Beruflich Dokumente
Kultur Dokumente
DIRECT CURRENT
TRANSMISSION
Volume
Inc.
Volume
Volume
ii
ni
Synchronous Machines,
fI:
WILEY-INTERSCIENCE
a Division of
John.Wiley &
Inc.
New York London SydneySons,
Toronto
PREFACE
s
'
'
Copyright
j
s
1
Printed in the United States of America
10 9 8 7 6 5
:-1
:n
i
VI
PREFACE
time render this book obsolete, but rather the continuing rapid development
of all phases of the art.
My interest in direct current transmission was awakened in 1962 when the
Bonneville Power Administration (BPA) asked me to teach two courses in
this subject for their engineers. The lack of an adequate text book for those
courses determined me to write one. Since 1962, Ihave taught three more
courses on dc transmission, and have been involved in studies of various
aspects of this subject. Iam indebted to the BPA for the opportunity to teach
these courses, to work on problems in the field, and to confer with my col
leagues, as well as for access to the BPA's excellent library services. However,
Iwish to make clear that this book is not an official publication of the BPA
nor one sponsored by it. It has been a spare-time project. Ialone am respon
sible for its contents, including any errors which may inadvertently appear in
it.
The large amount of essential information now available on direct-current
transmission and the time required to organize it led to the decision to divide
the work into two volumes of which this is the first. The proposed contents of
the second volume are indicated on page xi.
Units of physical quantities used herein are those of the International System
(SI) recommended by the I.E.E.E. and I.E.C.
Iam indebted to various engineers at the BPA and elsewhere for supplying
information, especially to Dr. John J. Vithayathil for many enlightening
technical discussions. Iam indebted to my wife, Ruth Merrick Kimbark, for
typewriting much of the manuscript and pertinent correspondence and for her
valued advice and encouragement.
CONTENTS
1-1
1-2
1-3
1-4
1-5
1-6
Historical Sketch
Constitution of EHV AC and DC Links
Kinds of DC Links
HV DC Projects from 1954 to 1970
Limitations and Advantages of AC and DC Transmission
Summary of Advantages and Disadvantages of HV DC
Transmission
Principal Applications of DC Transmission
Economic Factors
The Future of DC Transmission
1-7
1-8
1-9
Bibliography
1
9
11
12
19
32
32
33
35
36 .
2. CONVERTER CIRCUITS
49
Valve Characteristic
Properties of Converter Circuits
Assumptions
Single-Phase Converters
Three-Phase Converters
Pulse Number
2-7 Additional Six-Pulse Converter Circuits
2-8 Choice of Best Circuit for HV DC Converters
2-9 Twelve-Pulse Cascade of Two Bridges
Problems
Bibliography
2-1
2-2
2-3
2-4
2-5
2-6
vii
,
/
49
50
51
51
56
61
62
65
67
68
70
Viii
CONTENTS
4. CONVERTER CHARTS
4-1
4-2
4-3
Problems
Bibliography
5. CONTROL
5-1
5-2
5-3
5-4
5-5
5-6
5-7
5-8
Grid Control
Basic Means of Control
Power Reversal
Limitations of Manual Control
Constant Current versus Constant Voltage
Desired Features of Control
Actual Control Characteristics
Constant-Minimum-Ignition-Angle Control
Constant-Current Control
Constant-Extinction-Angle Control
Stability of Control
Tap-Changer Control
Power Control and Current limits
Frequency Control
5-9
5-10
5-11
5-12
5-13
5-14
5-15 Multiterminal Lines
CONTENTS
71
73
80
92
103
105
112
115
123
126
129
129
138
146
146
147
148
148
152
153
154
156
157
158
187
192
194
Problems
Bibliography
198
6. MISOPERATION OF CONVERTERS
7. PROTECTION
7-1 General
7-2 DC Reactors
7-3 Voltage Oscillations and Valve Dampers
7-4 Current Oscillations and Anode Dampers
7-5 DC Line Oscillations and Line Dampers
7-6 Clearing Line Faults and Reenergizing the Line
7-7 Circuit Breakers
7-8 Overvoltage Protection
Problems
Bibliography
IX
198
199
206
220
222
227
228
229
230
231
233
234
235
235
235
247
260
270
272
280
282
291
292
164
165
167
174
179
180
182
183
8-1
8-2
8-3
8-4
8-5
Summary
Characteristic Harmonics
Uncharacteristic Harmonics
Troubles Caused by Harmonics
Definitions of Wave Distortion or Ripple
295
295
296
318
323
325
CONTENTS
332
333
343
386
386
391
391
393
417
419
423
443
445
465
476
478
APPENDICES
484
9-1
9-2
9-3
9-4
9-5
9-6
9-7
9-8
11. DC CABLES
12. FORCED COMMUTATION
13. OPERATION OF A DC LINK AS PART OF AN AC SYSTEM
INDEX
496
i.'
xi
ABBREVIATIONS
A
ac
A.C.S.R.
A.E.G.
A.G.
Ah
Amer. Power
Conf. Proc.
ASEA
Assn.
AWG
B.E. & A.I.R.A.
B.I.C.C.
B.I.L.
BPA
B.T.S.
Bull.
C
CAB
cal/g
CAT
C.C.
C.C.I.F.
ampere
alternating-current
aluminum cable, steel reinforced
Allgemeine Elektricitatsgesellschaft
Aktiengesellschaft
ampere-hour
American Power Conference Proceedings (Illinois
Institute of Technology, Chicago)
Allmanna Svenska Elektriska Aktiebolaget,
Sweden
Association
American Wire Gage
British Electrical and Allied Industries Re
search Association (later known as E.R.A.)
British Insulated Callenders Cables
basic insulation level
Bonneville Power Administration (U.S. Dep't.
of the Interior, Portland, Oregon)
Bell Telephone System
Bulletin
coulomb, Celsius (temperature scale, formerly
Centigrade)
consequential arcback
calorie per gram
consequential arcthrough
constant current
XIV
ABBREVIATIONS
ABBREVIATIONS
C.CXT.
C.C.I.T.T.
C.E.A.
C.E.G.B.
C.G.E.
C.I.G.R.E.
cm
Conf.
const.
cos
cosh
cot
coth
CP
cosine
hyperbolic cosine
cotangent
hyperbolic cotangent
Elek. Stantsii
U.S.S.R.
electromotive force
Engg.
Engineering
Eq.
Eqs.
equation
equations
Electrical Research Association, Great Britain
E.R.A.
ETZ
E.u.M.
exp
F
Fe
Fe+ +
Fe(OH)2
ft
Gen. Elec. Rev.
GW
H
H
H+
H2
h
HV
csc
cosecant
dB
dc
deg.
Disc.
decibel
direct-current
degree (of angle)
discussion
free electron
Edison Electric Institute, New York
extra high voltage
Electrical
electrical degree
Electrical Engineering, formerly published by the
A.I.E.E.
Electrical World
Hz
ibid.
e"
ehv
Elec.
elec. deg.
Elec. Eng.
Elec. World
EMF
E.E.I.
XV
Id.
I.E.C.
I.E.E.
I.E.E.E.
J
Jour.
K
kA
kg
Elektrotechnische Zeitschrift.
Elektrotechnik und Maschinenbau (Vienna)
exponential function
farad
iron atom
ferric ion
ferric hydroxide
feet
General Electric Review
gigawatt
henry
hydrogen atom
1
Xvi
ABBREVIATIONS
kHz
km
kV
kVA
kvar
kW
k$
lb
lbf/in2
LC
In
log
mA
MCM
mH
MHD
MHz
mi
mm
MMF
ms
mV
MVA
ABBREVIATIONS
kilohertz
kilometre
kilovolt
rad/s
Ref.
natural logarithm
common logarithm
Rev.
Review
RLC
rms
resistance-inductance-capacitance
SCR
milliampere
thousands of circular- mils
millihenry
sec
S.E.V.
magnetohydrodynamic(s)
megahertz
mile
millimetre
magnetomotive force
S.I.L.
root-mean-square
sin
sinh
SW
hyperbolic sine
tan
tangent
T.H.F.F.
TIF
Trans.
Trans, and Dist.
U.S.
U.S.A.
U.S.S.R.
V
vs
versus
number
northwest
New Zealand
negative hydroxyl ion
W
w.r.t.
yd
with respect to
yr
year
Ontario
pP
microfarad
millisecond
millivolt
megavolt-ampere
MW
N
nF
nanofarad
newton
Nauchno-Izsledovatel'skii
Institut Postoyannovo
NW
rad
thousands of dollars
pound
megavar
N.Z.
OH"
Ont.
P.I.V.
Publ.
PVC
kilovar
kilowatt
megawatt
No.
Proc.
kilovolt-ampere
Mvar
N.I.I.P.T.
PA. &S.
XVII
sine
Southwest
Transactions
Transmission and Distribution
United States
yard
xviii
flS
pY
a
Q -m
ABBREVIATIONS
microsecond
microvolt
ohm
ohm-metre (unit of resistivity)
* Superior numerals and, in some chapters, including this one, superior letters alone or
followed by numerals refer to items or. to groups of items in the bibliography at the end
of the chapter.
The advent of the transformer, polyphase circuits, and the induction motor
in the 1880s and 1890s led to ac electric power systems.
The transformer, simple, rugged, and efficient, made possible the use of
different voltage levels for generation, transmission, distribution, and use.
Inparticular, it made long-distance, high-voltage power transmission possible.
The exploitation of water power, usually available at sites distant from
major load centers, gave impetus to such transmission.
The induction motor, especially the polyphase induction motor, is also
simple, rugged, and cheap and serves the majority of industrial and residential purposes. The commutators of dc motors and generators, in addition
to requiring maintenance, impose limitations on the voltage, speed, and size
of these machines. The voltage per bar of the commutator should not exceed
about 22 V lest excessive sparking occur. Thus a high voltage per commutator
requires many bars, resulting in a large diameter. A large diameter requires
a low speed in order that the commutator and windings may withstand the
centrifugal force. And a low-speed machine is heavier and more expensive
than a high-speed machine of equal rating. The advent of steam turbines,
which are best at high speed, gave a great advantage to ac generators.
When ac systems first appeared, there were heated arguments between the
proponents of dc and ac systems. Advocates of dc branded ac dangerous
because of the high voltages used. As a result of their advantages, however,
ac electric power systems became almost universal. Power was generated,
transmitted, distributed, and used as alternating current. If direct current was
needed for some particular purpose, such as adjustable-speed motor drives
or electrolytic processes, alternating current was converted to direct current
locally by motor-generator sets or synchronous converters or, later, by
mercury-arc rectifiers.
The last vestiges of dc distribution were the low-voltage networks in the .
centers of large cities and electric traction (streetcar, trolley bus, rapid transit,
interurban and suburban railways, and some long tunnels or mountainous
sections of main-line railways). Finally, however, low-voltage ac networks
replaced low-voltage dc networks, diesel locomotives replaced steam loco
motives and many electric locomotives, and gasoline or diesel buses replaced
most of the streetcars and interurban lines. Some dc rapid-transit systems still
remain. The victory of alternating current over direct current, however, was
almost complete.
Status of DC Transmission During the Ascendancy of AC Transmission
f
j
j.
"
f
,
f
|
1-1
had four water turbines, each driving four generators of 3.6 kV each. At Lyons
the greater part of the power received by hv direct current was converted to
alternating current and the remainder to 600 V dc for the street railway. The
over-all efficiency was 70.5%, which was considered satisfactory for a hydro
electric system.
In 1911 a second hydroelectric plant at La Bridoire, situated about halfway
along the line and rated at 6 MW, was added (in series). The line current was
then doubled (to 150 A). In 1912 a third hydro plant, located at Bozel, 7 mi
(11 km) beyond Moutiers, and rated at 9 MW, was added, raising the total
generating capability on the line to 19.3 MW. The maximum circuit voltage
became 125 kV and the route length 140 mi (225 km). Operation of the line
continued until 1937, when it was dismantled. Thury himself died in 1938.
The Thury system performed reliably in spite of the large number of com
mutators in series. The limitations of dc machines, already mentioned, how
ever, made it unsuitable to the larger amounts of power that had come to be
required. Further development of hv dc transmission required better con
verters than motor-generator sets.
Development of a Practical Converter
HISTORICAL SKETCH
1-1
ASEA valves.
About 1960, control electrodes were added to silicon diodes, giving siliconcontrolled rectifiers (SCRs), also called thyristors. At present these are not
capable of handling the highest voltages and powers required for hv dc
transmission. Their ratings have increased, however, with surprising rapidity,
and it seems certain that such valves will soon replace mercury-arc valves in
hv dc use.
Experimental DC Transmission Projects and First Commercial Lines
__
The line initially operated at constant current, the conversions from con
stant alternating voltage to constant current and vice versa being made by an
HISTORICAL SKETCH
transmission.
Two hv dc experiments were conducted in Germany during World War II
instance of the German Secretariat for Aviation.86,19 A 400-kV
three-phase line from the Alps to the Ruhr had already been planned, but the
at the
reparations.
In Sweden, where the principal new hydroelectric sites are in the north and
the principal loads are in the south, hv transmission is required ; and, because
of the development of valves by the Swedish firm of ASEA, interest was
aroused in the possibility of a hv dc transmission system as an alternative to
researches/3
1-2
at reduced voltage and power in 1962 and, beginning in 1965, was operated
at its full rating of +400 kV, 900 A, 720 MW.1 Other dc lines of lengths of
2000 to 2500 km and voltage of +750 kV are planned/6
1-2 CONSTITUTION OF EHV AC AND DC LINKS
ehv transmission links, superposed on a lower-voltage ac network, or inter
connecting two such networks, or connecting distant generating plants to an
ac network, are compared as to their principal components and the arrange
ments thereof, according to whether the line operates on ac or dc. The phrase
"transmission link" denotes the transmission line proper together with its
terminal and auxiliary equipment.
Figure 1a shows a single-circuit three-phase ac line. In general, such a line
in the categories already mentioned, one which might be competitive with
a dc link, requires transformers at both ends step-up transformers at the
sending end and step-down transformers at the receiving end although in
some cases they can be omitted at one or both ends. If the transformers are
operated as an integral part of the link, only low-voltage circuit breakers are
required.
Ac
system
system
Ac
system
Ac
system
nverter
Dc line
Ac
system
Ac
system
Ac
system
Ac
system
O-J
-o--c
10
1-3
KINDS OF DC LINKS
11
1
12
Rectifier
CO
"o
o
o. 3
I
03
a
O
D.
O cO
(U
VI
no
no
o
H
I
.S
8
u
o
V
if
On
no
o
o
On
If
u-
On
fS
w
rf
Tf
On
r-
cd
/N
ID
NO
ry,
if
-cO
cO
oo
ry
If
CllO
ID
r11
Tf
ry
if
r-
<D
NO
00
w
<o
stage
00
<o P
'C
SP -o
[S
NO
*0-
CO
CD
cy
CD
ID
r-
CD
CN
ry
'-3
<s>
CO
<o
'irst
t-
CO
"o
a O fl
a 2
<3 j- ,<u
<a
.g
a y
c"N
!s
c3
-o a
X)
<0
<y
h
ID
ID
D
ID
On
00
CD ID
?y
ID 00
CD
60
c
a
e
O
I
os
=
(c)
it-
cS
On
S -S
kj
r'0*
S~
OO
DN
v
IT)
If
ID
00
OO
On
fS
ID
<N
r-
Tj"
ID
CD
<y
CD
ID
OO
<B.
cy
O
N
no
60
CN
(N
ro -0"
Tf
St
3
tH
no
D
D
On
00
II
>
N in
>"
U
Q
>
Cascaded Groups
In each of these kinds of links there are usually several converters connected
in parallel on the ac side but in series on the dc side for obtaining the desired
OA >
CO :0
53 O
level of direct voltage from pole to ground. Each such converter consists of
a transformer bank and a group of valves.
"O
a
a
CO
CO
*o
CO
*a
cO
bb
O
60
<U
>
II
CO
i3
&
a
o
cO
a
*3
VI
>
p
o
>
eI
c3
Ph
c?
a a
o
CO
*3
CO
lej
14
different countries. Brief comments are made on these links, all of which
except Volgograd-Donbass were based wholly or mainly on ASEA
techniques.
Volgograd-Donbass Line1
When built, this was the longest dc line. It usually carries power from a
hydroelectric power plant on the Volga River at Volgograd to an industrial
and mining district in the Donets Basin. For such an operation generators
15
in the hydro plant may be disconnected from the ac bus and connected only
in the opposite direction
the inverter valve groups are connected to the ac bus.
It seems that the link did not offer any advantage in cost compared with an
ac link, but it was built to gain experience in dc transmission for longer higherpower lines that will be built in the future.
Each terminal has eight valve groups (four per pole), using single-anode
valves of Russian design, with two valves in series in each arm.
The year 1965 was called "the dc year" by the editor of Direct Current. Not
only was the Volgograd-Donbass link brought up to its designed voltage and
power, but also two additional dc transmission schemes (New Zealand and
Konti-Skan) and a frequency changer at Sakuma, Japan, went into operation.
A third transmission sdggme (Sardinia) was expected to go into operation, but
it was delayed untiHMlli&owinp year.
to a valve group of the rectifier. For power flow
16
1-4
The valves, manufactured by ASEA, are rated at 1.2 kA, 125 kY, and have
four anodes.
Konti-Skan LinkK
avoided.
2. The dc scheme can be built in two stages, and thus almost half of the
investment can be postponed. The first stage operates monopolarly with one
submarine cable and sea return at a power capability of 250 MW. In the
second stage the line will be a bipolar, metallic circuit for 500 MW, with
sea return used only in emergencies.
Four-anode, 1.1-kA, 125-kV valves are used.
HV DC PROJECTS FROM
1954 TO 1970
17
is the usual direction, although the opposite direction holds when the Sardin
ian plant is shut down. Power flow is regulated so as to keep constant fre
quency on the Sardinian ac system. The valves are similar to those of several
other schemes, are rated at 1.0 kA, 100 kV, and have four anodes.
Vancouver Island Scheme1*
lWO
ac link.
18
Kingsnorth0
The Central Electricity Generating Board of Great Britain is interested in
the use of dc links for reinforcing an ac system in areas of high load density
without increasing the interrupting duty of ac circuit breakers. A trial installa
tion of this kind is the transmission of power by underground dc cable from
the Kingsnorth thermal power plant, situated on the south shore of the
Thames River estuary, to two substations in London. This is a bipolar scheme
having three cables: one for each pole and a neutral cable. Each pole goes to
a different substation, with the result that, although the whole scheme has
three terminals, each pole has only two terminals. The Beddington substation
is 37 mi (59 km) from Kingsnorth, and the Willesden substation is 14 mi
(23 km) beyond Beddington. Whenever the loads of the two substations are
unequal, there will be neutral current. This current is not allowed to flow in
the ground for fear of damage by electrolytic corrosion to some of the
many buried metallic structures found in a metropolitan area.
The rating of this scheme is 266 kV, 1.2 kA, 640 MW. There are four
groups of valves at Kingsnorth and two groups at each substation, each
group being rated at 133 kV, 1.2 kA, 160 MW.
1-5
19
other. The distinctive feature of the Eel River station is that it is the first large
converter station designed to use thyristor valves initially and exclusively.
The rating of the station is 320 MW, 80 kV dc, 230 kV ac.
Noting the universal use of alternating current for electric power trans
mission, as well as for generation, distribution, and use, one naturally asks
what limitations ac transmission has that have led to the use of dc transmis
sion in some projects.
The limitations may beither technical something cannot be done or
economic it can be done more cheaply some other way. In most practical
cases the technical limitations are not reached, and economic limitations
dictate the final choice of design.
We are interested in limitations on the amount of transmitted power and
on the distance over which it can be transmitted. More exactly, we are inter
ested in the cheapest method by which a certain amount of power at a certain
load factor can be transmitted reliably over a certain distance. The power
depends on the current, voltage, power factor, and number of conductors.
pole.
Voltage Limits
20
1-5
1.7 times normal voltage. On overhead lines, the maximum working voltage
or the minimum conductor size is limited also by loss and radio interference
due to corona. In current ac practice, radio interference during foul weather
(rain, snow, or fog) is usually the limiting factor. Here dc lines have a distinct
advantage in that radio interference is slightly decreased by foul weather,
while interference due to ac lines is greatly increased by foul weather. In
cables, where the limiting factor is usually the normal working voltage, the
insulation will withstand a direct voltage higher than the crest of alternating
voltage, which is already 1.4 times the rms value of the alternating voltage.
voltage (kV)
surge impedance loading (MW)
1.05
On long ehv ac overhead lines and on much shorter ac cables, the produc
tion and consumption of reactive power by the line itself constitutes a serious
problem. On a line having series inductance L and shunt capacitance C per
unit of length and operating voltage V and ciirrent I, the line produces
reactive power
Ql = coLI2
(2)
230
345
500 .
43
130
300
830
700
.
1600
\. No
load
2. Natural load Pw
1.00
(1)
132
oljp line carrying its natural load, the magnitude of voltage is the same
everywhere, as shown in curve 2 in Figure 3, and the reactive power is zero
Qc = coCV2
21
0.95
20
Distance from sending end (elec deg)
10
30
per unit of length. The reactive power produced by the line equals that con
sumed by it, with no net production or consumption, if
Q
Pn
coCV2 = (oLI2
2. Natural load Pn
hence if
HW'2-
(3)
In this case the load impedance has the value Zs , known as the surge im
pedance of the line. The surge impedance of an overhead line with single
conductors is about 400 Q, and with bundle conductors, about 300 Q; that of
cables is only 15 to 25 Q.
The power carried by the line so loaded is
V2
P = VI = -
P+jQ
-0.5
10
20
(4)
and is called the surge impedance loading (SIL) or natural load. It is inde
pendent of distance and depends mainly on the voltage. Typical values for
three-phase overhead lines are as follows :
Most lines cannot be operated always at their natural loads, for the loads
vary with time. The most economical load on an overhead line is usually
greater than the natural load. If the load is greater than the natural load, net
reactive power is consumed by the line and must be supplied from one or
both ends. If equal voltages are maintained at both ends of the line, equal
amounts of reactive power are supplied from both ends (curve 3 in Figure 4),
22
1-5
and the voltage elsewhere sags, being least at the center of the line (curve 3 in
Figure 3). If the load on the line is less than the natural load, net reactive
power is produced by the line and is delivered to one or both ends. With
equal voltages at both ends, equal amounts of reactive power are delivered to
both ends (curve 1 in Figure 4), and the voltage everywhere else is higher than
at the ends, being greatest at the middle (curve 1 in Figure 3). In all cases, the
flow of reactive power through series inductive reactance is "downhill," that
is, in the direction of decreasing voltage.
Thus, to maintain constant equal voltages at both ends, reactive power
must be absorbed at light load and supplied at heavy load. The reactive
power required for a given variation of load increases with distance (see
Figure 5, curves 3 and 4).
j
(
j
)
23
blocks.
Figure 5 shows some other disadvantageous characteristics of long, un
compensated transmission lines up to one-half wavelength (180). One is their
power limit. For any given length Ithe maximum power that can be trans
mitted on such a line is shown by curve 1. It is
Pm=*PHsec0
(5)
9 = IsJtC
where
3
2. Voc/E
!
f'
V(T=7xT=T).
P= 1.5P,
1
/ 5:Vm/forP= 1.5P,
3- Qt/Pn for P= 1.5P
-2
0
30
60
90
120
150
180
:
Fig. 5. Characteristics of lossless line with equal terminal voltages E (except curve 2)
versus length up to one-half wavelength. Curve 1. Maximum power/natural power, PmjP.
Curve 2. Voltage at open end/sending-end voltage. Curve 3. Reactive power from both
ends/natural power for P=1.5P. Curve 4. Reactive power from both ends/natural
power for no load (P = 0). Curve 5. Voltage at middle/terminal voltage for P = 1.5P.
(6)
is the electrical length of the line, Ibeing the actual length and X the wave
length. For a greater load than the natural load, there is a maximum distance;
for example, for P = 1.5Pn this distance corresponds to a line angle of
sec"1 (PJPn) = sec-1 1.5 = 41.8 and is (41.8/360)2, or 360 mi at 60 Hz. As
the length of line approaches this value, the reactive power that must be
supplied to the line increases rapidly, as shown by curve 3, and the voltage at
the middle of the line drops rapidly, as shown by curve 5.
Another limitation of long lines is the high voltage at an open end (the
Ferranti effect), shown in curve 2. This is important when a line is being put
into service by first connecting one end of it to the main ac system, for it is
not feasible to close both ends at exactly the same moment.
Long-distance ac power transmission is feasible only with the use of series
and shunt compensation, applied at intervals along the line, as illustrated in
Figure 6. Series compensation of degree s reduces the effective series in
ductance from L by sL to (1 s)L and thus decreases the electrical length
Eq. (6)from /?/ to
s and at the same time decreases the surge
impedanceEq. (3) and increases the natural load by the same factor. The
reactive power produced by shunt capacitance of the line at light load may
still be excessive, requiring shunt compensation of part h of it. The effective
shunt capacitance is then reduced from C to (1 - h)C,and the electrical length
is reduced by the additional factor %/l h or by the total factor
= pi
/?/Vl
1
24
1-5
Fig. 6. Series and shunt reactive compensation for 750-mi (1200 km) 500-kV 60-Hz
single-circuit line delivering 1000 MW and having two 1780-MCM A.C.S.R. conductors
per phase. Sending-end voltage, 525 kV /0; receiving-end voltage, 500 kV / 29.5. The
series capacitors have an aggregate loading of 1507 Mvar.; shunt reactors, 1275 Mvar. ;
reactive power entering line from sending end, 193 Mvar. ; from receiving end, 355 Mvar.
Degree of series compensation, 73%; of shunt compensation, 58%. (Ref. D4, Appendix B,
Figure 7.)
In Figure 6 and in similar schemes shown in the Ref. D4 for other voltages
and distances, the compensation was chosen to limit the angle between termi
nal voltages to 30 and to limit the voltages at the ends and at compensating
points to not more than 1.05 times nominal voltage, except that the voltage
at an open end is limited to not more than 1.1 times normal voltage.
On representative long overhead ehv compensated lines operating at full
load, the total reactive power furnished from both ends of the line and from
intermediate series capacitors plus the reactive power consumed by shunt
1
1
500 kV
1ooo
I
&
700 kV
6O-Hz a
o
.2 v
F
200
400
Y
600
2000 MW-
Dc
800
1000'
Fig. 7. Reactive power requirements of long ehv overhead ac and dc lines at full load
as a function of the length of line. (That of ac lines is from data in Ref. D4.)
25
however, draw reactive power from the ac systems. It varies with the trans
mitted power and is approximately half of the latter at each end. It is in
dependent of the length of line. Usually shunt capacitors or synchronous
condensers are installed for supplying this reactive power.
Both ac and dc lines have the disadvantage of requiring adjustable supplies
(or sinks) of reactive power. For distances of more than 400 km (250 mi),
however, Figure 7 shows that the dc line requires less than the ac line.
On submarine or underground cables, the situation is different from that on
overhead lines. Cables are always operated at a load much below the surge
impedance load in order to avoid overheating. Consequently the reactive
power produced by charging the shunt capacitance greatly exceeds that con
sumed by the series inductance.02 In a 50- or 60-Hz cable, 25 to 50 mi (40 to
80 km) long, the charging current alone equals the rated current, leaving no
margin for load current. Shunt compensation theoretically could correct this
situation.03 Shunt reactors, however, would be required at, perhaps, 10-mi
(16-km) intervals. Since it is difficult to lay and repair submarine cable to
which shunt reactors are connected, the practical length of ac submarine
cables is only about 20 mi (30 km). Dc cables have no such limitation.
Stability
By the stability of an ac system is meant its ability to operate with all
synchronous machines in synchronism. If a long ac line is loaded to a certain
value, known as its steady-state stability limit, the synchronous machines at
the sending end accelerate and go out of synchronism with those at the
receiving end. This condition is analogous to a slipping belt or clutch in a
mechanical transmission system. The slipping electrodynamic system not
only fails to transmit the power that it should but also gives rise to objection
able fluctuations in voltage.
Even if a line is operated below its steady-state limit, the machines at the
sending and receiving ends may lose synchronism after some large disturbance,
notably a short circuit, unless the line is operated below its transient stability
limit, which is always lower than the steady-state limit. Practically speaking,
the steady-state stability limit is the transient stability limit for very small
disturbances.
The problem of stability or synchronous operation constitutes the most
serious limitation of a long ac transmission system.
The power transmitted from one machine to the other in a two-machine
lossless ac power system is given by
pMsin5
(7)
26
1-5
where Ex and E2 are the internal voltages of the two machines, <5 is the phase
difference of these voltages," and X is the reactance of the architrave of the
equivalent n circuit of the system joining the internal points. Each machine
is represented by an internal voltage "behind" an internal reactance. The
reactance X is very nearly the sum of the inductive reactances inside the two
machines, of the transmission line, and of the step-up and step-down trans
formers. An actual power system involving a long interconnecting line with
many generating stations at each end of the line may be represented reason
ably well by a two-machine equivalent system in which all the machines at the
receiving end of the line are replaced by one equivalent machine and all those
at the sending end by another.
A graph of power P as a function of phase difference 5 between internal
voltages is a sine wave. Maximum power occurs at 3 = 90 and is
A,= exe2
X
(8)
The transient stability limit is lower than the steady-state limit, and, as a
rough guide, we may take the former as half of the latter, corresponding to
a phase difference of sin-1 0.5 = 30 in the initial steady state. (This value
was assumed in Figures 6 and 7.)
In an uncompensated line operating at its natural load the phase of the
voltage varies directly with the distance, going through one cycle (360) per
wavelength. A 30 difference, then, corresponds to
wavelength. On a 60-Hz
line this is 3100/12 = 258 mi (416 km). On a 50-Hz line it is 310 mi (500 km).
A lighter load can be transmitted farther : a heavier load not so far.
The distance to which the natural load can be transmitted stably can be
extended considerably by placing synchronous condensers or, better yet,
synchronous generators at various intermediate points of the transmission
system. If both generators and loads are scattered along the transmission
system, this method of long-distance transmission is called transmission by
displacement. As shown in Figure 8, the over-all transmission can be regarded
"
500 MW
27
rT~
500 MW
500 MW
500 MW
500 MW
500 MW
500 MW
500 MW
500 MW
500 MW
1000 MW
wO
500 MW
1000 MW
28
1-5
29
Short-Circuit Current
The interconnection of ac systems through an ac line raises the shortcircuit currents, sometimes to an extent that exceeds the interrupting capa
bility of existing circuit breakers and requires their replacement by more
capable breakers. The interconnection of ac systems by a dc link, however,
does not increase short-circuit currents of the ac systems nearly so much, for
the dc line contributes no current to an ac short circuit beyond its rated
current.
On the other hand, the proper operation of a dc line terminal requires that
the short-circuit power of the ac system at the point of installation be several
(now at least five) times the rated power of the dc line, and sometimes this
requirement dictates increase of the ac short-circuit power by the provision of
synchronous condensers or additional ac connections.
The current in a short circuit on the dc line, after a momentary transient
due to a discharge of the shunt capacitance of the line, is limited by automatic
grid control to twice rated current. Nor do faults on the dc line draw ex
cessive currents from the ac systems.
Let us assume that an ac line and a dc line using the same conductors and
insulators are built. How does the power per conductor compare on the two
lines?
Assume that in each case the current is limited by temperature rise. Then
the direct current equals the rms alternating current.
Assume also that the insulators withstand the same crest voltage to ground
in each case. Then the direct voltage is 2 times the rms alternating voltage.
The dc power per conductor is
-J
(9)
Pi=VJ*
and the ac power per conductor is
Pa =
VJa COS 0
(10)
where Id and Ia are the curents per conductor, Vd and Va the conductor-toground voltages, and cos 0 the power factor. The ratio is
pd
VdId
_ Vd h
pa
VaIa cos 0
Vg
Ia
1
cos 0
_ 72
cos 0
(11)
30
1-5
Pa
3 2
. 31
Terminal Equipment
Pd ~ 2Pd
pa = 3p
=
Pa 3
(12)
Both lines can carry the same power. The dc line, however, is simpler and
cheaper, having two conductors instead of three. Consequently an overhead
line requires only |
as many insulators, and the towers are simpler, cheaper,
and narrower. A narrower right of way could be used.
Both lines have the same power loss per conductor. The percentage loss
of the dc line is only two-thirds that of the ac liner If the basis of comparison
is equal percentage loss, the power of the three-phase ac line is decreased to
2/3 that of the two-conductor dc line.
If cables are used instead of overhead line, the permissible working stress
(voltage per unit thickness of insulation) is higher for direct current than for
alternating current, and, in addition, the power factor for direct current is
unity and, for alternating current, considerably lower than that assumed
above. Both changes further favor direct current over alternating current by
increasing the ratio of dc power to ac power per conductor. The resulting
ratio might be from 5 to 10.
Because the power limit of overhead ac lines is. often determined by factors
other than conductor heating, the ratio of dc power per conductor to ac
power per conductor may be as high as 4.
Harmonics
The converters used with a dc line produce harmonic voltages and currents
on both ac and dc sides.These harmonics, especially in the extensive ac net
works, may cause interference with audio-frequency telephone lines. Filters
are required on the ac side of each converter for diminishing the magnitude
of harmonics in the ac networks. These increase the cost of the converter
stations. Fortunately the capacitors used in the filters also supply part of the
reactive power required by the converters. The cost of the filters and of the
additional reactive power supply should be regarded as a part of the cost of
a dc line terminal.
The power flow on tie lines interconnecting different areas under different
ownerships must be controlled in conformity with contractual obligations.
In addition, the frequency of the whole system, or the frequencies of the
parts connected asynchronously, must be controlled.
The control system is a little simpler if the tie lines operate on dc than if on
ac, but the difference is not important. This subject is discussed in Volume 2.
Ground Return
A two-conductor bipolar dc line is more reliable than a three-conductor ac
line, because, in the event of a fault on one conductor, the other conductor
can continue to operate with ground return during the period required for
repairing the fault. The operation of an ac line with ground return is not
feasible on account of the high impedance of such a circuit and the telephone
interference caused by such operation. Further information on dc ground
return is given in Chapter 9.
A monopolar dc line with earth return is still simpler than a three-phase ac
line and is equally reliable. It is especially suitable to submarine cable. A line
cart be built in stages with monopolar operation initially, later changed to
bipolar operation with doubling of the power rating.
Generating Units
Some hydroelectric generating stations connected to a load center through
long ac lines have generators with abnormally low transient reactance or
abnormally high moment of inertia specified in order to raise the stability
limit. These features raise the cost of the generators and would not be re
quired if dc transmission were used, for there would be no stability problem
with direct current. In addition, if such a station were connected to an ac
system only through dc lines, the speed of the prime movers could be allowed
to vary with the load or the head of water, perhaps giving a cheaper or a more
efficient prime mover, and the nominal frequency of the generator, no longer
confined to 50 or 60 Hz, could be chosen for best economy. Perhaps also, in
32
1-8
impractical.
5. And, of course, combinations of these factors occurring in the same
project.
Advantages
tion.
Disadvantages
hv
dc
33
Less corona loss and radio interference, especially in foul weather, for a
certain conductor diameter and rms voltage.
Synchronous operation is not required.
Hence distance is not limited by stability.
May interconnect ac systems of different frequencies.
Lcjw short-circuit current on dc line.
Does not contribute to short-circuit current of ac system.
Tie-line power is easily controlled.
ECONOMIC FACTORS
\
1
|
I
The cost per unit length of a dc line is lower than that of an ac line of the
same power capability and comparable reliability, but the cost of the terminal
equipment of a dc line is much more than that of an ac line. If we plot the
cost of transmitting a certain amount of power by one method or the other
as a function of the distance over which it is transmitted, the resulting graph
is similar to Figure 9. The vertical intercept of each curve is the cost of the
terminal equipment alone. The slope of each curve is the cost per unit length
of the line and of that accessory equipment which varies with the length.
The curve for ac transmission intersects that for dc transmission at an abscissa
called the break-even distance. If the transmission distance is shorter than the
break-even distance, ac transmission is cheaper than dc; if longer, dc is
cheaper than ac.
Estimates of the break-even distance of overhead lines, published in the
technical literature, range from 500 km (310 mi) to 1500 km (930 mi). Such
great variation can be explained, at least in part, by a simple modification
34
1-9
Break-even
|
-distance, -|
500 mi
|
-iTerminal costs
Tt
i
200
iii
400
600
Distance (mi)
1
800
1000
of Figure 9, shown in Figure 10. Here the cost of each line is assumed to
5% for the ac line and 10% for the dc line. The
true cost of each is assumed to be within the crosshatched area. (Greater
variation is assumed for the cost of dc transmission than for that of ac
because there has been less experience with dc than with ac.) Itis now apparent
that even such small variations in estimated costs make the estimated break
even distance vary over a range of 2 or 3 to I.
For cables the break-even distance is, of course, much shorter than for
overhead lines, lying between 15 and 30 mi (24 and 48 km) for submarine
cables and, perhaps, twice as far for underground cables.
The ordinate in Figures 9 and 10 might be either capital cost or annual
vary over a certain range,
400
600
Distance (mi)
1000
35
cost; it might be for a given amount of power or per megawatt. In any case
the curves would have the same form.
In view of the relative novelty of hv dc transmission, there is a prospect
for a greater decrease in the unit cost of dc line terminals with increasing
experience and volume of production than in the cost of ac equipment. The
result would be to decrease the break-even distance.
An economic comparison between ac and dc transmission made by an
international working party of C.I.G.R.E.07 and based on 1965 costs showed
average break-even distances of 1000 km (600 mi) for transmitting 1080 or
2160 MW on two overhead circuits and 77 km (48 mi) for transmitting 1080
MW on two shunt-compensated underground cable circuits. An assumed
future 20% reduction in dc terminal costs reduced the break-even distance to
830 km (515 mi) for the overhead lines and to 64 km (40 mi) for the under
ground cables.
In the great majority of dc transmission schemes already built, other
factors than the costs assumed in such comparisions play a significant role.
These other factors are long water crossings, frequency conversion, and the
advantage of asynchronous ties between large ac systems.
The increasing size and load density of metropolitan areas create problems
of right of way for hv overhead lines. The increased public demand for the
better appearance of electric lines and for the preservation of the natural
environment is putting pressure on the electric power companies for placing
transmission and distribution lines underground, out of sight, even where the
load density is not high. Dc cables are cheaper and more compact than ac
cables for the same power and are not so limited in the feasible distance of
transmission.
As Greber discerningly points out,E3 the basic problem of ac transmission
is that of inductive and capacitive reactance; the basic problem of dc trans
mission is switching.
It is the series inductive reactance of long overhead ac lines that causes the
synchronous stability limit. It is the shunt capacitive reactance of long ac
cables that overloads them with charging current. On long overhead lines,
the presence of both kinds of reactance causes excessive variation of voltage
with load.
Series and shunt compensation of reactance are used on long ac lines, but
they add to the cost and complexity of such lines. Reactive compensation is
not required on a dc line itself, but only on the ac side of the converters. This
fact gives an advantage to long dc lines over long ac lines. If cheaper and
36
BIBLIOGRAPHY
2.
A. General
i
f
j
s
\
1
j
j
j
)
I
BIBLIOGRAPHY
37
j
|
I
!
|
;
;
B. History
1. "Constant-Current D.C. Transmission," by C. H. Willis, B. D. Bedford, and F. R,
Elder, Elec. Eng., Vol. 54, pp. 102-108, January 1935. Disc., pp. 327-329 (March),
447-449 (April), and 882-883 (August).
2. "Power Transmission by Direct Current: Apparatus Used in 3000-kw 15,000-volt,
200-amp Pump-back Test," by B. D. Bedford, F. R. Elder, and D. H. Willis, Gen.
Elec. Rev., Vol. 39, pp. 220-224, May 1936. Tests preceding the MechanicvilleSchenectady experimental dc transmission.
3. "D.C. Transmission in France," Elec. World, Vol. 106, No. 19, pp. 1341-1342,
May 9, 1936. On 275-mi, 125-kV, 20-MW line from Moutiers to Lyon.
4. "The First Power Transmission at 50 kV D.C. with Mutators" (in French), by
P. Egloff and J. J. Felix, Electricity, Vol. 23, No. 58-59, pp. 237-240, July-August
1939. Baden-Zurich transmission.
5. "The D.C. Power Transmission at the Swiss National Exhibition" (in German),
by E. Kern, Bull, de V Association Suisse des Electriciens, Vol. 30, No. 17, pp. 481-482,
Aug. 18, 1939.
6. "H.V.D.C. Transmission," by F. Busemann, Elec. Times, Vol. Ill,Nos. 2881,2883,
2885, pp. 3640, 98-101, 169-170, Jan. 9 and 23, Feb. 6, 1947. Experience in Germany
during World War II,including Charlottenburg-Moabit experimental 110-kV dc trans
mission and Elbe-Berlin project. English translation from German report.
>
j
38
BIBLIOGRAPHY
'
39
3. Comparison of Direct and Alternating Current for High-Voltage Electric Power Trans
mission, Edison Electric Institute Publication No. 62-901, 1962.
4. "High-Voltage DC Transmission," Advisory Committee Report No. 20, published
on pp. 289-313 of NationalPower Survey, a report by the Federal Power Commission,
part II, "Advisory Reports," U.S. Government Printing Office, Washington,
October 1964.
5. "Cost of Electrical Energy Transmission by AC and DC Extra High Voltage,"
Advisory Committee Report No. 16, pp. 189-203, loc. cit.
6. "High Capacity D.C. Transmission in the U.S.S.R.," by A. Berkovski, N. Chouprakov, T. Izrailevich,A. Kolpakova, and S. Rokotyan, I.E.E. Conference Publication 22,
H.V.D.C. Transmission, Manchester, Sept. 19-23, 1966, Part 2, Paper No. 94,
pp. 126-129. Also in Direct Current , Vol. 11, pp. 145-149, November 1966.
7. "A Technical and Economic Comparison between A.C. and D.C. Transmission,"
by W. Casson, C.I.G.R.E., 1968, Report 42/43-01, 42 pp.
D. Special Problems of AC Transmission
j
1
i
I
j
j
|
\
1
I
I
j
)
|
40
12.
13.
14.
15.
BIBLIOGRAPHY
41
hydrogen, and helium, are compared with one another and with conventional pipetype cable. The nitrogen-cooled cable is found to be the most economical.
16. "Economic Assessment of a Liquid-Nitrogen-Cooled Cable," by S. B. Afshartous,
Peter Graneau, and John Jeanmonod, I.E.E.E. Trans, on P.A. & S., Vol. 89, pp.
8-13, January 1970. Disc., pp. 14-16. Cable of tubular aluminum conductors cooled
internallyby liquid nitrogen and supported by dielectric spacers inhigh-voltage vacuum
insulation.
F. Bibliographies
1. "Direct Current Bibliography 1," Direct Current, Vol. 1, pp. 50-52, September
1952. Covers years 1943 to 1952. Vol. 1, p. 97, March 1953, covers rest of 1952.
An
Annotated Bibliography of High Voltage Direct Current Transmission, 1932-1962,
2.
compiled by Eric Bromberg, I.E.E.E. Paper CP 63-388, January 1963. Also in D-C
Transmission, Publication S-155, pp. 76-214, I.E.E.E., June 1963.
3. An Annotated Bibliography of High Voltage Direct Current Transmission, 1963-1965,
complied by Eric Bromberg, I.E.E.E. Publication 31 S 60, iii-f 113 pp., I.E.E.E.,
New York, 1967.
4. High Voltage Direct Current Transmission: An Annotated Bibliography, 1966-68, com
piled by Val S. Lava, published by the Library, Bonneville Power Administration,
Portland, Oregon, December 1968, ii 90 pp.
5
G. Gotland Link
|
|
|
;
1. "The High Voltage D.C. Power Transmission from the Swedish Mainland to the
Swedish Island of Gotland," by Ake Rusck, B. G. Rathsman, and U. Glimstedt,
C.I.G.R.E., Report 406, 1950.
"D.C.
