
themsel ves functions 'of the unknown conunutation time, p.
~
The following section discusses the mat ching of the terminal
'
~l
condi6tions. 'This leads to an expression .for
(0) as a
function of, Il alone.
4.3
Matching of Initlal and Final Conditions
In order to solve for the ,unkn<?wns
~l
(0)
and !.2 (fI)
of equations ( 43) and (46 ) one uses:
a}
the flux continuity requirement at
= Il. ,
t}
the periodicity in
,j
T requirement given in
equation (37)
These requiremen ts are used at the points
and
4 3. l
=ft
O.
Ma tching the Termina l Conditions a t
=p.
From the flux continui ty requirement at
='!J!f
one can de ri ve the unknown initial conditions of subinterval
k
= '2,
!.2 (p.), from the final 'conditions of subint~rval
= l,
!.l (Il). This is possible because the currents in the
five elements of the 'statevector
()
~1
(t) ,
the
1,
:
.'
r
!l
,,
1
, 35
1
l
e1ern,nts of the
ement of
stat~vector ~7 (t).
i
'!1 (t), that is
interval
,=
(t)
However, the sixth ~ e1
ceases to
flo~
in the sub
2. Furthermore, ic (t) is not an el'ement of
~
!2(t). Therefore, oneohas:
(47 )
where
[cp
J
,
is the dimensiona1
compatil;:li~ity
matr'ix (5 x 6)
'.
of the forro:
""
[CJ;]
1
0
O.
0 '1
<7
Equation
(47)
(48 )
shows how the dimension
inco~pati
bi1ity
4
tion at
difficulty in matching the final and initial condit :P is overcome.
b:Q
and from the (T/6)  transposed periodi:city, one can derive.
"\
k:::~, ~l(O),
the initial conditions of subinterval
,<the final condition.s of subinterval
from/
2, ~2(T/6).
From th~ periodicityin  T requirement, ~equation
1
1
o
.~~
36
"
~~
'
Furthermore, from Appendix A , the statevector
of subinterva1
k = 12 has the following forros:
~12 (t) T
;I
i3 (t) ,il (t) ,i g (t)',i rt) ,iJ (t) ]
, (410)
,
Equation (49) at tT may be partltioned as follows:
~
~13
(T)
[~fl~:~]'
=,
Id
( 41~)
"
w~ere
.,
'elements of
,resents
the term
represents
th~irst\five
~13 (T). The 1ast row in equation (411) rep
the element lC(t) of equation (49) at
= Id
t=T. Thu5
Id because att=T, the ~a1ue of
this last element is
ic(T )
~13(T)
\,
fram Fig. 33.,
Now, using the
1
~
con~inuity
requirement at t=T
one has:
(412)
~h~re the matrix
[Cs J i5 needed
of the currents of, ~12(T)
,
ta match with those of
~13 ( 0 ) .
~
This matrix i5:
1 1
[Cs]
'
"
0
..... 
in rearranging" the or der
0'
,0
_\0
"
,,
1
~
(413r':
37
r
(r
At
= 0,
equation
~1 (0)
may be
partit.i~ned ,las
\
fo11ows:
(414)~
\'
5
?5.1(O)
where the term
of
J
~1(O).
represents the first fiv& e1ements
The 1ast row in
eq~ation
(414) is
Id
because
iJO):::~Id priori to the firing of V4.
1
App1ying continui tyin  T
and
requirement, at
=0
one obtains:
t ::: T
(415 )
!(O) = ~13 (T)
,r
( ']
One fur"ther substitutes (412) into (415) to
get:
,
~
:s.f(O) =
f (
[~~]
~12 (T)
(416)
"1
Now from the transposed periodicity in T/6 reequation (32) one has:
quir~ment,
'1
1
?5.12 (T)
=!2 (T/6)
(417 )
/'"
So that the equation (416) is rewritten as:
\
:s.i(O) =
[cs] ~f(T/6)
(418 )
o
/
'
38
As a summary of this section, the initial and
final conditions to he used are:
i
a)
b)
at
at
t
t
=0
= f.L
,!i<O)
~2 (,u)
= [cs']
= [cp]
!2(T/6)
(419)
~:
?!.l (Il)
(4:20)
In section 4.4 these results are used in order
to find an 'algebraic function of
!.l (0)
a:s a function of 1'
only.