Transmission from Swedish Mainland to Island of Gotland, by A. Rusck,
2.
B. G. Rathsman, and U. Glimstedt, Engineer, Vol. 190, No. 4931, pp. 92-93, July 28,
1950.
3. " High-Voltage D.C. Power Transmission Pioneer Project," by U. Lamm, ASEA
Journal, Vol. 23, No. 12, pp. 172-174, December 1950.
4. " Submarine Cable Project Will Operate at 100 kV D.C.," by B. G. Rathsman, Electric
Light and Power, Vol. 29, No. 8, pp. 108-109, August 1951.
5. "Gotland H.V.D.C. Link: Present Progress," by B. G. Rathsman and U. Lamm,
Direct Current, Vol. 1, pp. 2-6, June 1952.
6. "Gotland D.C. Link: Layout of Plant," by I. Liden, Sy Sviden and E. Uhlmann,
Direct Current, Vol. 2, pp. 2-7, June, and pp. 34-39, September 1954.
7. "The First High Voltage D.C. Transmission wigh Static Converters: SomeNoteson
the Development," by U. Lamm, ASEA Journal, Vol. 27, No. 10, pp. 139-140,
October 1954. "The Gotland D.C. Link: The Layout of the Plant," by I. Liden
and E. Uhlmann, pp. 141-154.
8. "The D.C. Transmission to Gotland: Initial Experience," by S. Ekefalk, ASEA
Journal, No. 10, pp. 123-126, 1956.
1. "The Possibilities of a Cross-Channel Power Link between the British and French
42
2.
3.
4.
5.
6.
7.
8.
9.
I. Volgograd-Donbass Link
1. "D.C. Transmission from Stalingrad Hydro-electric Station to Donbass" (in Rus
sian), by V. P. Pimenov, A. V. Posse, A. M. Reider, S. S. Rokotian, and V. E. Turetskii,
Elektricheskie Stantsii, No. 11, 1956, pp. 12-18.
2. "Transmission of Direct Current at HighVoltage according to Present-Day Concepts
and the Prospects for Their Application in the U.S.S.R." (in Russian), by N. M.
Mel'gunov, N.I.I.P.T., Vol. 1, pp. 21-38, 1957.
3. " The Transmission System Stalingrad Hydro-electric Station,Donbass "(in Russian),
by E. S. Grois, M. L. Zelikin, V. E. Turetskii, andE. A. Man'kin, Elektrichestvo,Vol.
77, No. 9, pp. 1-10 September 1957.
4. "Design Features of Stalingrad-Donbass 800 kV D.C. Line," by F. I. Butaev, E. S.
Grois, E. K. Levitski, E. A. Man'kin, A. V. Posse, A. A. Sakovitch, and V. E.
Turetski, Direct Current, Vol. 4, No. 2, pp. 59-66, September 1958.
BIBLIOGRAPHY
43
44
BIBLIOGRAPHY
11. "Valves, Valve House, and Indoor Equipment at the Converter Stations," by M. A.
Louden, Vol. 20, No. 10, pp. 393-402, October 1965.
12. "Layout of the Direct Current Switchyards," by D. G. Young, Vol. 20, No. 11,
pp. 472-478, November 1965.
13. "Harmonic Phomenena," by G.H. Robinson,Vol.21, No. 1,pp. 16-29,January 1966.
14. " Synchronous Condenser Installation at Haywards Substation," by L. S. Y. Gock,
Vol. 21, No. 1, pp. 29-35, January 1966.
15. "Power Line Carrier Communications," by F. R. Swan, Vol. 21, No. 2, pp. 45-55,
February 1966.
16. "Commissioning and Early Operating Experience," by H. R. Gunn, Vol. 21, No. 3,
pp. 93-101, March 1966.
17. "The 250 kV d.c. Submarine Power-Cable Interconnection," by A. L. Williams,
E. L. Davey, and J. N. Gibson, Vol. 21, No. 4, pp. 145-160, April 1966. Also pub
lished in I.E.E. Proc., Vol. 113, No. 1, pp. 121-133, January 1966.
18. "The New Zealand 500 kV High-Voltage Direct-Current Project," by P. W. Blakely,
Amer. Power Conf. Proc., Vol. 28, pp. 850-859, April 1966.
The following papers are from I.E.E. Conference Publication 22, Conference on High
Voltage D.C. Transmission, Sept. 19-23, 1966, Manchester, Part 1:
19. " Commissioning, and Early Operating Experience with the New Zealand HVDC
Inter-Island Transmission Scheme," by H. R. Gunn, Paper No. 5, pp. 30-38.
20. "Benmore Power Station: Special Features for H.V.D.C. Transmission," by H. C.
Hitchcock, Paper No. 19, pp. 101-103.
21. "The Synchronous Condenser Installation at Haywards Sub-station for the BenmoreHaywards H.V.D.C. Transmission Scheme," by L. S. Y. Gock, Paper No. 52, pp.
265-267.
22. "Communications (Power Line Carrier Systems)," by F. R. Swan, Paper No. 63,
pp. 306-311.
23. "The Cook Strait 250-kV Cables," by E. L. Davey, Paper No. 64, pp. 312-314.
24. "A Direct Current Transmission Line: The Design and Construction of the 600 MW,
500 kV D.C. Line Between Benmore and Haywards," by T. A. J. Dickens, Paper
No. 72, pp. 343-346.
25. "Some Features of New Zealand's Inter-island H.V.D.C. Transmission," by R. J.
Fyfe, M. A. Louden, J. Noble, and D. G. Young, Paper No. 78, pp. 375-383.
26. "The Benmore Land Electrodes for the Benmore-Haywards H.V.D.C. Transmission
Scheme," by D. G. Dell, Paper No. 82, pp. 415-418.
27. " The NorthIsland Sea Electrode for the Benmore-Haywards H.V.D.C. Transmission
Scheme," by D. G. Dell, Paper No. 85, pp. 427-430.
28. "Experience with Harmonics New Zealand H.V.D.C. Transmission Scheme," by
G. H. Robinson, Paper No. 89, pp. 442-444.
29. "Operational Experience of the Benmore-Haywards HVDC Transmission Scheme,"
by M. T. O'Brien, C.I.G.R.E., report 14-03, 1970, 11 pp.
K. Konti-Skan Link
1. "The Conti-Skan HVDC Project," by G. von Geijer, S. Smedsfelt, and L. Ahlgren,
I.E.E.E. Conference Paper CP 63-1056, January 1963.
45
2. "The Konti-Skan Project," by G. von Geijer, Direct Current. Vol. 8, pp. 149-51,
June 1963.
3. "The Konti-Skan H.V.D.C. Project," by G. von Geijer, S. Smedsfelt, L. Ahlgren,
and E. Andersen, C.I.G.R.E., 1964, Paper No. 408, 45 pp.
4. Direct Current, Vol. 10, pp. 10-12, February 1965. Unsigned news article.
5. " Operational Performance and Service Experience with the Konti-Skan and Gotland
H.V.D.C. Projects," by S. Smedsfelt, L. Ahlgren, and V. Mets, I.E.E. Conference
Publication 22, H.V.D.C. Transmission, held at Manchester, Sept. 19-23, 1966,
Part 1, Paper No. 1, pp. 11-16.
L. Sardinia.Link
2.
3.
4.
5.
6.
1968.
8. "Vancouver Island HVDC Transmission," by Gordon H. Dunn and Lars A. Bergstrom, ASEA Journal, Vol. 42, No. 2-3, pp. 29-31, 1969.
46
BIBLIOGRAPHY
3. Elec. World, Vol. 165, No. 15, pp. 19-20, Apr. 11, 1966. ,
4. Direct Current, Vol. 11, p. 53, May 1966.
5. English Electric Jour., Vol. 21, No. 3, pp. 2-3, May-June 1966.
6. Elec. Times, Vol. 149, No. 10, pp. 361-362, Mar. 10, 1966.
7. Elec. Times, Vol. 150, p. 479, Sept. 29. 1966.
8. " Kingsnorth-London D.C. Transmission Interconnector," by W. Casson, J.E.E.
Conference Publication 22, High Voltage DC Transmission, Manchester, Sept- 19-23,
1966, Part 1, Paper No. 9, pp. 56-57.
9. "The Kingsnorth, Beddington, Willesden D.C. Link" by T. E. Calverley, F. H. Last,
A. Gavrilovic, and C. W. Mott, C.I.G.R.E., Report 43-04, 1968, 14 pp.
47
Q. Miscellaneous Projects
1. "The Exploitations of Yugoslav Water Resources and the Possibility of Using
H.V.D.C. Transmission," by H. von Bertele, Direct Current, Vol. 2, No. 5, pp. 107109, June 1955.
2. "The Present Status of High-Voltage D.C. Power Transmission in Japan," by Naohei
Yamada, Appendix Ito "Report on the Work of the Study Committee No. 10:
D.C. Transmission at E.H.V.," C.I.G.R.E., 1956, Paper No. 407, pp. 2-7.
3. "Electric Power Transmission by H.V.D.C. Submarine Cables across the Adriatic
Sea from Yugoslavia to Southern Italy," by M. Visentini, A. Asta, and F. Trimani,
C.I.G.R.E., Report 210, 1958, Vol. 2, 22 pp.
4. "The Introduction of H.V.D.C. Transmission into a Predominantly A.C. Network,"
by W. Casson, F. H. Last, and K. W. Huddart, Elec. Rev., Vol. 178, No. 8, pp. 290295, Feb. 25, 1966.
5. "The Economics of D.C. Transmission Applied to an Interconnected System," by
W. Casson, F. H. Last, and K. W. Huddart, I.E.E. Conf. Publication 22, H.V.D.C.
Transmission, Manchester, Sept. 19-23, 1966, Part 1, Paper No. 13, pp. 75-83. Rein
forcement of an ac system with dc links which do not increase required circuit-breaker
interrupting ratings.
6. "High Capacity D.C. Transmission in theU.S.S.R.," by A. Berkovsky, N. Chouprakov, T. Izrailevich, A. Kolpakova, and S. Rokotjan, ibid., Part 2, Paper No. 94, pp.
126-129.
7. "Introductory Lecture," by F. J. Lane, ibid., Part 2, pp. 7-25. Describes proposed
transmission of 4500 MW by 500-kV dc from Churchill Falls, Labrador, to Boston
and New York, U.S.A., by the Atlantic route, comprising 1575 miles of land line and
92 miles of submarine cable.
48
Converter Circuits
It was indicated in Chapter 1 that the best kind of valve for use in the
converters for HV dc transmission is the mercury-arc valve with control grid.
In this chapter we examine various ways in which groups of such valves and
transformers can be connected to form a converter, and we determine which
of several possible circuits is best for hv dc transmission.
Voltage Current -
(b)
Anodej Cathode
Current
Voltage
Anode
voltage
Forward
Envelope
Anode
Control grid
Cathode
Inverse
Wii
Z-l
Current
Forward
Control
(grid or gate)
Current
Anode "'Cathode
(c)
Inverse
(d)
49
50
2-4
CONVERTER CIRCUITS
control.
For each circuit considered, we find the wave forms of the voltages and
currents and their magnitudes in terms of the direct voltage Vd and direct
current Id . Fromthese data we find the required volt-ampere ratings of valves
and transformers in terms of the dc power Pd = VdId.
The volt-ampere rating of a valve is taken as the product of its average
current and its peak inverse voltage (PIV), and the rating of a transformer
winding is the product of its rms voltage and rms current.
SINGLE-PHASE CONVERTERS
51
2-3 ASSUMPTIONS
Half-wave Rectifier
This is the simplest rectifier, having only one valve (Figure 2). The current
is inherently intermittent, and therefore the circuit cannot be analyzed in
i .
W
Ac
vi
Vl
d
=- Dc
'
vd
accordance with the assumption of steady current, although all other converter
circuits can be and are so analyzed. Both direct current and direct voltage
pulsate at the same frequency as the alternating voltage. If a transformer is
placed between the ac source and the valve, as is usually necessary for obtain
ing the desired magnitude of direct voltage, the direct current passes through
the secondary winding of the transformer, and its mmf may be sufficient to
saturate the iron core and make the primary current excessive. The circuit is
useful only for very small amounts of power.
52
CONVERTER CIRCUITS
T: 1: 1
Full-wave Rectifier
This has two valves and one transformer with center-tapped secondary
winding (Figure 3). The wave forms are shown in Figure 4. In Figure 4a, the
line-to-neutral secondary voltages et and e2 , having a phase difference of
one-half period (180), are plotted. The anode voltages of valves 1 and 2 with
respect to neutral point N are equal to et and e2 , respectively. The common
cathode voltage of both valves, being equal to the higher of the anode
voltages, consists of the positive half waves of ex and e2 . This curve, redrawn
in Figure 1b, represents also the instantaneous direct voltage vd on the valve
side of the smoothing reactor. The average direct voltage Vd is also shown.'
The difference vd Vd, which appears across the reactor, is represented by
the vertical shading in Figure lb. Its average value is zero, corresponding to
equal positive and negative areas between the curve and the horizontal line.
When valve 2 is conducting, the full secondary voltage et e2 appears
across valve 1; when valve 1 is conducting, e2 e1 appears across valve 2.
Figure 4c shows the voltage vx across valve 1. The valve currents, which are
also the currents in the halves of the secondary winding, are shown in Figure
4d. They are rectangular pulses of height Id and length 180. The mmf of the
entire secondary winding is porportional to ix i2 and has an average value
of zero; in other words, there is no dc component of mmf, hence no tendency
to saturate the core. The primary mmf must oppose the secondary mmf
(Tip = ix - i2), so that the primary current ip has the form shown in
Figure 4e.
Now let us compute the numerical values of the various circuit quantities.
The filtered direct voltage Vd is the average value of vd, and the latter consists of the positive halves of sine waves having crest value Em and frequency
/ cojln. Let 9 = cot.
2f
Vd = 71
71/2 .
2E.
71/2
2E
= 0.637F.
e =
(a)
(b)
>
Vd
(c)
(1)
(d)
Em = -Vd=\.51lVd
(2)
1P
(e)
UlT
IV?
->- bit
i.
54
2-4
CONVERTER CIRCUITS
3.1427>d.
If the two valves in Figure 3 were reversed, so that the anodes instead of
the cathodes were at a common potential, nothing would be changed except
the direction of the direct current and voltage. Note, further, that the
same transformer could feed two pairs of valves, one pair connected as
in Figure 3 and the other pair reversed, as just discussed. Each pair of valves
could feed a separate load with a common return to neutral point N; in other
words, the rectifier would now be feeding a three-wire dc wire system. If the
currents on the positive and negative sides were equal, the neutral conductor
and the center tap on the transformer secondary winding would no longer be
needed. The result of such omission is the single-phase bridge rectifier, shown
in Figure 5. It is also known as a two-way circuit.
SINGLE-PHASE CONVERTERS
55
The number of valves, the direct voltage Vd, and the dc power Pd of the
full-wave rectifier have been doubled, but the P.I.V. of each valve is un
changed. Two valves (1 and T) conduct in series for one-half cycle, then the
other two (2 and 2') in the next half cycle. The valve and load currents are
unchanged. The transformer primary current is doubled. The transformer
secondary currents now have the same wave form as the primary current,
and thus the required secondary volt-ampere rating is equal to that of the
primary, being less than doubled.
The wave forms are shown in Figure 6, and numerical values are given in
Table 1.
Although the bridge circuit may appear more complicated than the fullwave circuit because it has four valves instead of two, the secondary winding
is used more effectively, and the P.I.V. of each valve has been halved for a
given dc output voltage. The former feature is important in high-power
vs=vn/T
(a)
(b)
nwr-
(c)
Dc
(d)
Vd
Tip
h h
ls ~
(e)
.If
I1*
03t
Fig. 6. Wave forms of the circuit of Figure 5 : (a) primary and secondary voltages v
and vs\ (b) unfiltered and filtered direct voltages vd and VA\ (c) voltage across valve 1, vi;
(d) valve currents i2 and i2\ (e) primary and secondary currents ip and is.
56
2-5
CONVERTER CIRCUITS
converters, and the latter in high-voltage ones, where the valve P.I.V. may
be a limiting factor.
THREE-PHASE CONVERTERS
57
3
z*=i r cos 0 dd = -{3/sin. 6V/3
I = V3
- = 0.828
(3)
Em = 1.209Fd
(4)
(5)
Em
a,a
Xj 0
whence
Jo
7E
direct voltage is smaller in magnitude and higher in frequency than in singlephase converters and therefore easier to filter. The same is true of harmonics
on the ac side, but this is not shown until Chapter 8.
Dc
Vd
(c)
-<
saturates the transformer cores. This may be avoided by replacing the Y con
nection by the zig-zag connection, in which the dc mmfs of the two secondary
windings on the same core cancel out. The circuit as shown, however, is useful
as a step in explaining several other connections, and therefore it is analyzed
as it stands.
The three secondary voltages ea , eb , ec form a balanced three-phase set, as
shown in Figure 8a, and the anode voltages with respect to neutral point N
are equal to the corresponding secondary voltages. The common cathode
voltage vd coincides with the upper envelope of this set of voltages, as shown
by the heavy line. The average direct voltage Vd is given by
(d)
UdlT
1
3 Ci)t
(e)
lpa
1IdIT
\
-t)f
58
2-5
CONVERTER CIRCUITS
The wave form of voltage across valve 1, shown in Figure 8a and b, consists
segment, while
of three segments, each lasting one-third cycle or 120. In one
other
two segments,
the
zero;
is
in
the valve is conducting, the valve voltage
to one of the
equal
voltage
inverse
has
an
while the valve is nonconducting, it
by
the height
shown
is
8a
it
Figure
In
line-to-line voltages, first eab, then eac.
axis. The
horizontal
a
from
redrawn
b
is
it
of the vertical shading; in Figure 8
P.I.V. is
(6)
y/3tn = %/3 x 1.209K, = 2.094K,
and rms value
The transformer secondary voltage has crest value E,
= 1.209 Vd x 0.707 = 0.855Vd.
Each valve conducts during the one-third cycle when the associated
rectangular
secondary voltage is the highest one. The wave of valve current is a
and its
is
value
IJ3,
average
Its
8c).
(Figure
120
length
pulse of height Id and
same
the
is
secondary
current
rms value is IJsj3 = 0.511ld. The transformer
second
the
from
differs
8)
(Figure
as the valve current. The primary current
average value is zero. Its
ary current in having no dc component; that is, its
rms value is
EJ\/2
_ 72/0.471/,
lTT2Y7JT'IY
3
rV V3/ 3\ 3/
3T
0.333Id = 2.094Pd
Primary;
3 x 0.855TVd x
= L209Pd
(8)
(9)
(10)
the circuit in
This is known in Europe as the Graetz circuit (Figure 9), In
except
as
before
operates
Figure 7, if the three valves are reversed, the circuit
The
same
reversed.
are
that the directions of direct current and direct voltage
each,
valves
three
of
groups
transformer secondary windings may feed two
similarly except for
one group connected as in Figure 7, the other connected
load, the two loads
dc
separate
reversal of the valves. Each group may feed a
to the trans
connected
constituting a three-wire load with its neutral point
neutral con
the
currents,
equal
former neutral. Now, if the two loads have
point of the
neutral
the
Since
ductor carries no current and may be omitted.
THREE-PHASE CONVERTERS
59
*4
r~-1
'c
O.
*5
m.
Li2
60
2-6
CONVERTER CIRCUITS
eb
ec
eb
PULSE NUMBER
61
(a)
(13)
0.707m = 0.707
(b)
V3E;
(c)
VaE;
(d)
w t
.2
4
<
ia
(?)
~5l
cot
v3
= q.428 Vd.
The load current is always carried by two valves in series, one from the
upper half bridge and one from the lower. Each valve conducts for one-third
cycle, as in the one-way circuit. Commutation in one group, however, is stag
gered with respect to commutation inthe other group ;considering both groups,
commutation occurs every one-sixth cycle (60). In Figure 9, as well as in the
diagrams of other converter circuits, the valves are numbered in the order in
which they fire (begin to conduct). Commutation occurs from valve 1 to
valve 3, then from 2 to 4, from 3 to 5, from 4 to 6, from 5 to 1, and from
6 to 2. The current wave forms are shown in condensed fashion in Figure 10d.
The current in each phase of the Y-connected secondary windings is the
difference of the currents of two valves, the numbers of which differ by 3 ; for
example, ia = it iA (Figure lOe). Its rms value is Id\j2j3 = 0.816/d.
The aggregate valve rating is 6 x 1.047Fd x I
J3 = 2.094/V
The aggregate volt-ampere rating of the transformer secondary windings
is 3 x 0.428 Fd x 0.816/d= 1.047Pd. The rating of the primary windings is
the same as that of the secondaries.
If the secondary windings are connected in A instead of Y, the wave shape
times that of
of current is different (Figure 10/), and its rms value is 1/
current in the Y. The primary line currents have the same wave shape as
secondary Y currents if the transformer connection is YY or AA, and the
same wave shape as secondary A currents if the transformer connection is
\/3
(f)
Fig. 10. Wave forms of the circuit of Figure 9 : (a) secondary line-to-neutral voltages ea
eb, ec and, in heavy lines, unfiltered voltages of positive and negative dc poles with respect
to transformer neutral point; (b) secondary line-to-line voltages and, in heavy line,
unfiltered direct pole-to-pole voltage vd; (c) secondary line-to-line voltages and, in lower
heavy line, voltage t?i across valve 1; (d) condensed representation of valve currents;
(e) transformer secondary current ia=iiu', (/) primary alternating line current
iA = dc ~ h)lT.
Va
SEm
Vd is given by
n/6
COS
(11)
SEm = -Vd=\MlVd
/
(12)
AY or YA.
2-6 PULSE NUMBER
Pulse Number p
1
2
2
3
62
CONVERTER CIRCUITS
,/
Fig. 11,
Dc
transformer.
64
2-8
CONVERTER CIRCUITS
65
CONVERTERS
2-9
-ffi. "5
cn
cs
o
cj)
J. -a
cq
.
U o
IO
cs
O co
O CO
""i
o o
o o
N N,
o
o
VD
>3
u
a>
.is
"N
O W">
5-1
>
n?
t-
o tO vo
V> n
P-
v->
00
CB
00 CO
r- o
r- o
fN
N, N
n >-n
o
--
O co
o>
60
T3
-Tf CN
oo
00
5b
rf 5=
t~-
'
fck.
00 00
o o
o d
c-i
PS
r-
WO
Tf
o
En
ft, ft,
00
oo
<'
00
3;
ii
3"
O O
fsj
i
I
3
o o
OS
wo wo
wo wo
oo
Os
so o
u"i
Os
<N <N
Tf
23
CO
n, >n
o
00 CO
i>
00
<N
-N
*5
o o
o o
w->
w->
cs <s
o o
t
o o
CO
W"l
co
O co
r-i
w-l w-l
ri
r-
w->
ON 00
<N
2 8*
oo
g co
'
.-O
En
t>
O CO
O co
En
Cl
WN 00
pa
rs
J
3
OS CO
n "*5
o NO
oh
1. For a given direct voltage, the P.I.V. of the valves is only half that of
En
d ci
do
-'
Tj- OS
Os Q
r4
ft- ft,
<S
o o
o o
*-
J-
Cl
o
>
s
N N
r- o
o o
r- o
O
>n
En
0h
67
any of the other six-pulse circuits except 5 and 8, in which it is equal. Conse
quently for a given P.I.V., the direct voltage is twice that of some other
O"
&o
T3
cq
rm
ON
Ft
23
"n
o rvn
cq
En
i--
oo r-
00
g<
>
rl- Tf
(S
W1
$s
En
tJ-
En
-e-
CS ' i'
O fN]
N
En
a
Is
\o
3"
o
"7
"n >n
CN
TWELVE-PULSE CASCADE
circuits.
2. For a given power throughput, the volt-ampere rating of the transformer
secondary winding is less than that in any other circuit.
3. The volt-ampere rating of the transformer primary windings is equal to
or less than that of the other circuits.
4. The transformer connections are the simplest. Double or center-tapped
secondary windings are not required. This simplcity is important for obtaining
a sturdy and reliable design for operation at very high voltages having super
posed high-frequency transient voltages caused by commutation.
5. The secondary windings may be connected either in Y or in A. The
advantages of this will appear in Section 2-9.
6. The aggregate volt-ampere rating of the valves is lower than that of
circuit 8 and equal to that of the rest.
7. A very important property of the bridge circuit is that arcbacks* can be
suppressed by grid control and a bypass valve. Arcback is a random phenomenon that is unlikely to occur in more than one valve at the same time. Thus,
although grid control is ineffective in the valve with the arcback, it may
reasonably be expected that in all other valves of the same bridge such control
will be effective and may be used to prevent any of these valves from reigniting
after the current in them has once become zero. In the bridge connection
there are two valves in series across the dc line and two in series opposition
across each pair of ac terminals. If all valves but the maloperating one are
blocked by grid control, that one has no circuit through which current can be
furnished to it. This is in contrast to the six-phase star and Y-Y-interphase
connections, in which there is a path through the ac source, the maloperating
valve, and the dc load.
Because of these advantages, the bridge circuit is universally used for highpower hv ac-dc converters.
sr,
<a
i>
ft ea
O
g>
+- S?
G
>*
x>
'5
S
!>
..
iS
E "2p
.o
O
- ggB
S"3 2 1/1
5>>H
-
ft
"3
W5
Jo
its
Ifil3
.2
J 73
*5 Q
>
>H
"
<a
<o
s 6e cS -ft
a i?
& S 2
>
53
*S 9
8
<HH
68
PROBLEMS
CONVERTER CIRCUITS
3 4> Ac
11T.
5J[
9V
7 y.
11::
3 y.
Dc
-=
21'.
12::
43:
i
Fig. 15. Twelve-pulse cascade of two three-phase bridges.
voltage in one bridge are staggered with respect to the other, giving a 12-pulse
converter. Under balanced conditions, the twelfth harmonic is the lowest one
in the direct voltage, and the eleventh and thirteenth are the lowest pair in the
alternating current.
The two bridges could be fed from one three-winding transformer bank
13.
14.
PROBLEMS
1. Prove that the average voltage across the smoothing reactor is zero if
the reactor has no resistance.
2. Prove that the half-wave rectifier in Figure 2 can conduct only inter
mittently.
3. Analyze the half-wave rectifier circuit in Figure 2, finding the wave
shapes and the average and rms values of the current and the sev
eral voltages as functions of EJVd.
69
15.
16.
17.
18.
19.
70
CONVERTER CIRCUITS
Draw the circuit diagram and the wave shapes, and find the peak,
rms, and average voltages and currents and the aggregate volt-amperes
of each set of transformer windings of converters described as follows:
20. Four-phase star, one-way, four valves.
21. Two-phase, center-tapped, one-way, with interphase transformer.
22. Four-phase, two-way, eight valves.
23. Cascade of two single-phase full-wave converters, fed from a two-phase
line.
BIBLIOGRAPHY
1. High Voltage Direct Current Power Transmission, by Colin Adamson and N. G.
Hingorani, Chapter 2, " Type of Converter Circuits and Valve Connections."
2. Power Transmission by Direct Current, by Ya. M. Chervonenkis, 1948, translated from
Russian to English by the Israel Program for Scientific Translations, Jerusalem, 1963.
Chapter IV, "High-Voltage Rectifier Circuits," pp. 32-44.
3. "D.C. Power Transmission," Part II, "Basic Principles," by E. Openshaw Taylor,
Elec. Jour. (London), Vol. 164, pp. 25-26, Jan. 1, 1960.
4. Rectifier Circuits: Theory and Design, by Johannes Schaefer, Wiley, New York, 1965.
5. High Voltage Direct Current Convenors and Systems, edited by B. J. Cory, Macdonald,
London, 1965. Chapter 2, "H.V.D.C. Transmission Circuits," by B. J. Cory, pp.
14-21.
3
Analysis of the Bridge Converter
modified thus:
1. Power source (or sink) consisting of balanced sinusoidal emfs of constant
voltage and frequency in series with equal lossless inductances.
The inductance, which really is chiefly the leakage inductance of the trans
formers, is the cause of the overlap. The other assumptions made in Chapter 2
are retained. They are :
2. Constant ripple-free direct current.
3. Valves which have no forward resistance and infinite inverse resistance.
4. Ignition of valves at equal intervals of one-sixth cycle (60).
Figure 1shows the bridge circuit with the notation adopted. The instantaneous
(la)
(16)
(lc)
ea =
eba = eb
= fZEmsm<ot
eCb = ec
eb
(2a)
(26)
(2c)
72
3-1
Ld=
-nsw
l>3
2.3
wb 5
Vb
Lc
Vd
Vd
Id
-V4
-120'
ve
120'
180'
240'
0)t
V2
Fig. 1. Bridge converterschematic circuit for analysis. The valves are numbered in their
firing order.
Fig. 3. Waves of instantaneous e'mfs of ac source: line to neutral, ea> eb, ec; line to line,
aci &bo>
74
3-1
75
(a) u = 0
60a
u
(6; u = 20
>6
>[<
<|
cot
60a-|
\\
V
\
ri
(c) u = 40
>
(e) u = 80
(f) u = 100
(g) u = 120"
sources be opened while other sources are acting. We obtain, for the instan
taneous currents and voltages in Figure 1 or 5,
h = h = E = h = H = *3
(3a)
(36)
(3c)
(3d)
(3c)
h=
VX = v2 = 0
3 = eba =
y4 = y5 =
= eCb =
V3
vd=
V3
sin cot
V.3
vn
(d) u = 60
12
rui
(3/)
(3g)
(36)
(3i)
Equations for these other intervals may be found from Eqs. (3) by appropriate
changes of subscripts and phase angles.
It was shown in Chapter 2 that an uncontrolled valve ignites (begins to
conduct) as soon as the voltage across it becomes .positive. If valves 1 and 2
have been conducting, valve 3 ignites as soon as eb becomes greater than
ea; that is, at point R, Figure 3. This instant is taken as cot = 0. At the same
instant, under present assumptions, valve 1is extinguished (ceases to conduct).
Valves having control grids can be made to delay ignition but not to advance
it. The delay angle is denoted by a and corresponds to a time delay of
ajco sec. If delayed this long, valve 3 ignites when cot a ; valve 4, when
cot = ot + 60; valve 5, when cot = a + 120; and so on. The delay a cannot
exceed 180; for example, point T, 180 after R in Figure 3, marks the end of
the range over which valve 3 can ignite, for only in this range is eb greater
than ea.
The ignition delay affects both the direct voltage and the ac displacement
factor, as will now be shown.
The instantaneous direct voltage vd across the bridge on the valve side
of the dc reactor is composed of 60 arcs of the alternating line-to-line voltages,
as shown by the vertical lines of the shaded area A0 in Figure 3 for the period
in which valves 1 and conduct with a = 0, or as shown in a different way
in Figure 6a.
The average direct voltage Vd is found by integrating the instantaneous
voltages over such a period or any 60 period. For a = 0 (no delay) and with
cot called 6, it is
3 C
=Vd0=-A0
71
71
~
(3j)
The condition of this circuit endures for only one-sixth cycle (60) at a time
and is followed by one in which valves 2 and 3 conduct ; then valves 3 and 4,
then 4 and 5, then 5 and 6, then 6 and 1,and so back to 1and 2. See Figure 4a.
71
eacd9=-71
J -/3
sin (6 +
30)T
m 2 sin 30 =
/-
J 60
J 60
3 Em
71
= 1.65 Em
(4)
3-1
1
- 3E,.n
2.34Etn
Vdo =
LN =
'LN =
71
Instant when valve 3
ignites in an
uncontrolled rectifier
77
1.35Er.r.
E
h/lEr.r.
= 1-35
Ell -
(5)
LL
71
voltage.
For delay angle a, both limits of integration are simply increased by a, as
shown in Figure 6b to g, giving
I.
VA = V.dO
a 60
(b) a = 30'
(c) a
60'
(d) a = 90'
wt
(e) a = 120'
(f) a = 150'
(6)
Thus we see that one effect of delayed ignition is to reduce the average direct
voltage by the factor cos a. Inasmuch as a can range from 0 to almost 180,
cos a can range from 1 to 1, and Vd can range from Vd0 to VdQ. Since the
current Id cannot reverse because of the unidirectional property of the valves,
negative voltage Vd in conjunction with positive current Id represents reversed
power flow; that is, conversion from dc power to ac power instead of from
ac to dc. In other words, it represents inversion as opposed to rectification.
Obviously a converter cannot invert unless it has grid control.
Figure 6 shows the waves of direct voltage on the valve side of the dc
reactor for various delay angles. It is noteworthy that the ripple and hence
the harmonics in the direct voltage increase with delay up to 90 and then
decrease from 90 to 180.
With losses in the converter neglected, the ac power must equal the dc
power; that is,
3ELNILl cos <f) = VdId = IdVdo cos a
(7)
-n/3
(g) a = 180'
]+60
71
-60
Fig. 6. Instantaneous direct voltage (shown by heavy line) of bridge converter with
ignition delay angle a but no overlap.
60
2V3 L =1.11/,
sin 60)
71
(8)
78
3-1
(a) a = 0
Ea
tibial
(b) a m 30
(o) a = 60
Ea
(d) a. = 90
Ial
Ea
(e) a = 120
hi
Ea
I\+
a *\
i
(f) a = 150 /
la1
Inverter
Rectifier
Inverter with
forced commutation
Rectifier with
forced commutation
iLl =
id
= o.mid
n
79
in which cos <j> is the displacement factor or vector power factor and <p is the
angle by which the fundamental line current lags the line-to-neutral source
voltage.
From Eqs. (7) and (9) we see that the converter operates as a sort of trans
former having a fixed current ratio but a voltage ratio that varies with the
ignition delay imposed by grid control.
With no ignition delay, the rectangular pulses of alternating line current
are centered on the half sine waves of line-to-neutral voltage, as shown in
Figure 7a, and the fundamental sine wave of current is in phase with that
voltage wave. Ignition delay a shifts the current wave and its fundamental
component by angle = a, as shown in Figure lb, c, d, e, and f Thus the
converter rectifier or inverter draws reactive power Q from the ac system.
See the vector power diagram in Figure 8. The rectifier is said to take lagging
current from the ac system, and the inverter is said either to take lagging
current or to deliver leading current to the ac system. The signs of active and
reactive power in a rectifier are like those in an induction motor, and, in an
inverter, they are like those in an induction generator. The analogy between
the inverter and the induction generator may be extended a step further: both
work best in parallel with synchronous machines, but both may be made to
work with shunt capacitors instead.
A converter having valves with normal control circuitry can work only in
the upper half of the complex power plane (Figure 8), because ignition can
be delayed but not advanced from its uncontrolled position. Imagine a con
verter in which the valves are replaced by synchronously-controlled switches,
the closing times of which can be shifted in either the lagging or the leading
direction. By operating with the switches closing on negative instead of
Ea
Ea
(?)
= 2.34LN yields
(10)
80
3-2
Vb
Vi=
j
j
4?)-
At end (ft)/ = a + u 5) :
anc*
*3 = 0
(H)
and
i3 = Id
(12)
81
io
i>6
"n
At beginning (ft)/ = a) :
60
<?-TO CP
*b
But let us return to an analysis of normal operation with 0 < w < 60 and
with alternately two and three valves conducting. The interval in which valves
1 and 2 conduct was analyzed in Section 3-1, with the results given by
Eqs. (3). This interval ends at cot - a, when valve 3 ignites. In the next
interval, the effective circuit is that in Figure 9, with valves 1, 2, and 3 con
ducting. During this interval the direct current is transferred from valve 1 to
valve 3. Hence
die,
di,
r
r
= L ~ ~ L ~
dt
The emf in this loop, known as the commutating emf, is see Figure 2 and
Eq. (2b):
~ ea =
(13)
dt
Since
\/3 Em sin
(14)
ft)/
h = Id ~~ *35
then
dh _ q _ dh
dt
dt
V3 En
|
|
2Z,
where
sin co/ dt =
Is2- 2coLr
du
Id
(15)
(16)
Equation (15) shows that z'3, the current of the incoming valve during com
mutation, consists of a constant (dc) term and a sinusoidal term (see Figure
82
3-2
10). The sinusoidal term lags the commutating voltage by 90, as it should in
a purely inductive circuit, and has a crest value hi which is that of the current
in a line-to-line short circuit on the ac source. The constant term, which
serves to make i3 = 0 at the beginning of commutation, depends on a, and
for a = 0 it shifts the sine wave upward by its crest value. The current /
of the outgoing valve has a sine term of the same amplitude as that of i3 but
of opposite phase, and its constant term serves to make / = Idat the beginning.
Currents and i3 would reach highpeak values if the short circuit remained,
as infrequently happens because of an arcback (Chapter 6). Normally, how
ever, the short circuit is removed when the current in. the outgoing valve
reaches zero, for normal valve action prevents it from reversing.
During commutation the line-to-line voltage of the short-circuited phase is
90'
7s2 (cos a - cos cot)
"1"
va = vb =
~2
C0S
&c
/T7\
for the two phases of the source have equal impedances jcoLc and act as a
voltage divider.
During the overlap interval of valves 1 and 3 (a < cot < a + u), the instan
taneous values of all currents and voltages are given by the following
equations :
(18zz)
h=-h=-h
(18c)
*4 = *5 =*6 = 0
va = vb = vp 0.5ec = 0.5Em cos cot
(18tO
(18/)
v4 = v5 = v6 = vd =
83
zero, and the two line-to-neutral voltages are equal to each other and to the
average of the corresponding open-circuit voltages:
Ui = v2 = v3 = 0
cot
60
(18e)
(18g)
(18/r)
(18i)
At the end of the commutating period, Eqs. (12) hold, and substitution of
them into Eq. (15) yields
Id = Is2{cos a cos 5)
Fig. 10. Currents it and i2 during commutation of valve 1 to valve 3 as arcs of offset
sinusoidal waves of line-to-line short-circuit current of amplitude Is2, with (3 lagging 90
behind the commutation voltage eba.
(19)
This gives the direct current in terms of the ignition and extinction angles.
A similar equation will now be drived for voltage drop AVd due to overlap.
In Figure 12 the effect of overlap is to subtract an area A from the area A0
Table 1. Instantaneous Currents in Three-phase Bridge Converter with Overlap Less than 60'
(Alternately Two and Three Valves Conducting)
Valves
Con
ducting
1,2 cos a
2, 3, 4
3,4
3,4, 5
4, 5
4, 5, 6
5,6
I,2 cos 6
/s2 COS a
/s2 cos(0 60)
Id
Id
Id Is2 cos
+/s2cos(0 60)
0
Id
Id IS2 cos a
/s2cos(0 + 60)
Id
Id
Id I,2 cos a
Is2 COS 0
I,2 COS a
Id
0
0
6, 1,2
+1,2 0
1,2 COS 0
Is2 cos a+
-Id
1,2 cos
Id
Id
Is2 COS a
Id
+ IsZIdcos 0
Id 1,2 COS a
Id
Id
/s2cos(0 60)
Id
-h
Id
Id+h2 COS a
/s2cos(0 60)
0
Id
$2 cos a
Id~ I
/s2Cos(0 + 6O)
I,2 cos a+
1,2 cos(0-i-6O)
- Id + 1,2 COS a
+ 1,2 COS 0
1,2 COS a
1,2 COS 0
-Id
I,2 cos a
Id
Id I&2 COS a
1,2 cos (060)
60)
+1,2 cos(0
Id
-Id
Jd + /s2C0S
/s2cos(0+6O)
si cos a
h I
+I,2 cos(0
-Id
1,2 COS a
Id~Is2 COS
+1,2 cos(0+6O)
-h
COS
Id
2, 3
6, 1
I'd COS
+/s2 cos 0
1,2,3
5, 6, 1
1,2
+ 60)
1,2 cosa+
/s2cos(0+6O)
Table 2. Instantaneous Voltages in Three-phase Bridge Converter with Overlap Less than 60
(Alternately Two and Three Valves Conducting)
Valves
wt=6
conducting
V4-
Vs
V6
Va
vb
Gba
Gca
Gca
Gcb
Go
Gb
l.5ec
i.5ec
&ab
Gca
Gcb
-l.5eb
Gcb
6 be
Vi
Vz
1,2
1,2,3
vP
Vn
Gc
Ga
Gc
Gac
Gc
-ed2
Gc
1.5ec
Gcb
Gbc
Vc
Vd
8 60
/y
<JL
8
2,3
a
Ga
Gb
ec
Gb
Gc
-Gbl2
Gb
ebj2
Gb
es/2
Gab
Ga
Gb
Gc
Gb
Ga
Cba
l.5ea
ea
Ga
1.5ea
G ab
Ga
Gca
+ 60
'2,3,4
-1.5e l.5eb
8+60
3,4
Gab
1.5eb
a -1-120
3,4,5
l.5ea
1.5ea
Gac
Gac
S -!- 120
4,5
a
+ 180
8 + 180
a
+ 240
8 + 240
4, 5,6
Gb
-GcS2 -ec/2
Gac
Gbc
Gbc
Gba
Ga
eb
5, 6, 1
1.5e&
1.56
l.5eb
-Gbl2
Gb
6,1
Gbc
Gba
Gba
Gca
Ga
Gb
6,1,2
5,6
a +300
8 + 300
-1.5ea
l.5ea
1.5ea
Ga
Gc
Gc
ea
Gc
Gc
-eJ2
Gc
Gc
Gb
Gcb
Gb
1.5eb
Gb
Gab
-Gb/2 -ebj2
Gc
-Gail ea/2
Ga
Ga
-eaj2
1.5ec
1.5ea
3-2
60
87
(a) JO
Nli
>
ea + e&
ec
(at
*ab
*bc
eb)dQ= \"eJde
U. - e" +
2
A=
*ba
r u
sin G dQ =
V3 Em (cos a
3
A
AKd = 71
1.5e
bb
8S
OO
to 10
oo
CJ CM
Fig, 11. Instantaneous currents and voltages of three-phase bridge converter with
a = 15 and u= 15: (a) valve currents; (6) line currents; (c) alternating line-to-neutral
voltages; id) direct positive and negative voltages with respect to neutral point of ac source;
(e) direct voltage between poles; (/) voltage across valve 1. Voltages across other valves
have the same wave form but are displaced 60 apart; 4/ means valve 4 ignites; 2e, valve
2 extinguishes; etc.
-25
[-cs0J
cos 5)
.
= --(cos a cos 5)
2n
(20)
Without overlap, the direct voltage was Vd0 cos aEq. (6). With overlap,
it is
Fj0(cos a + cos 8)
Va = VAn cos a - AVa =
(21)
Comparison of Eqs. (19) and (20) shows that the voltage drop is directly
proportional to the current:
Vdo
2Is2
(22)
88
Hence
Vd = Vd0 I cos a - 21
(23)
sZ
- coLc = - Xe = 6fLc
Rc = 71
71
ohms
(25)
89
for given values in volts and amperes, respectively. However, the ratio
VJIs2, which is the base dc resistance, is independent of the alternating
voltage, and so is the per-unit value of commutating resistance R'c.
Second System. A more usual base direct current than
of course, is the
rated or full-load current Id. The per-unit current on this new base is Id =
IJIdn. For reasons that will appear later, it is desirable to adopt at the same
time a new base direct voltage Vd0n, which is the value of Vd0 under normal
full-load rectifier operation with terminal voltage Vd at its normal value
Vd. Vd0n must be distinguished' from Vd0, for only one of them can remain
constant while the direct current varies. The per-unit direct voltage on this
new base is Vd = VdjVd0n.
We shall show that in this system there is a simple relation between the
per-unit commutating resistance and the per-unit commutating reactance
based on the rated voltage and current of the converter transformer. These
transformer ratings must be expressed in terms of the base voltage and current
adopted for the dc circuit. The rated line-to-neutral rms valve-side transformer
voltage is, by Eq. (5),
1
Vd0
&LNn =
Rc = 6fL
(27)
vW
Vdo cos a
First System. Let Vi0 be chosen as the base voltage and Is2 as the base
current of a per-unit system. Division of Eq. (23) by Vd0 yields
(26)
' V'd = cos a - \Id
where
/ = 4,72/3 = 0.8164,
,b
Eln
4/1 6
Vd0
Id
(Z9)
V'd = VdjVd0 is the per-unit direct voltage, and l'd = Idjls2 is the per-unit
direct current on the chosen bases. On the same bases the per-unit commutat
ing resistance is R'c =
It may be noted that in this per-unit system the base quantities Vd0 and
Is2 both vary in direct proportion to the alternating voltage referred to the
valve side of the transformers, which voltage is proportional to that on the
other side of the transformers and to the turns ratio, which is adjustable.