,,
i .
4.4
Determination of 1' \,
Using the two sets of ini tial condi tio'ns deri ved
1
f
in section (4.3), that is equations 1 (47) ,and (419), an
')
is derived as a function of # alone.
a.
Furthe
solved by a
of
hqw a nonlinear equation in,u, F(,u) is
formula, in order to find the value
f
il
()
\
"
39
(:
4'.4.1
hlgebraic Expression of
Having matched the initial and final conditions
of subinterval
k ': land 2 as shown in equations (47) and
(419), one is further interested in finding explicit solutions to equations(41) and (44). The first step towards
this
g~al\is
~~(O)
to express
as a function of
alone.
This is done as follows:
a)
k = 2
The solution of subinterval
at
=p.
is obtained from eguations(47)and (43) as:
"2 (~)
[Cp ]{ "16 (~)
th (p) ( [~~~~:]  "ls (O)}
(421)
The statetransition matrix[1P (,u>] may be parti1
(
tioned
as follows:
5
5
\
[~l(P)J,
C (,u)
11
(422)
 
C (,u)
21
.....)1
1
1
1
22
(p.)
__
~
( 423)
where
Id[C12(,u)J~[CllUL)~C12(P.)J~lS(O) +[CPJ~lSI(~)
(424 )
r'
;;,
,r
>
40
i
b)
From the initial condition
5
~1(O)
an expression of
'\
i)
Substituting
= T/6
~2 (T/6) = ~2s (~/6)
a~
= 0,
,1
one obtains
as fo1lows:'
1
:,
in equation, (46) one has:
+ [ C1>2 ('!l/6P)]
{~2{P)
~2S{P)}
l'
(425) ,
ii)
~2(~)
Equatioh (423t is substituted for
and
the resulting equa~ion (42S) is substituted on
\
the right hand side of equation (419) to get:
!>i (0)
=[
cs]
{!>2 S
(Ji)
(~/6) . +[ 2(T/6">J
~2 ~ LU)}
Thus rearranging equation
~i (O)
=
x
([cu (P>] !>~
(4  26)
~426):
Jr
~ [I} [C S 2(T/6!l~ [ Cn (p~ f 1
[cs]
j1!.2S (T/6) + {'''2 (T/6 "
Id [C 12
 x 2s
("~
(P)] 
(0)
[C ll
(~)
: C
12
j ( [cp]
(p)]
!>ls (p)
::ls (O)
'(427)
The above system of equatioJs is a function of
i
on ly one unknown, the conunu ta tion time, P _ One is further
~nterested
in finding an explicit equation in p only in a
Il
form such that one may solve that equation in arder to find p
...
,,_
....
~~_
..
_.__. 
. ,,
"
, 41
,
4.4.2
A1gebraic Exprssion of
(~)
It should qe reminded that
commutating current
Furtherrnore
i5 the time tht the
i c (t) (see Fig. 23) , goes to zero.
(t) is the 1ast elemnt of the statevector
c
1. Thus, ohe
~l (t) (see equation 42) in subinterva1 k
defines ian equation
F (fl)
to be equa1 to the bottom row
of equation (43).
"', F (pl = [OOOOOlJ{~lS (Ji)
Q\ (p),
[!~(O)]
=~~(428)
5
where, as already been stated in eguation (427),
~1
(0)
is a function of ~ a1one.
The problem now reduces to solving for Ji in:
F(,u)
Since
=0
F(,u)
(429)
ls a nonlinear function in
Ji
.,
it
Jt
, has to be solved iteratively as discussed in section 4.5.
"
:,
';
'!
4. 5
Newtontype
Algori thm .
"
Thealgorithm to be used in so1ving
F(p)
i5 shown
,e~
l'
i
graphically in Fig. 41, ~~
l!
\;!
i
~
,~
~
'\
"
42
Fig. 41
Newtontype
C)
algo~ithm.
It is necessary to assume that the equation
(429) has a solution and that the Newtontype recursive
F(,u} ='
formula conv.,erges to the solution
o.
The formula used is:
= ,Il 1T 1
wh~re
,u 0
P.o
(430)
F(,uL+l)  FUlj),
 f' (~i of 1) ....k"'+...;l"_L_ __
t
1,2,3 ...
is the iteration number.