Hence a change in voltage affects the per-unit values of voltage and current
(28)
X,.
{7tRf\f6Vd0\
2R,
*=zrlrAJ-i=2*'
(30>
Hence
Y"
r"=t'
See Appendix A for more accurate values with overlap.
(31)
90
3-2
60
91
that is, the per-unit commutating resistance on the new base is half the per-unit
commutating reactance based on the transformer rating. Equation (23) becomes
which is exact only if u = 0 but which is true with a maximum error of 4.3%
at u = 60 and only 1.1% for u 30 (the normal operating range). It follows
that the displacement factor is
cos a -
, cos a + cos 8
-cos (p -
Vd = Vd0
or, dividing by Vd0 and putting
V'J =
I'd
Vd0n
(32)
Vd0/Vd0n = V'JQ,
V-cosa-l'd'
(40)
(33)*
cos (f>
cos a
' dO
(41)
' dO
36(cs + cos
(34)
P.=P*
(35)
Pa = 3ELNIL1 cos $
(36)
Pd=VdId
(37)
where
Vd s
Eln cos 0
The concept of the converter's having a fixed current ratio IJI and an
adjustable voltage ratio VJELN is still valid. Now, however, the current ratio
varies a few percent with load, and the ac displacement factor depends on
loadEq. (41) in addition to ignition delay angle.
Equations (35), (39), and (42) give approximate relations, not involving
the converter angles a and <5, for power, current, and voltage, respectively.
These three equations, in terms of per-unit quantities on any power base
equal on both sides and on any pair of current bases proportioned as in
Eq. (39), become
(43)
(44)
and
Equating the right-hand sides of Eqs. (36) and (37), substituting Eq. (34)
for Vd, and solving for the active alternating current, we get
\/cos a + cos
'cos*=(VX
2 j
(38)
(45)
Q = Pa tan <{>
(46)
where <j) is given by Eq. (40) or (41). Of course, there is no reactive power
on the dc side.
/I]S76Wmo
(42)
(39)
92
3-3
of the transformers, that is, the factor by which the ratio is off its nominal
value Tn.
Then Eq. (43) is unchanged, but Eqs. (44) and (45) become
()
(48)
cos a + cos 8
....
(49)
93
60
(1)
(k)
<s>
and
. ,
2u + sin 2a sin 25
77
cos 5)
4(cosa~
....
(d)
where 2u is in radians and ILl0 = (J~6}tc) Id. The phasor fundamental current
is
(51)
hi ~ hi cos <f> -jhi sin <j>
and the scalar fundamental current is
hi =
cos 4>f
(52)
= P tan
<f>
(53)
where
cos 2a cos 2o
2u + sin 2a
tan <p = -
sin 25
Fig. 14. Bridge converter with valves 1, 2, 3, and 4 (or 6, 1, 2, and 3) conducting.
a pole-to-pole short circuit on the dc terminals. When three valves are con
ducting (Figure 9), they constitute a line-to-line short circuit on the ac source,
as already considered in Section 3-2. In the four-valve subintervals, the direct
voltage is zero; in the three-valve intervals, it is an arc of a sine wave of
amplitude 1.5Em (the altitude of the voltage triangle in Figure 2).
-tt
The current i3 in valve 3 will be traced through the three subintervals that
constitute the interval of angular span u in which i3 rises from 0 to Id. During
this same interval ix decreases from Id to 0 and is always equal to Id i3.
Operation of the bridge converter with overlap angle in the range between
60 and 120 is abnormal, being encountered only under overload, dc short
First Subinterval. The first subinterval extends from cot = a to cot =5 60.
Valves 1, 2, 3, and 6 are conducting, and the upper one of each pair of
quantities in Figure 14 should be used.
94
3-3
,c-et-2Lcf-Le|=
(SO
I'X
(57)
3Lc
ea + eb + ec=0
(58)
Lc
dt
dt
(59)
Similar equations hold for currents in the other two phases. Of course, these
coLr
+ 60) d{cot)
cos {cot
(60)
_ 2
(b 1)
a.
60 and
(62)
cot = a
2coL
<5 _6Q'
fx
fx
(64)
Tcos
cos cot
(65)
By vector addition,
-cos (5 - 90)
cos {d
(66)
h = 7js3
jcs
(67)
cos cot
where
cot
<"'
Hence
-/,3[cosM-30)rw'
= /s3[cos (a 30)
V"3 E,
to which must be added the initial value, equal to the final value in the pre
ceding subinterval and given by Eq. (62). The sum is
rat
<3 =
(63)
Integration gives
Lc
95
f =zf=4rcos(ffl'-90O)
(55>
dt
60
dt
i3 =
cos
(a 30)
+ \ cos (5 + 30)
cos (a +
60)J
u-ijy/icos tan-1
[f
+ icos
0)j
(68)
cot = 5. Valves
96
3-3
subinterval except that valve 3 is now in series with phase b, so that the ac
component is 90 behind Eb . The dc component is different from what it was
in the first subinterval. The total current in valve 3 is
i3 = ib = Is3 cos (cot 150) 4-
(69)
73
/s3cosa tan1-y|+
j'3 - /s3j-~cosa tan-1
(70)
(71)
At the end of the subinterval and whole interval, cot = 6 and i3 = Id. This
fact enables us to obtain the direct current as a function of the ignition and
extinction angles.
jd = /s3[ cos (a - 30) + icos (S + 30) - cos (3 + 30)]
cos cot
(72)
(75)
(76)
Vl = v2 = v3 = 0
(77)
v4 = vs=v6 = vd
Va
II
Vc
= v =
II
\.5ec
ec = Em cos cot
oo
(79)
(80)
(81)
i2
(82)
(85)
(87)
7s3cos(a
h = is = h = 0
= v2 = v 3 = v4 = v5 = v6 = va = vb = vc = vp = vn = vd = 0
i2 ic = Id
cos (a
V3
ix = ia = Jd
II
97
h = K3[ cos (a
ii
60
= Is3[i cos
*3
cos cor
(73)
(74)
Table 3. Instantaneous Currents in Three-phase Bridge Converter with Overlap Greater than 60
(Alternately Three and Four Valves Conducting)
Valves
!l3
conducting
ii
ii
iz
is
is
ic
ib
la
y'
8-60
a
+ 60
1,2,3
Id A*
1,2, 3,4
Id -A
-Id
C-h
ld-B L -C
~D
-E
Id
-Id
h-G
Id
to
8
a
+ 120
+ 60
+ 180
,8
+ 120
+ 240
8 + 180
a
D-h
Id -D
2, 3, 4, 5
Id -E Id -F
3,4,5
3,4,5,6
h -H Id -K
K-h
h- H-K
4, 5,6
Id -L
Id
L-h
-L
Id
4, 5, 6,1
.0
Id M L-N
M+N-h
h-N
5,6,1
.0
h-P
Id
-Id
h-P
5,6,1,2
Id Q
Id-R
R Id
h- Q-R
Id-F
+ F-Id
+300
6,1,2
6, 1, 2, 3
2, 3,4
Id
Id-U
S
T
h-S
Id-T
-3
Table 3A
A = 7s3[cos (a - 30) cos (8 30) - (V3/2) cos 0]
B 7s3[ cos (a 30) i
cos (8 30) cos (0 30)]
C /s3[cos (a 30) cos (0 90)]
D = 7s3[cos (a - 30) + cos (8 30) - (Vl/2) cos (0 - 60]
E = 7s3[ cos (a 30) + cos (8 30) cos (0 30)]
F 7s3[cos (a 30) cos (0 150)]
G = 7s3[cos (a - 30) +* cos (8 30) - (VJ/2) cos (0 - 120)]
H 7j3[ cos (a 30) +% cos (8 30) cos (0 90)]
K = 7s3[cos (a 30) cos (0 210)]
L = 7,3[cos (a - 30) +icos (8 + 30) - (Vl12) cos (0 - 180)]
M= 7,3[ cos (a - 30) +h cos (8 30) - cos (0 - 150)]
N=Is3[cos(a -30) - cos (0 - 270)]
P = 7s3[cos (a - 30) J cos (8 30) - (V3/2) cos (0 - 240)]
Q = Jssti cos (a - 30) icos (8 30) -cos (0 -210))
R = 7s3[cos (a 30) cos (0 330)]
+
+
+
+
+
+
-S
. .. U.
cq
T+U-h
-T
13. lb
in
101
ii
v->
(a)
'
<a
'
(b)
11
1
fS3I
in
>n
<w
!Q
*1
is
(c)
J=
"jo-
Cu w
g $
HH
.S >
Q>
(A
hS
f1
eba
If
1.5e,
1.5ec 1.5<?2, /
Nr l'5ea 1.5ec
1.5eft
1.5ca
1.5e,
<
in_in_|n__
_-
(d)
1
H
in
in
in
$ .S
4-
1
g
o
o
3
!lII
CO
N-
<N
CO
CO
<N
CS
<N
oo
+a
CO
*t
'S-
>
+
to
<n
in
ro*-
co*
n
0
CO
SO
+
GO
oo
1-"
+a
GO
+a
it
<N
-T
cN
"
n"
in
Tf"
m*
0
oo
ii
CO
+a
GO
CO
-T
o
*
0
CN
GO
CO
co
Fig. 15. Instantaneous currents and voltages of three-phase bridge converter with
and u = 75 : (a) valve and phase currents; (b) alternating line-to-neutral voltage,
phase a; (c) direct positive and negative voltages with respect to neutral point of ac source;
{d) direct voltage between poles (broken line) and voltage across valve 2 (solid line).
a = 45
102
3-4
COMPLETE
CHARACTERISTICS OF RECTIFIER
103
voltage is, by Eq. (78), vd = \.SEm cos 6. The average voltage over 60 (rt/3 rad)
is.
/*a+60
Vd = -71
J <5-60
_]c5 60
/3
~|a+60"
qp
- 90)
1.5m cos e dd = cos(6
271
n/3
+ cos (5 + 30)]
(88)
with a' and S' as in Eq. (72). Equation (88) for u > 60 is analogous to Eq. (21)
for u < 60.
Direct Voltage as a Function of Direct Current for Constant Ignition Angle
By elimination of co S between Eqs. (72) and (88), we get
=V3coS(a-30)-
"dO
(89)
*s2
I.
Fig. 16. For explaining that the minimum ignition delay angle is 30 if the overlap angle
is 60 or greater.
y'd =
where
V'd = VdjVd0
and
Vd
cos (
30)
1.5/rf
(,
(90)
(91)
where V"d = Vd{Vd0n, Vd0 = KJ0/Fd0n, I"d =IJId, and X"c is the per-unit
commutating reactance based on the rating of the converter transformer.
Comparison of Eq. (90) with Eq. (26) and of Eq. (91) with Eq. (33) shows
that the equivalent commutating resistance for u > 60 is three times that for
u < 60.
Minimum Ignition Angle
If the overlap is greater than 60, ignition of a valve in one half bridge
must begin while the preceding commutation is still under way in the other
half bridge (Figure 4e), putting a line-to-line short circuit on the ac source.
For example, valve 3 must fire while valves 2 and 6 are still commutating and
putting a short circuit on phases b and c. The anode of valve 3 is at potential
vb = (fc + O/2 = -eJ2} and its cathode is at potential va = ea. See Figure 16.
Valve 3 cannot ignite until its anode potential becomes a little greater than
its cathode potential, that is, until vb > va, which, for balanced alternating
voltages, occurs first at point B, where cot = 30. If it were not for the short
circuit, vb would be eb, and ignition would occur at point A, where cot = 0.
Hence the minimum ignition delay angle is 30. This delay occurs spon
taneously even with no grid control.
The same phenomenon may be seen in the voltage v2 across valve 2 in
Figure 15d. This voltage becomes positive at cot = 30, which for valve 2
corresponds to a = 30.
.
104
Second mode
Three valves conducting.
a increases from 0 to 30.
u = 60.
I'd increases from 0.500 to 0.866.
V'd decreases on a curve from 0.750 to 0.433.
Third mode
Alternately three and four valves conducting.
a = 30.
u increases from 60 to 120.
I'd increases from 0.866 to 1.155.
V'd decreases linearly from 0.433 to 0.
A rectifier with grid control set for a = a0, where a0 is between 0 and 30,
has three modes similar to those of the uncontrolled rectifier :
First mode
Alternately two and three valves conducting.
a = a0.
u increases from 0 to 60.
Second mode
Three valves conducting.
<x increases from a0 to 30.
u = 60.
Third mode
Alternately three and four valves conducting.
oc = 30.
u increases from 60 to 120.
For a 30, the second mode disappears, and transition is made from the
first mode directly to the third, with a = a0.
The spontaneous increase of a in the second mode is sometimes called
auto phase control The incoming valve cannot ignite until the polarity of the
voltage across it is reversed by the completion of commutation in the other
half bridge. The situation is similar to that already described in Section 3-3,
page 102, for u > 60.
The direct current and voltage at transition points between modes can be
calculated by means of the equations already given for Id and Vd as functions
of a and 5. The relations will be clarified by study of the first chart described
in Chapter 4.
Normally the rectifier operates in the first mode. The other modes occur
3-5
INVERSION
105
3-5 INVERSION
General
Because the valves conduct in only one direction, the current in a converter
cannot be reversed, and power reversal can be obtained only by the reversal
of the average direct voltage Vd. The voltage then opposes the current, as in
a dc motor, and is called a countervoltage.
In Section 3-1 it was shown that, if there is no overlap, Vd reverses at
a 90 ; that is, rectification occurs for 0 < a < 90 and inversion for
90 < a < 180. Since in a practical case there is always some overlap, the
value of a at which inversion begins depends on the overlap, or on the
current, and may be found from Eq. (21) or (24). From Eq. (21), the transi
tional value is
(92)
discussed in Chapter 6.
The commutation voltage for hv dc inverters is furnished by synchronous
machines. If the ac system receiving power from the dc line has no generators,
or none running, a synchronous condenser is usedin Gotland, for example.
Some small single-phase and polyphase inverters use tuned LC circuits
instead.
106
3-5
inversion
107
h = Ucos y - cos p)
(96)
_ Fd0(cos y + cos p)
(97)
(98)
__ cos y + cos p
(99)
F,=
cot
1/r>
v*
'
Rectifier
X'cI'd
,
=o(cos y-cos/3)
(100)
P0(cos y + cos P)
2
(101)
For constant ignition advance angle P, Eqs. (24), (26), and (33) become,
respectively,
Fig. 17. Showing relations among angles used in converter theory and why the curvature
of the front of a current pulse of an inverter differs from that of a rectifier.
extinction angle 8 is measured by the delay from that same instant (cot = 0 for
valve 1). Although angles defined in the same way and having values between
90 and 180 could be used in inverter theory, commoner practice is to define
ignition angle p and extinction angle y by their advance with respect to the
instant (cot 180 for ignition of valve 3 and extinction of valve 1) when the
commutation voltage is zero and decreasing. See Figure 17. The relation
among the several inverter angles is as follows :
P = n-ot
(93)
y =n
(94)
u =8
a = 13 y
(95)
V'd
(102)
(103)
(104)
(105)
(The corresponding equations in per-unit quantities are like (103) and (104)
but with reversed sign of the last term.) Under this condition the equivalent
commutation resistance is Rc and is negative. It may be likened to the
voltage drop per unit of current due to the series field of a differentially com
pounded dc motor, forced to run at constant speed by driving an induction
generator.
108
Id-
.
5
Vdo COS B
T-
(a)
lA/
W9
VdO COS 7
(b)
Wave Forms
Inverter
Ignition
V~3 Em sin a
V 3 Em. sin ft
(106)
Extinction
V3 Em sin 8
3 Em sin y
(107)
I I
[
<&c1 1
O
lO <>
<Q
1 1
oo
'
co
co r~1
CM CM
~H
l l
CO
"S-
CM-CM
II
Fig. 19. Wave forms of currents and voltages of three-phase bridge inverter with a = 150,
u 15, and y = 15: (a) valve currents; (b) line currents; (c) line-to-neutral
alternating voltages; (d) positive and negative direct voltages with respect to neutral point
of ac source; (e) voltage across valve 2; (/) direct voltage between poles.
jS = 30,
aJ
TT
tTr?
( t.. 7.
Ct-ts*.
110
3-5
111
INVERSION
Voltage
Current
(a)
*60
-60'
(a) First range
Voltage
60s-
-60'
Fig. 21. Relations among inverter angles right after extinction of a valve: (a) /3 = 45,
u = 30, y = =15; (b) jS = 75, = 45, y=30, =15; (c) jS = 97.5, =52.5,
y = 45, =15. A 1.5Em sin(a>? -f 30); B = V 3Em sinatf; C l-Sism
(lot - 30).
(b)
Fig. 20. Wave forms of voltage across a valve in (a) inverter and (b) rectifier: 1. inverse
There are also four minor voltage jumps per cycle, two of which are half as
great as the jump at ignition, and the other two, half as great as the jump at
extinction. These occur at moments when other valves are being ignited or
extinguished.
Commutation Margin
A distinction is made between extinction advance angle y and commutation
margin angle . The two are often confused, because they are equal under
Commutation margin angle is the time angle between the end of conduction
and the reversal of the sign of the nonsinusoidal voltage across the outgoing
valve.
The voltage across an inverter valve (Figures 20a and 21a) has a positive
< 60c
=y
(108a)
(1086)
Second Range (Figure 216). The front of the dent is partly above and
partly below the horizontal axis.
60 <
< 90c
= 60 u = y (fi~ 60)
(109a)
(1096)
112
3-6
Third Range (Figure 21c). The front of the dent is entirely below the
horizontal axis.
P > 90
C = y-30
(110a)
(110b)
These laws can be verified by a study of Figure 21 inthe light of the follow
ing considerations. The ignition and extinction advance angles are measured
back from the instant of reversal of the commutating voltage. For the com
mutation that results in the extinction of the arc in the valve across which the
voltage is graphed in the figure, this instant is marked 0, and the extinction
angle y for this valve is indicated; however, the ignition of the incoming
valve of the same commutation is indistinguishable in the graph. The ignition
and extinction in the next commutation (occurring in the other half bridge)
occur 60 later, at the front and tail of the dent D. Hence its front is 60
after (or /? 60 before) the reference point O, and its tail is 60 y after the
reference point. The sine wave forming the upper limit of the dent leads the
main wave by 30.
;
\
anodes. Either scheme requires a current divider for forcing equal division of
the current among the several parallel paths. Otherwise, the first anode to
begin conduction would limit the voltage of the remaining anodes to the arc
drop, which is insufficient to ignite arcs to those anodes.
A precaution usually taken with multianode valves is to delay ignition by
a minimum angle of about 5 to 7, so that there is a definite and adequate
voltage from each anode to the cathode just before ignition. This increases the
probability that all anodes ignite simultaneously as soon as the control grids
are made positive. This can be done with little effect on the average direct
voltage, which varies as cos a ; the voltage across the valve just before
ignition varies as sin a; for example, with valves having peak inverse voltages
of 100 kV, by changing a from 0 to 7, the firing voltage is changed from the
bare minimum (about 2 kV) to 100 sin 7 = 12.3 kV, and the no-load direct
voltage is decreased from 95.5 kV to 95.5 cos 7 = 94.8 kV, or only by
j
j
1.
2.
3.
4.
transformers/ratio 1:4
\JL>
rwr
imn
VW
iJVW
Anodes
113
0.7%.
I
3-6 SERIES AND PARALLEL ARRANGEMENTS OF VALVES, ANODES,
OR BRIDGES
"
Cathode
114
3-7
The transformers and the resistors across their secondary windings main
tain almost equal currents in the several anodes. At first glance, the use of
transformers of direct current appears strange. It must be remembered,
however, that the current in each valve (except the bypass valve, to be dis
cussed in Section 6-2, page 199) is a train of pulses having both ac and dc
components. Transformation of the ac components, together with the clamp
ing effect of the unilateral conduction of each anode, suffices to divide the
dc components also.
The resistors further aid in maintaining equal current division by increasing
the voltage on the anodes carrying less than their share of the total current
and decreasing the voltage on the anodes carrying more than their share. On
the assumption that each secondary winding in Figure 22 always carries onefourth of the primary current, if each anode likewise carries one-fourth, there
are no currents in the resistors. If the anodes carry unequal currents, the
surplus or deficit of each anode current with respect to the normal current
flows through the resistor in the same or opposite direction, respectively,
introducing equalizing voltages into the anode circuits.
By making the resistors nonlinear they perform the additional function of
limiting the voltage across each transformer winding. This voltage is adequate
in amount and duration for ignition at the minimum delay angle. The
product of voltage and duration equals the change of core flux and is limited
by saturation. If the voltage were not limited, it would be much higher and
shorter at large ignition delays (near 90), and the transformer insulation
would have to withstand the higher voltage. The nonlinear resistors limit the
voltage at large delays, giving a pulse of magnitude and duration not too
different from that at minimum ignition angle and avoiding the need for
excessive insulation.*
The need for bridges in parallel has not yet developed in hv dc transmission.
Two bridges could be connected through an interphase transformer in the
manner shown in Figure 12 in Chapter 2 for parallel connection of two half
bridges. The two bridges would probably be fed through transformers con
nected for a 30 phase difference between one bridge and the other so as to
give 12-pulse operation of the converter.
Valves in series as opposed to bridges in series have the advantages of
fewer and larger transformers or transformer windings. However, they have
the following disadvantages:
No reduction in harmonics
More valve auxiliaries at more different insulation levels with respect to
ground
Uneven voltage distribution across the valves as a result of stray capacitances
MULTIBRIDGE CONVERTERS
115
In the present state of the art, two or more bridges in series on the dc side
needed for obtaining as high a direct voltage as required for
are usually
economical transmission.
The extension of converter theory to converters having B bridges in series
on the dc side requires consideration of (a) harmonics, (b) relations between
ac and dc quantities, and (c) the effect of mutual commutating reactance on
inverter operation.
Harmonics
116
3-7
Higher pulse numbers can be obtained in converters having more than two
bridges. For example, an 18-pulse 3-bridge converter and a 24-pulse 4-bridge
converter are possible. The transformer connections required, however, are
more complex than those for 12-pulse converters, and most hv dc engineers
think that it is more practical to build a 12-pulse converter provided with
additional filter capability than to reduce harmonics by use of a pulse number
higher than 12.
Harmonics are discussed further in Chapter 8.
Pr+JQr
I,
Pi +JQi
1\BrTr
cos a
IL1 = 0.780Sr/d
amperes
(111)
and
Vd=l35BTELLco$4>
volts
(112)
Still, of course,
Pa = Pd
watts
(113)
The equations in per-unit quantities eqs. (47), (48), (43) are unchanged
but must be interpreted with respect to new bases that are for the whole
converter instead of for one bridge.
The equivalent commutating resistance in ohms of a -bridge converter is
B times the value for one bridge, but the per-unit value of this resistance is the
same for B bridges as for one bridge and gives the same per-unit voltage drop
PJdIVdo
P-
L ~Rci
Kl
Rcr
\AAoyyV /*//>oT>
h>
cos y
<3
BiTi ' 1
Ini
VrfOr
cos a
XX
Vdr
Vdi
Vjoi
T
COS
X
Ac -
-o
117
MULTIBRIDGE CONVERTERS
X
-Line-
inverter-
Dc-
Ac-
Fig. 24. Equivalent star circuit of the ac circuit between two converter bridges (Br. 1,
Br. 2) and a common voltage source E.
its branches are predominantly inductive and are assumed to be entirely so.
Tr is an ideal transformer having a 30 phase shift. If the two bridges are fed
by separate transformer banks having equal voltage ratios and equal leakage
reactances, the two branches of the equivalent circuit adjacent to the bridges
are equal, and a single equivalent emf E is adequate. The third branch has a
reactance Xm due to the reactance of the ac system beyond the transformers.
It is the effect of this branch which we investigate.
118
3-7
Bridge 2
119
represented by triangle a"b"c, whose vertex a" lies between a and a! and
vertex b" lies between b and b'. The ratio of vertical lengths in the phasor
diagram depends on the ratio of reactances thus:
bb
X
/
= ""
bb'
ab"
b'b"
ab
bb'
aa
x=/c
aa!
aa
aa
Bridge 1
MULTIBRIDGE CONVERTERS
(H4)
X. X,
= 1 -k
(115)
(d)
b' b
Fig. 25. Distortion of triangles of line-to-line voltages on bridges 1 and 2 due to com
mutations (a), (b) on phase ab of bridge 1, (c), (d) on phase be of bridge 2.
//
M / / //
( / ///
cot
is occurring, the alternating voltages VB1 across bridge 1 are the same as those
of the source because constant current produces no inductive drop. The
voltages VB2 across bridge 2 are likewise balanced, but because of the different
transformer connection they are advanced 30 with respect to the voltages of
bridge 1; they are shown as a, b, c in Figure 25b.
Now assume a commutation to occur in bridge 1, placing a short circuit on
phases a and b and collapsing the voltage ab to zero. The voltages on bridge 1
are then a'b'c in Figure 2 5a. At star point M of the equivalent circuit, the
voltages are deformed to a lesser extent than at the bridge 1 terminals, being
Fig. 26. Voltage across valve of 12-pulse two-bridge inverter, showing additional dents
D' and D" due to. common reactance, y 15, u = 15, /S = 30.
to the additional dent D' that occurs 30 sooner after the extinction of the
arc than does dent D, whose effect on commutation margin was discussed in
120
3-7
>
at
hlM
Fig. 27. Decrease of commutation margin angle
however, /? > 30, as in Figure 21b, then D' encroaches on the commutation
margin and < y. More specifically, = 30 u, and y = (3 u. The
difference y continues to increase with increasing /? until the vertical front
of D' all becomes negative, after which y remains constant until the whole
dent has passed below the horizontal axis.
The decrease of commutation margin due to dent D' is objectionable
because, by requiring greater extinction and ignition angles, it decreases the
power capability of the inverter, increases its demand for reactive power, and
increases the voltage jumps on the valves. Moreover, the effects of this dent
occur at values of /? normally reached at less than full load.
Overlap Greater than 30. If overlap angle u < 30, only one commutation
MULTIBRIDGE CONVERTERS
121
occurs at a time (with five valves conducting in two bridges), and there are
intervals between commutations (with four valves conducting) when the
voltages are undistorted. In the foregoing, such values of overlap were
assumed.
If u = 30, only one commutation occurs at a time, but as soon as one is
completed, another one begins. All the time five valves are conducting.
If 30 < u < 60, there are intervals in which a commutation in one bridge
begins before completion of the next earlier commutation in the other bridge.
This condition, in which six valves are conducting, may be called double
overlap. The distortion of voltages during double overlap is worse than that
during single overlap, although not so bad as during the double overlap be
tween the halves of the same bridge (with u > 60), which places a threephase short circuit on the valve-side transformer terminals, as described in
Section 3-3.
In the present case there are at the same time two line-to-line short circuits
on different transformer banks (or different sets of windings of the same
bank). These short circuits are on terminals at which the open-circuit voltages
differ in phase by 30. The triangle of line-to-line voltages of each bridge
collapses to a straight line, which is an altitude of the equilateral opencircuit triangle. If Xm = 0, hence k = 0 Eq. (114)the voltage line (vector)
of one bridge is 30 from that of the other. As k increases, these vectors
shorten, and their phase difference decreases. At k = 0.42 the magnitude has
decreased from 1.50isAr to 1.32ELN, and the phase difference, from 30 to!7.
At k = 1, the two line-to-line short circuits become equivalent to a threephase short circuit, and all voltages on the valve side of the transformers
vanish. The circuit analysis for finding these and other voltages is straight
forward but laborious and is not presented here.
Double overlap in a 12-pulse converter with mutual reactance between
bridges has several objectionable effects, including a more rapid drop in
direct voltage and the likelihood of commutation failure. Double overlap is
not likely to occur in normal operation, for the overlap at rated voltage and
current generally is less than 30, say, 20 to 25; however, it can occur at
small overloads or at small undervoltages.
For overlaps between 30 and 60 two circuit conditions occur alternately.
In one, a commutation is in progress in one bridge; and, in the other, com
mutations are in progress in both bridges, as just described. The number of
conducting valves in the two bridges is alternately five and six.
With u = 60, there are always commutations in both bridges at the same
time, and six valves are conducting.
With u between 60 and 90, there are alternately six and seven valves con
ducting. When seven are conducting, there is double overlap in one bridge
and single overlap in the other.
122
With u between 90 and 120, alternately seven and eight valves conduct.
When eight conduct, there are double overlaps in both bridges.
Complete analysis of all these conditions is lengthy and not of much
practical interest. Such analyses have been made by Shekhtman13 and
Freris.24
Elimination of Mutual Reactance
The objectionable effects of mutual reactance in a 12-pulse two-bridge
converter, especially the effect in decreasing the commutation margin of the
inverter, must be eliminated; and this implies the elimination or compensa
tion of the mutual reactance itself. There are two methods by which this can
be done.
Filters. The mentioned objectionable effects are due to the influence of
commutations occurring in one bridge on the wave shapes of the alternating
voltages appearing on the other bridge. The wave shapes are changed in six
short arcs per cycle, each lasting less than one-sixth of a cycle. If the differ
ences between voltage wave shapes of such converters with and without
mutual reactance were analyzed into a Fourier series, it would be found that
the differences consisted principally of certain harmonic frequencies.
Harmonic filters (Chapter 8) are necessary in most converters and are
usually installed on the network side of the transformers. The effect of the
filters is to make the ac bus voltage substantially sinusoidal. The commutating
inductance, which can be defined as the inductance between the valves and
the nearest point in the ac network where the voltages are substantially
sinusoidal, is then the leakage inductance of the converter transformers. If
separate transformer banks are provided for each bridge, there is no mutual
commutating reactance between bridges. The system reactance beyond the
bus is still common to the several bridges as regards fundamental-frequency
voltage drop. However, as currents of higher harmonic frequencies are
effectively short-circuited by the filters and do not enter the ac network
except in amounts that are negligible for the present discussion, commutation
in one bridge does not alter the voltage wave forms in other bridges.
The fundamental-frequency voltage drop in the ac network is calculated by
ordinary phasor methods, not by converter theory. Since the filter appears as
a shunt capacitance at fundamental frequency, the "drop" may actually be
a rise if the reactive current furnished by the filter exceeds that consumed by
the converter. The ideal no-load direct voltage Vd0 is proportional to the
alternating voltage of the bus. The drop (or rise) of direct terminal voltage
with respect to Vd0 is calculated by converter theory : Vd = Vd0 cos a RJd,
or Vd0 cos y - RcId . For a given alternating voltage at the converter, this
drop is independent of the power factor.
problems
123
XT
-H
xm
xa
Valve
XL
groups
(bridges)
Reactor-""
XT
Y A
Fig. 28. Tapped reactor for compensating the coupling between valve groups due to
power-system reactance Xm. X, is self-reactance of each half of reactor winding and Xm,
mutual reactance between the halves; XT is leakage reactance of each transformer bank.
PROBLEMS
124
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21. Compute the rms value of the line-to-neutral voltage on the ac side of
the bridge (Figure 11c).
22. Draw the voltage and current wave forms of a six-phase, one-way
diametrical rectifier for a = 15, u - 30.
problems
125
23. Derive equations for the average direct voltage and current of a sixphase, one-way diametrical rectifier as functions of the ac source voltage,
commutating reactance, ignition delay angle, and extinction angle.
Confine to overlap less than 60.
24. Show that, in a three-phase bridge rectifier operating with no delay and
with 60 overlap, the direct current is half of the crest value of the lineto-line short-circuit current on the valve side of the transformer bank.
25. Find the maximum permissible commutating reactance, in per unit
based on transformer rating, to allow commutation to be completed in
an overlap angle of 60 at rated direct voltage and 1.5 times rated direct
current.
26. Find the ratio of the kVA rating of the transformer bank for a threephase, two-way converter to the rated dc power in kW if a = 15 and
X" = 0.15 pu based on the transformer's rating,
27. Prove that the two expressions for tan< in Eq. (54) are equivalent.
28. Draw curves as specified in Problem 30 but for a = 30, u = 75
(double overlap).
29. Plot pu direct current l'd versus pu direct voltage V'd from open circuit
to short circuit (a) for a = 0, (b) for a = 30, (c) for a = 60.
30. Draw curves of (a) instantaneous voltage of each dc pole of the bridge
with respect to transformer neutral, (b) the six valve currents, (c) one
transformer secondary current, Y connection, and (d) voltage across
one valve for a = 15, u = 60 (auto phase control).
3 1 Find the values of V'd,Id, and Pd of an uncontrolled rectifier at the point
of maximum power.
32. Plot a graph of reactive power versus active power on the ac side of an
inverter operating at a constant extinction angle y = 15. Use per-unit
values with base power Vd0Is2 . Terminate the curve at u = 60.
126
36. In the current divider in Figure 22, find the voltage across the non
conducting anodes with respect to the common cathode (a) when the
whole current flows in one anode and (b) when it is equally divided
among two or (c) three anodes.
37. Draw curves of one cycle of instantaneous line-to-neutral voltage of one
phase and of instantaneous line-to-line voltage of one phase at the
terminals of a Y-connected secondary winding of the converter trans
former feeding one bridge of a two-bridge 12-pulse converter with
coupling coefficient k = 0.422 between bridges. Take a = 30 and
u 10.
38. In a certain ac power system under light-load conditions, the reactive
power produced by the shunt capacitance of the hv ac transmission
lines exceeds that consumed by the series inductances of the lines and
transformers and leads to undesirably high voltage at and near a
rectifier station. The situation is further aggravated when the rectifier
also is operating at light load, because then the reactive power consumed
by the rectifier is less than that produced by its ac harmonic filters. One
method proposed for improving these conditions is to make the rectifier
consume additional reactive power by controlling it to operate at larger
ignition delay angle than its normal value of 15 but subject to the
limitation that neither the peak inverse voltage of the valves nor the
greatest voltage jump across a valve should exceed their respective
values at rectifier operation at rated alternating voltage, rated direct
current and voltage, and normal ignition delay. Under these conditions
the overlap angle is 25. Compute and plot the additional reactive
power consumption of the converter (in per unit of rated power) as a
function of per-unit direct current.
BIBLIOGRAPHY
1. "The Current and Voltage Conditions in Large Rectifiers" (in German), by
W. Dallenbach and E. Gerecke, Archiv fur Elektrotechnik, Vol. 14, pp. 171-248, 1924.
2. Principles of Mercury-Arc Rectifiers and Their Circuits, by D. C. Prince and F. B.
Vogdes, McGraw-Hill, New York, 1927.
3. " The Current and Voltage Relations in the Graetz Three-Phase Rectifier Circuit " (in
German), by K. Maier, E.u.M., Vol. 53, pp. 577-581, Dec. 8, 1935.
BIBLIOGRAPHY
127
6. The Transmission of Power by High Voltage Direct Current (in German), by Karl
Baudisch, Springer, Berlin, 1950, pp. 116-122.
7. " Extended Regulation Curves for 6-Phase Double-Way and Double-Wye Rectifiers,"
by I. K. Dortort, A.l.E.E. Trans., Vol. 72, Part I, pp. 192-198, May 1953. Disc., pp.
198-202.
8. " The Operation of Several Phase-Displaced Inverters on the Same Receiving Net
work," by E. Uhlmann, Direct Current, Vol. 1, pp. 106-110, June 1953. Additional
dent in voltage across valve due to common reactance decreases the commutation
margin.
9. "Influence of A-C. Reactance on Voltage Regulation of 6-Phase Rectifiers," by
R. L. Witzke, J. V. Kresser, and J. K. Dillard, A.I.E.E. Trans., Vol. 72, Part I, pp.
244-252, July 1953. Disc., p. 252.
10. " The Relations of Currents and Voltages in the Rectifier Connected in the Graetz
Three-Phase Bridge" (in Italian), by Walter Dallenbach, VElettrotecnica, Vol. 44,
No. 3, pp. 133-143, March 1957.
11. Recommendations for Mercury-arc Converters, International Electrotechnical Com
mission, Geneva, Switzerland, Publication No. 84, 1st ed., 1957. Definitions, symbols.
12. "The Current and Voltage Conditions in the Graetz Three-Phase Rectifier Bridge
Circuit," by Walter Dallenbach, Direct Current, Vol. 4, pp. 72-80, December 1958.
13. "Operating Modes and External Characteristics of a Twelve-Pulse Cascade-Bridge
Converter Circuit" (in Russian), by M. G. Shekhtman, N.I.I.P.T., Vol. 5, pp. 23-63,
1960.
14. "Bridge Rectifier and Inverter Parameters," Chapter 3 of High Voltage Direct
Current Power Transmission, by Colin Adamson and N. G. Hingorani, Garraway,
London, 1960.
15. " Current and Voltage Conditions from No Load to Short Circuit in Three-Phase
Bridge Circuits," by F. Hoelters, Direct Current, Vol. 5, pp. 112-121, 132, March 1961
16. "The Universal Characteristic of the Three-Phase Bridge Converter," by L. L. Freris,
Direct Current, Vol. 6, pp. 198-201, October 1961.
17. " Reactance Drop in Mercury-arc Power Rectifiers," by O. E. Mainer, Direct Current
Vol. 7, pp. 182-184, July 1962.
18. Schaltungslehre der Stromrichtertechnik (Converter Circuit Theory, in German), by
Th. Wasserrab, Springer, 1962, 466 pp. Especially pp. 241-256.
19. "An Analysis of the Three-Phase Bridge Converter," by L. L. Freris, Direct. Current,
Vol. 8, pp. 6-11, 19, January 1963. Discussion by J. R. G. Schofield, p. 136, May.
20. High Voltage Direct Current Converters andSystems, edited by B. J. Cory, MacDonafd,
London, 1965, Chapter 4, "Analysis of Bridge Convertor Operation," by L. L. Freris.
Also part of Chapter 8.
21. "Analysis of a Hybrid Bridge Rectifier," by L. L. Freris, Direct Current, Vol. 11, pp.
22-33, February 1966. Bridge with three, controlled valves and three uncontrolled
diodes.
22. " Voltage Regulation in 3-Phase Bridge Rectifier Connection from No Load to Short
Circuit," by J. A. Budek and A. H. Marchant, Direct Current, Vol. 11, pp. 38-41,
February 1966. Experimental check.
23. " Series Connection of Mercury Arc Valves," by G. E. Gardner, J. H. Holliday, and
D. J. Johnson, I.E.E. Conference PublicationNo. 22, High Voltage D.C. Transmission,
Manchester, Sept. 19-23, 1966, Part 1, Paper No. 43, pp. 211-214.
I
j
'1
\
j
j
128
Converter Charts
Two different charts are described, both of which show relations among direct
voltage Vd and current Id, the three converter angles a, <5, and u, and the ac
quantities, including active power P, reactive power Q, apparent power S,
displacement factor cos (j>, and alternating current Ia .
The first of these charts has rectangular coordinates x, y proportional to
direct current and voltage Id and Vd . The second one has rectangular coordin
ates proportional to active and reactive powers P and Q. Both charts are
based on the assumptions made in Chapter 3, and both use, for generality,
the first system of dimensionless or per-unit variables described there.
Rectifier and inverter operation are represented by different regions of the
charts.
4-1 CHART 1 WITH RECTANGULAR COORDINATES OF DIRECT
CURRENT AND VOLTAGE
It can range from zero to 2/ 3, its value during a short circuit on the dc
terminals. It cannot reverse.
The vertical coordinate y is the ratio of the direct voltage under general
conditions to its value with no load and no delay of ignition:
v_
v*
v30
"
n_
r'io 3,/3E
(2)
130
CONVERTER CHARTS
4-1
CHART
131
+ 1.0
Loci of Constant a and 8. Coordinates a and y are related to the ignition
delay angle a and extinction angle 8, as shown in Eqs. (19) and (21) in Chapter
3; thus
cos 8
(3)
(4)
x cos cc
Vd
+ 2y = 2 cos a
x + 2y = 2 cos 8
Rectification
+0.5
Rectification
!
0.5
1.0
Rearrangement gives
-0,5
y = cos a x
(5)
.y = cos <5
(6)
+4*
Equation (5) represents the locus of constant ignition angle a in the xy plane.
It is a straight line of slope , intercepting the y axis at cos a and the x axis
at 2 cos a. There is a family of such parallel lines, one for each value of a
(Figure 1). Similarly, Eq. (6) gives the locus of constant extinction angle 5 in
the same plane. It is a straight line of slope +, y intercept cos <5, and x
intercept 2 cos 8. Again, there is a family of such lines, all parallel (Figure
2). The values of a and 8 may range from 0 to 180. Since p = n a, the loci
of constant a are also loci of constant p, although the numerical values of a
and p differ from one another on the same locus. Similar remarks apply to y
and 5. For inverter operation, ft and y are commonly used instead of a and 8.
Inversion
nversion
-1.0
-1.0
Substitution of Eqs. (3), (4), (7), and (10) into Eqs. (8) and (9) yields
u
o =o
2y
2 cos - cos 2
2
, u
. X- sin
~ sin
x = 2
2
2
(7)
+ a) cos -J()5 a)
cos 8 cos a = 2 sin (<3 + a) sin -],-(<5 a)
8 +a X
cos 5 + cos a = 2 cos \{8
Let
(8)
(9)
(10)
cos - =
and
cos u}2
2 sin uj2
sin - =
""
(11)
2
sur
x
+ cos2 - = 1
(12)
gives
r+/ >
-*
2 sin uj2J
\cosw/2
=1
(13)
132
CONVERTER CHARTS
4-1
CHART
133
For this range of overlap we have from Eqs. (72) and (88) in Chapter 3,
instead of Eqs. (3) and (4), the following:
Rectification
(16)
2 \
= \y = cos (a - 30) + cos (5 + 30)
(17)
V3
y]3x = x'
=/
we
a -30 = a'
5 + 30 = 8'
2/
= cos a'
(18)
+ cos 5'
(19)
(20)
Inversion
(21)
(22)
=1
(23)
The new (primed) variables are now replaced by the old (unprimed) ones.
Equations (21) and (22) become
-1.0
Fig. 3. Loci of constant overlap angle in the I'a, Vi plane for u < 60.
3
yj3 cos (a 30) -x
y=
V 3 cos (8 + 30) + -x
(14)
which is the standard equation of an ellipse with center at the origin, hori
zontal half axis a, and vertical half axis b. In the present case,
a = 2 sin -
b ~ cos -
(15)
(24)
(25)
These represent straight lines, as before, but with different intercepts and
with slopes three times as great. Figure 4 shows the lines for a = 30 in both
ranges of u. The lines for a = 30 and 8 = 150 are shown in Figure 3. They,
with the elliptical arc for u = 60, form the boundaries of the region of
134
CONVERTER CHARTS
0.5
1.732
-\
0.4
y
0.3
0.433
s=
60
0.2
u = 120'
1.0 1.155
0.1
Fig. 4. Broken line for a 30 formed by segments of two straight lines having the
following properties:
--intercept
x-intercept
Range of u
0.9
Slope
1.0
1.155
(120)
0 < < 60
Vi 1.732
v'3/2 = 0.866
60 <u< 120
2/V3
V3 = 1.732
1.155
-1/2
-3/2
double overlap, shown in more detail in Figure 5. This region has a small
area and is characterized by high current and low voltage.
The loci of constant u in this region are found by replacing the primed
variables in Eq. (23) with the respective unprimed ones :
1 +[-7=
] =1
[12 sin~3x
(u + 60)/2J
LV 3 cos (u + 60)/2j
-
0.2
(26)
This has the form of Eq. (14) and represents an ellipse with semiaxes
2
. + 60
a = -7=. sm -
V3
and
w 60
,,
b = \j/-3 cos +-
2
(27)
-0.5
For u = 60, which is the boundary between two regions of the chart where
different equations are applicable, Eqs. (24), (25), and (27), for u 60, give
the same results as do Eqs. (5), (6), and (13) respectively, for u g 60.