In the next chapter, the
ex~ple
discussed up to
hers is implernented on the digital computer. It will be
shown that in fact the Newtontype algorithm has a very fast
convergence to ,u.o~
,,
l
1
43
f
1
CHAPTER,
RESULTS
In order to dernonstrate, its effectiveness, the
for
"sernianalytical" ~ethod developed is applied to solve
the steadystate of the il1ustrative exarnple shown in Fig.
, 1
21. The following points are
a)
highlig~ted:
Rapid Convergence of the Newtontype algori thm
used in solving for the commutation time,
b)
Correctness of the "semianalytica1" method.
This is established by showing that it predicts
the saroe commutation ang1e,w~
a numerica1 integration
t
1
m~thod.
as the prediction based on
In the numerica1 integration
methC2.d, the steadystate solution is reached
after inteli
grl,ating for a sufficiently long time sc that ail the tran ,,l'
sient components
have subsided. '
,
The inverter circuit, Fig. 21, under examination
h~s
the fo11owing parameters:
\
i)
Constant
de
current
Id"" .5Amps.
ii)
Transformer:
= .0531
LT
R.r=
iii)
LOad:
H.
\
. 0005 ohms .
R ,.. 425 ohms.
= ,.6986
"
H.
"', .
','
44
r'
!,
1
i
i
iv) Synchronous
LS= .0796 H.
machine~
RS'" . 001 ohms,
E
,m
= lO~ Volts
a =211"/3 radians
= 120
degrees.
where a is the firing angle of each valve, which for
"
an inverter lies within the range of:
<
11"
2'
5.1
<
11"
Convergence of the Newton~type Algorithm
,
The effectiveness of the "semianalytical" method'
discussed in this thesis is partiy due to
th~
rapid conver
gence of the recursiv formula. The Newtontype algorithm
(equation 430) used in solving the nonlinear equation in
\
~,
F(~),
is based on two initial values' which are:
w~
'1
11"
lB radians
11"
w Il 2
= '9
= 10
de~rees
radians = 40 degrees
It should be noted that
Il
must lie within the
, range:
o<
C)
<
11"
(51.)
45
As is well known
WfJ>..!.
occurs for,
[31J
commutation
fai~ure
The' two initial values employed
wer chosen with the constraint of equation (51) in mind.
1
10
,
f
l,
t,
2
Fig. 51{a)
()
...
"""~_ __._~,.._.~
c?nvergence of F(fJ) towards zero as a function
of the i teration number, L .
l'
1
.,1
:.
()
'J,\
(initial guess,
w~2)
,, \ ,
,1
1
1
,,
,,1
,
,
,
\
,,
\
\
\
\
1
1
\
\
1
1
,,
,
,
,
,
,
1
,i
,
,,
,,
.~
T
Fig.S1 (b)
1.
2.
Commutation angle, WIL, against the iteration
number~i.
o
)
47
()
Fig.
5~
displays the rapid convergence of the
I!
Newtoritype'algorithm. Fig. 5l(a) and (b) plot respectively
the function,
F(~),
iteration number,
and the commutation
l.
angle,w~,
against the
From these results, 'it is evident that
a very high accuracy is obtained within three or four iterations.
,
(
At this stage one may wonder as to the correctness
of the method. One asks for a basis against w~ich the results
may be checked.
In the section 5.2 a comparison i5 made
fetween the "semianalytical" method and a purely "numerica,l"
method.
5.2
"
Numerical Correlations
1
As seen in Fig. 51, the "semianalytical" method
results in a very rapid convergence of
1l
~,
F(f) towards zero.
Thus after only four iterations the 'value of the commutation
time,
~,
is found to be acceptable.
In order to check the ,correctness of the "semi
analytical" method, the problem is solved also by a traditional method. This method is based on the following steps:
~
The initial conditions of the ,system are\guessed.
2.
Numerical integration is perfc;>rmed w!th a step
s'ize, h:w.t=2 degrees, until a value of the commuta tion time, p., is reached. Th'is is the time
where reversaI of the phase current, ic(t), occurs.
48
3.
The final conditions of one
compl~te
cycl'e are
found, using the (T/6) transposed periodicity
requirement.
4.
5.
These final~conditions
~13{T)
ttie initial conditions,
xl (Or.
are compared with
f
New initial conditions are obtained from ,the
final conditions.
The steps (2.) through (5.) are repeated until
\''
the peruni tized closure norm:
lI~l (0)  ~13 (Tjl/ Il ~l
is acceptably small. In the test the value of
(52)
(0) Il
is:
'
.01
Of course,this
, method is more suited for the
,~
""
study of transients: It represents the behaviour of the
system
unde~transient
conditions. In order to arrive at
the steadystate solution, it is necessary to run the program for a sufficiently long time, so that the transient
part of the
solutio~,
r'
dies out. Thus this method is inher,
ently expensive.
The,problem was solved in the context of the
commutation angle,wp, versus the firing angle,. The switching
tion
(J
ang le,
fai~ure
Cl'
was varied from
occurd,Le.
wp
n /2
radians until commuta
)n/3 radians. Fig. 52 shows
the study of WIJ, versus optained in twodifferent ways:

.1
/
o
49
/
~I
a)
The "semianalytical" method is represented by
the full line.
b)
The numerical method is represented ,by the crossed points.
The slight discrepancy is attributed to having
terminated the numerical integration at an early stage be!
cause of slQw convergence.
Thus, it
rna~
be concluded that the "semia.nalyti,
~
cal" method yields in fact the correct solution of the systern. Yet, this novel method arrives at the solution very
rapidly, narnely arter ,only four iteration steps.
,l,
w J..L
o
60
50
30
20
~;~~~~~_"~I~I~~
o
90
Fig. 52
()
'0
100
120
140
commutation angle,wp , as a fuction of the switching
angle, a. Line: "semianalytical" method.
Crossed points: numerical lntegration.
~~~~~
..
~.~~~'"
 ._...,_'".....
, ...
50
JI.'
CHAPTER VI
"
, ;
,,
..
LL
CONeLUSIONS
A new and fast semianalytical method for solving
1
Steadystate" problems i.n HVD systems using the "digi tal computer has been presented.
The rapid convergence of the
method has been tested. Its correctness has been verified
against the results of a numerical integration method.
To the best of the author's knowledge the method
is a new and original contribution ta knowledge.
The method overcomes the basic shortcoming
conv~rge
the nurnerical integration in not being able to
""
of
,
J
;;:
~
to
wards the .steadys,tate sol.ution rapidly. As an extensive
>
part 'of any analysis is concerned with the steadystate "SO
~
, f
'. f
lution, 'this semianalytical technique will contribute :O
1
1
wards making the digital computer into a viable alternative
to analog
simi~ators.
Further work in the development of the semianalytic~l
a)
method should consider:
Automatic compilation of system equations from
E1ystem <~ine diagrams of any confi<iuration.
b)
Unbalanced phases.
c)
Fault conditions.
'd)
Multiterminal HVDC
systems.~
.'
..
{
l
)
 l'"
51
1
1 /
APPENDIX
A
o
CIRCUIT CONFIGUAATt'ON AND STATEVECTORS
IN THE
A.l
Subinterval
l2 SUBINTERVALS
=1
(Al)
V3
J
V'l
('l'
Fig. Al
1
r
A.2
"
Circui t
Subinterval
k= 1.
configura tian dring subinterval
k
=2
I
[
!2 (t') T
"1
= [i2
(t),
i3 (t), 'iS
(t>, i7 (t),
i9
(t>]
(A2 )
1i
' '
"14
()
Fig. A2
Circuit
configuration during 'subinterva~
1
jo
k=2.
52
1
A.3
Subinterval
=3
~3 (t) T =
[il (t), i 2 (t), i (t), i (t)
7
9
.f
iS (t), i (t)]
b
(A3)
1/
A.4
Circuit
Fig. A3
Subinterval
configuration during
subinterval
';
"
k=3 _
~4 (t)
=4
(A4)
t'
f
V4
Fig,_ A.;.4
1
Circuit
configura1tion during, sUbinterval" k.4.
Or
  
..,

~
 
~
._
1
j
;:;i.~ ___ ___ ._
53
A.5
Subinterval
k.= 5
(A5)
,
 eo.. +
Ls
\
1
Fig. AS
Circuit
configuration during sUbihterval
!