136
CONVERTER CHARTS
4-1
137
+ 1.0
namely,
For power :
For current:
VaT'l'd = T'x
(29)
v'd
E'a cos < = =
(30)
'
CHART
Rectification
(28)
where T
of the
is ::-
turns.
cos <f>Vd = y
* = jyss'=707T'
(31)
Hence the per-unit alternating voltage on the network side of the transformer
is
(32)
It is independent of the direct current and voltage and, therefore, does not
appear on the chart.
Loci The following loci, all shown in the chart in Figure 6, are independent
of transformer ratio :
EX=y.T'x = x = Ii
(34)
given by
(35)
* The
Fig. 6. Loci of constant active power P', reactive power Q', apparent power S', and
displacement factor cos <f>.
Loci of Constant Active Power P'. These loci are rectangular hyperbolas
* = S"
Inversion
xy = P'
or
P'
(36)
Those in the first quadrant are for positive power (rectification), and those in
the fourth quadrant are for negative power (inversion). The locus for P ' = 0
consists of the y axis and the positive x axis.
138
4-2
CONVERTER CHARTS
q'
x = -y===
(37)
r = S' =
or
These curves are of the fourth degree and are unnamed. All values of
positive, except on the vertical axis, where Q' = 0.
The following loci depend on the transformer ratio :
Q' are
Approximate Loci of Constant Alternating Current I'a. These loci are vertical
lines at abscissa
X
(39)
=~
At nominal ratio ( T ' 1) they coincide with the loci of constant S' of the
same numerical value. For other values of T' they are shifted; the corre
sponding loci do not appear in the charts.
Complete Chart
The loci of constant ignition angle a, constant extinction angle <5, and
constant overlap angle u are best derived from the exact equations for perunit vector power in terms of these angles. From Appendix B the equation
for 0 u 60 is
Loci
(40)
(4i)
rb
Pb = Vd0Is2 =
9E
of Constant
+ 30) +j2u]
(42)
=a
(45)
+ u, factoring out
(46)
(44)
120, it is
P'+jQ'=j?,I+ l-0/2
4 -
where
~/2S + j2u)
The same per-unit system is used with this chart as with Chart 1. The
rectangular coordinates are the following :
y=Q' =
(43)
For 60
Angle <j), measured counterclockwise from the positive x axis representing the
phase angle by which the fundamental alternating current l[ lags behind the
alternating source voltage E.
x = P' =
139
//-r\ 2~|V2
Q'
chart
sin m
u 90
=2 1-
(47)
With constant u, the only variable factor in Eq. (46) is /2a. With increasing a,
140
4-2
CONVERTER CHARTS
CHART
141
Rectification
inversion
(2m
sin u
2u
Reference
-12u
90,
locus of constant u is the arc of a circle of radius (sin u)j2 and center at
0 +juj2 with central angle extending from u 90 to 270 u.
For u > 60 Eq. (45) is used instead of (44) and with a similar rearrange
ment. The loci of constant u are still circles, and their centers are still at
0 +ju/2. The radius of each is given by \ sin (u + 60), and the valid arc
extends from central angle u - 30 at a = 30 to 210 ~u at a = 150 u.
The loci for u 0 and u = 120 are single points, the former at the origin
and the latter at 0 +yVt/3. .
Loci of Constant Ignition and Extinction Angles, Let us consider first the
locus of a = 0. For a = 0, Eq. (46) becomes
-0.5
+0.5
the vector represented by the second term of this equation rotates counter
clockwise through angle 2a, and its tip traces a circle of radius (sin u)J2. The
valid arc of this circle extends through 360 2u, from u 90 at a = 0 to
270 u at a = 180 u, and is symmetrical with respect to the imaginary
axis. The first term in Eq. (46) represents a vector locating the center of the
circle on the imaginary axis at distance w/2 rad above the origin. Thus, the
P'+jQ'=i+j~i/2u
(48)
142
CONVERTER CHARTS
them.
The most practical way of constructing these loci seems to be the following.
First draw the circles of constant u, and on them mark the point for a = 0 and
the equally spaced points for other values of a, using constarit increments of
a. The loci of constant a may then be drawn by connecting, by a smooth
curve, points for the same value of a but different values of u.
Each such point, representing certain values of a and u, also represents a
value of <5 = a + u that may be marked on each point. The loci of constant 5
are then drawn by connecting all points for the same value of 8. The pairs of
loci for a
and 8 = n (or y = at) are mirror images with respect to the
imaginary axis. Hence the loci of constant <5 are also arcs of cycloids.
4-2 CHART 2
0.3
cy
1.0
H,
Inversion
Rectification
0>
Q.
|0.2
O
"N
it
S 0,5
= 30'
o.i
20'
0.3
0.4
O (u = 0)
0.5
Fig. 9. Converter Chart 2 loci of constant converter angles in P', Q' plane.
+ 0.5
equations.
Loci Basedon Approximate Relations. Per-unit
direct current l'd is approxi
mately equal to per-unit fundamental apparent power
S and per-unit direct
voltage Vd is approximately equal to displacement
factor cos </>. Hence the
loci of constant direct current based on an approximate
relation, are, like the
loci of constant alternating current and
constant apparent power, circular
t44
4-2
CONVERTER CHARTS
arcs having radius l'd and their center at the origin. These loci can be drawn
with a compass. The loci of constant direct voltage, based on an approximate
relation, are straight lines through the origin, having slope tan (j>, where
(j) = cos-1 V'd. These lines can be drawn with a straightedge.
The two sets of loci together constitute a polar-coordinate grid (Figure 7).
Loci Based on Exact Relations. Loci of this form are related to those of
constant a and constant <5 by the following equations :
(49)
(50)
chart
whence
cos oc = Vd -j-
\l'd
(51)
cos 5 = V'd-\I'd
Ignition angle a
or extinction angle y
90>
20
(52)
The exact loci of constant l'd and constant V'd can be constructed as follows :
Values of l'd and V'd are selected, and corresponding values of a and 6 are
calculated by Eqs. (51) and (52), and u by 8 a. Values of P ' + jQ' are then
calculated by Eq. (44) or (45), depending on whether u is less than or greater
than 60. Finally, lines are drawn connecting points of equal l'd and other
lines connecting points of equal V'd . The former are approximately, but not
exactly, concentric circles; the latter are approximately, but not exactly,
straight, radial lines. See Figure 11.
Rectification
Inversion
145
146
BIBLIOGRAPHY
CONVERTER CHARTS
147
a given pair (Id,Vd) in the P ' + jQ' plane according to the approximate
relations has the same abscissa (P' V'dl'd) as the point representing the same
pair according to exact relations. The approximate ordinate Q\ however,
given by
Qsfd-Jl-(V'df
(53)
is greater than the exact one by an amount that varies from 0 to nearly 8%
and is greater along the outer boundaries of the chart (a = 0" and 5 = 180)
than on the vertical axis. In the normal operating range, with a = 15 or
y = 16 and u between 0 and 25, the error in Q' does not exceed 3 per cent.
See Figure 12 for further information.
BIBLIOGRAPHY
1. "The Theory of the Control Problem of H.V.D.C. Transmission with Rectifiers and
Inverters in Bridge Circuit," by F. Busemann, Publication Z/T74, B.E.&A.I.R.A.,
Mar. 2, 1948, Figures 15, 19.
2. High Voltage Direct Current Power Transmission, by Colin Adamson and N. G.
Hingorani, Garraway, London, 1960, Chapter 3.
3. "A Chart Showing the Relations between Electrical Quantities on the A-C and D-C
Sides of a Converter," by E. W. Kimbark, I.E.E.E. Trans on P.A.&S., Vol. 82, pp.
1050-1054, December 1963. Also in I.E.E.E. Publication S-155, D-C Transmission,
June 1963, pp. 34-43, and in Direct Current, Vol. 8, pp. 166-169, June 1963. Chart 2
with different scales than herein.
4. Schaltungslehre der Stromrichtertechnik (Converter Circuit Theory, in German), by
Th. Wasserab, Springer, Berlin/Gottingen/Heidelberg, 1962, pp. 337-340.
PROBLEMS
:(
:1
1,
5-1
5
Control
\
I
This chapter discusses the control of dc transmission under normal con
ditions. Grid control of the converter valves, however,is used not only for the
control of normal transmission but also for clearing faults in the converters
and on the transmission line. These aspects of protection are discussed in
Chapters 6 and 7.
|
I
Critical Grid Voltage. The critical grid voltage is that which separates
grid voltages which permit ignition from those which block ignition. The
critical grid voltage varies oppositely from the anode voltage and is roughly
proportional to it (Figure 1), but the exact value varies somewhat with
temperature and vapor pressure in the valve. In addition, it depends on the
geometry of the valve.
|
Let us begin the discussion by recalling the conditions under which the
conduction of valves can be initiated and terminated or, in other words,
under which the arc can be ignited and extinguished.
In mercury-arc valves, the control grid can delay ignition but, after con
duction has begun, cannot stop it. The same limitation applies to the control
of thyristors by a gate signal. There are other kinds of valves, such as vacuum
(or "hard") electronic tubes and transistors, in which both ignition and
extinction can be controlled. Some devices that can conduct in either direction,
such as mechanical switches and Marx arc switches, also can control both
ignition and extinction. None of these devices, however, is commonly used
in high-voltage high-power converters; therefore, we confine our attention
to the kinds of values that can control ignition only and, in particular, to the
mercury-arc valve.
Such a valve begins to conduct if the following two conditions are satisfied:
(a) if the anode voltage, with respect to the cathode, is positive and (b) if the
control-grid voltage, also with respect to the cathode, is more positive than
the critical value. For brevity we express the second condition by merely
saying that the grid is positive.
The cessation of conduction of such a valve depends primarily on the
external circuit. The current must become zero through action of that
148
circuit. It would be better to say that, if the valve were not a valve but a plain
conductor, the current through it would reverse. Since, however, the valve is
a valve, the current through it cannot reverse but becomes and remains nil.
Instead, the voltage across the valve becomes negative. Next, this voltage
must remain negative long enough for the adequate deionization of the arc
(requiring 50 to 400 fis in a hv mercury-arc valve, or 1 to 8 of a 50- or 60-Hz
wave). After deionization, negative grid voltage blocks reignition, even if the
anode later becomes positive.
Practically, in a converter, the circuit condition that causes extinction is
the firing of the next valve of the set and the completion of commutation to
that valve.
As we saw in.Chapter 3, the need of adequate time for deionization is more
critical in an inverter than in a rectifier, for the anode remains negative for a
much shorter time after extinction (see Figure 20 in Chapter 3).
Form of Grid Pulse
Valve Characteristics
149
|
j
GRID CONTROL
I.
I
I
|
I
Front of Pulse. For assuring ignition, the grid voltage should rise from a
value well below the minimum control value to one well above the maximum
critical value. For accurate timing of ignition (independent of variations of
critical voltage), the grid voltage should rise very rapidlyideally, instantly.
150
5-1
CONTROL
cot
(a)
(b)
GRID CONTROL
151
152
CONTROL
Vdoi
-=-
Vd()i cos a
Vdl
Vd2
5-3
POWER REVERSAL
Vd02 COS
-=-
153
Vd02
X
Rectifier -
-Line-
Inverter
Fig. 4. Equivalent circuit of dc transmission valid for average currents and voltages in
the steady state.
resistances are fixed, the current is proportional to the difference of the two
internal voltages and is controlled by controlling these voltages. The direct
voltage at any designated point of the line, as well as the current, can be
controlled by controlling the two internal voltages; for example, if the line is
uniform and if the two commutating resistances are equal, the voltage at the
midpoint of the line is the average of the internal voltages. The direct voltage
at any other point of the line is a weighted average of the internal voltages.
More generally, any two independent quantities, for example, power and
voltage, could be controlled by the two internal voltages.
Each internal voltage can be controlled by either of two different methods :
grid control or control of the alternating voltage. The internal voltage of the
rectifier is written in Figure 4 and Eq. (1) as Vd01 cos a. Grid control, delaying
the ignition angle a (time ocjco), reduces the internal voltage from the ideal
no-load voltage VdQ1 by the factor cos oc. (It will be remembered that the
voltage drop due to overlap is represented by the voltage across the com
mutating resistance Rcl.) The ideal no-load voltage Vd01 is directly propor
tional to the alternating voltagesee Eq. (5) in Chapter 3. The alternating
voltage in some exceptional cases could be controlled by generator excitation,
but it is usually controlled by tap changing on the converter transformers.
Grid control is rapid (1 to 10 ms), but tap changing is slow (5 to 6 sec per
step). Both these means of voltage control are applied cooperatively at each
terminal. Grid control is used initially for rapid action and is followed by
tap changing for restoring certain quantities (ignition angle in the rectifier or
voltage in the inverter) to their normal values.
The current in a dc line operating in the steady state is given by Ohm's law
as the difference in its terminal voltages divided by its resistance. By incorporatingthe equivalent circuits ofthe converters ("given in Figures 13 and 18 in
Chapter 3) with that of the line, Ohm's law may be extended to the internal
voltages, embracing the commutation drops as shown in Figure 4. The current
Id in the line is then given by
r
(i)
In this equation cos fi is used in the numerator and + Rc2 in the denominator
if the inverter is operated with constant ignition angle (3; cos y and Rc2 are
used if the extinction angle y is constant. For present purposes, the former
mode of operation is assumed, because it is the ignition angle [3 that can be
directly controlled; the extinction angle y is controlled indirectly through
controlling (3 to values computed from the direct current Id, the commutating
voltage, and the desired extinction angle, as discussed in Section 5-10.
Direct current Id, then, depends on the voltage dropnumerator of
Eq. (1) divided by the total resistance (denominator). Since in practice the
It should be noted that in Eq. (1), Id and hence the difference of internal
voltages are always positive because of the unilateral conduction ofthe valves
(symbolized by the diodes in Figure 4). If it is desired to reverse the direction
154
5-4
CONTROL
of power transmission, the polarity of the direct voltages at both ends of the
line must be reversed while maintaining the sign of their algebraic difference.
Station 2 then becomes the rectifier and station 1 the inverter. The ter
minal voltage of the rectifier is always greater in absolute value than that
of the inverter, although it is lesser algebraically in the event of negative
voltage.
In some hv dc cable lines it is desired not to reverse polarity of conductor
voltages (see Volume 2, Chapter 11, for further discussion). It then becomes
necessary to have reversing switches between each converter and the line.
A method of reversing current, sometimes used in low-voltage dc industrial
practice, by having two converters in parallel, one for each direction of
current, is prohibitively expensive for hv dc transmission.
Further discussion of power reversal by voltage reversal will be found in
Section 5-7, page 163.
155
-Inverter
-Rectifier
0.5
1.5
2.0
2.5
IdRdn
Normal
Low
Alternating
voltage
Lower
The direct voltage at either end of a transmission line can vary in a sudden,
unexpected, and undesired manner because of short circuits or other dis
turbances on the ac systems' or because of faults in the converters. It is then
the responsibility of the rapid grid control to maintain or restore the desired
conditions on the dc line as far as possible with the available range of control.
This use of rapid, automatic control is more important than the effecting of
a rapid change of conditions on the dc line.
The need for rapid, automatic control will be shown by first supposing that
each converter has only manual control of the ignition angle. Viewing the
transmission line from its midpoint, draw a graph of the characteristic curve
of voltage at this point versus current for the converter and half line on each
side (Figure 5). Both characteristics are straight lines. For the left-hand
(rectifier) side, the intercept on the vertical axis is the internal voltage of
the rectifier, Vd0l cos a, and the slope is (Rc i + RJ2). For the right-hand
(inverter) side, the intercept is the internal voltage of the inverter, VdQ2 cos fi,
and the slope is +(Rc2 + RJ2). Since, at a given place and time, there can
be only one value of voltage and one value of current and since both these
values must conform to the characteristics of both converters, they must be
given by the coordinates of the intersection of the two characteristic lines. Let
the intercepts of the two lines be adjusted so that their intersection lies at
rated current and voltage (Jdn ,Vdn) at point N. The slopes of the lines have
been drawn on the assumption that at rated current the voltage drop due to
Fig. 5. Changes of direct current due to decreases of alternating voltage at either terminal
of the dc line when both converters are operating at constant ignition angles.
+ 9/2 = 12.5%.
9% of the
same.
Now suppose that the alternating voltage at the inverter drops by 12.5% of
Vd . Then the inverter characteristic drops to that represented by the first
parallel broken line below the solid line, and its intersection with the rectifier
characteristic moves to point A, corresponding to 1.5/d. A further 12.5%
voltage drop increases the current to 2Idn (point B). If the alternating voltage
at the rectifier drops by 12.5% of Vdn while that of the inverter is normal, the
rectifier characteristic drops to the first parallel broken line below the solid
line, the intersection moves to point C, and the current drops to 0.5Idn . An
additional 12.5% drop of alternating voltage lowers the characteristic to the
second parallel broken line and moves the intersection to point D: the current
becomes nil.
In this example, a dip in alternating voltage produces a percentage change
in direct current of four times the percentage change of voltage. Such large
fluctuations of current cannot be tolerated. The high overcurrents are
especially undesirable, for they may lead to arcbacks of the rectifier (Section
6-3), commutation failures of the inverter (Section 6-5), and damage to the
valves. The need for rapid control of current is indicated.
156
CONTROL
The same two methods could be used for ac transmission and distribution.
In the constant-current system, the various loads and one or more power
sources are connected in series ; a load or source is turned off by bypassing it
after bringing its emf to zero if it has one. This method has been widely used
for street lighting circuits, and was used on several of the earlier dc trans
mission projects. In the constant-voltage system, the various loads and sources
are connected in parallel; a load or source is taken out of service by opening
the respective branch. This system is in general use in ac transmission and
distribution systems and dc distribution systems.
In a hv dc transmission system having only two terminals (none to date
have had more), the distinction between series and parallel connection of the
two converters (rectifier and inverter) disappears. The choice between con
stant current and constant voltage must be made on other grounds. The chief
5-6.
157
occurs
7. Keeping the voltage at the sending end of the line constant at its rated
value insofar as possible in order to minimize losses for a given power
8. Controlling the power delivered or, in some cases, the frequency at one
end.
Many of these features have already been discussed, but others need some
discussion before the scheme actually used is described so that it may better
be appreciated.
There are four reasons for keeping the power factor high; two concerning
the converter itself and the other two concerning the ac system to which it is
connected. The first reason is to keep the rated power of the converters as
high as possible for given current and voltage ratings of valves and trans
formers. (The voltage rating of the transformer and the voltage across the
valves when not conducting are both proportional to Vd0 , but the power is
proportional to Vd .) The second reason is to reduce the stresses on the valves
and damping circuits. The third reason is to minimize the required current
rating and the copper losses in the ac lines to the converter. The fourth reason
is to minimize voltage drops at the ac terminals of the converter as its loading
increases. The last two reasons apply to any large ac loads.
The power factor can be raised by adding shunt capacitors, and if this is
158
CONTROL
done, the disadvantages become the cost of the capacitors and switching them
as the load on the converter varies.
The power factor of the converter itself is
cos <f>
Vdo
y[cos a + cos (a + )]
(2)
159
Inverter
(const 7)'
voltage
for a rectifier or
cos </> = i-[cos y
+ cos (y + w)]
(3)
.Rectifier
(const Id)
(4)
160
5-7
CONTROL
rectifier station. The dc reactors at both ends of the line tend to prevent rapid
changes of current, thus easing the duty of the current regulator.
It may be said with fair accuracy that the rectifier controls the direct current
and that the inverter controls the direct voltage. The statement would be
exact only if the rectifier characteristic were truly vertical and if the inverter
characteristic were truly horizontal. Otherwise, each control affects both
current and voltage, although it affects one of them but slightly and the other
greatly.
If the inverter voltage is raised, as described, the rectifier voltage must be
raised by an equal amount in order to keep the current constant. This can be
done quickly only by the current regulator, which increases cos a by electronic
control. This can continue only until a = 0 and cos a 1. The valve cannot be
ignited earlier, and so it is spared the vain effort of trying to increase cos a
by making a negative. The rectifier voltage can be increased further only by
changing taps on the rectifier transformer. If the inverter voltage is being
increased by its tap changer, the rectifier voltage can be increased just as
rapidly by its, assuming that both tap changers are of similar design. The
rapid electronic current regulator need only be able to raise the rectifier
voltage by as much as, or a little more than, it is raised by one step of the
inverter tap changer.
In practice, the rectifier tap changer is automatically controlled so as to
bring a into the range between 10 and 20. These values represent a com
promise between (a) keeping the power factor high, which requires small a,
and (b) having a margin for quick increases in rectifier voltage, which
requires great a.
In the foregoing we have considered the behavior of the control system for
slow changes of voltage. Equally important is its behavior for rapid changes
of voltage, especially for rapid decreases of voltage due to short circuits on the
ac systems, to electromechanical oscillations ("swings") between ac genera
tors, or collapse of the voltage of one valve group (bridge). These changes
may occur at either end or both ends simultaneously, usually in unequal
degrees.
Consider first a reduction of inverter voltage. The inverter characteristic is
shifted downward (from CD to FG in Figure 6). The new operating point is
H. The line then operates at reduced voltage but at substantially the same
current as before; the power is reduced in proportion to the voltage. The dip
in voltage may be of short duration, in which case operating conditions are
restored to substantially the initial ones. If the low alternating voltage is
maintained, the inverter tap changer raises the direct voltage until either it
becomes normal or the limit of the tap-changer range is reached.
Now consider a decrease of alternating voltage at the rectifier. If cos a were
constant, the direct voltage at the rectifier would decrease proportionally.
161
Rectifier
(a = an)
Normal voltage
Inverter
/(r=7)
Reducedj
voltage
Inverter
Rectifier
(const/,)
(const Ia)
Now a big dip in rectifier voltage shifts the rectifier characteristic down to
A'B'H, which does not intersect the inverter characteristic. Consequently,
the current and power drop to zero after a short delay due to the dc reactors.
In order to avoid such a.great change of current and power caused by a
moderate dip in alternating voltage (exaggerated in Figure 7), the inverter is
also equipped with a fast current regulator, but it is set at a lower current than
the rectifier's regulator. The inverter characteristic is now DFG, consisting
of two segments, one of C.E.A.(y = y) as before, and one of constant current.
It intersects the new rectifier characteristic at L. It may now be said that the
inverter is controlling the direct current, and the rectifier, the direct voltage
an interchange of functions from those pertaining to normal voltage con
ditions. (More generally, the station having the lower value of VdQ1 cos a0
RclId or Vd02 cos y (Rc2 Rt)(Id AId) controls the voltage and the other
station the current.)
The difference between the current command of the rectifier and that of the
inverter is called the current margin and is denoted by A Id. It is generally 15%
of the rated current, although it could be made smaller. It must be great
162
5-7
CONTROL
enough so that the two steep constant-current lines do not cross each other
in spite of errors of current measurement.
During a large dip in rectifier voltage, the power is reduced not only in
proportion to the voltage but also because of the reduction in current A Id.
Thus a voltage dip at the rectifier end reduces the power more than does an
equal dip at the inverter end. This, however, is much better than having the
power suddenly become zero.
Moreover, if the dip in power is objectionable, a supplementary current
control can be added, which, whenever the current command exceeds the
measured current by approximately the magnitude of the current margin,
automatically increases the current command first to the rectifier and a little
later to the inverter by the amount of the current margin, and thus restores
the current to its correct value in a few tenths of a second. Later, if the
measured current exceeds the new current command by a similar amount, that
command is reduced by that amount first at the inverter and a little later at
the rectifier.
Under the abnormal condition just considered (low rectifier voltage with
current controlled by the inverter), the rectifier current regulator sees that the
current is too low and tries to raise it by raising the rectifier voltage by de
creasing the ignition delay. It is unable to do so, however, either because the
delay is already zero or because the minimum a control overrides the current
control. The inverter, in order to control the current, is operating at a greater
extinction angle than the minimum specified value. This occurs because the
current regulator fires the inverter valves before the C.E.A. control has a
chance to do so.
Under normal voltage conditions or low inverter voltage, the inverter
current regulator sees that the current is too high according to its own setting.
It tries to lower the current by raising the inverter voltage. To do so, it must
decrease the extinction angle y by decreasing the ignition advance angle /?.
It cannot do so because the C.E.A. regulator is already igniting the valves
before the current regulator would do so.
Suppose that, when a change of current is desired, the new current com
mands are to be set manually in response, perhaps, to telephoned orders. The
setting at one station must be changed before that at the other station. If the
current is to be increased, the current setting is raised first at the rectifier and
second at the inverter. But if the current is to be decreased, the current setting
should be lowered first at the inverter and second at the rectifier. In either
case, the current margin is first increased and later decreased to its normal
value. In this way there is no danger of accidentally changing the sign of the
current margin and thus suddenly reversing the power. If the change of
current is to be made automatically, the order of events is the same as
described, but the changes can be accomplished more quickly.
163
Converter 2
Vd
/
Converter I
/'Converter 2
If
U
1
1I
if1
C.EA
I I C.C. |
l
>1
>1
1
i
c.c.
1'
Hn
JL
J 1
C.E.A.
Converter 1
Converter 2
(b)
164
5.-9
CONTROL
A/d
seconds
(5)
165
-J
TC
CONSTANT-CURRENT CONTROL
If the measured current in a rectifier is less than the set current, a must be
decreased in order to increase cos a and thus raise the internal voltage of the
rectifier Vd0 cos a. The difference between the internal voltages of the rectifier
and the inverter is thereby increased, and the direct current is increased pro
portionally see Eq. (1). If the measured current exceeds the set current, a
must be increased instead of decreased, and all the quantities mentioned
above are changed in the opposite sens,e.
In the inverter, if the measured current is too low, the internal voltage must
be decreased instead of being increased as in the rectifier in order to increase
the difference of internal voltages. This refers, however, to the absolute value
of the inverter voltage. If we consider the inverter voltage to be negative,
which is usual if the same converter sometimes rectifies and at other times
inverts, the algebraic value of inverter voltage must be increased, as in a recti
fier; and to accomplish this, a must be decreased, as in a rectifier. The curve
166
5-10
CONTROL
Con.
PS
Dc line
Tr;>
+1
s = Ids
dVj = -(K+Rc)
dL
-l
Fig. 9. Curve of cos a versus a.
where
167
= Kt
+ T
dt
(6)
Ts
+1
(8)
(7)
168
CONTROL
5-10
the time when the commutation voltage reverses. Furthermore, after conduc
tion ceases, the voltage across the outgoing valve must be negative and must
remain so long enough for adequate deionization of the arc path. Let both
time t and the corresponding time angle cot be measured from the instant when
the commutation voltage of the valve in question first turns positive. The
instantaneous value of the commutation voltage is then
eba=/3Emsmmt
(9)
Commutation can begin after cot = 0 and must be completed before cot = n.
Let it be required that, under normal conditions, commutation be completed
at cot = n y, where y is the normal extinction advance angle. For accom
plishing this, commutation must begin at an ignition angle cot = a = n P,
which depends on the commutation voltage (crest value 3 Em), the direct
current Id to be commutated, the commutation inductance Lc , and the
desired yn . For the present, all four of these are assumed to remain constant
during a particular commutation, but the voltage and current can vary from
one commutation to the next.
The relation among the five quantities (a, y, Em, Id, and Lc) was derived
in Chapter 3. It depends on the fact that the time integral of the commutation
- V3 Em cos cott
1,2
ebadt = y/3Em
Jn
-v/s E-(cos
=CO
C0/2
>f3Em
(12)
(13)
cos ut
90 lag
circuit
Polyphase
1_
rectifier
En
-Id
/""N
dld
dt
Differentiator
Pot.
-V3E,m COS yn
O
cos yn
Comparator
-N
-<o
dt
Summer
and trigger
-2XM
JL
2XC
(10)
COS COfi)
volts
i-
Pot.
-coscof-!'2
,
.
f2sin
cot dt = *j3Em&
Jn
Ja
or
voltage is equal to the change of magnetic flux linkages produced. The latter
is 2LcId. The former is
169
valve
Pulse
forming
circuit
where
cot2 = 7i
Fig. 11. Schematic circuit of analog computer for C.E.A. control. Pot., potentiometer.
The signal dhldt is explained in connection with Figure 15.
J3 E[cos (n
co
y) cos cotj
/s E
-( cos yn
=-
CO
volts
(11)
170
CONTROL
that of the commutation voltage and a phase 90 behind that voltage. When
the two quantities become equal, a pulse is generated that initiates the grid
pulse to ignite the valve in question. The quantities involved in the analog
comparison are shown in Figure 12.
The wave of commutation voltage is obtained from one phase of a bank
of control transformers (potential transformers) connected to the ac bus. The
control voltage should be proportional to the open-circuit ac voltage on the
valve side of the main converter transformers, but it cannot be obtained from
the valve side because of the objectionable notches in the voltage wave caused
by commutations. The ac bus voltage has a good wave form if adequate ac
harmonic filters are provided. It does not remain proportional to the valveside open-circuit voltage, however, when the tap changers are operated.
A solution to this dilemma, if the taps are on the winding on the network side,
is to have the control transformers connected to a fixed tap on that winding.19
171
R<
fi,
fisut
El
JR
fiut
-o
Eu
-o-i
Eb
+ V3m
(b)
-jR =
jwC
-vVA
- V3 Em cos ait
Fig. 12. Wave forms of voltages in C.E.A. analog computer.
The negative cosine voltage wave required for C.E.A. control can be derived
from the sine wave of commutation voltage in either of two ways. One way is
to use an RC or RL phase-shift circuit adjusted for an output lagging the
input by 90 (Figure 13). The other way is to integrate the commutation
voltage.19,36 The analog integrator (Figure 14) yields the negative of the
time integral of the input voltage.
For the correct operation of the C.E.A. control under unbalanced alternat
ing voltages, such as may occur during short circuits on the ac system, each
valve should have its own analog computer. The negative cosine wave is
derived from the sine wave of commutation voltage for the valve in question
fin
fi,
taken from the secondary winding of the control transformer. The proper
phases are shown in the following table:
Outgoing valve
(to be extinguished)
Commutation voltage
Incoming valve
(to be ignited)
ebo
3
eca
6c b
6 ab
ebc
172
CONTROL
173
It would be easy to obtain the negative cosine waves from the polyphase
supply, but with unbalanced voltages these waves would be incorrect in both
magnitude and phase.
The signal y/3 Em cos y should likewise be taken from a rectifier con
nected to the phase listed above. From the two waves derived from the same
phase, a three-phase voltage can be obtained for feeding a six-pulse rectifier,
whose output can be better filtered than that of a single-phase two-pulse
rectifier. This signal from one phase can be used for the computers of the two
valves of a pair.
The signal proportional to 7dcan be used for all the valves of the converter.
Up to now we have assumed that Em and 7d were constant during a single
commutation, although they might vary between successive commutations
of the same valve. Now let us briefly consider the effect of their varying while
a commutation is in progress. Increased direct current requires more voltagetime integral for commutation, and decreased commutation voltage provides
less voltage-time integral between the same angular limits. Obviously, after
commutation has begun, its time of beginning cannot be altered by anything
that might happen afterward. Increased direct current, or decreased com
mutation voltage, or both together, require more time (and time angle) for
completion of commutation, and if enough time is not available for completion
and deionization before the sign of the commutation voltage reverses, the
direct current that began to be shifted from one valve to another is shifted
back to the valve it came from. This result is called a commutation failure and
is discussed further in Section 6-5.
Decreased alternating voltage, caused, perhaps, by a short circuit on the
ac network, lowers the direct voltage of the inverter proportionally. Con
sequently the direct current through the inverter increases at a rate deter
mined principally by the inductance of the dc reactor. The chief purpose of
this reactor is to decrease the probability of commutation failure of the
inverter caused by a disturbance occurring too late for making the necessary
correction in firing angle. For further discussion see Section 7-2, page 236.
Some commutation failures can be prevented by the introduction of a
supplementary input signal proportional to the rate of change of direct
current, dlfdt. This can be obtained in several ways, two of which are shown
in Figure 15. The method thus modified, however, is still ineffective unless the
increase of current occurs long enough before the time when commutation
otherwise would begin. This control should be arranged so that, although
increasing current advances the firing angle, decreasing current does not
retard it. The diodes in Figure 15 accomplish this.
The usual value of y is 16. Only 1 to 8 of this is required for deioniza
tion, the larger angles applying to the higher-power valves.30 The remainder
is margin for variations in direct current and in commutating voltage occur
ring too late to permit adequate correction of /?.
W\r
eR = Rid
Ux
-T
eL~L
P
Id
dld
dt
(a)
"H
eout
(b)
Fig. 15. Circuits for obtaining a voltage signal proportional to dljdt: (a) RL
circuit(b) RC circuit.
The foregoing discussion has neglected the effect, described in Section 3-5,
page 110, of the dent in the wave of voltage across the outgoing valve due to
the subsequent commutation. At large values of l'd, which is proportional to
Id/Em in Eq. (12), this dent decreases the commutation margin for a given
extinction angle. Hence, in order to prevent commutation failures at high
direct currents or low alternating voltages or both, the extinction advance
angle y must be greater than its normal value y , and the ignition advance
angle must be made accordingly greater. Curves of required (3 and y as
functions of l'd for constant commutation margin angle ( = 15 are shown in
Figure 16.
The C.E.A. control system should be modified so that at high values of it
Id
functions to increase y and to attain approximately constant commutation
margin .
It may be noted that the C.E.A. control as usually performed and as
described above is not a feedback control system as is the constant-current
control. A method of control based on the measurement of the extinction
angle, comparison of the measured angle with the desired angle, and altera
tion of the ignition angle in the direction to decrease the error has been
proposed.34 It has the obvious shortcoming that, after sudden changes of
circuit conditions, the correction is made too late. The standard method
described is superior in that it can prevent most commutation failures that
would occur without it. Another proposed methodis to superpose a correction
based on measurement upon the standard method.19
174
5-11
CONTROL
STABILITY OF CONTROL
175
135
120
105
Dc reactor
Line
Line
45'
15'
L3rd a-L*4thJ
2nd
1st range
0.5
1.0
/3
Fig. 16. Ignition advance angle and extinction advance angle y versus per-unit direct
current I'd in an inverter operated at constant commutation margin angle 15. Broken
modified form of control37,38 in which the valves are ignited at equal time
intervals and the ignition angles of all valves are retarded or advanced
equally so as to obtain the desired current, in C.C. control, or the desired
minimum extinction angle in the valve for which that angle is least, in C.E.A.
control.
5-11 STABILITY OF CONTROL
The dc line, together with the dc reactors at each end, constitutes a lightly
damped system that can be set into oscillation by various disturbances, such
as line-to-ground short circuits, converter faults, and improperly controlled
energization of the line. The natural frequencies of these oscillations usually
lie in the range of 10 to 100 Hz. Overvoltages on the line insulation may
result from improper energization or from converter faults that impress a
component of power-frequency alternating voltage on the dc line. It is
desirable that the oscillations be rapidly damped so as to limit the overshoot
of voltage to a moderate value.
(b)
Fig. 17. Circuits for damping line oscillations: (a) in parallel with dc reactor; (b) in
shunt with dc line.
* The term used here as in the theory of feedback control systems has a somewhat different
meaning from that of the terms "power system stability" or "synchronous stability."
5- 11
176
STABILITY OF CONTROL
177
CONTROL
Vd
Zi Zm
-Zm z2
\
Inverter
=0
(16)
or
(C.E.A. control)
zLz2-zl =0
(17)
Rectifier
K + i?cl
Ldi
Dc reactor
Dc transmission line
Dc reactor
Rdi
Li/2
Li/Z
Rll2
Ril2
Rdi
R2 + L2s + )
Rx +Lxs -+
-Ra
=0
il
il
(14b)
Ri
It
(b)
Fig. 19. Equivalent circuit of dc link for analysis of stability of control (a) before com
bination of line with terminal equipment; (6) after combination.
Z1I1 ZJ2 = Ex
ZmIx + Z2I2 = e2
y+
where
Zx
= Rx
+ Lxs + Cs
1
(15a)
Z2 = R2 + L2s + Cs
(15b)
z = -L
Cs
(15c)
(19)
(14a)
(a)
Li
+ R2) = 0
This is a cubic equation in s, and its roots are values of s that characterize the
transient phenomena. A cubic equation with real coefficients has three roots,
at least one of which is real. The other two may be real or a conjugate com
plex pair. From our knowledge of the nature of the circuit, we know that
the latter alternative is more probable. The corresponding solution in the
time domain, found from the inverse Laplace transform of Eq. (19), would
have a damped direct current and a damped oscillatory current. If R2 is.
negative, one or both of these terms may have negative damping; it is more
likely that only the oscillatory term has it.
If numerical values of the coefficients were given, the cubic polynomial
could be factored into a linear one and a quadratic. Factoring of the algebraic
cubic is very cumbersome. So let us make another simplifying assumption.
Since we know that ordinarily
> Lxs, let us put Lx = 0. Equation (19) then
simplifies to the following quadratic:
Li
Ri
(18)
=0
+ i442
l2rx c
(20)
-Wv
Ri
(15d)
L2 RXC
z.C
E2 0 and Li = 0.
178
CONTROL
. 5-12
may further simplify Equation (20) by putting R2 0 in the last term, which
then becomes 1/L2 C. Equation (20) has the form
or
(21)
s2 + 2(o)ns + co2 = 0
(22)
(23)
and the damping coefficient is
i/r2
(24)
It is the average of two reciprocal time constants, the first of which, containing
R2, may be negative. For positive damping a > 0; hence, if R2 < 0,
R -.
L% _
\aJl2c- cr2
(26)
TAP-CHANGER CONTROL
179
The rectifier tap changer is controlled so that if a becomes less than 10, it
raises the direct voltage by raising the transformer ratio T; and if a becomes
greater than 20, it lowers the direct voltage by lowering T. As a result, a lies
between 10 and 20 in the steady state unless the tap changer has reached one
of its limits.
The inverter tap changer is controlled so that the direct voltage at some
designated point of the line, preferably the sending end, is close to its rated
value. More precisely stated, the tap changer raises the direct voltage by
raising T if that voltage is lower than the desired value by more than a specified
amount and lowers the direct voltage by lowering T if the voltage is higher
than the desired value by a specified amount.
The direct voltage at the point of the fine where it is to be held constant is
computed by adding to the direct voltage measured at the inverter station the
RtId drop in the dc line, obtained to scale by sending the current output of the
dc current transformer through a resistor representing the line resistance to
scale. This arrangement may be called a line-drop compensator. It is subject
to some error attributable to changes in the fine resistance with ambient
temperature and current.
In order to avoid hunting of the tap changer, the dead band must be wider
than the size of the tap step. Otherwise, the voltage might be too high on one
180
5-13
CONTROL
tap step and too low on the adjacent step, with the result that the control
would force the tap changer to oscillate between the two adjacent steps.
In the rectifier, the voltage band between a = 10 and 20 is Kd0(cos 10
- cos 20) = 0.045 Vd0 , or 4. 5 % of Vd0 An allowance of 2 in measure
ment of the angles reduces the voltage band to KJ0(cos 12 - cos 18) =
0.027Fd0 = 2.7% of Vd0 . Hence the tap step must be less than 2.7%, say, 1.3
to 2.0%. In other words, the dead band is approximately 1.3 to 2 tap steps.
The inverter transformers have the same tap step as the rectifier trans
former, especially because most dc lines are intended to transmit in either
direction. With due allowance for errors in voltage measurement and in linedrop compensation, the voltage band should be at least +1.3 to 2.0%,
depending on the size of tap step.
181
groups.
Power control could be accomplished in either of two ways. The first and
most obvious way would be to measure the power on either the ac or dc
circuit, to compare it with a power command, and to use the difference (error
signal) to advance or retard the ignition of the valves. In order to retain the
advantages of current control, however, including its speed and the inclusion
of current limits (to be discussed later), the power control does not replace
current control but supplements it. The power command and the measured
direct voltage are fed into an analog divider, whose output signal represents
the direct current Id = PjVd. This becomes the current command, which is
one of the input signals to the current control.
The voltage measurements at both terminals must be corrected by current
compensation to a common point so that equal power commands at both
terminals result in equal current commands. The current margin signal is still
subtracted from the current command of the inverter.
Increase and decrease of power command, if made manually, should be
executed at the two stations in the order already described for current
control.
0.25 Vd,
Loci o
Ida
/max
Id
*
182
CONTROL
Sf=kAf
183
gf -kAf
(a)
(b)
Fig. 22. Frequency error signal
versus frequency /: (a) frequency-controlled network
exporting; (b) frequency-controlled network importing.
All the dc lines built heretofore have been essentially two-terminal lines,
having one rectifier station and one inverter station. Nevertheless, multiterminal lines have been proposed, and control schemes have been developed
for them. One of these is a constant-voltage scheme with all converters connected in parallel. Another is a constant-current scheme with all converters
connected in series.
There are schemes for example, Kingsnorth in which a bipolar line
forms a trunk from which monopolar branches go to different terminals
(Figure 23).49 Each pole is controlled independently, and the neutral con
ductor of the trunk carries the difference of the currents of the two poles.
The two truly multiterminal schemes to be described may be either bipolar
or monopolar. If bipolar, each pole may serve some different stations. The
schemes are described as monopolar.
?
j j
;
!I
j
wm
184
5-15
CONTROL
Inverter 1
Rectifier 1
Rectifier 1
Vdi
\/
Rectifier 2
Inverter 3
Rectifier 2
h
?dl
Inverter 4
Vds
Vd2
Inverter 2
185
MULTITERMINAL LINES
"T
Vat
[*-13
vd
AI
I
Id2
-Ir
H'
h
-hi
-Id2
(a)
Vdi
A line having four terminals two rectifiers and two invertersis assumed
for illustration (Figure 24). The control scheme is merely an extension of that
already described for a constant-voltage two-terminal line. Current com
mands Ix , I2, h > ft > having an algebraic sum equal to the current margin
A I, are sent from a control station to the respective converter stations.
Rectifier currents are taken as positive; inverter currents, as negative. The
station having the lowest ceiling voltage, that is, the lowest Vd0 cos oc0 or
cos y , controls the line voltage. This station is normally one of the inverters
operating at constant extinction angle. The other three stations operate on
constant current and at voltages lower than their respective ceiling voltages.
The current at the voltage-controlling station is the algebraic difference of
the current command and the current margin. (In the inverter this is an
arithmetical sum.)
vd
t*
tfe
I-h
y
L 1
Fig. 24. Four-terminal dc line with converters in parallel.
1
*-
h p*
i
hi
Vdi
VdZ
vd2
Id.2
i4 -
-JdS
-hi
(b)
Fig. 25. Control diagrams of four-terminal dc lines. Line voltage is controlled (a) by
inverter station 3; (b) by rectifier station 2.
several stations,* but the relationship among the four currentsthat their
sum is zerois less apparent.
If the ceiling direct voltage at one of the stations operating on constant
current fell (say, because of a drop in alternating voltage or because of a
converter fault), that station would become the voltage-controlling station
and its current would decrease by the current margin. Figure 25b shows the
current-voltage relationships with rectifier station 2 controlling the voltage.
The current of the former voltage-controlling inverter station 3 rises (alge
braically) by the current margin. The currents of stations 1 and 4 are un
changed.
If any of the converters is to be changed from a rectifier to an inverter or
vice versa, its current must be brought to zero, a reversing switch must be
operated there, and, finally, the current must be increased to the desired
value. The voltage polarity of the line could not be reversed unless it were
desired to reverse the sign of power at every terminal at the same time.
186
5- 16
CONTROL
MEASURING DEVICES
187
the voltages of both bridges are swiftly changed without changing the net
voltage of the pair.
Additional ideas on control of multiterminal lines are set forth in Refs. 51
and 54.
Measurements of direct current are required for the current regulator and
one of the inputs to the constant-extinction angle control of the inverter. For
both these purposes very fast response is necessary.