J
A.6
Subinterval
k: 6
k:5.
1
" f
,1
~6 (t) T =' [i~_~~L~ iJ~~_~(~LJS<~)' i:iCt~1
(A6)
Fig. A6
()
Circuit
configuration during subinterval
k.6.
54
()
A.7
Subinterval
=7
/11
(A7)
vs
 eo.. + Ls
V6
Fig. A7
A.
Circuit
Subinterval
configuration ,during subinterval
k=7.
k = 8
~8(t)T ~[i2(t),
i (t), is(t), i 7 (t),
3
~i9(t)J
(Aa)
Vi
Fig. A8
()
.Circuit
\ \
configuration during subinterval
k=8.
"
55
A.9
Subinterval
=9
1
;
~9 (t) T =
[ 1 (t)',
i 2 (t), i 7 (t), i 9 (t), i 5 (t), ib (t)]
(A9 )
Fig. A9
Circuit
qonfiguration during subinterval
,
A.1D Subinterval
k=9.
= 10
(A..,.lO)
VI
Fig. AlO Circuit
()
configuration during subinterval
k=lO.
56
A.ll Subinterval
k: 11
(A11)
V3
,,
,\
'1
,~
1
FIg. A~ll Circuit
configuraton during subinterva\ k=ll
A.12
Subinterval
~'12
~12 (t) T :;: [i3 (t), :ii (t), i 9 (t), i 5 (t)" i 7 (t>]
(A12)
L,. R,.
V3
V2.
~ig.
A12 Circuit
configuratio~
during
subinterv~l
k:12.
()
,
1
1
1.
"
 ~"'""~""_+1""o;.., ...~~~~ ~~..1.....':If;<,
,,
57
Cj
APPENDIX
PARTICULAR INTEGRAL
In the Kth subinterval, the linear differential
syst;ern of equations describing the net~ork is:
11
(Bl)
where the forcing function
~k
is rendered in the
(t).
form:
(B2)
where:
1
~
(,
E
i,
1
!!.k
In
sin
Em cos
E
In
(B3)\
sin (5 n /6
Em cos (n/6 +
~k
Em sin
Em cos \ (5 1f/6
(B4)
a)
()
I~
'
,
58
o
o
~=
(B5)
The ,particular integral
~s(t)
of the equation
(34) is of the for.m:
!ks (t)
cos
.9:k + sin
w t
t ~ + ~k
(B6)
where the constants
~k' ~k
and
~k
are of dimension nk
These constants are found by substituting the eqution (B6)
into the eguation (Bl), and eguating the coefficients af
the left hand side and right hand side of the r,esul ting
()
~k'
equation. Thus
~k
and
~k
are the solutions to the
linear simultaneous equations:
~
 [A.Jl
(B7),
k
(B8)
The inverse matrices involved are calculated
using the subroutine, called
LEQTIF, available 'at McGill
University computer library. The subroutine is in the IMSL
bibrary l Fortran IV System /370360.
(J
~"
...'
59
APPENDIX
COMPLEMENTARY FUNCTION
The
homogeneo~s
part of the solution of equation'
D
(34) is composed by the statetransition fatrix [r.ilk<t)] and
the unknown integration constants ~k: The statetransition
matrix is calculated as follows:
,
[ tr>k <t)] = exp
[~Jt
(Cl)
or
i'
[ r.ilk(t>l= [
where
~Jexp [AkJt[MkJl
(C2 )
.~
[~J is the modal matrix which cmtains the eigepvectors
1
of the matrix
[~}
columnwise. [AkJ is a diagonal matrix
containing the eigenvalues of the matrix [~J. The matrices
[~J and [Ak] satisfy the following equation:
(C3 >
t,
~.;:
.,
;
In the problems where
[~](k=1,2)
are not
chang~d,
the evaluation of the eigenvalues and eigenvectors is an
effective way of computing the statetransition matrix through
equation (Cl) .
, The eignvalues and eigenvectors of
~e matriX[~ ]
are calculated numerically usihg a subroutine, called EIGRF,
avail~ble
at McGill University computer
li~rary.
The sub
routine is in the IMSL Libtary l Fortran IV System /370360.