A direct current proportional to but less than the direct current in the main
circuit can be obtained, in a circuit isolated from the main circuit, by means
of a dc current transformer .58-64 This device consists of two saturable reactors
(transductors), each having two windings, an ac supply usually fed through
a small step-down transformer, and a rectifier employing several small diodes.
The reactor cores have a sharp saturation point and a very low mmf Fm
for saturation (Figure 27). One winding (the primary one) carries the direct
Magnetic flux <j>
Bridge 1
-Fm
0,5
MMF
s, -0.5
Bridge 2
-1.0
-2.0
-Fs
-1.5
1.0
0.5
-0.5
0
-1.0
Net voltage of converter/rated voltage per bridge
-F
1.5
2.0
Fig. 27. Idealized magnetization curve of a high-permeability core for a dc current
transformer.
188
5-16
CONTROL
Primary bar
MEASURING DEVICES
189
The flux< in each reactor core is the ordinate of the respective point on the
magnetization curve. If both reactors are saturated, the fluxes in both cores
are almost equal. If one is saturated and the other unsaturated (points R and
S), the fluxes are unequal.
The relation between these fluxes and the applied alternating voltage e will
now be developed. If the resistance of the secondary circuit, including second
ary windings of transformer and reactors, rectifier, and load, is negligible, the
applied voltage is consumed in the inductive drop almost all of which is in the
two reactors :
The negative sign arises from the opposite connection of the two secondary
windings. The time integral of this voltage is
->
source
(29)
Saturable
reactors
Load
Rectifier
e = Em cos cot
Then the flux linkage of the secondary circuit is likewise sinusoidal and lags
90 behind the voltage (Figure 29a), because
'
E
= Ns<f> = e dt = sin cot
o
that, in the absence of a secondary current, drives the cores of both reactors
far into saturation, producing flux $ and mmf F represented by point P in
Figure 27. The secondary current ts has a mmf Fs NJS that is added to Fp in
reactor A and subtracted from it in reactor B. Thus the total mmfs are
(21a)
FA = Fp + Fs = NpId+Nsis
(27b)
and
Fb = Fp ~ Fs = NpId Nsis
The points on the magnetization curve representing these conditions are
always equidistant horizontally from P, as, for example, are M and N.
Points M and N are for a very small secondary current, and both reactor
cores are still saturated. A larger secondary current drives these points farther
apart (points Q or S and R), causing reactor B to become unsaturated.
(30)
(31)
At every instant the net flux linkage of the two reactors must conform to the
value on this sine curve. The crest value Em of the applied voltage must be
chosen so that the crest value of the flux linkage EJco is somewhat less than
the height of the hysteresis loop, making it impossible for the two reactors to
be saturated in opposite directions. With no primary current, both are un
saturated throughout the cycle. With primary current having mmf greater
than Fm , one reactor or the other is unsaturated during most of the cycle
(points R and S). Only for short time intervals near the zeros of the fluxlinkage wave (near crests or valleys of the voltage wave) are both reactors
saturated (points Mand N). Hence during most of the cycle the mmf Fs of the
secondary current is is very nearly equal to the mmf Fp of the primary current
Id, differing by Fm, depending on whether |0| is increasing or decreasing.
While \<fi\ is increasing the operating point of one reactor is moving down
ward on the left-hand branch of the magnetization curve and Fs = Fp + Fm ;
190
5-16
CONTROL
MEASURING DEVICES
191
saturable reactors.
Ripples in the direct current are transmitted to the secondary circuit through
the unsaturated reactor acting as an ac current transformer unless special
means are used to suppress the ripple.
Another feature of this current transformer is noteworthy: the output
current is independent of the direction of Id. This is not necessarily a short
coming in applications in which the primary current is always in the same
direction, as it is in a converter.
two
Direct Voltage
(c)
while \<p\ is decreasing, the operating point is moving upward on the righthand branch of the curve and Fs= Fp~ Fm. If Fm is very small, this differ
ence is unimportant, and, in any event, the average of these two quantities is
the desired value Fp .
The wave shape of the secondary current is is shown in Figure 29b. This
current is rectified to form the output current i0 shown in Figure 29c. The
output current faithfully follows the magnitude of primary or input current
except for two minor features :
1. Narrow notches in small time intervals when both reactors are unsatur
ated (near zeros of iJ/s or peaks and valleys of e)
2. Differences of +Im occurring before and Im after a step at crest
values of \\j/s\ (zeros of e\ Im being the magnetizing current corresponding to
FmThe notches are the chief source of error, for Im is very small if suitable
material is used in the cores. The notches are undesirable in applications in
which the output signal must operate fast-acting devices.
The notches can be obviated by the use of two or three devices like that
described supplied with alternating voltage from different phases of a poly
phase source and with their output terminals paralleled. The final output
current is at every instant equal to that of the unit having the greatest
current.
Measurements of direct voltage are required for the control of the inverter
tap changers, for one of the inputs to a power controller, for current limitation
at low voltage, and for indication of abnormal voltages.
At the very high voltages used in dc transmission, measurement of the line
voltage is difficult and expensive. A resistance voltage divider is enclosed in
an oil-filled porcelain insulator. Its total resistance must be high for limiting
the power converted to heat to an amount that can be dissipated without
undue complication and expense. As a result, only a fraction of a watt is
available at ground potential, and an amplifier is required between the voltage
divider and the several instruments and controllers using a voltage signal.
The combination of the high resistance of the divider and its unavoidable
stray capacitance results in some time delay in the response of the output
voltage to changes in the input voltage. The time constant is of the order of
50 ms.
Power
Dc power can be measured by forming the product of the measured direct
current and voltage. A Hall effect device can be used for this purpose.. Its
output voltage is very low and must be amplified.
Ac power measurements can be used to give a reasonably accurate measure
ment of dc power, because the losses in the converter are small.
For reasons stated in Section 5-13 the direct measurement of power is
seldom used in the control of the power of a converter.
Frequency Deviation
In frequency control, a direct voltage proportional to frequency deviation
is required. The circuit used for this purpose is called afrequency discriminator.
It has a parallel LC circuit tuned to the normal frequency. The quadrature
192
PROBLEMS
CONTROL
positive deviation (frequency above normal) and lags behind it for negative
deviation (frequency below normal). This current must be compared with
another alternating current or voltage in quadrature with the voltage im
pressed on the tuned circuit.
Ignition Delay Angle
this angle is
1. Less than 10: direct voltage should be lowered.
2. Between 10 and 20: no action.
3. Greater than 20: direct voltage should be raised.
This angle can be measured in a number of ways. One way would be a
measurement of the time between the zero point on the sine wave of cornmutating voltage and the front of the grid pulse, which would then be com
pared with the times 10/360/and 20/360/. Another way would be to measure
the maximum instantaneous positive voltage across the valve, normally the
voltage just before it ignites, with 3_Em sin 10 and 3 Em sin 20 obtained
from a voltage divider fed from
Em sin cot, rectified. Of course, these
low
of voltage.
level
a
at
comparisons would be made
V
\/3
Input
PROBLEMS
Output
Input
(a)
CV
AC
v CC
(b)
AC
CC DC
Line
Rectifier
CC AC,
CV
AC
Inverter
(c)
Fig. 30. Monocyclic squares: (a) single-phase; (b) three-phase; (c) their use in constantcurrent dc transmission.
194
CONTROL
f
BIBLIOGRAPHY
1. "The Theory of the Control Problem of H.V.D.C. Transmission with Rectifiers and
Inverters in Bridge Circuit," by F. Busemann, Technical Report, Ref. Z/T74,
B.E. & A.I.R.A. , Mar. 2, 1948.
2. "The Stability of Frequency of A.C. Systems Connected by a D.C. Transmission
Line " by F. Busemann, Technical Report, Ref. Z/T75, B.E. & A.I.R.A., Mar. 5, 1948.
3.
14. " Control Systems," by U. Lamm, Appendix VI to C.I.G.R.E., 1960, Report No. 417,
"Report on the Work of Study Committee No. 10, D.C. Transmission at E.H.V.,"
pp. 29-33.
15. "Semiconductor Circuitry and Arrangements for the Protection and Control of
Bridge-Connected Current Convenors," by Colin Adamson and N. G. Hingorani,
Direct Current, Vol. 5, pp. 52-64, September 1960.
BIBLIOGRAPHY
195
16. "Control of Power Flow in a D.C. Link," by L. L. Freris and B. J. Cory, Direct
Current, Vol. 5, pp. 72-77, December I960.
17. "Corrector for the Compounding Device," by G. V. Smirnov, N.I.I.P.T., Vol. 6,
pp. 149-154, 1960. In Russian.
18. "The Scope of Semiconductor Devices in the Control and Protection of H.V.D.C
Systems," by N. G. Hingorani, Direct Current, Vol. 6, pp. 180-181, September 1961.
[
j
26. High Voltage Direct Current Convertors and Systems, edited by B. J. Cory, Macdonald,
London, 1965. Chapter 5, "Control of H.V. D.C. Systems," by L.L.Freris,pp. 73-93;
Chapter 8, " Some Design Aspects of the Cross-Channel Power Link," by L. A. Harris,
pp. 185-187, 199-211.
27. "Infra-red Radiation Pulses for the Control of a Converter," by R. Feinberg, M. E
Roberts, and J. Armstrong, Direct Current, Vol. 11, p. 72, May 1966. Pulses trans
mitted by semiconductor crystal lamps to photodiodes.
28. " System Stability and Control-Protection System of A.C.-D.C. Link," by T. Yamada,
S. Fujii, and T. Horigome, C.I.G.R.E., Report 325, 1966.
29. " Principles of Control for H.V.D.C. Transmission," by J. D. Ainsworth and C. J. G.
Martin, I.E.E. Conference Publication No. 22, High Voltage D.C. Transmission,
Manchester, Sept. 19-23, 1966, Part 1, Paper No. 30, pp. 158-160.
30. Discussion by Uno Lamm, I.E.E. Conference Publication No. 22, High Voltage D.C.
Transmission, Manchester, Sept. 19-23, 1966, Part 2, p. 74.
31. " Operation of the Control and Protection System of the Volgograd-Donbass Link,"
by K. Gusakovsky, A. Posse, and A. Reider, Contribution No. 95, I.E.E. Con
ference Publication No. 22, High Voltage D.C. Transmission, Manchester, Sept. 19-23,
1966, Part 2, pp. 130-135.
32. "EHV-DC Simulator," by J. E. Hudson, E. M. Hunter, and D. D. Wilson, l.E.E.E.
Trans, on P.A. & S., Vol. 85, pp. 1101-1107, November 1966.
33. " Simplified Study of the Stability of Regulation of a D.C. Link," by J. Clad6 and
A. Lacoste, Direct Current, Vol. 12, pp. 9-12, 14-15, 18-22, February 1967.
196
CONTROL
34. "A Method to Detect the Deionization Margin Angle and to Prevent the Commuta
tion Failure of an Inverter for D.C. Transmission," by T. Machida and Y. Yoshida,
I.E.E.E. Trans, on P.A. & S., Vol. 86, pp. 259-262, March 1967.
35. "A Method of Automatic Frequency Ratio Control by a D.C. System," by T.
Machida, Y. Yoshida, and H. Nakamura, I.E.E.E. Trans, on P.A. & S., Vol. 86, pp.
263-267, March 1967.
36. "A New Constant Extinction Angle Control for AC/DC/AC Static Convertors," by
Narain G. Hingorani and Philip Chadwick, I.E.E.E. Trans, on P.A. & S., Vol. 87, pp.
866-872, March 1968. Disc., p. 872. (I.E.E.E. Paper 31 TP 67-497.)
37. " The Phase-Locked Oscillator A New Control System for Controlled Static Con
verters," by J. D. Ainsworth, I.E.E.E. Trans, on P.A. & S., Vol. 87, pp. 859-865,
March 1968. Disc., p. 865.
38. "Present and Future Status of HVDC Transmission," by H. M&rtensson and
B. Skoglund, I.E.E.E. Publication 68 C 57, E.H.V. Conference held in Montreal,
1968. Contains a brief discussion of equiangular control.
39. " Untersuchung von Schutz- und Regelungsproblemen in der Modellanlage Rheinau "
(Investigations of Control and Protection Problems in the Rheinau Model Plant, in
German), by Reiner Foerst, Gerhard Heyner, Karl-Werner Kanngiesser, and Michael
Becker, ETZ, edition A, Vol. 89, No. 9 pp. 213-218, Apr. 26, 1968. English translation
in HGt)-Schriften-reihe, pp. 47-52, published by the Arbeitsgemeinschaft HGt).
40. "Der Loschwinkel-Regelkreis einer HGt) als Abstatsystem " (The Extinction-Angle
Regulating Circuit of a HVDC Transmission Link as a Sampled-Data System, in
German), by Reiner Foerst, ETZ, edition A, Vol 89, No 16, pp. 381-385, Aug. 2,
1968. Based on dissertation done at Darmstadt Technical College, 1967.
41. "Single-Phase Bridge Converter with a New Control and a Corresponding ThreePhase Converter Part II, Method of Control," by N. G. Hingorani and Philip
Chadwick, Direct Current (new series), Vol. 1, No. 1, pp. 31-35, April 1969.
42. " Control of the Extinction Angle of the Inverter of a High-voltage d.c. Transmission
System," by Winfried Muttelsee and York Rogowski, Direct Current, Vol. 1 (new
series), No. 2, pp. 49-53, August 1969. Feedback control of extinction angle using
low loop gain in steady operation but high loop gain when the extinction angle is
too small, as during a rapid increase of direct current.
43. "A Refined HVDC Control System," by Ake Ekstrom and GoteLiss, I.E.E.E. Trans.
on P.A. & S., Vol. 89, No. 5/6, pp. 723-730, May/June, 1970. Disc., pp. 730-732.
Provides equally spaced ignition pulses.
44. "Stability Analysis of the HVDC Transmission Control System Using Theoretically
Calculated Nyquist Diagrams," by Kjell Eriksson, Gote Liss, and Erik V. Persson,
I.E.E.E. Trans. onP.A.&S., Vol. 89, No. 5/6, pp, 733-739, May/June, 1970. Disc.,
pp. 793-740.
45. "Stability Analysis of the DC Power System," by Dennis P. Carroll and Paul C.
Krause, I.E.E.E. Trans, on P.A. & S., Vol. 89, No. 6, pp. 1112-1118, July/August,
1970. Disc., pp. 1118-1119.
46. "Control Methods for Improving the Reactive Power Characteristics of HVDC
Links," by Karl W. Kanngiesser and H. P. Lips, I.E.E.E. Trans, on P.A. & S., Vol. 89,
No. 6, pp. 1120-1125, July/August, 1970.
Pulse Transformers
47.
BIBLIOGRAPHY
197
I
i
49. "The Place of H.V.D.C. in the Power Transmission Field," by G. D. Breuer and
E. M. Hunter, I.E.E.E. Publication S-155, D-C Transmission, June 1963, pp. 5-18.
Disc., pp. 222-223.
50. "Series Connection of Converter Stations in an H.V.D.C. Transmission," by John
Reeve and J. Arrillaga, Direct Current, Vol. 10, pp. 72-74, 76-78, May 1965.
51. "Possibility of H.V.D.C. Transmission with More than Two Stations," Section
5.3, pp. 84-91, of Ref. 26.
52. " Universal Control-Protection System for Multiterminal D-C Power Transmission,"
by T. Yamada, H. Kondo, T. Horigome, and T. Sekine, I.E.E.E. Paper No. 31
PP 67-57, Winter 1966.
53. "Simulator Studies of Multi-terminal H.V. D.C. Systems," by H. M&rtensson and
T. Adielson, I.E.E. Conference Publication No. 22, H.V. D.C. Transmission, Man
chester, Sept. 19-23, 1966, Part 1, pp. 134-145. Three-terminal system controlled
sccording to the principles of Ref. 48. Oscillograms of voltage and currents during
blocking and deblocking of valve groups, etc.
54. " Multi-terminal D.C. Transmission," by K. W. Huddart and W. G. Watson, I.E.E.
Conference Publication No. 22, H.V. D.C. Transmission, Manchester, 1966, Part 1,
Paper 17, pp. 94-98.
55. " Multiterminal Operation of HVDC Converter Stations," by Reiner Foerst, Gerhard
Heyner, Karl Werner Kanngiesser, and Hermann Waldmann, I.E.E.E. Trans, on
P.A. & S., Vol. 88, pp. 1042-1050, July 1969. Disc., pp. 1050-1052.
56. " Model Tests on Three Terminal HVDC Systems," by R. Foerst and G. Heyner,
C.I.G.R.E., report 14-02, 1970, 10 pp.
Measuring Devices, Including DC Current Transformers
57.
58.
59.
60.
61.
ffo.
63.
64.
I
i
j
j
I
|
6-2
6
Misoperation of Converters
BYPASS VALVES
199
General
Most of the valve faults that are not self-clearing with the valve in service
are cleared by relieving the valve from current for a fraction of a second. This
is the purpose of the bypass valve.
Normally each converter bridge has a bypass valve across its dc terminals,
as shown in Figure 1. Since the bridge already has six main valves, the bypass
30 Ac
/10
/10
Commutation
198
Fig. 1. Addition of bypass valve (7) and switch (8) and disconnecting switches (9 and 10) to
a three-phase converter bridge.
200
MISOPERATION OF CONVERTERS
6-2
Positive dc line
BYPASS VALVES
201
both rectifier bridges is increased. Next, assume that a rectifier bridge instead
of an inverter bridge is bypassed. The rectiffer emf is almost halvedit would
be exactly halved if the remaining rectifier bridge did not decrease the ignition
delay angle to its minimum valueand, therefore, the current decreases.
The constant-current control of the inverter, however, increases the ignition
advance angle so as to prevent the current from decreasing by more than the
current margin.
r-j
(a)
180*
60
-60
Inverter
AY
YA
(b)
YY
YY
CM
1,2,7
Fig. 3. Commutation from a normally operating rectifier bridge to the bypass valve:
(a) voltages; (b) currents.
202
6-2
MISOPERATION OF CONVERTERS
a blocking signal (that is, the positive grid pulses are turned off) and that
valve 7 (the bypass valve) is unblocked; that is, its grid is made positive. At
this moment the anode voltage of valve 7, v7 = vdvn vp, is negative
and it cannot ignite. At instant D valve 3 would normally ignite but cannot
for lack of a grid pulse. Valves 1 and 2 continue to conduct, and the direct
voltage, vd = vp vn , declines. At instant E, the direct voltage of the bridge
becomes zero and starts to reverse. Immediately the bypass valve ignites.
Thereupon the effective circuit (consisting of the conducting branches in
Figure 1) becomes that in Figure 4. Its left-hand mesh has a line-to-line emf
BYPASS VALVES
203
Normally
positive for rectifier
negative for inverter
et
= Vh
Fig. 4. Equivalent circuit for commutation from regular valves 1 and 2 to bypass valve 7 of
a rectifier or inverter bridge.
eca , inductance 2Lc , and three valves in series. The bypass current i7 and the
commutating voltage are both zero at the moment when valve 7 fires (a 0
for the bypass valve). Taking this moment as t = 0, the voltage is
eCa
3 Em sin 03 1
(1)
i7 =/s2(l - cosrof)
where
IS2 =
(2)
(3)
cos u7 = 1
*s2
(4)
Fig. 5. Commutation from normally operating inverter bridge to bypass valve: (a)
voltages; (b) currents.
204
MISOPERATION OF CONVERTERS
The circuit diagram in Fig. 4 is still applicable, but with the difference of
polarity of direct voltage shown thereon.
its ignition delay angle at the normal value (around 15), its voltage builds up
rapidly with normal polarity. The voltage across the bypass valve under these
conditions is negative, as already noted; and current is transferred from this
valve after which it must cease conducting. Preferably the bridge voltage is
raised more slowly (by appropriate control of the ignition delay angle) for
limiting the overshoot of voltage due to the lightly damped natural mode of
the line and its reactive terminations.
In an inverter, on the other hand, the normal voltage is opposite to that of
the same converter when rectifying and would constitute a positive voltage
across the bypass valve. That valve would therefore continue to conduct even
if its grid had a blocking (negative) voltage, for the grid can exert no control
on a valve that is already conducting. The bridge, therefore, works into a dc
short circuit. Its current output is not limited by constant-current control.
Instead, because the current for this control is measured on the line side of
the bypass valve, it tries to raise the bridge voltage in an attempt to reduce the
line current, which now exceeds the inverter current setting by even more than
the current margin.
When the transfer of current from the bypass valve to the main valves is
required, the control should not only block the bypass valve and unblock the
main valves but also should reduce the ignition angle /? enough to make the
inverter bridge temporarily work as a rectifier. We have already noted that
the polarity of voltage of a rectifier is that required to transfer the line current
from the bypass valve to the main valves. After the completion of this com
mutation and adequate deionization of the arc in the bypass valve, the
decreased) so that the bridge
ignition of the main valves may be delayed
voltage builds up with proper polarity for inverter operation. The bypass
valve is now prevented from reigniting, despite its positive anode voltage, by
its negative grid voltage.
On examining these phenomena in more detail, we note that, with ft = 90,
the instantaneous open-circuit voltage of the bridge, which is still actually
short-circuited by the bypass valve, is positive about half of the time and
negative the other half (see Figure 6d in Chapter 3). This alternating emf,
in series with the relatively small commutating inductance 2Lc, can in a small
fraction of a cycle produce a short-circuit current in the amount and direc
tion to force the net current in the bypass valve to become zero. (The current
supplied from the bridge does not pass through the dc reactor.) Thus, the
value of fi required to force extinction in the bypass valve may be somewhat
less than 90 and, thus, be in the zone considered, from the standpoint of
205
average voltage and current, to be inversion rather than rectification; that is,
the required rectification occurs in about one-twelfth cycle (30) or less.
Special Auxiliaries
The bypass valve requires a different kind of current divider and a different
kind of grid-pulse supply from those used with the other bridge valves. The
reason is that the bypass valve is required to carry current continuously for a
longer time than the other valves normally do, and, also, its grid voltage must
be positive for a longer time.
The current divider for a multianode bypass valve differs from that for
other multianode valves (Figure 22 in Chapter 3) only in having equal resis
tors in series with each anode.4 A current divider like those of the main valves
gives nearly simultaneous ignition of the several anodes and equal current
division during pulses as short as those normally used (one-third cycle).
However, it cannot maintain good current division for times up to several
seconds. The added resistors do this. The power loss in the resistors would be
intolerable for continuous operation but is tolerable for a few seconds. If a
bridge must be bypassed for a longer time, a switch is closed in parallel with
the bypass valve, relieving that valve and its resistors of current. The voltage
drop in the resistors at rated current is of the same order of magnitude as that
in the arc.
The grid pulses for the regular valves are usually formed at ground poten
tial and are transmitted to the grids through pulse transformers. Such trans
formers cannot transform direct current and, therefore, work well only with
short pulses. Bypass valves do not always ignite immediately after they are
unblocked. Moreover, a positive grid voltage maintained for the duration of
desired conduction is a safeguard in the event of arc quenching.
A method of furnishing grid signals to the bypass valves is to connect a
free-running square-wave generator to the primary winding of the pulse
transformer. The secondary voltage of that transformer is rectified, and the
resulting direct voltage is applied to the control grid.5
Because bypass valves are sometimes ignited at a low anode-cathode
voltage, it is less certain that all the anodes of a multianode bypass valve will
ignite.
Use of Main Valves for Bypassing
Proposals have been made for the elimination of the bypass valve and
using, instead, a pair of the main valves on the same phase of the ac supply
for bypassing the bridge.6,7 This proposal has been adopted for thyristors,
206
6-3
MISOPERATION OF CONVERTERS
which do not arc back, but the bypass valve is still required for clearing
arcbacks in hv high-power mercury-arc converters. Even on such converters,
however, it is proposed to use main valves as backup in the event of failure of
the regular bypass valve to ignite.
6-3 ARCBACK
Causes
Arcbacks are the commonest and best known as well as the severest kind of
malfunction of rectifier valves but are less frequent in inverters. Since arcback
is reverse conduction, it can occur only when there is inverse voltage across a
valve. In rectification each valve is exposed to inverse voltage during approxi
mately two-thirds of each cycle but to forward voltage for a much shorter time
and with a lower crest value. In inversion the opposite is true (see Figure 20
in Chapter 3).
Arcbacks are a random phenomenon. We know some of the factors that
tend to increase the average frequency of occurrence but do not know how
entirely to eliminate them. An average occurrence of one or two arcbacks per
valve per month is considered satisfactory.
Among the factors that tend to increase the occurrence of arcbacks are the
following:
1. High peak inverse voltage
2. High voltage jumps (Section 3-5, page 108), especially of the jump at
arc extinction
3. High rate of change of current at the end of conduction
4. Overcurrent
5. Condensation of mercury vapor on anodes
6. Impurity of materials in anodes and grids
7. High rate of increase of inverse voltage
Most of these factors can be controlled. Factors 1 and 2 can be reduced by
decrease of rated valve voltage; 3 and 4, by decrease of rated current. These
measures, however, reduce the power handled per valve and, hence, raise
ARCBACK
207
the cost of the converter per unit of power. Factors 2 and 3 can be improved
by the use of small converter angles (a, /?, y, S); but these angles must be
increased to large values temporarily in such control operations as starting
up, maintaining constant current during dips of alternating voltage, or in
causing the transfer of line current from the bypass valve to the main valves
of an inverter. These operations do increase the incidence of arcbacks.
Factor 5 is minimized by maintaining the anodes at a higher temperature than
the cathodes. Factor 7 is minimized by the use of RC damper circuits in
parallel with each valve (discussed in Section 7-3). Factor 3 can be made to
occur less frequently by not allowing operation with very small overlap (see
Sections 5-13, page 181, 7-2, page 240, and 7-4, page 266).
i
I
Consequences
|
j
|j
5;
j
I
m
6-3 ARCBACK
209
Narrative
A.
B.
C.
id
IV
3V
Vp
'P
(a)
Fig. 6. Equivalent circuits for study of arcback of valve 1 with blocking and bypassing:
(a) first period, with valves 1, 2, and 3 conducting, and second period, with valves 1, 2, 3,
and 7 conducting; (b) third period, with valves 1, 3, and 7 conducting.
Crest
+ 18.8 kA
inverse voltage step immediately appears across valve 1, which has just
stopped conducting. Assume that valve 1 arcs back, renewing the short
circuit on phases a and b of the transformer; vp , which had momentarily
jumped by eb, now reverts to the average of ea and eb . v continues equal
to ec . Currents it and i3 continue on the same offset cosine waves that
they were on during normal commutation. Valves 1, 2, and 3 are con
ducting; see the equivalent circuit in Figure 6a. Blocking of further grid
pulses is assumed to occur promptly at the latest, before instant E.
D. cot = 60 + a. Valve 4 would normally ignite now. However, even if its
grid were pulsed now, it could not ignite until instant E (at intersection
of curves ec and eJ2), because its anode-cathode voltage is still
negative. Hence, from Cto Eis. the time available for blocking.
E. cot = 90. Direct voltage vd becomes zero and tends to reverse. Bypass
valve 7 ignites and begins to commutate with valve 2. Second period
begins, in which valves 1, 2, 3, and 7 are conducting; see Figure 6a.
They put a three-phase short circuit on the transformer. The bypass
valve short-circuits the dc terminals. All valve voltages, as well as volt
ages of ac and dc terminals, are now zero.
F. End of commutation. Current in valve 2 becomes zero and remains so.
Valve 7 current is ld. The third period begins, in which valves 1, 3 and 7
are conducting; see Figure 6b. Valve 7 keeps the dc terminals shortcircuited. Valves 1 and 3 constitute a line-to-line short circuit on the
transformer.
cot = 180. The commutating emf reverses. Currents in valves 1 and 3
reach their maxima and begin to decrease.
H. Currents in valves 1 and 3 reach zero. If valve 3 extinguishes, the original
arcback is finished. Henceforth only the bypass valve conducts. See also
Section 6-10, page 231, for the case in which valve 3 arcs back.
G.
i7 (bypass)
ojt, deg
ii (in valve
arcing back)
-10
~20i
D E
Instants: ABC
Period:
Valves
1st
1,3 k- 1,2,3
conducting:
F
2nd
S,2, 3, 7
*-b = - -ec
Va = Vb = Vp = -
(5)
vc = vn = ec
(6)
= v2 = v3 = 0
(7)
210
MISOPERATION OF CONVERTERS
6-3
cos a)
(9)
~h~ hi
cos a + la
30) 0.5 V3
(13)
(14)
Narrative
(10)
(11)
(12)
A.
2
= ic
+ 1] + 7d
(15)
\l3
3 =
(16)
i5 = i6 = 0
(17)
+ 1]
(18)
T *-b
Cc
= --
(19)
(20)
(21)
=iec
(22)
ii = z3 = ia =
(23)
h=h=0
(24)
h = Id
Numerical
(25)
e,
Lc
ij
-H-
-& ea
va
eb
Vb
ui 114
N'Qy
wId
f5=6
Example
Q-
n"
vc = ec
~v2 = v5
cot = 0. eaeb.
V4 = US = l>6
va = vb = vn = vp=
Is2 = 10.0 kA
and i3
211
(8)
h = ia
ARCBACK
-&
-w-
Vn
212
c.
6-3
M1S0PERATI0N OF CONVERTERS
wt = 60. Valve 4 normally would ignite at this time but cannot because
ARCBACK
213
-Q-nsw*" w
N
Id
-M-
Ic
L,.
la
1b
-Hic
v"
Lc
vc 5
eb
J.
2
ev. eb = Vb
Id
-wVZ
v$
"b
w-
I.
Id
V2 = V$ = U4
E.
la
214
6-3
MISOPERATION OF CONVERTERS
ARCBACK
215
Equations
Same as Eqs. (5) to (12) for the previous case, with cos a = 1.
2ndperiod, D to G, valves 1, 2, 3, and 5 conducting (Figure 10):
Fig. 14. Equivalent circuit 7: valves 1, 3, 5, and 6 conducting.
(26)
*"i = 4 = 4a cos (cot - 30) +Kt = 11.6 cos (cot - 30) - 13.9 kA
(27)
(28)
1*2 = 4= 1.8 kA
i3 = ib = /s3 cos (cot - 150) + Kz = 11.6 cos (cot - 150) + 4.2 kA (29)
i5 =ic + 4 -4 [cos (cot + 90) + I] = 11.6(sinco/~ I) kA
(30)
4 = *6 = 0
(31)
V5 =
- 1)2
(32)
(33)
(34)
(35)
(36)
(37)
ea + ec
va = vc = vp = j
(38)
v = eb
(39)
*h = v5 = v6 = 0
(40)
-vd=%eb
4 = 4 = 42 cos (cot - 60) + K5 = 10.0 cos (cot - 60) - 10.9 kA
(41)
% =
'
eb
v2 =v3 =v4 =
(42)
7\
2 16
MISOPERATION OF CONVERTERS
6-3
(43)
h = h = *4 == 0
*5
(44)
Valves:
+25
(45)
+20
(46)
+ 10
217
1,2,3,6
1,2,5,6
Instant:
ARCBACK
GiH
1,2,3
1,2,3,5
1,5,6
1,3,5,6
300'
360'
1,2,3
1.4 kA
(47)
(48)
i2 = i4 = 0
150) + 1] kA
23.0 kA
15.8 kA
120'
180'
240'
420'
(49)
(50)
*6 I
d = 1.8 kA
(51)
h = h ~ Id
($2)
-25.5 kA
-20
+E,
(53)
(54)
h~ I
d~ h
(5(5)
U = i5=0
i6 =Id- i2
(57)
(55)
360'
180'
360'
-E,
O>
(58)
180'
-E,
Use Eqs. (5) to (8), 10), and (12) from the previous case.
(59)
it = ia = ls2 cos at + Ktl - 10.0 cos at 12.4 kA
(60)
i3 == ib = Is2 cos at + K12 = 10-0 cos at + 14.2 kA
Crest Values of Currents. In the first cycle the crest values in the numerical
example are f = 25.5 kA,/3 = 15.8 kA, and i5 = 23.0 kA. Those for and
i5 are higher than the crest currents with successful blocking. The greatest
crest current is that of the valve which arcs back. The fact that |z A| and
i3 are greater at L (at = 450) than at D (at = 90) indicates that still higher
crest values may be reached in the second cycle, and, perhaps, in succeeding
cycles. However, decay of the dc components due to resistance, which was
Fig. 17. Arcback in uncontrolled rectifier without bypassing: (a) valve currents; (b)
alternating emf's and voltages of dc bridge terminals with respect to ac neutral point N;
(c) valve voltages.
218
6-3
MISOPERATION OF CONVERTERS
ARCBACK
219
at /, the same valves (1, 2, and 3) would be conducting from Kto L. Detailed
analysis of the other choices is left as an exercise (Problems 3 and 4).
Table 1 gives these crest values in terms of Is2 , Isl , and E/X. For the numeri
cal example, with hz = 10.0 kA, these crest values are 20.0 and 27.3 kA,
respectively. E[X =
is the effective value of three-phase short-circuit
current.
Study of the foregoing analyses of arcbacks shows that the crest currents
depend somewhat on Id and a, being greatest for Id = 0 and a = 0, the condi
tions for a fully offset cosine wave. Figure 18 illustrates these conditions for
two kinds of arcback.
For the controlled arcback the crest current in valves 1 and 3 in the first
cycle is clearly
(61)
ti=2Is2
/s3/>/2
f//.3
f//
iKEIX)
Controlled
Uncontrolled
1.73
2.37
2.00
2.73
2.45
3.35
(62)
An inverter valve can arc back only in the small time interval in which the
voltage across it is negative, that is, with cot between 180 y and 180. The in
verse current is of short duration, and its crest value is small, perhaps even
less than rated current. Because the valve suffers very little excess heating in
this case and because arcing back is a random phenomenon, there is little
likelihood of the valve's arcing back again after its next commutation or of a
consequent arcback. Inverter arcbacks are cleared without need of blocking
ait
270
180
90
Arcback in Inverters
_1L
-1
and bypassing.
The phenomena after an arcback are similar to those after a commutation
failure (Section 6-5).
-2
(a)
90" 120'
Is3[cos (ait
300'
30) 1.366]
(b)
Fig. 18. Computation of crest arcback current in valve 1: (a) with blocking and bypassing;
(b) with no blocking nor bypassing.
220
MISOPERATION OF CONVERTERS
6-4
alternating current greatly exceeds the direct current. The difference can be
used to give the blocking signal. The alternating currents may be taken from
the secondary windings of current transformers in the three transformer
bushings or valve-hall entrance bushings, while the direct current is taken
from a dc current transformer at or near ground potential. The secondary
alternating currents and rectifiers are required for each bridge and so are
fewer than the number of valves. The same direct current can be used for all
the bridges of a pole.
As will appear later (Section 6-9), the same comparison of direct current
and rectified alternating current can be used to indicate various inverter
|
|
i
A short circuit on the line side of the dc reactor is a line fault and is con
sidered in Section 7-6. A short circuit on the valve side of the dc reactor is a
converter fault and is considered here. It is assumed to be a fault on the dc
terminals of one bridge. It is similar in many respects to an arcback, and the
protection against it is the same as that against an arcback, that is, blocking of
221
the main valves and unblocking of the bypass valve whenever the rectified
alternating currents of a bridge greatly exceed the direct current.
faults.
I
?
1
j
j
Assume that the fault occurs while valves 2 and 3 are conducting and that
the remaining main valves are blocked before valve 4 is due to fire. There is
a phase-to-phase short circuit on phases b and c through valves 2 and 3 and
the fault. This continues for about half a cycle, during which the valve cur
rents reach a crest value of approximately hi and then decline to zero. In the
absence of arcbacks in both valves, which is very improbable, these currents
cannot reverse, and so the valves extinguish and are blocked. The bypass
valve cannot ignite until the fault current reverses, which occurs a few degrees
before the currents in valves 2 and 3 become zero, and can do so only if the
fault is an arcing one. Before then the polarity of voltage across the bypass
valve, due to the fault arc, is negative.
A dc short circuit is bound to occur, if it does at all, when the phase-tophase emf associated with the two conducting valves is near its maximum
value, and, consequently, the dc offset of the sinusoidal short-circuit current
is small. An arcback, on the other hand, occurring soon after the end of
commutation when the commutating voltage is small, is likely to produce an
almost fully offset sine wave of current.
Another difference between an arcback and a dc short circuit is that in an
arcback one of the two valves through which the short circuit occurs is
conducting backward; in a dc short circuit both valves are conducting
forward.
In multianode valves, still another difference is that the dc short-circuit
current is shared by all the anodes.
DC Short Circuit on Uncontrolled Rectifier
T*
\
222
MISOPERATION OF CONVERTERS
6-5
insufficient to ignite any of the nonconducting valves. When the current in any
valve decreases to zero, however, and is extinguished, a positive voltage
appears instantly across the other valve on the same phase, igniting it; for
example, when valve 2 extinguishes, valve 5 ignites, both of these valves being
connected to phase c. Thus the three-phase short circuit is maintained except
during extremely brief time intervals when there is a phase-to-phase short
circuit.
The dc components decay but usually too slowly to change the crest
current significantly during the first cycle.
COMMUTATION FAILURE
223
as 15 and the ignition angle as 40. For simplicity in computation, the failure
is assumed to result from late firing of valve 3, with direct current and alter
nating voltage remaining unchanged, although in practice changes in the last
two quantities are likely to be involved either as causes or effects. Is2 is
assumed to be 10 kA and Idto be 2 kA.
-L
(a)
These are given below in the same manner as the arcback currents in
Table 1. They are less than the corresponding arcback currents.
Table 2. Crest Currents in DC Short Circuits
Type
Controlled
Uncontrolled
///.3
Hhz
0.87
1.00
(loo)' 2.31
mix)
(b)
2.83
(c)
180
20
Causes
Instant
DE G
M N
Valves
conducting
Fig. 19. Wave forms of voltages and currents with failure of commutation of an inverter :
(a) phase emf's and dc pole voltages with respect to neutral of ac source; (b) direct voltage,
and (c) valve currents. Valve 1 fails to commutate to valve 3.
Narrative
Analysis
B 105
C 145
224
MISOPERATION OF CONVERTERS
225
D 180
J 345
K 360c
L 380
M 440
Fig. 20. Equivalent circuit when valves 1, 2, and 4 are conducting.
ve
ec
<S)
va
ea
Vb
eb
O-
H 260
/
320
N 465
P 500
Q 525
Symptoms
From Eto / (a period of 145) the dc terminals are short-circuited, and the
direct voltage of the bridge is zero; for 205 after E it is below normal.
During this time, the direct current, which in the foregoing analysis was
assumed constant, would really increase at a rate determined mainly by the
inductance of the dc reactor. From G to I
(105), there is no current in any
of the ac terminals. Thus, there is a time interval in which the direct current
is greater than the rectified alternating current, which is a condition opposite
to that during arcback of a rectifier valve.
226
6-6
MISOPERATION OF CONVERTERS
Cure
After the occurrence of a commutation failure, the succeeding commuta
tion is initiated earlier by the C.E.A. control system (Section 5-10) and is
usually successful. If the failure is caused by low alternating voltage, the
reappearance of normal alternating voltage helps prevent further failures. In
the event of persistent commutation failures, the bridge in which they occur
should be blocked and bypassed.
Double Commutation Failure
227
6-6 ARCTHROUGH
240
ARCTHROUGH
300
360
420
480
540 600
Causes
Arcthrough is conduction during a scheduled blocking period. It can be
caused by failure of the negative grid bias, by a defect in the grid circuit, by
the too early occurrence of a positive grid pulse, or by a sufficiently great
This arcthrough merely reduces the ignition delay angle from its normal
value (typically 15) to a smaller value or zero. Its effect on the wave shapes
of current and voltage is small. A persistent arcthrough in one valve, however,
introduces a dc component into the transformer currents, because this valve
conducts longer than does the other valve of the pair connected to the same
ac terminal, and the preceding valve conducts for a shorter time than its
paired valve. For the same reason, the harmonic components of both the
alternating currents and the direct voltage are altered.
Arcthrough in an Inverter
228
6-8
MISOPERATION OF CONVERTERS
D F
STU V
MN
J
I
WX
QUENCHING
229
(a)
Misfire in Rectifier
-U
Vd
60'
240
360
420
480
540
(b)
Fig. 24. Arcthrough of valve 3 of inverter: (a) valve currents; (6) direct voltage. At C
valve 3 arcs through while valve 6 is conducting; at Mit arcs through again. Valve 6 ignites
at P and extinguishes at F and T.
3 has not been removed, however, that valve arcs through again at M as soon
as the voltage across it becomes positive. When valve 6 ignites at P, the short
circuit of the dc terminals is reestablished and persists longer than it did
previously, until valve 6 extinguishes again at T, after which the direct voltage
is reestablished. The short circuit reoccurs once per cycle until the cause of
arcthrough is removed or until the bridge is bypassed.
Protection
6-8 QUENCHING
Causes
Prevention
6-7 MISFIRE
Causes
Misfire is the failure of a valve to ignite. It may be caused by failure of a
grid pulse or by extinction of the cathode spot by failure of the excitation
system or by low or reversed net cathode current.
Analysis of Misfire in Inverter
Suppose that valve 2 fails to ignite. Then obviously commutation from valve
6 to valve 2 cannot take place, and valve 6 continues to conduct. The con
sequences are much the same as if the commutation had begun but failed.
230
MISOPERATION OF CONVERTERS
23 1
Persistence
Similarities
Misfire
Arcback
Arcthrough
Quenching
Single commutation failure
Figure 25 shows the wave forms of direct voltage of the affected bridge for
each of these faults. The wave forms differ only in minor details, some of
which are caused by the calculations having been made for different values of
direct current. Part of the wave form for misfire, drawn in broken lines, has
fewer notches than the corresponding parts of the other three curves. The
first cycle of arcthrough (Figure 24b) may differ from the subsequent cycles,
depending on the instant of the cycle when arcthrough occurs.
The direct voltage during double commutation failure (Figure 23) has a
substantially different wave form from those of the other inverter faults in
that the direct voltage reverses for about half a cycle. It, too, however, has an
interval (50 in the example) of dc short circuit, and during this interval not
only is the direct voltage zero but also the maximum absolute value of phase
current is less than the direct current.
Misfire or arc quenching in a rectifier exhibits the same characteristics as
does a dc short circuit, but these are of shorter duration than are most inverter
faults.
60" intervals
co t
3,4
3,4
Fig. 25. Direct voltage during inverter faults: (1) arcback, /d = 0.18; (2) single commuta
tion failure, Is 0.2; (3) arcthrough, I'd = 0.1 ; (4) misfire or quenching, Is = 0.1.
Cause
During an arcback at least two valves carry excessive current the valve
that arcs back and carries reverse current and another valve that carries
forward current. Let us call the latter valve the partner valve. These two
valves constitute a phase-to-phase short circuit on the transformer. The short
circuit cannot be extinguished by the bypass valve alone.
The arcing-back valve is out of control and must be relieved of current as
soon as possible, so that it may recover. The partner valve, in spite of having
been subjected to excessive current, can usually function normally; that is, it
can cease conducting when its current falls to zero and it can withstand there
after a voltage of either polarity. The approximately equal sharing of the
overcurrent by the several anodes of the partner valve favors its normal
behavior, in contrast to the other valve, where the arcback is usually con
fined to one anode that carries the whole short-circuit current. Normal
behavior of the partner valve ends the short circuit.
Occasionally, however, the overheated partner valve misoperates and fails
to interrupt the short circuit. Its misoperation, being a consequence of the
arcback in the other valve, is called a consequentialfault.
232
MISOPERATION OF CONVERTERS
PROBLEMS
E F
\60
H IJ
'ii
Protection
CAB's are more serious than simple arcbacks, because they cannot be
cleared by grid control. They must be cleared by a slower method; hence the
valves involved become hotter and may, therefore, require more frequent
overhaul. The reduction of dc power lasts longer. In order to clear a con
sequential fault, the transformer of the affected bridge must be deenergized
by an ac circuit breaker. Reclosure should be delayed about 40 sec. Not only
is this much slower than grid control, but also it causes a greater reduction in
transmitted power than that of the affected bridge unless a circuit breaker is
provided for each bridge. A circuit breaker per bridge is more expensive
than a circuit breaker per pole and might not be required if consequential
faults were very infrequent. If the power rating per bridge is great, however,
a circuit breaker per bridge might be justified on other grounds.