()
)1
I_~____~ \___
\
/
1
1
, i,
In system parametric studies where
[~J (kl,2)
a're altered at each parameter chahge, alte.rnate methods tnay
have to be considered. A discussion of the\comparative rnerits
ll
of various nurnerical rnethods and an extensive bibliography
appe~rs in the review paper [34].
1t
o
\
\
o
. ,
o
\'
61
( ',
D
APPEN'oIX
DERIVATION. OF
D.l
~TRICES ~]_, ~kl
Derivation of the matrices
AND THE VECTOR .!!k (t)
~lJ'~lJ
and the Vector u (t)
1
When three valves are conducting
k l, Fig.. A  i.
i~
the subinterval
,from KVL and KCL the following system of
equati9ns is written:
where:
2L
L
L
L
o
o
L/R
L/R
L/R
o
o
o
o
L/R
o
o
o
o
o
o
o
o
o
R
R
()
L 5
2L
,1
(D2)
(03 )
\
~
... _ _ _
____
_ _ _ _ _ _ ......., _ _ ....... \0<_
~_.,...
. _ , .......... "_"'~_" ________ _
'
~~.J_
62
~1 (t) T
= [i2 (t), i3 (t), i5 (T), i7 {t), i9
(t.),
ic (t)]
(04 )
[G1]
}!1 Ct)
1
1
1
1
1
1
(05)
R.r
I(
[e a (t)
e (t), e c (t) ,
b
Id]
1\
where:
~
Ct') = E sin (wt +7t/6 +
m
Cl' ),
t "
eh Ct)
e
(06)
(t) :,E
' m
sin (wt  7t/2 + Cl'
+ Cl'
sin (wt, + Snj6
(07)
"
Equation (01) may he rewritten as follows:
!.1 (t)
= [AIJ~1 (t) +
where:
r
[Al]
[B 1]
~l(t}
1
=
[Lll
(08)
(I:~~9 )
[R1]
1
[Bl] = ,[Ll] [G1]
(010)
~
"<;)
"
'
63
i
1
t
i
'1
The inverse of the matrix [L1] is calculated
using a subroutine, called LEQT1F, avai1able at McGi1l
Universi ty compu1ter 1ibrary The subroutine is in the IMSL
Library 1 Fortran .IV System /370360.
./'
0.2
Derivation of the Matrices
!!.2 (t)
and the Vector
[A 2]' [B 2]
When two valves are conducting in the subinterva1
'\
k.2, Fig.
I\'l.
, from KVL and KCL the following system of
equations is wri tten: .
dt
[R2J!2
, [LJ9. !2<t)
:;1
(t)
[GJ}!2 ('t)
(D11)
where:
()
,
[L~ =
Ls
2Ls
L
L
s
s
L
L
L/R
L/R
(012)
c.
0
1(
Rs
R
R
o 1
0
0
>~
q
0
R
R
[R2]
.. \
(D13 )
i \ C)
\
\
.
\
1
1
1._____________._
Il
\
1
1
.j
~2 (t) T =
[i2 (t)" i3 (t)., iS (t), i7 (,t) , ,i9 (t~
1
\.
,,<,
."
j
[G 2]
,
!!2
1
"0
R
R
0
1
(t)?' ~ [e a
,1
""'\
...,
(DlS)
J
(t) "
'"
e b (t)
"
e c (t)
Id]
, :
(016)
fil>
whre e (t) , e (t) and e' (t) are as decribed in equation (07) .
a
b
\
c \
\
l
~
. Equation (011) rnay be rewritten as follows:
,,
/
~2 (t)
(.
= .[A2]?i2 (t)
+ [B2J.~2 (t)
(017)
. where
T,
J'
1
[A 2]
[BJ
<l
  [L2] [R2]
(018)
= [L~1 [G 4]
(D19)
'j
1
\
,
J
"
"
l1
 '
1
)
,
1
,
'.
w~.'~~~~i~.iii~W,l~'ft:f74_~
l'
65
,
REFERENCES
IITransmission Line Reference Book HVDC to :t 600 KV",
Published
Ca 1 iforni
EPRI, 3412 Hillview Ave., Paolo Alto,
a.
G., A.' Hofma, G. L. La Barbera, N. E. Reed, L. A. Shillong,
liA High Speed HVDC Circuit Breaker with crossfield
Interrupters", IEEE
[3]
T~ans.,
PAS95, 1976, pp 11821193.