//
u
-J-
h
120
LM
CAT-/
77
233
110 f
\ /AB
//
\
i
240
PROBLEMS
//CAB
l\
300/ 3 10
//
V\
C>N420
1
480
\\
\\
5 40
i
600
660//
wt, deg
\\
//
//
\\
\\
\\
720
//
//
234
MISOPERATION OF CONVERTERS
8. Analyze arcback in an inverter. Assume that valve 1 arcs back at the end
of its commutation with valve 3 at y = 15 and that there is no blocking.
Id = 1.8 kA, Is2 = 10.0 kA.
9. Analyze a short circuit on the dc terminals of a bridge of an uncontrolled
rectifier with no bypass valve under the following assumptions: The
short circuit occurs at cot = 35. Id = 1.0 kA, Is2 = 10.0 kA.
10. Same as Problem 9 except that the rectifier has grid control with a 15
and has a bypass valve. Blocking of the main valves and unblocking
of the bypass valve occur at cot = 45.
11. Analyze a double commutation failure in an inverter that normally
operates with y = 15,u = 15, and, hence, = 30. Is2 = 10.0 kA. The
failure occurs because valves 3 and 4 fire 5 late, that is, with fi = 25.
12. Analyze arcthrough of valve 3 of an inverter at cot = 75 while valves 1
and 2 are conducting. Id= 1.0 kA, Is2 10.0 kA, y = 15.
13. Analyze misfire of valve 2 of an inverter for which Id = 1.0 kA,
Is2 = 10.0 kA, and y = 15.
14. Analyze misfire of valve 4 of a rectifier operating with Id= 1.0 kA,
/s2 = 10.0 kA, and oc = 15.
7
Protection
7-1 GENERAL
4.
5.
6.
7.
8.
9.
10.
7-2 DC REACTORS
236
7-2
PROTECTION
'
In Chapter 6 it was shown that nearly all inverter faults lead to the collapse
of the direct voltage of a bridge. This voltage remains at zero for about
three-eighths of a cycle, which is several times greater than the time available
for commutation. The countervoltage of the affected inverter pole falls to
(b 1)fb times its normal value if there are b bridges in series; for example,
if there are two bridges per pole, the voltage of the affected pole drops to
half of its normal value. Consequently, the direct current through the inverter
rises, tending toward a new steady value such that the new RIdrop in the line
is the difference between the rectifier voltage and the reduced inverter voltage.
The rise of current prolongs commutation and, if excessive, causes commu
tation failure, which, in turn, leads to an additional large decrease of inverter
voltage, which further increases the rate of rise of direct current. Thus the
failure is likely to spread to all the bridges of the affected pole, decreasing
the power transmitted by the dc line to half of its prefault value.
It is good practice to limit the magnitude of such a disturbance to the dc
and ac systems by confining it to one bridge. To do so, the rise in direct
current during a commutation in one bridge, caused by a collapse of direct
voltage in another bridge, must be limited to a value that does not cause
commutation failure.
The factors governing the amount and rate of rise of direct current are,
in addition to the dc reactor, the surge impedance of the dc line, lumped
capacitances at the inverter station, and current regulators.
The surge impedance Zs limits the increase of direct current A Id drawn by
the inverter from the line due to a decrease of inverter voltage AVd to
A Id = A VdjZs. This limitation holds for the time of round-trip wave travel on
the line. On a short line the reflected wave returns so soon, and on a cable
the surge impedance is so low, that the rise of current is not adequately limited.
DC REACTORS
237
Even the surge impedance of a long overhead line would not limit the
discharge of lumped shunt capacitances, such as the surge capacitor see
" Protection against Lightning," page 285 through the inverter.
The current regulator at the inverter only prevents a decrease of direct
current in excess of the current margin. The current regulator at the rectifier
has too long a time constant and, even if it were instantaneous, would be
powerless to prevent discharge of shunt capacitance of the line and of the
inverter station through the inverter.
Practically speaking, then, a dc reactor is always required, and its induetance is the principal factor limiting the rate of rise of direct current. The
magnitude of rise during commutation depends on the overlap time.
In calculations of the inductance required for sufficiently limiting the rise
of direct current after the collapse of the direct voltage of a bridge so that a
commutation failure does not ensue, the following assumptions are made :
1. The direct voltage on the line side of the dc reactor remains constant
initial value. This assumption may be slightly pessimistic: its corollary
is that only the dc reactor limits the rise of direct current.
2. The alternating voltage at the inverter remains constant. A sudden drop
in this voltage would make commutation failure more likely, but it is improb
able that such a drop would coincide with misfire or arcthrough of a valve.
3. The tap changer does not move.
4. Initially the direct current has its rated value Idn. This requires greater
overlap than would a smaller value of Id and thus gives more time in which
Id can increase.
5. The collapse of direct voltage occurs simultaneously with the beginning
of commutation. This is the worst time for it to occur. If it occurred earlier,
the C.E.A. control could advance the ignition angle some. If it occurred later,
there would be less time left in which the direct current could increase before
the completion of commutation.
6. Because of error in the C.E.A. control, ignition is assumed to be 1
I
j
j
j
at its
late.
Let the normal extinction angle be yn - 16, and let the minimum extinc
tion angle to be permitted as a result of collapse of direct voltage be
= 8.*
The normal ignition advance angle ft is calculated from Eq. (96) in
Chapter 3 :
/
cos ft, = cos y--~
(1)
52
an angle of 5 to 8.
'
jl
238
PROTECTION
7-2
and 1 is subtracted from it. Next, the greatest permissible current at the
end of commutation, that which reduces y to ym, is computed from a valid
modification of Eq. (1) in which Idn is replaced by the arithmetic mean of
the values of Id at the beginning and end of commutation, hp and /dv, respec
tively. The modified equation is :
(2)
= /,2(cosy-cos/?)
AId = 2x
By Eq. (6),
37.6- 8.0
ym,
hp = hn,
and
hy = hn + &h-
=L37X1"
S6C
By Eq. (5),
Then
(3)
Example
(4)
AVsAt
La=AL
(5)
then
where
200 x 1.37 x
= 036
10-3
= -76 H Ans-
Find the smallest fraction of normal value to which the balanced alter
nating voltage at the inverter in Example 1 can fall suddenly without causing
a commutation failure. Assume that the fall of voltage occurs, as in
Example 1, immediately after the beginning of a commutation at fi = 37.6
with Id = 1.8 kA. Assume also that a commutation failure occurs if y < 1.0.
SOLUTION
1 - y*
At fin-
360/
Example
29.6
W60=600
y
239
By Eq. (3),
Let us put
fi=fin-\,
DC REACTORS
(6)
By Eq. (1),
Commutation is made more difficult not only by increase A Idof the current
to be commutated but also by a decrease in the commutating voltage, which
is proportional to Vd0. Initially
'
dO ~
2Va
cos y
+ cos fi
2 x 400
cos 16 + cos 38.6C
800
= 459 kV
0.961 + 0.781
Let the alternating voltage fall from E to kE. Simultaneously, Vd0 falls to
kVd0 and Is2 falls to kls2. The direct terminal voltage falls from =400kVto
E2S
v'-kv
yd~KVdO
= k 459
= k 459
y'
+ cos F
2
1.000 +0.792
---= k 411
7-2
240
PROTECTION
The drop in Vd is
360
\f)
J = 1.70xl0- sec
360 x 60
A/id
':
As a check, let us calculate A Idby putting this value of k into both expressions :
This ripple concerns us in two ways. First, it may induce noise in telephone
circuits that parallel the dc line. This is discussed in Chapter 8. Second,
since the magnitude of the ripple is almost independent of the average value
of direct current, the ripple sets the lowest average current at which the
current is continuous. Below this, the current becomes discontinuous or
intermittent. In twelve-pulse operation, the current is then interrupted twelve
times per cycle. Such operation of a high-power converter is objectionable
on account of the high overvoltages {L dijdt) induced in the windings of the
converter transformer and the dc reactor by the high rate of change of
current at the instants of interruption. These overvoltages are related to
those caused by arc quenching.
Actually, operation, even with continuous current, is objectionable if the
DC REACTORS
241
overlap is too small. Two reasons may be given. The first is that the two
jumps in the direct voltage at the beginning and end of commutation merge
to form one jump twice as great. Consequently, the overshoot of direct
voltage caused by a high-frequency oscillation is twice as great and may flash
over the protective gap placed across the dc terminals of each bridge. See
Section 7-8, page 286. The second reason is that the current oscillation pro
duced in a valve that is beginning to conduct (see "Arcback," page 266) is
capacitively coupled into the valve that is ending conduction. If the overlap
is short, the oscillation does not die out before the latter valve ends conduc
tion and may produce a high rate of change of current at the end of conduc
tion. This is a condition conducive to arcback and should be avoided, because
it tends to increase the erosion of the anodes.
The minimum current limit is introduced into the control system of the
converter for this purpose. It should be set at about twice the critical current
below which there is no overlap.
Calculation of Ripple. Let the following assumptions be made for a study
of the effect of the dc reactor on the ripple in the direct current:
1.
2.
3.
4.
did =
dt
(7)
Therefore
1
"~L.
(P. - Vt) dt
(8)
242
PROTECTION
7-2
Ld
Id
w
VdVd
Vd
DC REACTORS
243
0.2
0
Vd
(a)
-0.2
Fig. 1. Circuit for analysis of relation between current in and voltage across the dc
reactor.
0.8
Max
0.6
I
0.4
ld
Em=V,
(9)
dO
0.2
The equation of the first such arc after the origin of time is
vd =
for a cot a + 60
In the absence of overlap, the voltage on the line side of the reactor is
Fig. 2. Ripple in direct (a) voltage and (b) current for = 30.
fi
Vd = Vd0\
- cos (cot
30) cos a
(12)
[sin (cot
(cot a) cos a
(13)
L=
dO
id d(cot) = coL,
Id = f (0.0931 sin a)
coLd
This is plotted in Figure 2a for a = 30. Since Vd is the average value of vd,
the average voltage across Ld is zero.
The instantaneous direct current id is found by substitution of Eq. (12) into
Eq. (8) and performing the indicated integration over the range of cot from
atoa + 60. The result is
h=
<x>t
(11)
Vd = Vdo cos cc
vd
75
(10)
71
71
(14)
(15)
"/
244
7-2
PROTECTION
Example
Find the inductance of the dc reactor required to keep the direct current
continuous when the converter in Example 1 is operating with one bridge
per pole at
SOLUTION
DC REACTOR
245
reactor and the total shunt capacitance of the station, and the dc line is
represented by one of the approximations shown in Figure 3. A long over
head line may be represented by a Thevenin equivalent circuit for transients
comprising the open-circuit voltage and surge impedance (Figure 3a). A
cable may be represented by a nominal T or n circuit with series resistance
and charged shunt capacitance; the n circuit is shown in Figure 3b, where
the rectifier is represented by a constant-current source.
Ld =
(Oldm
Vd0 at light load may be taken equal to the no-load direct voltage of one
bridge, 200 kV.
sin a = sin 90 = 1
co 2nf = 27160 = 377 rad/sec
idm for barely continuous
Dc hne
Terminal
200 x 0.0931 xl
L'-iTr
Dc cable
Termina
A4nTr
= 0'49H
R,
-vW-
i,r
Ld -LCi
' 2
2
Zg Cg
_L
\Ld
(c)
(d)
Fig. 3. Approximate circuits for calculation of discharge of (a) overhead line and (b)
cable through bypass valves BP. Simplified circuits for transient component of discharge
current :(c) overhead line, {d) cable. Ch line capacitance; C station capacitance; Ed, line
voltage; Ld, dc reactor; Ri, line resistance; Zs, surge impedance.
In either case, the current through the bypass valves consists of a steadystate (dc) component and a transient component.
For the circuit in Figure 3a, the dc component during the first round-trip
wave-travel time is EdjZs. The transient current, found by analysis of the
simplified circuit in Figure 3c, is a damped sinusoid. The current through
z.il~s
cos cot
coL,
CO
]}
1 sin cut
(16)
246
7-3
PROTECTION
where
-zfa
co =
(I7a)
Vtojl a2
(17c)
SOLUTION
Ld (H)
0.1
0.3
0.5
1.61
1.11
1.013
0.3 H/
/ 0.5
' '
"""
-/
/ / /
/ / / /
/ / / /x
/ / /
_ _ -- _
'
/ // /
\ // /
I// /
Example 4
\\
_ ___ _
---
V
\\
/ o.l H
/
/
//
For the circuit in Figure 3b, the steady-state current is Id. The transient
current consists of a dc term with exponential decay and a damped sinusoidal
term.
The effect of the dc reactor on the crest value of discharge current may be
seen from the following example.
I
\
~~
(17b)
247
'/
11
III!
Time t, ms
Fig. 4. Discharge current of overhead dc line as affected by the inductance of the dc
reactor. The broken lines are the envelopes of iL for Ld = 0.1 H.
In order to avoid too frequent arcbacks, one must limit the rate of rise of
inverse voltage and the peak inverse voltage across mercury-arc valves to
values depending on the valve design. Damping circuits are used for satisfy
ing these limitations rather than derating the valve. Such circuits are required
also across thyristors for avoiding their breaking down on inverse voltage
exceeding the rated value.
The crest current does not exceed the rated current of the bypass valves
(1.2 kA) except fot Ld = 0.1 H.
The round-trip wave-travel time is 2ljv = 2 x 1300km/(300 km/ms) =
8.7 ms. The crest currents occur earlier (1 to 4 ms); hence the line represen
tation is valid.
Vje =
y/3Em sin 5
(18)
248
PROTECTION
7-3
249
1.5 E,
kSk
-VJ2Fig. 5. Theoretical voltage across rectifier valve, showing voltage jumps. Drawn for a
= u = 15: Vji =\3Em sin a is ignition voltage jump; Vjc = \/3 Em sin 8 is extinction
voltage jump. The smaller jumps (F}</2 and VJc/2) occur at ignition and extinction of
other valves.
by solid connections.
2D
Valve
Switch
Active network
Passive network
Fig. 7. Simplification of the converter circuit of Figure 6 for finding the transient voltage
across valve 1 immediately after the end of its conducting period. Block D represents an
RC damper circuit; 2D represents two such circuits in parallel.
5
i
i
j;
j
j
j
j
'
250
7-3
PROTECTION
bo
2D
2 S
IOC
(b)
(a)
D=
Cs
= admittance of one damper
RCs+ 1
(21b)
Figure 9b, differing from 9a only in the detailing of the equivalent dampers
1 and 2, makes the complex natural frequencies apparent. They are
bo
(22a)
(22b)
$2* 4 =
2D
(21a)
and
(19)
3 L,
X:2C
where
5. During the short time that is of interest, say, one cycle of the natural
frequency, it is sufficiently accurate to represent the source current as a ramp,
h-t
251
where
a*
31,
and
j2
i
JL = 0.fal
(23a,b)
co, =
Fig. 8. Circuit equivalent to the passive network of Figure 7 with respect to terminals
a,b, c after eliminating node N.
65 +
10Z> + 25 + 2D
85+ 12D
(25 + 2D)(6B + 10D)
(20)
"'~V3Z+
and
co, =
"2~~j5Lrc
= 0.775co,
(24a,b)
VcoJi - <7?
and
co2 =
y/a>i2
(25a,b)
V1(4 = ZJs)I1(s)
(26)
252
PROTECTION
where Zab is given by Eqs. (20) and (21), and IL is the Laplacian transform of
Eq. (19), which is
cr,
= 0.915
/,=
cox
(27)
e'CTl'cos coxt
i(0 =
sin
Example
Hz
1=1810
2n
= 1640 Hz
f1= 1M?5
2n
= 0.613
COt
The voltage is
253
since,*) 1
o>2
(28)
Calculate the wave form of voltage across a valve and its associated damper
at the Lydd terminal of the Cross Channel link. The damper parameters are
R =3500 Q
C = 25 nF
Lc = 56 mH
A curve plotted from the foregoing equation is plotted in Figure 10. The
calculated overshoot is 22.25%. If the high-frequency resistance of the trans
former windings had been taken into account, the calculated overshoot would
have been somewhat less. If stray capacitances across the valves had been
taken into account, the curve would have been shifted to the right a little
and joined to the origin through a short arc.11 The inclusion of stray capaci
tance in the analysis complicates the mathematics but does not appreciably
modify the rate of rise and crest value of the reverse recovery voltage across
the valve.
In order to put Eq. (28) into a more general form, introduce the dimensionless variables
*'
Vj = 100 kV
SOLUTION
0.1
-Jlcc
0.5
By Eqs. (23),
1=
<j
3500
= 10,420 sec-1
6 x 56 x lO"3
-50
By Eqs. (24),
= (3 x 0.056 x
o)2
=0.775o
-100
By Eqs. (25),
Overshoot
<ox
-130
(29)
254
7-3
PROTECTION
from which
255
SOLUTION
cr't'
m=t.
at
alt=,
f'V 3 (<r')~2
t'
----
/
= = ,*'
V3'
and
t'
V5'
3!
sin co'.t
103/25 x 10-9\1/2
(56
lO-3/
'
-=
--
, 3.5
t'5-(a')2
+ 0.25e"<,r''5cos2
sin j/ 1
(30)
The form of the curve of v' versus (' depends only on the parameter a'. A
graph of overshoot as a function of a' is given in Figure 11.
Example
The power lost in the valve damper circuits is important both (1) because
of its effect on the overall efficiency of the converter station and (2) because
it determines the heat to be dissipated from the damper resistors, and hence
affects their size, cooling requirement, and cost.
Reasonably accurate formulas for the loss have been presented by Ainsworth11 and are as follows :
Pd = Phf + Ptf
where
Find the voltage overshoot in Example 5 by use of the curve in Figure 11.
(31)
fx
(32)
(33)
= fundamental frequency, Hz
C = damper capacitance, F
Cs = stray capacitance, F
R = damper resistance, Q
a = ignition delay angle
6 = extinction delay angle
u overlap angle, rad
Pd = power loss in damper for one valve, W
Phf = high-frequency component of Pd
Pis = low-frequency component of Pd
100
75
50
256
7-3
PROTECTION
jump and (b) a sinusoidal component is resulting from the sinusoidal arc of
voltage.
The transient component of current actually has the form of a rapidly
damped high-frequency oscillation, as discussed in Section 7-3, page 247,
but, according to the assumption made at the beginning of this derivation,
it would be a nonoscillatory exponential wave of short time constant RC.
Under either assumption, this component vanishes long before the occur
rence of the next voltage jump.
The sinusoidal component of current is an arc of fundamental frequency
leading the corresponding voltage wave by nearly 90, since at this frequency
the reactance of the capacitor predominates.
The instantaneous power consumed by the resistor is
p = Ri2 = R(jt
watts
(34)
In the last member of this equation, the first term is the high-frequency loss,
and the sum of the second and third terms is arbitrarily called the lowfrequency loss, although the second term depends on both components of
current. It is obvious that the first and third terms are always positive but
that the second term may have either sign. The expression must be integrated
with respect to time over the eight parts of one cycle; and the result must then
be multiplied by /j in order to find the energy dissipated per second, which is
the power loss. The energy entering the capacitor over a whole cycle is zero.
Whf
= 0.5CZV]
jumps.
Example
Find the damper loss per valve and per bridge of the converter described
below for both rectifier and inverter operation at full load. Each converter
is rated at 1.8 kA, + 400 kV, 1440 MW, and has three bridges per pole. The
resistance of the bipolar line is Ri = 18.95 U per pole. The sending-end
voltage is kept at rated value. The equivalent commutating resistance is
7.65 Q per bridge. The rectifier normally operates at a = 15 and the inverter
at y 16. The fundamental frequency is 60 Hz. The damper circuits have
R = 1200 Q, C = 0.100 /iF. Neglect stray capacitance.
SOLUTION
x 1.8
=ra==152-5kVperbridge
Vll=1~M1.35
LL
U3.3 kV
= 0.966
cos 5 = cos a. --
Vd0
8 = 38.2
--
TFFs152.5
-66
0.181 = 0.785
joules (35)
257
*J2VLL
\J2VLL
Vdo cos y
258
7-3
PROTECTION
135.8
135.8
=
= 141-3 kV per bridge
Vdo 6
cos y pTnTi
0.961
= 2940(2.46
259
Vll =
cos /? = cos y
104'! kv
2RcId
2x13.8
-
= 0.961 = 0.961 - 0.195 = 0.766
v do
141.3
P = 40.0
+ sin2 38.2)
= 1.138 x
= 1.138 x
+ sin2(180 - 16)]
The high-frequency loss is slightly greater than the whole damper loss.
The damper loss in rectificationis473/240,000 = 1.97 x 10 ~3 pu = 0.20% of
the rectifier power. In inversion it is 334/220,000 =1.52 x 10~3 pu = 0.15%
of the inverter power.
= 2nf\VtL2C2R[2A6
If valves were developed that could withstand a more rapid rise of inverse
voltage and still have an acceptably infrequent occurrence of arcback,
smaller capacitances could be used in the dampers, producing a correspond
ingly smaller power loss in the damper resistors.
The damper loss during inversion, although normally less than that during
rectification, becomes greater under the abnormal condition of low direct
voltage together with the inability of the tap changer to lower the alternating
voltage proportionally because of having reached its limit. The required
ratio of direct voltage to alternating voltage is then obtained by advancing
the ignition angle of the valves, leading to bigger voltage jumps and to higher
damper losses. The inverter is then operating on its constant-current charac
teristic instead of on minimum extinction angle. One cause of low direct
voltage at the inverter is the bypassing of a bridge at the rectifier. If this
condition is sustained, the corresponding bridge of the inverter should be
bypassed also. Even so, the direct voltage per remaining bridge on the same
pole is, at rated current, lower than its normal value, because the line voltage
drop is divided among fewer bridges. The inverter tap range should be
designed to permit continuous operation of the inverter at rated current
without the damper loss's exceeding its value for rectification at rated current
and voltage, although the extinction advance angle may be somewhat greater
than its normal value.
Even if the tap range were adequate for allowing continuous inversion at
normal extinction angle and highest alternating voltage, both rectifier and
inverter would operate at abnormally large ignition angles for short times,
such as during start-up or blocking or unblocking of a bridge. The damper
260
PROTECTION
resistors, which are usually oil-cooled, have a long enough thermal time con
stant to endure such brief overstressing without overheating. The damper
capacitors, however, cannot withstand even such a brief application of a
All these effects depend on the magnitude of the ignition voltage jump and are
therefore greater during temporary operation at abnormally large converter
angles than during normal operation.
Radio interference is discussed in Vol. 2, Chapter 16. The other two effects
are discussed in what immediately follows.
Extinction of Cathode Spot
7-4
261
current. If the net cathode current (the sum of the currents from the main
anodes and the excitation anodes) becomes zero, the spot becomes unstable
and could be extinguished. Then the valve cannot conduct again until the
spot is reestablished by the ignition electrode. Extinction would occur if,
during an oscillation, the main current goes negative by the amount of the
excitation current.
Because one-way conduction is the distinguishing characteristic of a valve,
one might doubt that the main anode current could ever reverse except when
an arcback occurs. A negative current of brief duration, however, can flow
because of the appreciable time required for deionization of the space between
the main anode and the control grid. This time is ample for a half cycle of
a high-frequency oscillation, although in the same time the power-frequency
current could change but little.
Oscillations of the current of the incoming valve at the beginning of com
mutation can be calculated reasonably well by use of the equivalent circuits
in Figure 12. In circuit a, the stray capacitance C is initially charged to
La ,
~\SLr
(a)
(b)
Fig. 12. Basic circuit of current oscillation in incoming valve: (a) stray capacitance and
inductance only; (b) anode reactor and resistor added.
voltage V}i. When the valve is ignited, the capacitance discharges through
the stray inductance L and the valve. A high-frequency oscillation occurs, as
illustrated by curve a in Figure 13, in which the negative main current
greatly exceeds the excitation current, causing spot extinction at point P.
Spot extinction is avoided, and radio interference is decreased, by the
addition of the circuit components shown in Figure 12b:
* The resistor across the anode reactor of the bypass valve is only 200 to 400 Q.
262
7-4
PROTECTION
160
Example
263
> -50
03
>
(A)
-160
<
C
OJ
0)
Time, ms
_>
SOLUTION
-25
Vji =
-50
(B)
Fig. 13. Current of incoming valve at beginning of commutation: (a) without anode
reactor; (6) with anode reactor; (c) with anode reactor and resistor; (d) power-frequency
component of valve current; (e) negative of excitation current.
The reactor lowers both the frequency and the amplitude of the current
oscillation, as illustrated by curve b in Fig. 13. The resistor increases the
damping of the oscillation and thus further decreases the reverse main current,
as illustrated by curve c in Figure 135. Such a moderate reversal of the main
current, having crest value less than the excitation current, is harmless.
The reactor is connected on the anode side of the valve and is therefore
called an anode reactor. The combination of reactor and resistor may well
be called an anode damper. If the damper were connected to the cathode side,
it would be less effective, because it would be bypassed by a greater stray
capacitance than that on the anode side. In some converters, dampers are
connected on both sides of the valve.21
\/2Ell sin a
V2 x
113 x 1 = 160 kV
litLC
=
1 _
x 10-10
2tiV10-4
264
PROTECTION
7-4
where
265
= 0.5
whence
Then
m~
= 1000n
tt = In 2 = 0.693
160,000
1000
and
= 0.22
C=
Also
amperes
---
104
Vnt = -160 x = 3.8
2T,
2 x 21.1 x lO"3
106r
2.
R=
amperes
is plotted in Figure 13 as curve a. The current has a crest value equal to 3.2
times the excitation current and results in extinction of the cathode spot
and cessation of anode current.
160,000
= -J = ! = 4000
40
The value of the stray inductance, which is included in this value, should be
subtracted to find the inductance of the anode reactor, which is then
4 = 1.6-0.1 =
1.5 mH
MHz
1.6
f= -
--
0.22
A;==il=18A
Vn
4
2R
0.8 x 50 = 40 A
2R
whence
2R\C)
amperes
9.1
106 rad/sec,
and
C = 0-22
z't = 18 A
and
dit_Vn_
dt
L .
160 xlO3
=1()8A/sec
1.6 x 10-3
'
it = [45 sin (2.44 x 106t) + 18 cos (2.44 x 106r)] exp(-0.55 x I06t) amperes
The current i= it + is is plotted in Figure 13 as curve c.
The significance of the damping of the high-frequency oscillation will be
more apparent from what follows.
266
7-4
PROTECTION
267
Arcback
Current oscillations occur not only in the current of the incoming valve
but also in that of the outgoing valve of the same commutating pair. The
current oscillation in the outgoing valve is caused by capacitive coupling
between the two valves, represented in its simplest form by the circuit in
Figure 14 for commutation from valve 1 to valve 3. If the direct current is too
small and the voltagejump too great, the amplitude of the high-frequency cur
rent oscillation is great enough to force the current in the outgoing valve to
zero prematurely with a high rate of decrease, as sketched in Figure 15, curve
ix . Extinction of the current in the outgoing valve is immediately followed by
a high-frequency voltage oscillation across that valve, induced by the same
current oscillation in the incoming valve. This voltage oscillation is super
posed on the inverse recovery voltage already discussed in Section 7-3 so that
the initial rate of increase of that voltage is thereby increased.
A high rate of decrease of current in the outgoing valve as that current ap
proaches zero, followed immediately by a high rate of increase of inverse
->
ik 3
Co
<
o
.20.
-10 1
-1
Time, jus
43
11
Fig. 15. Currents in a pair of valves (1 and 3) undergoing commutation with short
overlap (1 or 46 /us): h, current of outgoing valve (valve 1); /3, current of incoming valve
(valve 3); i[ and iz ,theoretical currents of valves 1 and 3, respectively, with high-frequency
oscillations neglected.
voltage across that valve, combines two of the factors listed in Section 6-3 as
conducive to arcback.
The minimum current limit in the control system (Section 5-13) must be
set for an acceptably low incidence of arcback. Not only must the minimum
current setting exceed the initial amplitude of current oscillation in the out
going valve, but also the damping of that oscillation must be adequate to
prevent the current's reaching zero before the normal end of commutation. If
the high-frequency current oscillation is damped out before the end of com
mutation, both the high rate of decrease of current and the high rate of in
crease of inverse voltage are prevented.
The valves may operate momentarily with low current and short overlap
during start-up, fault clearing, etc., but they should not be permitted to work
thus for a long time.
Example 9
a =15
u= 1
Lc = 55.5 mH
La = 2.5 mH
C0 = 1.0-nF
Cm = 1.0 nF
i? = 7000Q
268
7-4
PROTECTION
Calculate and plot the currents in each of the twp valves during commuta
tion from valve 1 to valve 3. Also calculate the rate of decrease of current
in valve 3 as it becomes zero.
where
CO<X
CO,
SOLUTION
power-frequency and DC phenomena.
= Eab
co =
where
e =
2E
2nfx2(Lc + La)
154 1Q3 _
*
377 x 0.116
Ty
Id
x 3500 = 16.0 A
currents.
n/co2ny - o2y
(41)
(42)
(43)
LyCy
1
(44)
2RCX
(45)
2RC
(46)
L=L
Hence
= 0.0046
(40)
yjLxCx
r52 -
e\
at ignition is
Vj>
LL<
La+Lc
(47)
CX = C0
(48)
Cy=C0 + 2C
(49)
dhJdt = VjJL,, .
ht = ix + iy
(36)
ht = ix ~ iy
(37)
Ly =
where
ix
iy =
sin
exp(~o
Co
C0 + 2C
(3)
(x)
exp( o-ji)
(39)
269
(y)
LaLc
La+Lc
270
7-5
PROTECTION
I-
ji
r|
j
i'|
Causes of Oscillation
Oscillations may be caused by impressed alternating or step voltages or by
short circuits. Step voltages and short circuits cause oscillations at the natural
frequency or frequencies of the line. Alternating voltages cause oscillations
of both natural and impressed power frequency. If the natural frequency is
near the impressed frequency, oscillations tend to build up by resonance.
Alternating voltage may be impressed because of the following:
1. Blocking the main valves and misfire of the bypass valves. Then two
terminals of the ac source are connected continuously to the dc line through
the valves that were conducting when blocking took place. The rms ac
1.
2.
3.
4.
Starting up a rectifier
Bypassing or unbypassing a bridge
Recovery from reduced alternating voltage caused by an ac short circuit
Reenergization of the dc line after a short circuit
i]
271
*2
|
I
Damping of Oscillations
|
|
Control Circuitry, The current regulator of a converter tends to hold constant current and thus to restrain oscillations of current that accompany
Ld
Ri
'T~ C
I
f
|
i
i
(a)
(b)
(c)
272
PROTECTION
-V\A/-1(
7-6
273
the fault. Thus the voltage profile would shift from curve 1 to curve 2 in
Figure 20. Note that the inverter would have to reverse its voltage and
become a rectifier in order to maintain its current. With both converter
currents maintained, the final steady current in the fault would be the current
margin, which is typically 0.10 to 0.15 times rated current.
w
Ld
Fig. 20. Voltage profiles on a dc line: 1. Normal condition. 2. Short circuit at center,
currents maintained. 3. Fault clearing: a = 135 at rectifier; a= 100 at inverter.
274
7-6
PROTECTION
series inductance of the line and of the dc reactors, and deliver the energy
to the ac networks. The difference of terminal voltages required for both
converters to invert is opposite to that for maintaining the direct current.
They produce a gradient, somewhat like that in curve 3 in Figure 20, which
decreases the current and tends to reverse it. Valve action prevents reversal.
Therefore, the line current becomes and remains zero. The fault current does
likewise. Because the faulted conductor was grounded through the fault until
or nearly until the line current became zero, the conductor is left at or near
ground potential so that the recovery voltage across the arc path is negligible
and the arc does not restrike.
In order to establish terminal voltages of the correct polarity for fault
clearing, two steps are taken:
1. The inverter normally has the correct polarity, but in order to prevent
reversal of polarity, as described above on page 272, its ignition delay
angle is given a lower limit of about 100 (fi 80), which allows the inverter
voltage to run down to a low value but not to reverse.
2. The polarity of the rectifier voltage must be reversed so that it inverts.
As soon as the fault is detected, the ignition delay angle is shifted to some
value considerably beyond 90, say, 135.
Fault detectors are ordinarily used to initiate the increase of a, and these
are described shortly.
A method described by Uhlmann5 but not in general use obviates the
need for fault detectors.
Fault Detectors
The time delays and voltage settings are intended to provide selectivity
between dc line faults and ac faults or converter faults. Such selectivity is
desirable because, although a dead time is necessary for arc deionization of a
dc line fault, it serves no useful purpose in the event of an ac fault or a converter
'
275
fault but unnecessarily delays the recovery of the rectifier to normal operation.
Low alternating voltage at either converter station is a cause of propor
tionally low direct voltage. The rate of decrease of direct voltage on the line
from this cause, however, is slower than that caused by a dc line fault, because
the discharge of the dc line is delayed by the dc reactors. Of course, the rate
of decrease of dc line voltage is also slow in the event of a dc line fault
through a high resistance ; and this is the reason for providing the voltagelevel unit. The unit may be actuated by ac faults.
One-phase-to-ground faults are by far the most frequent ac faults, but they
cannot reduce the direct voltage by much more than 33%, which is not a
great enough reduction to actuate the voltage-level unit. Multiphase ac faults,
however, may actuate the voltage-level unit unless either the time delay set
ting of the unit is longer than the clearing time of such faults, which is
normally 67 to 100 ms, or the unit is blocked by low ac voltage. If such
blocking is provided, it may be set to last beyond the backup clearing time
of ac faults (200 to 400 ms). It can be actuated directly by low alternating
voltage at the rectifier station and, through a telecommunication channel, by
low alternating voltage at the inverter station.
Converter faults also cause low direct voltage, but, again, the dc reactors
limit the rate of fall of the voltage to a value that does not actuate the voltagederivative unit. Converter faults affecting only one bridge out of two or more
on the same pole do not actuate the voltage-level unit. Faults affecting
several bridges simultaneously may actuate the unit unless some blocking
scheme is provided. Blocking for inverter faults, if provided, would reach the
rectifier station by way of the telecommunication channel used in the event
of low alternating voltage at the inverter. Simultaneous faults on several
inverter bridges would most likely have low alternating voltage as a con
tributing factor, and these would block the voltage-level unit. The clearing
time of converter faults is another factor that determines whether the voltagelevel unit would be actuated.
Reenergization
276
7-6
PROTECTION
277
U
U
Recttfier
ed
Fig. 21. Circuit for calculation of current in a short circuit of a rectifier beyond the dc
reactor.
The circuit diagram is shown in Figure 21. The wave of direct voltage ed
under normal conditions is shown by the upper envelope in Figure 22a.
The fault is assumed to occur when cot = 60 + . Because of the increased
delay in igniting the next valve, ed follows the heavy curve, whose equation is
(50)
Fault Current. Initially the fault current is limited only by the surge imped
ance Zs of the line to 2VJZS> which is considerably greater than the current
margin. For example, on a dc overhead line rated at 375 kV and 1.8 kA and
having a surge impedance of about 300 Q, the initial fault current is
2 x 375/300 = 2.5 kA, but the current margin is only 0.15 x 1.8 = 0.27 kA.
This initial current results from the discharge of the shunt capacitance of the
line through the series impedance of the line, and the current regulators at
the terminals cannot possibly limit it. In some cases the crest current in the
fault exceeds this initial value.
Rectifier Current. Overshoot of the rectifier current above the set value
depends on (a) the inductance of the dc reactor, (6) the gain and time con
stant of the current regulator, (c) the distance of the fault from the rectifier,
(d) the wave form of the direct voltage, and (e) the instant on this wave at
which the fault occurs.
The general effect of some of these factors is well shown in a paper by
Peterson, Phadke, and Reitan.25'32 They assume conditions that tend to give
the greatest crest current : (a) a six-pulse converter with (b) a = 0, (c) negligible
commutation reactance and hence no overlap, (d) fault at the rectifier end of
the line just beyond the dc reactor, (e) occurring immediately after a commu
tation has started. The current control is idealized by assuming that the fault
is detected immediately and causes the ignition of the next valve to be
delayed 180, thereby reversing the direct voltage on the valve side of the
dc reactor.
n s= 1
120"
n= 2
240'
180
aif
360
300
.
420'
278
PROTECTION
7-6
The current is
h = ho + A/d = ho +
where
j*
ed dt
(51)
Ido is the prefault current and Aid is the increase due to the fault :
=
V3
r
a>Ld
(52)
60
where
h=
(53)
Aidm when ed = 0 at
(54)
The foregoing analysis shows that even the best imaginable control of the
direct current cannot prevent a considerable overshoot of that current if a
fault occurs soon after a commutation has occurred. If control action is
delayed longer, the overshoot of current is correspondingly greater.
Suppose now that the first valve to be ignited after the occurrence of the
fault is not delayed at all but that the second valve is delayed by 180. Then
the voltage, after the firing of the first valve, follows the sine curve (labelled
n 1), which lags 60 behind the curve followed in the previous case. The
area under the curve of voltage from the time of the occurrence of the fault
until the voltage becomes zero was Ax + A2'm the previous case, and in the
present case is augmented by A3. An additional delay of one-sixth cycle in
the control action adds area Aa. The relative areas are Ax = A3 = A4 and
A2 = 0.5Ax. The crest currents are 1.5Ib, 2.5Ib, 3.5Ib, etc. In general, for a
control delayed by a time between n and n + 1 sixths of a cycle and then
giving full shift of a, the crest current is
A?'dm = (+ 1.5)4
(55)
The cited paper shows that the effect of normal ignition delay a is to make
= 1 +(
(56)
which, for a = 15, gives 1.26. It also examines multibridge converters and
the effect of commutating reactance. A 12-pulse 2-bridge converter has the
possibility of faster control action, because the normal firing interval is one-
279
twelfth cycle, and the overshoot with a = 0 and n = 0 does not exceed 1.22Ib,
a value somewhat less than that for a six-pulse converter. The effect of com
mutating reactance on overshoot is shown to be small.
Discussion of the paper brought out that a design criterion is that, in
12-pulse operation, the control system should not allow more than one com
mutation after the occurrence of a short circuit on the dc line : that is, n = 0
or 1. Further, the inductance to be considered is not merely that of the dc
reactor but, in addition, the sum of the commutating reactances of all the
bridges between the faulted pole and ground and the reactance of the ac
system supplying the rectifier.
Field tests have shown crest short-circuit currents of about 1 pu, which,
added to initial load current of 1 pu, gives a total crest current of about 2 pu.
Although this value is well above rated current, it should be emphasized that
it is indeed small compared with the short-circuit currents obtainable in an
ac system.
A later paper38 refines the representation of the constant-current control
and considers short circuits in various locations, including short circuit of one
bridge.
Fault Clearing on a Multiterminal DC Transmission System
280
7-7
PROTECTION
CIRCUIT BREAKERS
281
(a)
(b)
D
(c)
The lack of a suitable circuit breaker for hv dc power systems has been
alleged to be a disadvantage of such systems -and, perhaps, really is. The
preceding section has shown that, on a system with two terminals, faults
can be cleared by grid control of the converter valves and that, on a system
of three or four terminals, transitory faults can be cleared by grid control
only and permanent faults by a combination of grid control, fault locators,
and isolating switches. The times required for fault clearing and reenergization of a dc line are not much different from those for an ac line.
Experience with ac systems has shown that they tend to grow into complex
networks. Under present conditions there is little expectation that dc systems
of comparable extent will develop. Dc lines are merely adjuncts to ac
systems. A good hv dc circuit breaker could change this picture.
What is needed is not so much a circuit breaker for interrupting shortcircuit currents, for such currents can be limited by grid control to the mag
nitude of rated currents, but rather a switch capable of interrupting load
currents in circuits at high potential with respect to ground. With such
00-r
282
7-8
PROTECTION
The rate of rise of recovery voltage across M is dVJdt = IJC. The non
linear resistor R absorbs energy without greatly adding to the voltage
across M.
Another kind, tested in a model for 20 kV, 10 A, uses the negative-resistance
characteristic of an arc shunted by a capacitor to increase the amplitude of
current oscillation until a zero occurs.14
Still another proposed circuit breaker uses an oil blast through the arc.13,31
hv dc breakers appear to be technically feasible, but their cost may be
excessive.
OVERVOLTAGE PROTECTION
283
284
PROTECTION
Plain Spark Gap. The rod gap is simple, cheap, and rugged. It is not
damaged by discharge of great currents. However, it has two disadvantages:
The sparkover voltage of a sphere gap is more nearly constant under good
atmospheric conditions but is influenced by dust and water drops.
Screened rod gaps have been developed that have consistent sparkover
voltages and are not sensitive to dust, etc. The screened rod gap and the
sphere gap, however, like the plain rod gap, short-circuit the power system.
Lightning Arresters have nonlinear resistors in series with precise spark
gaps. The resistors keep the voltage across the arrester within satisfactory
limits for a wide range of discharge currents: the voltage neither falls to zero
nor rises greatly above the sparkover value. The arrester " seals off," that is,
becomes nonconducting, when the voltage across it drops to a certain value
higher than the normal working voltage. Ac lightning arresters, like ac circuit
breakers, depend on the normal current zeros for arc extinction and cannot
extinguish a dc arc. Hence special lightning arresters are required for dc
circuits. Such arresters have been developed recently, and are described on
page 288. Before dc lightning arresters were available, ac arresters were
sometimes used on hv dc circuits to give precise sparkover, and the arc
was extinguished by grid control.
7-8
OVERVOLTAGE PROTECTION
285
286
7-8
PROTECTION
OVERVOLTAGE PROTECTION
287
The sparkover voltage of the gaps of the lightning arresters across the dc
terminals of each bridge should be set high enough to prevent sparkover at
zero overlap, although normally this condition is prevented by the minimum
current limit (Section 5-13).
The overshoot of the voltage across a valve at the end of its commutation
and the diminution of this overshoot by valve dampers were considered in
Section 7-3.
The voltage across the dc terminals of a bridge, being always equal to the
sum of the voltages across any pair of valves on the same phase, has voltage
jumps with overshoot. In each cycle of the power-system frequency there are
12 jumps in the direct voltage of the bridge, six associated with ignitions and
six with extinctions. The two kinds of jumps generally differ in magnitude,
but as the overlap decreases, the two magnitudes approach equality. At very
short overlaps, the extinction jump occurs before the voltage oscillation due
to the ignition jump has died out, and at exactly zero overlap the two jumps
merge into one jump twice as high. The corresponding overshoot is the same
percentage as that of a single jump but twice as high in absolute value.
288
7-8
PROTECTION
OVERVOLTAGE PROTECTION
289
Protection failure
DC Lightning Arresters23,27
Two special difficulties are encountered in applying lightning arresters to
the protection of the dc part of converter stations against overvoltages :
1. The current through the arrester does not have natural current zeros,
as alternating current does, to aid in resealing the arrester against sustained
direct voltage.
2. The greater part of the apparatus requiring protection and having
direct voltage is buffered by significant lumped inductances from the sources
of overvoltage surges. Overvoltages reaching the converter from the dc line
must pass through the dc reactor; those originating in the ac system, through
the transformers.
With respect to the first difficulty, it should be noted that one kind of modern
lightning arrester used for the protection of ac systems has, in series with
blocks of nonlinear resistance material, spark gaps so designed that they
develop a countervoltage, increasing with time and helping to bring the
arrester current to zero very soon after the subsidence of the applied overvoltage and even before the occurrence of a natural current zero. This same
kind of gap, called an active gap, is used in one kind of dc arrester for
forcing the arrester current to zero. If the overvoltage protection is not to be
jeopardized, however, the current must be brought to zero before the arrester
voltage becomes higher than the desired protective level.
Here the second difficulty appears. The arrester current is supplied through
a reactor (Figure 26), and the current through the reactor is proportional,
not to the voltage across the reactor, but to the time integral of this voltage.
The result is illustrated in Figure 27. The broken line vL there represents the
incoming voltage surge on the line side of the reactor. As long as there is
no changing current in the reactor, there is no voltage across it, and the same
voltage appears across the arrester. When this voltage at time attains a
sufficient value, however, the arrester's gap sparks over. Thereafter, the
voltage va across the arrester, shown by a solid curve, differs from the voltage
vL, and current ia flows through the arrester. The difference between the
L.A.
VL
Fig. 26. Overvoltage surge entering the converter station through an inductance.
Protection level
-<Resealing failure
Fig. 27. Single-stack lightning arrester failing to protect and reseal for a switching surge
entering the protected zone through a reactor. Area I> area II. (From Ref. 27, by per
mission.)
ordinates of the two voltage curves represents the voltage across the reactor,
and the area between the curves is the time integral of this voltage to which
the arrester current is proportional.
The curves in Figure 27 show that the arrester voltage can exceed the
desired protective level before the arrester current is suppressed. Moreover,
they show that, because the arrester voltage reaches a peak and then declines,
it may drop below normal operating voltage before the arrester current
becomes zero, with the result that the arrester fails to reseal and is destroyed
by overheating.
A special dc lightning arrester has been developed,27 having two or more
stacks connected in parallel, each stack containing nonlinear resistors and
active spark gaps. The several stacks carry current in sequence, with results
as illustrated in Figure 28. When the voltage across the first stack to conduct
reaches a certain value a little below the protective level, the second stack is
triggered and begins to conduct. Because this stack has a countervoltage
lower than that of the first stack, all the current is immediately transferred
from the first stack to the second. When the voltage across the second stack
reaches a value approaching the protective level, the current is transferred
either to a third stack or back to the first stack, depending on the number of
stacks provided. (This number is chosen so as to provide the energy-absorbing
capacity required by the particular application.) While one stack is conduct
ing and being heated, the others are being cooled. Thus it is possible to
extinguish the arrester current without the arrester voltage's ever exceeding
the protective level and without overheating, which would otherwise cause
erosion of the gap electrodes and consequent change of their sparkover
voltage.
PROBLEMS
291
Protection levei
Normal
operating
voltage
-t
The usual locations of lightning arresters and gaps are shown in Figure 29.
t
Fig. 28. Multiple-stack lightning arrester successfully protecting and resealing against
a switching surge entering the protected zone through a reactor. (From Ref. 27, by
permission.)
(va vL) dt = 0
Dc reactor
/TOTV
Oc line
Converter
transformers
Ac bus
rr
Electrode line
Fig. 29. Locations of lightning arresters and gaps on one pole of a converter station
having two bridges per pole (single-line diagram) : 1. Ac arresters on ac side of transformers.
2. Dc arresters, phase to ground on valve side of transformers. 3. Ac arresters, phase to
phase on valve side of transformers. 4. Dc arrester, neutral to ground on valve side of
transformer. 5. Gaps across valves. 6. Dc arrester, pole to ground. 7. Ac arrester across
dc reactor. 8. Arrester from electrode line to ground. 9. Small inductor. 10. Surge capacitors.
PROBLEMS
1. Analyze the ripple in the direct current during 12-pulse operation, and
verify the numerical coefficient given in the text.
2. Analyze the ripple in the direct current during mixed 6- and 12-pulse
operation (two bridges with YY-connected transformers and one bridge
with YA connection).
3. Analyze the ripple in the direct current during six-pulse operation with
both ignition delay and overlap taken into account.
4. Show that when a capacitance C, initially charged to voltage V0, has its
charge and stored energy modified by a change of voltage + VJ} the
energy dissipated in any series resistance is CVj/2.
5. Show that the wave form of the high-frequency oscillation of voltage
across the bypass valve immediately after a voltage jump across it is the
same as that across a working valve.
6. Investigate the natural frequencies in the wave of voltage across a valve
immediately after ignition of another valve. Does it matter which other
valve ?
7. Assuming that a phase-to-ground short circuit does not change the other
two phase-to-ground voltages, find the reduction in direct voltage of a
rectifier at light load.
8. Find the power loss per valve damper of the inverter in the dc link in
Example 7 on a pole operating with two bridges instead of three at each
end of the line. Assume rated current and, at the rectifier, rated voltage
per bridge. Assume also that the alternating voltage on the valve side of
the inverter transformers is the same as in the example. Does this loss
exceed the damper loss during rectifier operation at rated current and
voltage ?
292
PROTECTION
BIBLIOGRAPHY
BIBLIOGRAPHY
1. "Arc Quenching in Switchgear for Disconnecting High-Power EHV DC Circuits,"
by G. A. Kukekov and P. G. Sorokin, Direct Current, Vol. 1, pp. 213-216, March
1954. Translated from Elektrichestvo, No. 10, pp. 20-24, October 1953.
2. "Protective Means for H.V.D.C. Transmission," by E. Uhlmann, U.S. Patent No.
2,730,667, Jan. 10, 1956.
"Switchgear
for H.V.D.C. Lines," by G. A. Kukekov, P. G. Sorokin, and N. A.
3.
Shipulina, Direct Current, Vol. 4, pp. 123-126, June 1959. From Elektrichestvo,
Vol. 79, pp. 24-27, March 1959.
4. High Voltage Direct Current Power Transmission, by Colin Adamson and N. G.
Hingorani, Garraway, London, 1960, Chapter 6, "Protection of H.V.D.C. Systems,"
pp. 85-114, 184-186.
5. "Clearing of Earth Faults on H.V.D.C. Overhead Lines," by Erich Uhlmann,
Direct Current, Vol. 5, pp. 45-47, 65-66, September 1960. Correction on p. 103,
December 1960.
6. " Lightning Overvoltages on Long D.C. Cables Connected to Overhead Transmission
Lines," by S. Rusck, Elteknik (Stockholm), Vol. 3, No. 7, pp. 107-109, September
1960.
7. "Disconnection of Branch Lines from D.C. Transmission Lines," by N. D. Leshukov
and N. A. Shipulina, Direct Current, Vol. 7, pp. 39-46, 58, February 1962. Reviews
previous work on dc circuit breakers; recommends a certain scheme, and gives cost
estimates.
8. "A New Approach to H.V.D.C. Switchgear: Part I, Theoretical Appraisal of a
Rotating Plasma Machine as a Switch," by J. J. Matthews, Direct Current, Vol. 8,
pp. 184-189, July 1963; "Part II, Experimental Investigations on a Rotating Plasma
Machine," by J. J. Matthews, R. L. May, and H. Taylor, pp. 216-220, August 1963.
(The names of the last two authors were omitted, and a correction was printed in
Vol. 9, p. 25, February 1964.)
9. " Commutation Phenomena in a Static Power Convertor," by R. Feinberg and W. Y.
Chen, l.E.E. Proc., Vol. Ill,pp. 125-134, January 1964. Experimental and theoretical
study of current and voltage oscillations in a three-phase one-way rectifier.
10. "Mercury-Arc Valves for High-Voltage D.C. Transmission," by Uno Lamm, l.E.E.
Proc., Vol. Ill, No. 10, pp. 1747-1753, October 1964. Includes some information on
valve dampers and anode reactors on p. 1752.
11. High Voltage Direct Current Convenors and Systems, edited by B. J. Cory, MacDonald, London, 1965. Chapter 7, "Filters, Damping Circuits and Reactive VoltAmps in H.V. D.C. Convertors," by J. D. Ainsworth, pp. 137-174.
12. "Some Design Aspects of the Cross-Channel Power Link," by L.A.Harris, ibid.,
Chapter 8, esp. pp. 181-195, 205-206, 211-215.
13. "Testing a New Type of Circuit-breaker for H.V.D.C.," by K. Mollenhoff, D.
Morghen, J. Salge, and W. Pucher, Direct Current, Vol. Ill,pp. 3-6, February 1966.
Oil blast, experimental, 645 A, 85 kV.
14. "H.V.D.C. Circuit Breakers Using Oscillating Current Techniques," by Naohei
Yamada and Tung Hai Chin, Direct Current, Vol. 11, pp. 87-89, 92-98, and 107,
August 1966.
|
i'
'1
i
:
|
!
|
f
!
>
i
293
294
PROTECTION
36.
37.
38.
39.
|
).
8
Harmonics and Filters
!
\
8-1 SUMMARY
h=pq 1
(2)
on the ac side, q being any integer. Most hv dc converters have pulse number
6 or 12 and thus produce harmonics of the orders given in Table 1. The
Table 1. Orders of Characteristic Harmonics
Pulse No.
p
6
12
j.
DC Side
AC Side
pq
pq\
296
-8-2
CHARACTERISTIC HARMONICS
297
1. The alternating voltage has no harmonics except the first (the funda
mental).
2. The direct current has no harmonics.
There can be higher harmonic currents on the ac side and harmonic
voltages on the dc side, however, and deductions are made concerning these.
Because of assumptions 1, 3, and 4:
3. The overlap angle is the same for every commutation.
4. The ripple of the direct voltage has a period of one-sixth that of the
alternating voltage.
5. Hence the harmonics of the direct voltage are of order 6 and its multiples
12, 18, 24, etc.
6. The alternating currents of the three phases have the same wave shape
but are displaced by one-third cycle in time (120 of the fundamental).
7. The alternating currents have positive and negative parts of the same
shape except that one is inverted; that is, F(9 + 180) = F(9).
8. As a result of deduction 7, there are no even harmonics in the alternating
current.
9. As a result of deduction 6 and the fact that the phase difference for the
Ath harmonic is h times that for the fundamental, the ac harmonics have the
following phase sequences :
Sequence
Orders (A)
Zero (0)
Positive (1)
Negative (2)
pulse converter :
Harmonic analysis of the wave shape of the alternating current shows that
AC Harmonics at No Overlap
298
8-2
CHARACTERISTIC HARMONICS
299
F(d) =
L
00
(3)
where
A0=~rF(e)d(9)
(4)
71 0
-7c
fin
71
Bh = -71
7A
Fig. 1. Wave forms in a six-pulse bridge: line-to-neutral voltages ea, eb, ec and line
currents ia, ib, ie with YY-connected transformer; also line current IA with AY-connected
transformer.
lines show qualitatively how overlap would modify the fronts and tails of the
current pulses.
Valve Currents and Line Currents on Valve Side. The line-current wave
forms at no overlap are a series of equally spaced rectangular pulses, alternately
positive and negative. Fourier analysis of such a wave shape, for finding the
characteristic alternating-current harmonics in this limiting case, is very
simple; it also serves to illustrate several features of these harmonics. How
ever, let us take an even simpler starting point: the analysis of a train of
positive rectangular pulses of unit height and arbitrary width w radians, that
is, of duration w/co sec (see Figure 2). These pulses might represent the
current through one valve.
F(d) cos hd d9
(5)
F(d) sin hd dd
(6)
Jo
The limits of integration in Eqs. (4), (5), and (6) can be taken more generally
as o and <r + 2n, where o is any angle. A0j2 is the average value of the function
F;Ah and Bh are rectangular components of the hthharmonic.The correspond
ing phasor is
(7)
Ah-jBh=C[&,
where
and
Ah
If, in the analysis of the wave shown in Figure 2, the origin of d is taken at
the center of a pulse, F(d) is an " even " function, and Bh = 0 for all h; that is,
the series has only cosine terms. Their amplitudes are found by Eq. (5) thus:
I r
Ah = 71-
J
F(d) cos hddd = -
J -n
1 T . hw
71
.(
hw
T2
C+H-/2cos hd dO
J -wjl
=
2 . hw
(8)
300
8-2
Also
T-i/*_>-
3w
4w
----
(10)
In general, this series has a constant term and cosine terms of every harmonic
frequency. For certain pulse widths, however, certain cosine terms vanish.
This occurs if
hw
2qn
w=-
or
=qn
n\
--4
. w
h sin cos 9
2
.
-2 sin
3w
2
2w
H sin cos
5w
cos 50 H
} (13)
P '
--
2 v3
=-
7t
cos 130
(14)
ho =
= ~h= ojmd
Ij
V2 n
(15)
(12)
(16)
ho =
This result can be obtained in at least two ways: (a) by putting the new
function into Eqs. (4), (5), (6) and performing the indicated operations or
(b) by appropriate changes in series 10. These changes are the following:
(1) Shift the pulse n radians; this shifts the fundamental component n rad and
shifts the higher harmonics by hn rad. If h is even, cos (0 hn) = cos 0; but
if h is odd, cos (0 hn) = cos 0. Hence the signs of all odd harmonics are
changed. (2) Invert the pulse. This changes the sign of every term. The net
result is to change the signs of all even-order terms, including the constant
term.
3w
cos 29
\
1
4w
30 - sin cos 40 +
4
2
/
4/
= =1-103/,
271
so that if h = 3, 6, 9,
2/
F2(9) = - 1
--
Next, let us analyze the train of alternately positive and negative rectangular
pulses. Its Fourier series is
(11)
For example, the pulses of valve current in the three-phase bridge current
have width
W =
301
CHARACTERISTIC HARMONICS
I-f
(IV)
Series 14 represents the ac line current of phase a on the valve side of the
transformer (Figure 1) if the origin of 0 is taken at the center of the positive
pulse (axis la). The currents ib and ic in the other two phases have the same
wave shape as ia but are displaced 120 (= 2nj3 rad) behind and before ia,
respectively. Their Fourier series, if written for 0 = 0 at axes Ib and Ic, re
spectively, are the same as that for ia written with respect to axis Ia. Likewise,
these series are independent of the ignition delay angle a. If any wave is
shifted forward by an angle (f> , measured for the fundamental period, the hih
harmonic is shifted by h<}> measured for the shorter harmonic period, being
shifted forward if of positive sequence and backward if of negative sequence.
Line Currents on Network Side of Six-pulse Group. If the transformers
are connected YY or AA and have ratios 1: 1, the line currents on the network
302
side have the same wave shape, hence the same harmonics, as those on the
valve side. If, however, the transformers are connected YA or AY, the wave
shape on the network side is different from that on the valve side.
Let the transformers be connected in Y on the valve side and in A on the
network side, and let the ratio of each individual transformer be 1:1. Then
the currents in the delta-connected windings are the same as those in the
corresponding Y-connected windings. Each line current on the delta side is
the difference of two delta currents ; for instance,
ico
P lio
iro
00
If)
len Ico"
T-J"
w
> >
> >
a
Ss
<
T-
*
w
O
co
/s
(18)
*'a = h - ic
im
.O
00
co
IT)
en
(N
+ + +
en
<N
On
+ + +
m
p.
r-H
en
1i
i
.
en
F5
11
/"S
r-
On
N
<B
>fi
(19)
&
O
%
It differs from Eq. (14) only in the signs of the terms representing the fifth, seventh, seventeenth, nineteenth, etc., harmonics. Although all the harmonics
in Eqs. (14) and (19) are equal in amplitude, the two series represent different
wave shapes because of the difference in signs (or in phase) or certain orders
of harmonics.
o
o
ON
CI
CO
CO
O
ID
r-
i/1
<*3
CO
~"
co
>-< c
83
I-.
a*
-H
CO
ON
(S
-i
-<
OO
<-
co
co
1
j-V1
</->
rv-i
OO
00
4)
*o
p-
304
--
) (20)
305
AC Harmonics at Overlap
In Figure 1 the wave shapes for positive overlap appear as better approxi
mations to sine waves than do the wave shapes for no overlap. Hence we make
the qualitative deduction that the effect of overlap is to decrease the ampli
tudes of the harmonics.
Quantitative results are computed from the following formulas. They are
valid only for characteristic orders h. For overlap not exceeding 60, the
complex rms value, with phase referred to the respective commutation
voltage E is
Ih KtFx(a, 5, h)
(21)
amperes
where
=
(E\ = V64
<22>
and
j (h l)a / (A 1)5
/-(h + l)a-l-(h + l)5
- --Fx = h 1
h 1
(23)
-f=K2Ft
ls2
(24)
(25)
where
base
= K3Fx
per unit
(26)
306
8-2
CHARACTERISTIC HARMONICS
307
where
where
**=k
(27)
-f =
(28)
/fsin[(A - l)w/2])2
Mr
MO
where
-2/hr
<29>
i
= K5F
(30)
ln
OD
ho
where
K> =
i
t = K6Fi
(32)
*'=23
(33)
h~
[sin[(A + l)uj2]}2
h+ 1
) +(
h- 1
-2(sinp-l)M/2]}{sin[(lM2]|cos(2g + K)y/2
(41)
The last equation has the same form as the law of cosines for the length of
one side of a triangle in terms of the lengths of the other two sides and the
included angle.
Computed results for 4/4 0 versus u are plotted in Figures 4 to 11. Results
for l'h = 4/4ase versus l'd at the usual value of a or y, 15, are plotted in
Figure 12.
Overlap Greater than 60. In the region bounded by 60 < u < 120,
a > 30, and 8 < 150, Eqs. (21) to (41) apply if a is replaced by a' and d by 8',
where, as before (Eq. (20) of Chapter 4),
a' = a 30
8' =8 + 30
m' = m + 60
(42)
where
Direct-voltage Harmonics
where
a
+ <5 sin -u = ,
A formula for complex values of the harmonics of the direct voltage is the
following:
= iF3(,M)
"
(34)
(43)
dO
4 = 2K1F2(u, u, ft)
V6is
2K2F2=
ir2KA=-h
4
A. = 2K,FZ=
2KS2=%"
-'base
2K=
i=2
amperes
F-
\-1
Vdh = FJa, u, h)
(35)
(36)
(37)
(40)
(44)
V.do
where
l(h+l)a+l(h+l)8 /(h-l)a+l(k-l)8
TM
h+ 1
/(ft+i)(i+/(ft + i)) /(ft -i)g(i+ /(ft -i))
ft - i
ft + l
_ r(cos [(ft - 1)/2]Y (cos [(ft + 1)m/2]Y
Fs =
(38)
(39)
4 = 2K F =
-?
2K6F2
LI
cos
A-l
[fl -
+l
l)/2]yos P + l>/2]j
cos(2g +
K)j1/2
(46)
16
h= 5
12
\\\
\v
\\
\ \
\\\
V\15\
\\\
30/\\
h=7
10
wv\\\1
\\
-
\\
W
Vv\
w\ \
24
309
\5#
\
60 to
.a = 0
V
90'\\
W
K
\
V
\\ \\
\\\\
\\
'
Fig. 4. Fifth harmonic current o-f a six-pulse converter as a function of converter angles.
\\
\\
Vv \
\\
\.
90/"
60
10
20
30
40
50
60
10
8-2
h= 11
7
6
k\ \
w\
\\\\ \\
w
\\\ \
v\
\
v\ \N
V\
\\\
32
V\\\
V\
60 to
\a = 0
0"
15
60
30
20
10
60>/
90
A = 17
=-> 5 -
311
CHARACTERISTIC HARMONICS
40
60
50
II
\n
XO
V\l5
-0
\60
20
XN
-
II
cn
'
V\5
156-v_
a=0
Nsj_6Q>/
30
40
50
60
, 60*
\\ NV5L
10
8-2
CHARACTERISTIC HARMONICS
313
h= 1
1
h= 5
a = 15' or 7 = 15"
h = 23
6R
\a = 0
1
Q
6C)Ki5
60
50
40
30
20
10
60
40
20
60
70
Fig. 12. Characteristic ac harmonics as functions of the direct current fora = 15: I't is
in percent of Is2 ; IL is in percent of /base-
Tile equation for F4 is like that for F2 if the sines in the latter are replaced
by cosines. Again, only the characteristic values of h are applicable.
Graphs of VdhjVd0 for h = 6, 12, 18, and 24 are given in Figures 13 to 16,
v
respectively.
It is notable that, unlike the ac harmonics, the direct-voltage harmonics
depend on a even if u = 0.
h- 25
. \a = 0
i
Zero sequence:
va+vb
v0 = -
J*
Positive sequence:
vx
va
vb
= -
314
8-2
25
CHARACTERISTIC HARMONICS
315
12
70
-6020
_
k=6
10
50-N
-/90
40
30
15
90
50/
30
10
8 40
jC// /
20'
A
/// /s'/
151.
(AO
20
1/130
/
//
[////
20
vf
1
40
10
50
60, 9C
1
20
30
40
50
60
Fig. 13. Sixth harmonic of direct voltage of a six-pulse converter as a function of con
verter angles.
Sequence
Zero
Positive
Harmonic Orders
Effective Number of
Bridges per Pole
, 12# + 6
6, 18, 30,
0, 12, 24, ..., 12q
1
b
The zero-sequence voltages are caused by only one bridge per pole, because
the remaining bridges per pole (an even number) comprise pairs in which the
zero-sequence voltages cancel. If the number of bridges per pole is even, there
are no zero-sequence characteristic harmonic voltages. This is the preferred
10
20
30
Overlap angle u, deg
40
50
60
316
8-2
CHARACTERISTIC HARMONICS
317
80
60
60
60'
h- 18
50
60
40
50
40
50.
'30
'20
30
20
10
30
10
-V
\
(b)
0
Overlap angle u, deg
\\
Fig. 17. Alternating-current wave shapes, six-pulse rectifier with negligible commutating
and dc reactances and with resistive load: (a) a = 0, (6) a = 30.
6
60
60
30
60
40
40
40
60
30
20
30
20
30
10
10
20
10
20
10
10
50
0
Overlap angle u, deg
7 ,
11
13
17
19
23
25
+1.000
-0.226
+1.000
-0.200
-0.113
-0.143
+0.091
+0.065
+0.091
+0.077
-0.0567
-0.0588
-0.0454
-0.0526
+0.0412
+0.0349
+0.0435
+0.0400
+13
-21
0
-16
-4
-14
-5
-i3
318
The ignition delay angle of a rectifier is usually measured from a zero of the
commutating voltage. If the three-phase alternating voltages are unbalanced,
their zeros are not equally spaced, and, consequently, the valves are not fired
at equal time intervals. Probably, even with balanced voltages, there is some
"jitter" in the electronic circuitry of the current regulator that produces un
characteristic harmonics. The variation of firing angles from their normal
values is usually cited as 1 or 2. Reeve and Krishnayya58 state, however, that
on the Cross Channel link the variation was +3 for rectifier operation and
1.5 for inversion.
It was shown in Section 5-11 that the combination of high gain and short
time constant in the current regulator would cause alternate early and late
8-3
UNCHARACTERISTIC HARMONICS
319
Consequences
320
8-3
Suppression or Diminution
UNCHARACTERISTIC HARMONICS
321
sums are respectively 2 cos he and 2 sin he times the respective harmonics of
one train. For small he the decrease of magnitude of the odd harmonics is
negligible; the even harmonics except those of order 6q, which do not appear
in the individual trains, increase from zero to a nonzero value, which we
proceed to estimate.
The ratio of an even harmonic of order h to the fundamental wave at small
overlap is
--
+ TTo(e)5
h{\
%{he)2
T1~2hcose~
+ {hef
If, 2 sin he
he %(hs)3
= [1 "+ i(Ae)2
+ -(hs)4 -\
]=e
* *
radians
(47)
Triple AC Harmonics. It was shown in Section 8-2 by Eq. (11) that a train
of rectangular pulses of normal width (120) has no triple harmonics. If,
however, the pulse width is longer or shorter than normal, triple harmonics
are generated. Again let us estimate the magnitude of such harmonics as a
function of the angular ignition error. Suppose that the ignitions of two valves
connected to the same phase are late by e and that the other four valves of
the bridge ignite on time. Then the alternating current of that phase consists
of positive and negative pulses both of which are shorter than normal by e.
The current of the phase leading that one has pulses longer than normal by e,
and the current of the remaining phase has pulses of normal length. Assume
zero overlap, so that the series of Eqs. (10) and (12) are applicable. In each
of these series and in their sum, the ratio of odd harmonic to fundamental is
h _ sin {hw}2)
h h sin (w/2)
SinL5ge
3tf[( V3/2)cos(e/2)
sin (e/2)]
(49)
322
8-4
For small e, cos (e/2) 1, sin (e/2) = 0, sin 1.5gs 1.5<?e, and
323
= *0.577*
h 3q3}2
i
(50)
V3
List of Troubles
oscillograms.
Table 4. Harmonic Currents Measured on AC Side
of Converter at Lydd39,49
To an arbitrary scale
Order of
Converter
12-pulse
400-A DC
6-pulse
Harmonic
Blocked
Operation
Operation
29.7
9.3
25.9
10.2
10.9
21.6
92.5*
6.2
66.5*
44.3
I00-A DC
2
3
4
5
6
7
8
9
10
11
12
13
2.0
2.1
0.3
6.0
1.3
4.0
0.8
0.1
0.5
2.0
0.6
1.0
26.4
9.2
16.2
31.7
57.8
22.3
119.6*
67.9
3.8
19.2*
0.3
15
0.2
16
0.04
18.4
17
18
0.2
13.4
4.4
11.1
7.5*
0.3
19
20
0.08
10.4
8.6
11.7
3.4
5.4*
4.9
* Characteristic harmonics.
43.6
75.3*
14
0.1
23.8
21.5*
28.4
17.9
15.0
Noise on telephone lines is the most difficult trouble to eliminate and forms
the subject of most of the rest of this chapter. There are reasons for concern,
however, over the effects of harmonics on the power system itself. Item 4 was
discussed in Section 8-3, page 318. Items 1, 2, and 3 are briefly discussed
below.
j
i
I
:
324
8-5
the harmonics would divide among all these plants in any definite ratio with
out putting undue restrictions on the design of the future generators, ac
lines, etc.
Although filters are provided at the rectifier station, the frequency generated
by the plants connected to this station is subject to large variations during
system disturbances. The filters are then out of tune, and a greater part of the
harmonic current output of the rectifiers passes through the generators,, which,
however, were designed, at small additional cost, to withstand the temporary
additional heating caused by these increased harmonic currents.
Measurements of losses caused by harmonics in a cylindrical-rotor syn
chronous generator of normal design, loaded by nonlinear reactors were
reported by Easton.50 Measurements of harmonic impedances of a 27.5-MVA
generator were reported by Gardiner.51"
Extra Losses and Heating in Capacitors
fc(tanKFf
(51)
h=2
where
325
C = capacitance
tan 5 = loss factor
8-5 DEFINITIONS OF WAVE DISTORTION OR RIPPLE
The dielectric stress is proportional to the crest voltage, which may be either
raised or lowered by the harmonic voltages.
The total reactive power, including fundamental and harmonics,
CO
Q=1Q
A=1
(52)
constant nor
their magnitudes.
For some purposes, such as filter design, it is convenient to have a more
concise expression a single number serving as an index of the degree of
departure of the wave from its ideal shape. Several such indices have been
defined.
Total RMS Harmonics
/&Y
VSa2
Hi = -z-= -
Ii
per unit
(54)
per unit
(55)
-JYlYJi Yi2~i}
H2 = -- = -.-
326
8-5
327
values of all the harmonics divided by the crest value of the fundamental
wave. Of course, rms values can be used instead of crest values in both
numerator and denominator, giving
ff6 = Lh=2
h
j
h
(59)
This index weighs all harmonics equally. It is not used for dc quantities. In
practice, we can neglect harmonic orders above the twenty-fifth with very
little error.
jl
Zj | max
riz = --hm
._
per unit
(56)
(57)
dl max
Ha = --where
per unit
i =instantaneous current
z'j = instantaneous value of fundamental current wave
Ilm = crest value of fundamental wave
Id average direct current
Hs =
'min
-'max
Id
per unit
(58)
328
Reference frequencies
C-MSG
CCITT
800 Hz
1000 Hz
/
n/ / /
'V
20
30
//
'/c- Vl< G
g?
"n"
//
c6
t-4
no
no
in
Ti in
$ <5
NO N CO
NO CS oo It NO cs OO
it f in NO NO l> r- oo On ON
ro CO CO CO co CO CO if
if
on on
NO CS
.1
Coupling Factors
(ZJ
NO NO f- C-
r-
CS in n
in in in in in
1-4 44
rH
1-4 44 44
44
44
OO
44
'
""
1-4
n in n NO
r-;
44
C9
1
.?
i
g
NO t 00 On
1-4 1-4
1-4
If
cs cs cs'
Nro'tvir-coONO-iNf23
u-l
SwoNOinvNNOiNONONomiflOiogwNOMh/ico
iTfOOONMNOlD
N Tf NOON COf-NhCOCOWOir-rtNNcOM't'flfNlfll'llfllON01OlONO
m' oi i no
i
If)
NNmcO(OMCNNOONlqfs, t":
oooiO'tmNrtddddddoddbo-H
.a
44
ON cS n OO
r-,'Qin"n0
ao
t/i
rtNoPt8njv>o\t-->nr-rj(Nr-r-eor--3;cnci-
Snoo'nN-:<ONOOi?NOONO\o-HiogoooNNNON
000-,fiClfl|nrCOO'ONONOOONONONONONONOO
dddoddddeddddo'dd-HodcJoooo
NO N CO
s:
(60)
in in
NO N h
oo in
If SO NO if CO cs
ON ON ON
IONONONMCOONONO--<NNfOtlOlONOOOOOOlO
Eo(/Ft/2
co ON it On if
co <n oo co
On on On
NO (S OO it O
'ONoo'fONONoo'tONON'tgNgNco.;;.-;.;;;,_
1"N
t/N CS <S
Tir
44
1
!
l
I
1
o1
is iq iq N
defined by
>n
72
'* "
io
rtOQrt-HHrt-HrtHHrtHrtrt-MN'tNOO'fNCOI'N
MOOOOOOCOOOCOOOCOOOOOOOMCOCOOOCOCOOOIhrNqNONq
cs r~
m oo o
t- 00 OO
><
*o
NO N CO if O O
N ci to in
N
'
in
t Nf "t 4if
62 63 3 65 99 67 89 69 70
in>ninininninnnNONONONO
S.
On i'
On
NN N N
N N
m on N O On h m
r- On in cS no co in
ON t*
cs oo no m if cs -"-i co m
ii on m
OO in NO
CS CS
m if if it if on fS
' ' ' ' *' ' '
ON
HNClTfl/lNONCOONO-vn'l
10
23
1
1
Frequency, kHz
40
50
0.1
CC1
10
on oo t- OO NO <s oo
oo
NO
N h- Tf O NO if rNO n <n
f- NO
ON CO 00 OO 00 r-
OOOOOO
fOm'fNflnNONOf-'hCOQNONOO
N N co m
rsif,nNotccONcsifinNor--ooON44
cs
M N M N
u- m
(S
oo
330
8-5
/(Hz)
1
2
3
50
100
150
200
4
6
8
10
12
16
dB
fpf
0.71
63.0
0.000044
20
8.91
35.5
-41.0
0.00111
24
-29.0
-21.0
0.00665
30
40
50
1000/)/
300
400
295
-10.6
0.0223
0.111
484
-6.3
0.242
500
600
800
661
794
1000
-3.6
-2.0
0.0
0.413
0.595
1.000
89.1
60
70
80
100
/(Hz)
1000/7/
1000
1200
1500
1122
1000
2500
861
708
617
3000
525
3500
4000
5000
2000
dB
fpf
+ 1.0
1.40
1.50
1.62
0.0
-1.3
-3.0
1.77
-4.2
1.93
-5.6
1.97
376
-8.5
178 -15.0
15.9 -36.0
1.65
0.89
0.10
kf = 5000(//1000) = 5/
(61)
pf = C-message weighting
(62)
(63)
In the C.C.I.T.T. system the result is called telephone harmonic form factor
(T.H.F.F.) and is defined by a similar expression with
kf
800
Pf
psophometric weighting
1000
(64)
331
psophometric voltage:
y
where
V=
-JY]Lp,v,?
(66)
I-T=
VE(r? = /-(TIF)
(67)
(65)
Generally the sums are for a finite number of discrete frequencies, which
include the power frequency (50 or 60 Hz) and its multiples.
Transverse
I*T product
kV T product
C.C.I.T.T.
Psophometric weighting
Telephone harmonic form factor
(THFF)
Psophometrically weighted voltage
Psophometric voltage
Equivalent disturbing current
Equivalent disturbing voltage
332
Fig. 19. Voltage vector diagrams of autotransformers for shifting three-phase voltages by
angle (drawn for = 15): (a) ring; (b) zigzag.
<f>
8-7
TELEPHONE INTERFERENCE
333
harmonic output with three groups in service would consist of both the
12-pulse and six-pulse harmonics of three groups. Obviously, the 12-pulse
converter has some advantage over the six-pulse converter even when one
bridge is out of service though less than when all are in service.
Filters
load.
In addition to the ac harmonic filter at the converter stations, such filters
could also be placed in any sections of transmission line giving rise to especi
ally bad telephone noise. This is seldom, if ever, done, because it is usually
cheaper to modify or relocate the telephone line.
<f>
General
334
8-7
5
TELEPHONE INTERFERENCE
335
If the ground has finite and uniform conductivity, the foregoing statements
are still substantially true with respect to the electric field. The magnetic
field, however, can now penetrate the earth, and its effect on self- and mutual
inductances is as if the image conductors were lowered to a greater depth
History
below the surface of the ground. The equivalent depth of ground return is
proportional to the skin depth-, both of these depths vary inversely as the
square root of the frequency and of the conductivity.
If the ground is nonuniform, the foregoing is still true qualitatively.
Because the distance between the overhead conductors of a ground-return
circuit and their image conductors are much greater than the distances
between conductors of a metallic circuit and because the two conductors of
a ground-return circuit are in a vertical plane, the coupling between two
ground-return circuits is very much greater than the coupling between two
metallic circuits separated by the same distance as the ground-return circuits.
Although ground-return circuits were used for dc telegraphy and voice
telephony when these arts were new, they are seldom if every used now
because of the severe problems of noise and cross talk. Power circuits are
also all metallic except for some hv dc lines, because the power loss and
telephone interference from ground-return ac circuits are both high. It would,
therefore, appear that the coupling to be calculated is that between a metallic
power circuit and a metallic telephone circuit. But, on the contrary, the
practice is to calculate the coupling between a ground-return power circuit
and an open-ended ground-return telephone circuit. In other words, one
calculates the longitudinal voltage induced in the telephone circuit by residual
current in the power circuit. The reasons for this practice are now given.
336
8-7
TELEPHONE INTERFERENCE
337
approximation.
The mutual impedance between two groundreturn circuits is usually computed
by Carson's formula,1 which assumes a homogeneous earth and parallel
conductors. Krakowski54 has extended Carson's work to crossing conductors.
Riordan and Sunde4,11 have extended this work to a two-layer earth, that is,
to surface layer of one resistivity separated by a horizontal plane from an
infinite volume having a different resistivity.
If the separation between power and telephone circuits is much greater
than that between conductors of the same circuit, the conductors of each
circuit may be replaced by one equivalent conductor at the center of gravity
of the several conductors. The coupling between widely separated groundreturn circuits increases with the resistivity of the ground.
balanced.
338
8-7
TELEPHONE INTERFERENCE
339
The C.C.I.T.T. Directives35 state that the psophometric emf should not
exceed 1 mV. The psophometric emf is twice the psophometric voltage
(psophometrically weighted noise voltage) measured across a noninductive
resistance of 600 Q, terminating the line at the place of measurement, when
the other end of the line is terminated in its characteristic impedance. The
corresponding noise power in a 600-Q resistive termination is
V2/R = (0.5 x
1. Calculate the harmonics of the direct voltage on the valve side of the dc
assumed operating conditions. Among these conditions
are both rectification and inversion with all bridges in service and with one,
or more bridges out of service. In a bipolar line, monopolar operation with
either metallic or ground return should be studied if such operation is con
templated. The control angles a of the rectifier and y of the inverter should be
assumed to have their normal values ; for example, a = 15 and y = 16, and
values of overlap u corresponding to several values of direct current between
minimum and maximum values may be tried to see which produces the
greatest TIF. In the usual monopolar scheme, all harmonics have earth
return. In a bipolar scheme, the harmonics should be divided into zeroand positive-sequence components. Generally, only the zero-sequence
components are important. Both characteristic and uncharacteristic har
monics should be included. Uncharacteristic zero-sequence harmonics
corresponding to characteristic positive-sequence harmonics of the same
order are not so much the result of conversion from positive to zero sequence
by unbalance of the line, for most bipolar dc lines are inherently well balanced,
but are mainly due to generation of uncharacteristic harmonics by the con
reactor for various
jPR =
V0.5 x
10"lox 600 =
%/3 x
verters.
2. Draw the terminating circuits, including dc reactor and dc harmonic
filters and surge capacitor if used, and calculate their impedances and
Khk-hPh)2
residual cur
weighted volts
(68)
ph = C-message
Different ac frequencies at
340
8-7
Most of the harmonic frequencies produced at one end will be different from
those produced at the other end. Since each harmonic frequency is to be
considered separately, the resultant values of suchharmonics are not calculated
at this stage. Certain frequencies, however, such as 300 Hz and its multiples,
are generated at both ends, being characteristic of the 50-Hz end and un
characteristic of the 60-Hz end. If the uncharacteristic harmonic is negligible
in comparison with the characteristic one of the same frequency, it can be
neglected. If not, it can be calculated as in case b.
b. Same ac frequencies at both ends but no parallel ac tie; hence, asyn
chronous operation. The same nominal harmonic frequencies are generated
at both ends, but the phase difference between them varies continually and
randomly. In a sense, it may be said that the corresponding harmonics
generated at the two ends have slightly different frequencies but still have the
same noise weighting. Hence, at any given point of the line, the components
of the same nominal frequency should be combined by taking the square root
of the sum of the squares.
c. Parallel ac tie, synchronous operation. There is still a variation of phase
difference between fundamental voltages at the two ends, according to
loading of the parallel ac tie, and this difference is greatly multiplied in the
higher harmonics. Even for the lowest characteristic dc harmonicthe sixth
a 180 change of phase difference accompanies a 30 change in the funda
mental phase difference; and the 180 change is enough to make the vector
sum of the two components vary from an arithmetic difference to an arith
metic sum. Hence the components may reasonably be combined, as in
case b, by the root-sum-square method.
TELEPHONE INTERFERENCE
341
in the telephone line due to harmonic currents in each conductor of the power
line should be computed separately and then added vectorially. A computer
program for doing this has been described.60
Calculation of Telephone Noise Due to Three-phase Line
Outline
I. Reduce harmonics on the power system by filters.
II. Reduce coupling between the power system and the telephone system.
342
8-8
HARMONIC FILTERS
343
Purposes
The ac harmonic filters serve two purposes: (1) to reduce the harmonic
voltages and currents in the ac power network to acceptable levels and (2) to
provide all or part of the reactive power consumed by the converter, the
remainder being supplied by shunt capacitor banks, by synchronous con
densers, or by the ac power system. The dc harmonic filters serve only to
reduce harmonics on the dc line.
Types
344
HARMONIC FILTERS
r-nm~
(<0
(b)
For these two reasons shunt filters are used exclusively on the ac side. On
the dc side, the dc reactor, which is obviously a series element, constitutes all
or part of the dc filters. It must withstand high direct voltage to ground
and high direct current. It serves several additional functions (Section 7-2),
however, that require series connection. The remainder of the dc filters (if
used) consists of shunt branches.
Ac filters could be A-connected, but this connection offers no advantage;
therefore, the Y connection with grounded neutral is used.
Sharpness of Tuning. Two kinds are used : (a) the tunedfilter (high Q filter),
which is sharply tuned to one or two of the lower harmonic frequencies, such
as the fifth and seventh, and (b) the damped filter (low Q filter), which, if
shunt-connected, offers a low impedance over a broad band of frequencies
embracing, for example, the seventeenth and higher harmonics. The second
kind is also called a high-pass filter. Figures 21 and 22 show typical circuit
diagrams and characteristics of the two types. They are analyzed under
"Design of Tuned Filter," page 355, and "Design of High-pass Damped
Filters," page 375, respectively.
Cost of Filters
The capital cost of ac filters is in the range of 5 to 15% of the cost of the
terminal equipment.* This is high enough to justify careful design from the
standpoint of economy as well as adequacy. The cost of losses should also be
taken into consideration. The cost of filters may be partly charged to reactive-
* For example, the cost of the filters of the New Zealand scheme was said to be 12%.52
Fig. 21. Single-tuned shunt filter: (a) circuit; (b) impedance versus frequency.
2Rf~
0
(a)
Frequency f
(b)
Fig. 22. Second-order damped shunt filter: (a) circuit; (b) impedance versus frequency.
power supply and partly to filtering though there is no logical basis of the
division.
A minimumfilter is one that adequately suppresses harmonics at the least
cost and supplies some reactive power but perhaps not all that is required.
A minimum-cost-filter is defined under " Minimum-cost Tuned Filter," page
368. It may or may not give adequate filtering.
About 60% of the capital cost of the filters is that of the capacitors. Hence,
substantial savings are possible by judicious choice of kind of capacitor.
346
8-8
HARMONIC FILTERS
347
Converter
Ac network
Fig. 23. Circuit for computation of harmonic currents and voltages on the ac side of a
converter.
current
Y*/=Z'hf
and y
'hn
-A
7
&
hn
V =
ZhfZhIhc
Yftf+Y,
lhn
hf
and the harmonic currents in the network and filter, respectively, are
In =
V*
Zh,
V,
=
2W
Zhf+Zhn
Zh/hc
hf +
Z*hnhc
Zhf
Yft/ + Yhn
Yh/Iftc
Yfl/ + Yfin
(69)
(70)
(71)
348
Since tuned filters are customarily provided for the low characteristic
harmonics and since the impedance of such a filter at the frequency to which
it is tuned is a low resistance, severe parallel resonance of filter and network
to such a harmonic is unlikely unless the filter passband is too narrow and
unless either the system frequency is abnormal or the filter is detuned. Such
resonance is likewise unlikely at the higher frequencies for which the highpass damped filter provides a low impedance and high power factor. It is more
likely to occur at a low uncharacteristic harmonic. It is unlikely to occur at
more than one harmonic frequency at the same time although, because of
changes in the network, it could occur at another harmonic frequency at
another time.
The severity of resonance depends on the amount of damping due to
losses both in the filters and in the network. Therefore, some knowledge of
the response of the network to harmonics is desirable.
8-8
HARMONIC FILTERS
349
350
8-8
HARMONIC FILTERS
351
500
500
5613
550
540
250
530
610
*250
250
1085
109Q
250,
r300
1278
i5?3
1630
1 100
.1101
.633
650.
850'
'
iG50
950
>625
1120
710
760'
670
720
595
-250
-500
0
250
500
750
1000
R(fi)
Fig. 25. Calculated impedance of 220-kV network at minimum load with one line dis
connected. (From Ref. 38 by permission.)
500
250
500
750
Fig. 24. Calculated impedance of 220-kV network at maximum load. (From Ref. 38 by
permission.)
4. Resonances in the ac network are the rule rather than the exception.
5. The harmonic impedances bear no relationship to the fundamentalfrequency short-circuit level.
6. Loads provide some damping; that is, they make resonances less sharp.
This effect is more apparent on circuits of low voltage than on those of high
voltage.
7. This damping increases with frequency.
8. On cable systems, the impedances to higher harmonics (fifteenth to
twenty-fifth) are lower than on overhead-line systems.
It can be seen that the loci in Figures 24, 25, and 26 would fit closely
within circles on the right side of the vertical axis and tangent to it at the
origin. In addition, a very great part of all these loci would be to the right of
straight lines making angles of +80 with the horizontal axis. At lower
voltages the entire loci would lie to the right of such lines. Figure 28a shows
a boundary adopted by C.E.G.B. for its studies of 132-kV filters for the
Kingsnorth project,44 the radius being 55 Q. A similar boundary, but with
500- (1 radius, was adopted for the 400-kV system. Such diagrams facilitate
filter design by limiting the range of values of network impedance to be
assumed.
Figure 28b shows the boundary in the complex admittance plane corresponding to that in Figure 28a in the impedance plane. .
352
540
500
550
520
250
270,
970
'950,
50"V85,1000
(IxT -415 V
300 1010"
32?>330 .
i
'920
1020
310
11280
650
620
560
.1028
-250
1040
1050
Fig. 27. Measured harmonic impedance of the 132-kV 50-Hz system at Lydd. (From
Ref. 49 by permission.)
568
'660
570
-500
250
500
750
1000
1250
*(B)
Fig. 26.' Calculated impedance of 220-kV network at minimum load with two lines dis
connected. (From Ref. 38 by permission.)