J. Reeve, J. A. Baron, P. C. S. Krishnayya, "A General
Approach to Harnpnic Current Generation by HVDC ConJerters",
IEEE Trans., PAS88, No. 7, July 1969,' pp. 989995.
1
1
~. W. Huddart, W. G. Watso~, "Multiterminal d.c. Transmission",
1l,
in "Hlgh Voltage dc Transmission" IH Conf. Rep. Sere 22,
1966, pp. 9498.
[5J
,R. Foerst, G. Heyner, K. W. Kanngiesser, H. Wa1dmann,
IIMu 1 ti termi na 1 Ope rat i on of HVDC Converter Sta ti ons Il ,
IEEE Trans., PAS88, 1966, pp.
l042~1052.
J. P. Norton, B. J. Cory, "Digital
Simulation 1 Program for
,
Multiterm;nal HYDC, Systems", IEE. Proc. 1967, No. 114,
pp. 19171924.
G. K. Carter, C. E. Grund, H. H. Happ, R. V., Pahl, "The,
!
0
Dynamics of ACIDC , Systems with Controlled Multiterminal,
HYDe Transmission", IEEE Trans., PAS96, No.2, March/Apr.
1
1977, pp. 402410.
1.
( )
66
J
(
[8
J,
J. Mi1iasArgitis, G.
Giannakopou~os, G. Galanos, "Dynamic
,
"
Simulation for Multitenninal HVDe Systems", IEEE Trans
, PAS97. No. 2;
, March/April J978, pp. 587593.
[9 ]
J. L. Hay, N. G. Hingorani, "Dynamic Simulation of Multi
converter HVDC Systems by Digital Computer", IEEE Trans.,
\
PAS80, No. 2. Feb. 1970, pp.218228
J. Rerve, "Multitenninal HVDC Power Systems", IEEE paper
, i
1
~nted at PES Winter meeting in New York, Feb. 38.
l'
1
1980, F80 1875.
[11]
H,. Martensson. T. Adie1son, "Simulator studies, of mu1ti~
terminal hvde systems", in "High Vo"ltage de tran.smission"
IEE Conf. Rep.
()
~er.
22,1966, pp. 115118.
,L. Bergstrom, L. E. JiJhlin, G.
study of, Multiternlinal HVDC
liS;'
SYS~I
G. A. Nunan, "Simu1ator
Perfonnance", IEEE
'1
Trans., PAS97. Nq. 6. Nov/Dec 1978. PP. 10511067.
J. P. Bow1es;' H. L. Nakra,. A. B. Turner, "A smal1 series
tap on an HVDe 1ine": Paper F80 2709, IEEE Pts Wi nter
Meeting 1980
N~w
York.
H. Sato. J. Arrill'aga, IIImproved load flow techniques for
integrated acde systems", Proe. IEE. Vol. 116, No. 4,
April 1969, pp. 525532.
()
'\
1
J
"J
_,,""
,,~_, .c~"'" d'"~,,!....'I"oV"~ _~.; ~ tr;~"""",IO'li",~~...,.~"..,.,t::!.'<::
67
{15 J
'J. Reeve, G. Fahmy, B. Stott, "Versatile Loadflow Method
for Multitenni na 1 HVDC Systems". IEEE Trans., PAS96,
No. 3,
[16J
'
'.
'
May/Jun~
1977, pp. 925933.
J. Arr111aga, P. ~odger~ "Integration of HVDC links with
fast deeoupled 10adf1ow solution", Proe. IEC Vol. 124,
\
No. 5, May 1977, pp. 463468.
, [17J
B. J. Har:ker, J. Arrillaga, "3phase ac/de 10adflow",
Proc. IEE, Vol. 126, No. 12, Dee. 1979, pp. 12751281.
. [18J
N. G. Hingorani, T. L. Hay, R. E. Crosbie, "Dynamic
Simulation of HVDC Transmission Systems on bigita1 Computers",
. Proc. IEE. Vol. 113, No.5, May 1966, RP. 793802.
[19J
1
C)
N. G. Hingorani, R. H. Kitchin, J. L. Hay, "Dynamic Simulation
of HVDC Pow~r Transmission Systems on Digital Computers Generalized Mesh Ana1ysis Approaeh", IEEE. Trans., PAS87,
No. 4" April 1968, pp. 989996.
[20J
S. C. Katiyar, S. M. Peerman,' T. S. M. Rao, "Dynamic
Simu1ation\of HVDC Systems
o~
Digital Computer", J. Inst.,
Eng (India), E1ee. Eng. Div., Vol. 51, pt EL4, ND. 8,
April 1971, pp. 206211.
[21J
Y. Murakami. N. Kosaka. M. Nishimura. N. Sakuma, "A Simu1tion
program for Thyristor Circuits and its Applications",
Eng. Japan, Vol. 91, No. 4, 1971, pp. 5159.
~lec.
68
J. S. C. Htsui, W. Shepherd, "Method of
Digital Computation
;:
of thyristor Switching Circuits", Proc. IEE, Vol. 118,
~o.
8,
Aug. 1971, pp. 993998.
[23]
N. Matsui, N. Kidera, Y. Tsunehiro, Y. Adachi, "Ana1ysis
of Rectifier Circuits by StateVariable Method". Elec. Eng.
~apan,
Vol. 92B, No. 5, 1972.
"
P. K. Dash, R. W. Menzies. R. M. Mathur, "Digital Simulation
of a Rectifierloaded'Synchronous Machine'" E1ec. Machines
/
and Electromechanics: An International Quarterly, '1977, Co.
1977 by Hemisphere Pub1ishing Corporation, pp. 281294.
A. A. AbdelRazek, M. POloujadoff, "Ana1ytical Approach to
the Operation of a Synchronous Machine Associated with a
\
Thyristor Bridge", Electric Mahines an/d Eleetromechanics:
An 1nternati\ona l Quarter1y, 1978, pp. 167184.
,
J. J'. Aprille, T. N. Trick, "Steadystate ana1ysis of
non1inear'circuits with periodic inputs", Proc. IEEEE,
Vol. 60, No. l, Jan. 1972, pp. 108114.
\
M. S. Nakhla, F.. H. Branin, Jr." "Detennining the periodic
response of nonl inear
sy~)~~ms
by a gradient method",
Proe. 12th Ann. Allerton Conf. Circuit and System Theory,
Uniy. Illinois, Chicago; 1974, pp. 703704.
'00:;
\ )
\ 1
l
\
>
,
",~.r ~~ ;~, ,"1... ,"*;1,' ,(;I1>\.....r~~!:,''Y,')r~ J4k..,.~\...'r
69
c
"
J. C. Lindenlaub, "An approaeh for finding the sinusoidal
steadystate response of nonline'ar systems", Proe. 7th
Ann. Allerton Conf. Circui~. and System Theorj,
udiv.
Illino;s, Chicago 1969.'
M. S. Nakhla, J. Vlach, "A piecewise harmonie balance
technique for determination of periodic response of
nonlin,ear systems", IEEE Trans. on Circuits and Systems,
Vol. CAS23, No. 2, Feb. 1976, pp. 8591.
Samford M,. Roberts, Jerome S. Shipman, )\Twopoint Boundary
Va 1ue Prob lems: Shooting methods", Ameri can El esevier
publ. Comp., New York, 1972.
E. W. Kimbark; "Di rect Current Transmission", Vol. 1,
Willey Interscience, 1971.
Ct
R. Saucedo, E. Shiring; "Introduction ta continuous and
\
digital control systems",
Ma~Millan,
New York, 1968, p. 596.
C. R. Wylie, Jr., "Advanced,Engineering Mathematics",
1,
McGraw Hi 11, 1951. p. 495'.
C. Moler, C. Van Loan, "Nineteen
dubio~s
ways to compute the
exponential of a matrix": S,IAM Review. oVol. '20" No. 4,
1
(
Oct. 1978, pp. 801836.
B. 1. Ooi. N. Menemenlis, H. L. Nakra', "Fast SteadyState
1.
Solution for HVD,C'Analysis".
1
Paper submitted to IEEE.
Feb. 1980.
()

.~
_.
_~~~'I~;
...
"""~~
. ."'..,:t_....
.......,~
._ 
~~__;""'~4r~
.... .ii,:("t.I>'t;",
~ i.!".,r~~~~