<
Aims are (a) to achieve adequate harmonic reduction and (b) to supply
the required reactive power at fundamental frequency, (c) achieving both at
minimum cost.
O'f
(a) Z plane
(b) Y plane
Fig. 28. Bounds of loci of (a) impedance and (b) admittance to harmonics in an ac
network.
......
=-
354
8-8
The higher harmonics have smaller magnitudes, and the frequency ratio of
successive harmonics approaches unity. A great many tuned filters would be
required, and their passbands would overlap anyhow. A damped high-pass
filter is more economical for these higher harmonics.
The number of tuned filter arms varies from one dc link to another, the
commonest number being four. Logically, the combination of tuned arms
and high-pass arms should be the cheapest one that satisfies the filtering
requirements. Provision of tuned filters for the seventeenth and nineteenth
harmonics may depend on the number of bridges in the converter.
The relative magnitudes of characteristic harmonics of 2- and 4-bridge
12-pulse converters are shown in Table 8 for all bridges in service and for one
Table 8
Number of
Bridges
Zf = R+j[coL- )
a>C
2/2 or 4/4
3/4
1/2
=1
1.000
0.750
0.500
11
13
17
19
23
25
0.091 0.077 0
0
0.044 0.040
0.050 0.036 0.068 0.058 0.015 0.013 0.033 0.030
0.100 0.071 0.046 0.038 0.029 0.026 0.022 0.020
(72)
co =
Current
(73)
Vlc
X0 = conL = =
when
bridge out of service. Filters for the fifth, seventh, seventeenth, and nine
teenth harmonics are needed only when a bridge is out of service.
Shunt capacitors are used mainly for varying the reactive power when the
load on the converter changes. They also improve the filtering of high
harmonics.
Size. The size of a filter is defined as the reactive power that the filter
supplies at fundamental frequency. It is substantially equal to the funda
mental reactive power supplied by the capacitors. The total size of all the
branches of a filter, including shunt capacitors, is determined by the reactivepower requirement of the converter station and by how much of this require
ment can be supplied by the ac network and by synchronous condensers,
if any.
The size of individual arms depends on filtering requirements, but seldom
is it less than the size for minimum cost (see "Minimum-cost Tuned Filter,"
page 368).
The design of tuned filters involves selection of their size and sharpness of
tuning (0, and is discussed immediately below.
The design of high-pass damped filters involves selection of their size,
sharpness of tuning, and resonant frequency; it is discussed beginning on
page 375.
355
Fundamental
_
HARMONIC FILTERS
(o
co (75)
<0=co(l+5)
(77)
c=5o=vRe
(78)
COn
(79)
(On
Z/Rl+jQd
(80)
For the small frequency deviations (<M 1) in which we are now interested,
the impedance is given very nearly and more simply by
Zf s R(1 +j2d 0 =
|Z/| Wl + 4SZQ2 = Z0 V Q-2 + 4S2
(81)
(82)
!|
356
8-8
HARMONIC FILTERS
357
~ 1 +J25 Q)
fR(
|Yf I
l-j2SQ
Q-J28Q2
~
R(1 + 482 Q2) X0(l+ 48ZQ2)
(83)
~ _ 1_ _
Wl + 452Q2 X0/l +482Q2
f~R{\+ 482Q2)
f =
28Q
Q
Z0(l
X0(l
+ 432Q2)
28 Q2
+ 482Q2)
(85)
(86)
A,
1 (XL
A C\
Graphs
of Impedance.
|Zy| versus frequency deviation 8. Curves A and B are for the same R; they
have the same minimum impedance. Curves B and C are for the same X0 ;
they have the same asymptotes D (corresponding to R = 0). The equation of
the asymptotes is \Xf\= 2X0\8\. Curves A and Careforthe same
th.y
have the same passband PB. From Eq. (81) the edges of the passband are at
8 1/2Q, and the width of the passband is IjQ.
From these curves it is apparent that the impedance of the filter at its
resonant frequency can be decreased by decreasing R. In order to keep the
impedance low over a frequency band bounded by the points of maximum
0
S, percent
Fig. 29. Impedance of tuned filter as function of frequency deviation. Curve D consists
of the asymptotes of curves B and C.
Curve
A
B
C
R
{&)
10
10
5
X0
(a>
500
250
50
25
50
250
Passband
(PB)
2%
4%
2%
|Y|
|YV + Yhn\
-J
358
8-8
HARMONIC FILTERS
359
cfiy
vh = i\x,a,
L_ Asymptote,
slope = 2
V, =
0
The variables that are not under the control of the filter designer are chosen
pessimistically; that is, so as to give the highest Vh . Then the variables that
are under his control are chosen optimally to give acceptable Vh. The
variables for which pessimistic values are assumed are the frequency devia
tion S and the network impedance
Harmonic voltage is shown to in
crease with <5; hence, the pessimistic value of 5 is the greatest value that is
expected to persist: <5m. The network impedance is taken as the worst value
within reasonable limits (see "Impedance of the AC Network," page 348).
The variables that the designer can vary, within reasonable limits, are the Q
and the " size " of the filter. There is an optimum value of Q that gives mini
mum harmonic voltage for the assumed network conditions, and this value,
denoted by Q0 should be used. It is independent of filter size. Then size
is chosen for acceptable harmonic voltage and for the desired amount of
reactive power. Since Q0 depends on the assumptions about the network
impedance, several cases must be examined.41
Case 1. Infinite Network Impedance. In this case the resultant impedance
is merely that of the filter: Zh =Zhf. By substitution of Eq. (82) for \Zf\ into
Eq. (88), the harmonic voltage is given as
Q = Qo = CO
V (Ghf
Ihc
(92)
In the present case we may put Ghn = 0 and also, on the assumption of reso
nance, Bhf + Bhn = 0. Then, simply,
x = QS
,_
(89)
(90)
v = -
'
G>,f
(93)
V = X0Ihc(Q-'
+ 4dlQ)
(94)
This is minimized if
Q=Q.=X
20m
(95)
Vh=.45mX0Ihc
(96)
360
8-8
As before, the greatest value of <5, <5m, must be assumed, and optimum
must be found, this being the value that maximizes |YA|.
ZV=o(e_1+725J,
with fixed Xf 28mX0 and variable Rf = X0jQ as a horizontal line in the Z
plane. In the Y plane (part b of the figure) this line, inverted, becomes a
semicircle of diameter l((25mX0) tangent to the G axis at the origin. Points on
the semicircular locus in the Y plane, corresponding to points on the recti
linear locus in the Z plane for the same values of Q, are found by drawing
radial lines from the origin of each plot with equal but opposite slopes; for
example, the points for Q = 11(25J lie on lines of slope 1 (angle 45).
Vectors from the origins to points on the loci represent filter impedance and
admittance, Zv and Yhf, respectively.
In Figure 316 vector Yv is tentatively taken as that for Q l/(2<5m), and a
tentative vector Yh is added to it to give Y* . The terminal points of Yhn and
HARMONIC FILTERS
361
Yh must lie within or on the boundary of the shaded area, drawn there for
4>m 60. It is readily seen that the tentative choice of vectors does not give
minimum Yh . For the assumed Yhf the shortest vector Yh is perpendicular to
sb
the boundary and terminates on the boundary above the vertex. Furthermore,
it may be seen that the tentative choice of Yhf is not that which maximizes Yft
with respect to Q. The proper Yhf is that which ends on the semicircle at a
point where the boundary at angle +(j>m is tangent to the semicircle.
In Figure 32 the vectors are redrawn so that Yhf maximizes Yh, and Yh
X,
2 snlx,
Zhf/
26mX0
R,
(a)
Fig. 32. Construction for finding optimum Q and worst network admittance Yh. Drawn
for <f>m = 60".
2<5_X
Fig. 31. Loci of (a) filter impedance Zhf and (6) filter admittance
Yhf with constant X
and varying R and Q; (b) also shows tentative choices of Yhf and Yhn
minimizes it. The truth of the statement made earlier, that \(j>\ =</> and that
the sign of < is opposite to the sign of 5, is proved by the vector diagram,
drawn for positive 5 = 8m and for negative $ = <pm . (Remember that <t> was
defined as an impedance angle and that ~4> is the corresponding admittance
angle.) In addition, the effect of varying 5 may be shown. Decrease of 5
increases the diameter of the semicircular locus of Yhf, thereby increasing
Yh and decreasing Vh. Hence \5\ =8m is more pessimistic than |<5| <dm.
Negative 5 turns the figures upside down.
Formulas for optimum Q and for the corresponding Yhf , Yhn, and Yh
can be derived by trigonometry, starting with the known diameter of the
semicircle, the known angle <f>m , and other angles related to (f>m Triangle
362
8-8
OCD is isosceles, with the two smaller angles COD /_ CDO 4>J2. In
right triangle ODF, OD = OF cos (0m/2); hence
|Y-! =
cos ((bml2)
i&r
(97)
K- |YV|
IY ! IY I coco.
|Y,|
0S2 (0/2)
and
IYJ
- YW|
|
sin
os
ZOmA0
363
(f>m on the optimum Q of the filter and on the maximum guaranteed har
monic voltage Vh. In particular, limitation to +75 reduces the harmonic
voltage for a given size of filter, or the size of filter for a given harmonic
voltage by about 21% from Case 2 (purely reactive network, <j)m = 90).
Typical values of Q in practice range from 30 to 60 with series resistors.
+1
C0S
HARMONIC FILTERS
m
W
_ srnm
40mX o
Yv = |Yw|/-90MJ2
Yt = 17-90 + ,
(101)
(102)
(100)
La
\Lb
Ra
Rb
c3
The value of Q corresponding to the chosen Xft/ is found from Figure 31a:
o3>
and from Figure 316,
tan <j)f = cot ((f>J2)
(a)
Equating the last terms of the two equations for tan <pf , we find the optimum
value of Q to be
cot(&,/2)
Qo = -
= IT*-r-r-
25m
cos (j)m+l
(104)
'
25m sin
\Yh\
4Wc_
cos (f>m + 1
8mQo
Vh!8mx0lhe
co
2.00
30 45
1.87 1.21
2.14 2.35
those tuned
seventh harmonics. The equations for the parameters of the
double-tuned filter are given by Ainsworth.41
The advantages of a double-tuned filter over two single-tuned filters are the
following:
Such filters are used at both terminals of the Cross Channel link.42 The
impedance of that at the French terminal is given in Figure 34a and 6. Its
parameters are
Table 9
15
3.80
2.03
Table 9 shows the effect of limiting the impedance angle of the network to
(b)
Fig. 33. Transformation from (a) two single-tuned filters to ( b) double-tuned filter.
60
0.87
75
0.65
2.67
3.17
80 85
0.60 -0.55
3.41 3.68
90
0.50
4.00
R2
*3
= 4.2 Q
= 1.656 Q
=2.11 Q
Cj = 1.51F
C3 = 12.08 /iF
Lj =208 mH
L2 -- 24 mH
600
8-8
HARMONIC FILTERS
365
'i
500
C3
400
8.
300
<8 200
u
100
150
200
250
300
350
400
450
Frequency f, Hz
Fig. 34a. Double-tuned filter at Echinghen for fifth and seventh harmonics: calculated
scalar impedance versus frequency.
'
400 J L500
300 _ .450
200
285
N90
>400
/270
100
350C3
x-
0
240'
-100
i375/
\ /26O
1295
250
J 340
/ \330
T220 \320
-200
"200
-300
"180
/300
310
305
Capacitors. Capacitors account for the major part of the cost of filters.
They are composed of standard unitstypically rated at 100 or 150 kvar,
8 to 14.4 kV, at 50 or 60 Hz connected in series and parallel for obtaining
the desired overall voltage and kvar ratings.
Each unit consists of several rolls made of alternate layers of aluminum
foil and sheets of insulation, tightly enclosed in a sheet-steel box filled with
an insulating liquid. The solid insulation consists of either (a) several sheets
of special paper impregnated with the liquid or (b) a sandwich of one sheet of
such paper placed between two sheets of thermoplastic material. There is also
an unavoidable thin film of liquid between the solid insulation and the metal
foil; the thickness of the film depends on the pressure with which the rolls
are formed. Three kinds of liquid impregnants are in use: (a) mineral oil,
(b) trichlordiphenyl, and (c) pentachlordiphenyl. The last two are generically
called askarels. Thus the dielectric properties depend on those of the paper,
the impregnant, and the plastic (if used) and on the amounts used of each.
The density of the paper can be varied, ranging from 0.8 to 1.2 g/cm3, and
the paper may be impregnated with any one of the three liquids. Of these,
trichlordiphenyl is the most used at present, having superseded pentachlor
diphenyl, which has a higher freezing point, a lower dielectric strength, and a
lower dielectric constant.
Two of the most important properties of the capacitors are (a) temperature
coefficient of capacitance and (b) reactive power per unit of volume. The
latter is usually proportional to the dielectric constant and the square of the
maximum safe voltage gradient. Other important properties are (c) power
loss, (d) reliability (or life), and (e) cost. Approximate values of some of these
properties are listed in Table 10 for various dielectric materials. These values
should be interpreted as indicative rather than exact, because they vary with
the density of the paper, the thickness of the liquid film, the temperature at
which they are measured, quality control of the materials, etc.
A very low temperature coefficient of capacitance is desirable for tuned
filters in order to avoid detuning caused by change of capacitance with
ambient temperature or with self-heating of the capacitors; but this property
is unimportant for damped filters or for power-factor capacitors. Capacitors
filled with mineral oil can have either positive or negative temperature
coefficients, depending on paper density or film thickness, and, by proper
design, can be made to have essentially zero coefficient. In the past such
capacitors have been used for tuned filters almost to the exclusion of other
kinds. Low temperature coefficient is obtainable also by use of high-density
366
HARMONIC FILTERS
367
surges. The required insulation level may be greatly reduced, with attendant
savings in cost, by protecting the inductor by connecting a lightning arrester
Table 10. Typical Properties of Capacitors for Power Systems partly based
on Ref. 46
Solid
Dielectric
Paper
Density
(g/cm3)
Coefficient
of Capaci
tance (10-6
Impregnant
per deg C)
Dissipa Relative
Dielectric tion
kvar
Constant Factor per Unit
at 20C
(%) Volume
Overall
Paper
1.0
Mineral
oil
+250
3.6
0.17
41
Paper
1.2
+400
4.2
0.19
39
Paper
0.8
Mineral
oil
Pentachlor-
0.20
87
0.28
:V9
0.20
100
i\
diphenyl
Paper
1.0
Pentachlor-
-460
5.2
-50
5.4
5.5
diphenyl
Paper
1.2
Pentachlordiphenyl
Paper
0.8
Trichlordiphenyl
-750
1.0
Trichlordiphenyl
-500
5.5
Paper
1.2
Trichlordiphenyl
-100
5.5
Trichlordiphenyl
-710
3.1
paper
Paper
Plastic and
>!
0.28
66
Tuning
0.10
150
200
Capacitors obtain their high reactive power per unit volume by having low
losses and operating at very high voltage stress. For this reason, prolonged
operation at moderate overvoltage must be avoided to prevent thermal de
struction of the dielectric; and even very brief operation at high overvoltage
must be avoided to prevent destructive ionization of the dielectric.
The required reactive-power rating of a capacitor is calculated as the sum
of the reactive powers at each of the frequencies to which it is subjected.
Inductors. These are built with nonmagnetic cores. The inductance usually
has a fixed value. The Q at the predominant harmonic frequency may be
selected for lowest cost and is usually between 50 and 150. If lower Q is
desired, a series resistor is used. The cost of the inductor depends mainly on
the maximum rms current and the insulation level for withstanding switching
'
;
;
Skan.48
The converter current should be at least 0.8 of rated current. The fifth,
sixth, and seventh-harmonic filters (also seventeenth, eighteenth, and nine
teenth, if used) should be tuned while the converter is in six-pulse operation.
The eleventh, twelfth, and thirteenth-harmonic filters should be tuned during
12-pulse operation.
Operation with One Tuned Branch of Filter Out of Service
=5
368
8-8
system at the frequency for which the disconnected branch is tuned should be
assessed. If this possibility appears unlikely, the converter may be operated
with this branch disconnected while the harmonic voltage of this frequency is
measured to determine its acceptability. Perhaps it will be necessary to run
the converter at reduced load in order to make this harmonic voltage
acceptable.
HARMONIC FILTERS
369
100
/2
PrC = Vlcot C +
h(D
Total
80
Minimum
Harmonic
Fundamental
component, AS
B/S
megavars
(107)
The rating of the inductor may be assumed to depend similarly on the sum
of the fundamental and harmonic reactive powers :
20
PrL~h2+
Size of filter, Mvar/phase
Fig. 35. Cost of filter versus its size, illustrated for fifth-harmonic filter for 600-MW
4-bridge 12-pulse converter.
K=AS + BS~1
where
V2I2
+ A6
8 60
component
=S
K = cost (kj)
S = size (Mvar)
A, B = constants (k$/Mvar and k$-Mvar, respectively)
(106)
(108)
hS
The uncertainty of this assumption is considerably offset by the facts that the
fundamental reactive power of the inductor is much less than its harmonic
reactive power and that the total reactive power and cost of the inductor are
smaller than those of the capacitor. The cost of the inductor is greatly
affected by its insulation level, but this component of cost should be inde
pendent of filter size.
Neglecting the cost of the resistor, the total cost of the filter is
K = PrCUc + PrLUL
(109)
where XJC and UL are the unit costs of capacitor and inductor, respectively.
Substitution of the values of PrC and PrL from Eqs. (107) and (108) gives
K=s(uc
l(Uc+ UL)
= AS
+ BS~\
(110)
370
8-8
dK
-
= A-BS-2 = 0
dK/dS to
(ill)
(B\1/2
Smin = b)
(112)
and substitution of this value of S into the equation for cost gives the minimum
-I
Smin = (AJ
/3.32 x 105\1/2
..
J = 9-32 Mvar Per Phase
~(
=
\3.82 x 103
iin = 2fAB = 2
= $71,200 per
V3.82 x
phase
$332,000
= $38205+ -2
O
9-32
-
a>iV\
= 1.34 x
10-6F= 1.34/F
Its inductance is
SOLUTION
Pn = VdJdn = 600 x
L=
1.00 = 600 MW
IL1 = 3
(113)
Example 1
371
105 $- Mvar
= 3.32 x
as
Kmla = 2AB
---= -
'BV/2
whence
. cost
"
+ UL)
HARMONIC FILTERS
Pd
Vt
600
A = Uq +
= 3820
tl
8.00
3.50 + 2 = 3.82 $/kvar
$/Mvar
'
(D)
CihwJ2
1.34(5 x 377)2
= '21 H
3,
Q0
(L2
Qo\Cj
0-210 V/2
1 /
32.5 \1.34 x 10~6/
395
32.5
= 2.22 kV
372
373
1
]
zone.43
|
|
8-8
filter increases sharply with a decrease of size below that for minimum cost
and, on the other hand, that the cost increases more slowly with an increase
of size above that for minimum cost. Moreover, the quality of filtering in
creases with size. Thus, it is reasonable to assume that a filter smaller than
that for minimum cost is almost never used and that ones greater than that for
minimum cost are often used.
It was shown earlier (" Design of Tuned Filters," page 355) that for each
tuned filter branch there is an optimum value of Q, depending on the assumed
values of maximum frequency deviation 5m and maximum network imped
ance angle <f>m. Of these two variables, Sm is the one that has the greater
effect on Q, for Q varies inversely as <5m . Thus high Sm requires low Q, which
increases the continual power-frequency losses and which either impairs the
filtering by increasing the harmonic voltage or requires a greater capacitance
and, consequently, greater cost for maintaining the same quality of filtering.
Only partially offsetting this increase of cost is the fact that low Q decreases
the harmonic current at resonance and thus decreases the reactive-power
rating of a of a given capacitance.
One can get cheaper or better filtering by limiting the equivalent maximum
frequency deviation. A large part of this equivalent deviation is caused by
variation of capacitance with temperature. This part of the deviation can be
limited by using capacitors with low temperature coefficient of capacitance,
but this feature increases the cost of the capacitors.
Two methods have been proposed for limiting the equivalent frequency
deviation.
One of these maintains the average temperature of the capacitors nearly
constant by cooling them with air currents from a fan controlled thermo
statically or by capacitance measurement. The other method varies either the
inductance or capacitance by small steps so as to maintain the frequency
deviation at small values ideally at zero.
HARMONIC FILTERS
The advantages of the automatically tuned filter over a fixed filter are that,
for equally good filtering :
1
|
]
|
|
|
|
Advantages 1 and 2 reduce the cost of the capacitor, which is the most
expensive component of the filter. Advantage 3 reduces the cost of the resistor
and the cost of the system losses. These cost savings are offset partially by the
cost of the tuning control.
In some cases, advantage 1 cannot be realized because the filter must
supply a large amount of reactive power at fundamental frequency. In such
cases, however, advantages 2 and 3 are still realizable, and, in addition, the
quality of filtering is improved.
Filter design follows the procedure already outlined for fixed tuned filters
except that a smaller <5m is specified, which depends on the accuracy of the
automatic tuning.
Figure 36 compares the reactive-power rating of the capacitor of an auto
matically tuned filter for <5m = 0.01 to that of a fixed-tuned filter having the
same filtering performance, that is, the same harmonic voltage at maximum
frequency deviation.
Example
Calculate the rating of the capacitor required for the fifth-harmonic filter
bank of Example 1 (a) if the filter has fixed tuning and if the maximum esti
mated detuning varies from 1 to 5%; and (b) if the filter is automatically
tuned so as to limit the detuning to 1%. In both cases the fifth-harmonic
voltage is to be limited to 1% of the fundamental voltage, and the maximum
angle of the network impedance at 300 Hz is assumed to be 85.
i
I
374
8-8
HARMONIC FILTERS
375
950<5m +
0.0352
Mvar
Automatically tuned
3
2
Maximum detuning 5m,
Ox
Fig. 36. Comparison of capacitor ratings required for fixed-tuned and automatically
tuned filters.
SOLUTION, Part a
X,n =
F5(cos <f>m
~4dJs
+ 1)
"
ohms
c=
1
5 Xq
1
5g)j Xq
V\(QtC =
vfa
_ V\ _
y\K
5!,
51
5X0
5 x 3.86
=(i35.5)2aw
19.3
950<5m Mvar
Prcs = 40 = (95.5)2 x
I
(a)
(b)
(c)
Fig. 37. High-pass damped filters: (c) first order, (b) second order, (c) third order.
_ 1 (I J- w
2nd~jcoC+\R+j(oL
(114)
Z3rd y'coC,.+
\R +
+
l/jcoC jcoL
_i
(115)
376
8-8
377
HARMONIC FILTERS
12
- Q=
2.000
0.707
- 1.414
1.000
J2
1.000
1.414
0.707
3
2
Frequency //per unit
0.707
Asymptotes
2.000
Fig. 38. Scalar impedance of second-order high-pass filters. Per-unit /' and Z' are defined
by Eqs. (116) to (120).
introduced:
45
Fig. 39. Scalar impedance of third-order high-pass filters with equal capacitors.
1
"
-Jlc
(116)
(117)
-r
Q=
Z' =
Xo
Z
(118)
(119)
(120)
These definitions are consistent with those used in connection with the tuned
filterEqs. (73) and (75) except that Q is defined as the reciprocal of the Q
Z2nd=;r + (e+jr)
M
iX-1
1
1 , (
jf'+\Q + m'+jf)
(121)
(122)
8-8
HARMONIC FILTERS
379
2.0
2
= 2.00
1.5
= 1.41
Q = 0.71
1.0
1.00
1
1.41
2.00
0.5
>
0.3
/' =
= 1.41
0
Frequency f, per unit
= 2.00
-1.0
2.00
0.5
1.5
2.0
G'
Fig. 42. Second-order high-pass filtersloci in Y' plane.
1.41
Asymptotes
= 0.71
1
1.00
1.41
.00
0.5
1.0
2.5
per unit
Fig. 41. Scalar admittance of third-order high-pass filters with equal capacitors.
380
8-8
HARMONIC FILTERS
381
+ 1.0 --6#0.6
1.41
Q = 0.71
+ 0.5
= 1.41
-0.5
steps.
3. It provides more stability of the alternating voltage.
Q = 2.00
-1.0
0.50
1.00
1.50
2.00
2.50
G'
transformers.
Final Design
This can take the form of (a) minimizing the cost (and size) of the filters
for a specified quality of filtering, according to one or more specified criteria,
or (b) optimizing the quality of filtering for a given total size of filter. In the
second form, the best distribution of size among several filter arms must be
determined. In either case the Q of each tuned filter must be optimized, and
the corresponding guaranteed-maximum harmonic voltages must be found.
In finding optimum Q, each tuned branch may well be considered by itself.
In finding the harmonic voltages,, however, it seems better to consider the
382
8-8
entire filter as a unit after those <2's have been, determined. For those har
monics for which tuned branches are provided, the design value of maximum
detuning (frequency deviation) should be assumed.
An additional refinement is now introduced in finding the harmonic
voltages; namely, by assuming that the possible load admittance is not only
bounded by extreme angles +4>m but is also limited by the minimum conduc
tance shown in Figure 28b, corresponding to the circular locus in Figure 28a.
At each one of the harmonic frequencies the vector admittance of the whole
filter lies in one of the five regions demarcated on the admittance plane in
Figure 44. At any one of the frequencies for which a tuned branch is provided
Region 1
Region 2
HARMONIC FILTERS
383
pessimistic, that is, which give the lowest resultant admittance. From the
geometry of the figure, it is apparent that the resultant admittance vector is
as described below:
In region 1 or 5 it is perpendicular to the nearer angular limit.
In region 2 or 4 it terminates on the nearer corner of the boundary.
In region 3 it terminates on the vertical (minimum conductance) line of
the boundary.
In the determination of the proper size for each filter arm for a 12-pulse
seventh-harmonic filters
for acceptable deviation from a sine wave and divide the remaining size
among the remaining branches so as to obtain the lowest telephone inter
ference factor. The fifth and seventh harmonics have very little effect on
TIF. The total number of tuned arms may vary from two to six.
The optimization of filter design requires many successive trials and is
therefore accomplished most efficiently by a digital computer. A suitable
program for this purpose is outlined by Parker.45
A comparison between calculated and measured harmonics on the KontiSkan link, as reported by E. Uhlmann,516 is given in table 11.
Table 11
Region
of Yn
Region 3
Region 4
Region 5
Harmonic order
Calculated value (%)
Measured value (%)
11
0.4
0.33
13
0.3
0.24
23
0.28
0.10
25
0.28
0.10
Design of DC Filters
384
8-8
used for dc as for ac. The ratings of the capacitors are determined primarily
by the maximum direct voltage and by the required capacitances. The
harmonic voltages and currents in dc filters are relatively smaller than in ac
filters because of the effect of the dc reactor.
Under extreme conditions of telephone interference, the inductance of the
dc reactor may be increased, or two reactors may be connected in tandem.
In the latter case, the shunt branches should be connected to the point between
the two reactors.
Summary of Data or Asssumptions Required for Design of Filters
I. Ac network
HARMONIC FILTERS
385
Protection of Filters
Inductors. The voltage division across a tuned filter is as follows :
1. The power-frequency voltage appears almost wholly (and direct voltage,
wholly) across the capacitor.
2. The harmonic voltage appears equally across the capacitor and inductor.
It is the predominant voltage across the inductor but a minor one across the
capacitor.
3. Surge voltages appear almost wholly across the inductor.
Thus the only component subjected to transient overvoltages much greater
than the steady voltage is the inductor. Direct-current links constructed
before 1970 use filter inductors insulated for the same level (B.I.L.) as the
circuit to which the filters are connected. The cost of an inductor, however,
depends very largely on its insulation level. Considerable saving of cost is
possible by protecting the inductor by a lightning arrester connected in
parallel with it. The arrester must be able to "seal off" (stop conducting)
against the steady-state voltage, which may be assumed as the arithmetic
sum of the crest values of the power-frequency voltage and the harmonic
voltage. This voltage is very much lower than the crest surge voltage deter
mined by lightning arresters on the main power circuit. The arrester across
the inductor may be an ac type even if used on a dc filter, because the harmonic
voltage alternates. When the early dc links went into service, commercial
low-voltage lightning arresters capable of withstanding the required crest
currents and of dissipating the required energy were not available. For
satisfying these requirements, arresters of higher voltage rating were neces
sary, and, consequently, the economy in the inductor could only be partially
realized.
Similar considerations hold regarding protection of the inductors of highpass filters except that the steady-state voltage across these inductors is so low
that a simpler form of arrester or protective gap is feasible.
;
i
j
j
386
BIBLIOGRAPHY
in the same capacitor bank. The bank must be taken out of service for
replacing units.
General. If the dc ground is distant from the converter station, with con
nection to it by an electrode line, the dc filters are connected to the electrode
line instead of to ground. A dc lightning arrester or gap must be connected
from the electrode line to ground at the station. This must seal off against the
direct voltage caused by IR drop in the electrode and electrode line.
The ac filters as a whole are usually connected to the three-phase supply
circuit through a circuit breaker, and individual filter branches have dis
connecting and grounding switches. Use of resistance-step circuit breakers
reduces transient overvoltages during energization of the filters.
In some converter stations, faults in the filter capacitors are detected by
relays in which the fundamental currents are compared in two filter branches
connected to the same phase.48
PROBLEMS
1. Prove that the wave form of the network-side line current of one six-pulse
bridge is the same whether the transformer bank is connected YA or AY.
2. Make a Fourier analysis of the ac wave shape in Figure 3c.
3. Make a Fourier analysis of the ac wave shape in Figure 176.
4. Bipolar dc links are sometimes tested in round power. This means that
direct current is circulated in the two conductors, as usual, but both poles
have the same polarity of voltage. Therefore power flows in one direction
on one pole and in the opposite direction on the other pole. At each ter
minal one pole of the converter works as a rectifier, the other, as an inver
ter. Determine the phase sequence or sequences of the six-pulse and
twelve-pulse harmonics of the direct pole-to-ground voltages if the con
verters have one bridge per pole and the two bridges are supplied with
three-phase voltages (a) with no phase difference between them and (b)
with 30 phase difference. Which of the these arrangements is preferable
with respect to telephone interference (c) from the dc line and (d) from
the ac network ?
BIBLIOGRAPHY
1. "Wave Propagation in Overhead Wires with Ground Return," by John R. Carson,
Bell System Tech. Jour, Vol. 5, pp. 539-554, October 1926.
2. Engineering Reports of the Joint Subcommittee on Development and Research of the
Edison Electric Institute (formerly the National Electric Light Assn.) and the Bell
Telephone System (American Telephone and Telegraph Co.), New York, 5 volumes,
July 1926 to January 1943.
387
11
15. " Inductive Coordination Aspects of D-C Systems Supplied by Rectifiers," A.I.E.E.
Committee Report, A.I.E.E. Trans., Vol. 70, Part 1, pp. 1034-1055, 1951.
16. "The Influence of the Number of Phases Used in a Converter on Harmonics," by
F. Busemann, Direct Current, Vol. 2, pp. 65-72, December 1954.
17. " Contributions to the Study of Rectifier and Network Harmonics and Their Evalua
tion for Tracing the Source of Wave Distortions," by E. Friedlander and W. G.
Thompson, C.I.G.R.E., Report 312, 1954.
18. "Higher Harmonics in a Power System Supplying Mercury Rectifiers," by B. G.
Orkina, translated by B. F. Kraus, Direct Current, Vol. 2, pp. 115-121, June 1955.
From Elektrichestvo, No. 2, 1955.
19. Interference between Power Systems and Telecommunication Lines, by Hans Robert
Julius Klewe, London, Edw. Arnold, 1958. An E.R.A. report.
20. "Telephone Interference and Other Effects Caused by the Gotland H.V.D.C. Trans
mission," by G. A. Pettersson, L. Ahlgren, and H. Forsell, C.I.G.R.E., Report 324,
1958.
388
BIBLIOGRAPHY
22. "The Proceedings of the Direct Current Study Committee of C.I.G.R.E., Leningrad,
1957, Part III," Direct Current, Vol. 3, pp. 267-272, March 1958. Harmonics, pp.
267-269.
23. "Harmonics in Electrical Systems and Their Reduction through Filter Circuits," by
E. Bornitz, M. Hoffman, and G. Leiner, C.I.G.R.E., Report 304, 1958.
24. "Capacitors in Power Systems with Rectifier Loads," by A. Schmidt, Jr., C. E.
Rettig, and H. A. Gauper, Direct Current, Vol. 4, pp. 116-119, March 1959. Pre
sented to C.I.G.R.E. Study Committee 10, 1958. Based on A.I.E.E. Tech. Paper
53-24, revised. Parallel resonance between shunt capacitors and power-system reac
tance excited by harmonics from rectifiers may cause overloading of capacitors and
telephone interference.
25. "Evaluation of Message Circuit Noise," by A. J. Aikens and D. A. Lewinski, Bell
System Tech. Jour., Vol. 39, pp. 879-909, July 1960.
26. The Telephone Influence Factor of Supply System Voltages and Currents, Supplement
to Engineering Report No. 33, Joint Committee on Development and Research,
Edison Electric Institute and Bell Telephone System, E.E.I. Publication 60-68,
Sept. 12, 1960.
27. High Voltage Direct Current Power Transmission, by Colin Adamson and N. G.
Hingorani, Garraway, London, 1960, Chapter 10, "Harmonics."
28. "Problems Arising from Higher Harmonics in H.T. D.C. Power Transmission," by
E. Schulze, Appendix II to C.I.G.R.E., Report 417, 1960. Also in Direct Current,
Vol. 4, pp. 129-130, June 1959.
29. "Reactive Power Compensation and Harmonic Filters," by A. Stalewski, Appendix
III to C.I.G.R.E. Report 417, 1960. Also in Direct Current, Vol. 4, pp. 130-133,
June 1959.
30. " Computation of the Harmonics of Rectifier Current in the Single-Conductor Mode
ofD.C. Transmission" (in Russian), by Yu. S. Kraichik, N.I.I.P.T., Vol. 5, pp. 180192, 1960.
31. " Telephone Influence Factor (TIF) and Its Measurement," by W. C. Ball and C. K.
Poarch. A.I.E.E. Trans., Vol. 79, Part 1, pp. 659-664, January 1961.
32. Schaltungslehre der Stromrichtertechnik (Converter Circuit Theory), by Theodor
Wasserab, Berlin, Springer, 1962, Chapter VIII, "Oberwellen der Gleichspannung
und des Wechselstromes " (Harmonics of the Direct Voltage and of the Alternating
Current), pp. 361-393.
33. "Some Problems in Connection with the Commissioning of the Lydd Converter
Station," by L. Csuros and G. S. H. Jarrett, Direct Current, Vol. 7, pp. 114-121,
May
1962.
34. "D.C. Interconnection between France and Great Britain by Submarine Cables:
Behaviour of the Boulogne-Echinghen Converting Station from the Point of View of
Telephone or Radio Disturbances," by P.-G. Laurent, C. Gary, and J. Clad6,
C.I.G.R.E., 1962, Report 331, 27 pp.
389
42.
47.
48.
49.
50.
51.
"The Design of Capacitor Components of Large High Voltage A.C. Filter Net
390
9
Ground Return
Mostdc transmission lines use a return path through the ground or seawater
or both, either continuously or for short times of emergency. For brevity,
such return paths are called ground returns even if the sea constitutes all or
part of the path. The ground path has a very low resistance and corre
spondingly low power loss in comparison with a metallic line conductor of
economical size and equal length if the ground electrodes are properly
designed. The resistance is low because direct current in the earth in a steady
state, unlike transient or alternating current, does not follow closely the
route of the metallic conductor but spreads over a very large cross-sectional
area in both depth and width. The resistance of this path is essentially
independent of the length of the line and may be regarded merely as the sum
of the resistances associated with each electrode unless the electrodes are near
one anotherwhich certainly would not be true in long-distance transmission.
These resistances can be made low.
It has been stated frequently in the literature (e.g., Ref. 35) that a groundreturn dc line is more economical than a line with metallic return, because
ground return saves most of the cost of one metallic conductor and the
losses in it. (It is, of course, understood that these savings are slightly reduced
by the cost of the ground electrodes and the power loss in the ground re
sistance.) This contention is true only if a ground-return circuit is compared
with a monopolar metallic circuit that has the metallic return conductor
grounded at one point; but the latter kind of circuit is almost never used.
A fairer comparison would be one between a monopolar ground-return
circuit and a bipolar metallic circuit on the basis of cost of line per unit of
power transmitted. Let us assume that both lines have the same current per
conductor and the same voltage from conductor to ground. The bipolar line
has twice as many conductors and can carry twice as much power at a
slightly higher efficiency than the monopolar ground-return line. The cost of
the bipolar line is about twice that of the monopolar line if submarine cables
391
392
GROUND RETURN
are used; but if an overhead line is used, its cost is surely less than twice that
of the monopolar line.
The advantage of the ground-return line is not its capability to transmit
power at a lower unit cost. There are, however, two definite advantages of
ground-return fines.
The first is that a dc fine can be built in two stages if the power capability
required in the initial stage is half or less of that required ultimately. It can
operate in the first stage as a monopolar fine with one metallic conductor
(overhead or in a cable) with ground return and in the second stage as a bi
polar fine with two metallic conductors and normally without ground
current. Thus a considerable part of the ultimate capital cost is postponed
until the second stage. The initial expenditure is somewhat more than half of
the total expenditure. The additional costs for the first stage are for the
electrodes and the electrode fines and, in the case of an overhead fine, for
the towers and right of way.
After the completion of a bipolar line, a second advantage is enjoyed;
namely, that, in the event of an outage of one pole of the line or of a converter,
the fine can be operated temporarily at almost half of its rated power by the
use of the other pole with ground return. For this reason the reliability of a
bipolar dc fine is substantially equal to that of a double-circuit three-phase
line, although it has only two conductors instead of six.
A homopolar fine with two or more independent conductors of the same
polarity has the same advantages as a bipolar fine. The difference between
these two kinds of line is that the homopolar fine with two poles in operation
has twice the ground current that it has with only one pole in operation; the
bipolar fine with both poles in operation has much less ground current
(theoretically none) than it has with one pole in operation.
The Gotland, Konti-Skan, and New Zealand fines were built, or intended
to be built, in two stages. The Nelson River fine, which is to have several
bipolar circuits, is being built in several stages.
Ground return has its problems as well as its advantages. The problems are
(a) the design of the ground electrodes for low resistance and low cost of
installation and maintenance, (b) the location and screening of the electrodes
so that ground currents cause negligible electrolytic corrosion of buried or
immersed metallic structures and harmless step and touch voltages, and (c)
prevention of the ground current's interference with the operation of other
services, such as ac power transmission, railway signals, or ship's compasses.
Before considering these problems in detail, we consider the distribution of
ground currents, especially in the vicinity of an electrode. An understanding
of the current field in the earth is necessary for understanding the problems
caused by earth currents.
9-2
THE
393
The current field at any point in the earth is described in terms of two vector
quantitiesthe current density J and the electric field strength $, and one
scalar quantity: the electric potential V with respect to remote earth. The
field as a whole is characterized by how these quantities vary in a system of
space coordinates having as its origin the ground electrode that is conducting
into the ground. From a practical standpoint we are interested
a current I
mainly in these field quantities at or near the surface of the earth.
Uniform Earth
The only property of the earth that determines the field of direct current in
it is its resistivity. Hence, in this connection, "uniform earth" means an earth
of uniform anisotropic resistivity.
Actually, the earth is very nonuniform, having a core of radius 3500 km
(0.55 of earth's radius) consisting of hot molten magma of low resistivity,
surrounded by a mantle 2860 km thick of hot but solid material, also of low
resistivity. The mantle is overlain under the continents by a relatively thin
crust about 40 km thick made of rocks of higher resistivity, but this crust is
thin or absent under the oceans. The crust is also nonuniform, consisting of
strata of different rocks, overlain in places by various soils and by bodies of
water. The surface layers of the land have many local irregularities of differing
resistivity, such as rivers,marshes, deserts, and buried pipe fines. Approximate
resistivities of some materials are listed in Table 1.
In spite of this general nonuniformity, there may be relatively uniform
volumes of earth in the vicinity of an electrode. Since the assumption of
uniform earth gives the simplest case for analysis, it is a good starting point.
It also affords a basis with which the various cases of nonuniformity can be
compared.
The following assumptions are made in addition to that of uniform
resistivity :
394
GROUND RETURN
9-2
Material
Copper
Typical Resistivity
(Q m)
1.72 x 10-8
Remarks
1 x 10-7
1 to 7 x 10-7
2.2 x 10-7
Lead
Carbon
0.3
0.2
Seawater
Rocks
(2)
V=
temperature
Fresh water
Soils
Average land surface
-2
7tr
3 x lO"5
Granulated coke
395
where I
= current (A) entering ground from electrode
r = radius (m) of hemisphere
27tr2 = area (m2) of hemisphere
temperature
Iron
Steel
5 to 300
10 to 1000
150
103 to 108
[ idr=~-[
)r
2n)r
%r
= ~~
2nr
volts
(3)
These values hold at the surface of the earth as well as along vertical oi
inclined rays. Figure 2 shows equipotential lines on the surface of the earth
I
77Wm
Fig. 1. Current flow from hemispherical electrode in surface of uniform earth is shown
by solid lines. Broken lines are intersections of equipotential surfaces with vertical plane.
hemisphere in the earth having its center at the center of the electrode. This
through the hemisphere divided by the area
of the hemisphere :
current density is the total current
/ = -z2
0)
10 m
Fig. 2. Plan view of equipotential circles on surface of the ground near a hemispherical
electrode of 1-m radius on which 100 V is impressed.
396
GROUND RETURN
9-2 THE
397
100
P
I 2na
ohms
(4)
Half of this resistance is that of the earth between the electrode and a hemi
sphere having twice the radius of the electrode. Low grounding resistance is
required for hv dc transmission and can be obtained by using a large electrode.
For practical reasons, very few, if any, electrodes are hemispherical.
However, the formulas given, except Eq. (4), are still valid for distances
exceeding several times the greatest dimension of the electrode but much
smaller than the distance to the other electrode.
>>
6
Equation (4) indicates that low earth resistivity aids in obtaining low
resistance to ground, but, under our former assumption of uniform earth,
this would imply lowering the resistivity of the whole planet. A more realistic
assumption would be to locate the electrode, still assumed hemispherical, at
the center of a hemispherical volume of material of resistivity px much lower
than the resistivity p2 of the rest of the earth (see Figure 5). This assumption
5
10
15
Distance r from center of electrode, m
Fig. 3. Current density J, electric field intensity 3, and potential V in uniform earth of
resistivity 100 O- m near hemispherical electrode of 1-m radius conducting current of
O)
i-2
,"3
,-4
10
102
103
104
105
106
preserves the symmetry that makes analysis very simple. The current density
is still given by Eq. (1), and the gradient by Eq. (2) with the appropriate value
of p. The potential for r > b is still given by Eq. (3) with p =p2 and for
a < r < b by
'bdr
~J2
a'
+P2
.
dr
Pi } Pi ~Pi
2n\ a
(5)
398
GROUND RETURN
9-2
The current density, electric field intensity, and potential for this case are
shown in Figure 6 for a = 1 m, b = 10 m, pt = 10 Q m, and p2 = 100 O m.
10'
10
'
10"
10"
however, in decreasing not only the grounding resistance and potential of the
electrode but also the heating of the earth, the power loss, and the step voltage
near the electrode.
10"
399
Buried Electrode17
\\ \ \
\ \ \
10"
\a
\
_
2?
>& S
CO
.- c
THE CURRENT
Image
10
102 10 104 105 106
Distance r from center of electrode, m
Fig. 6. Log-log plot of current-field quantities produced by 6.28 A into ground from
hemispherical electrode embedded in hemisphere of 10-Q, m soil of 10-m radius, which
in turn is embedded in 100-H m earth. Broken lines are for uniform 100-H m earth.
The curves are like those in Figure 4 for uniform earth at distances greater
than b but differ at smaller distances, that is, within the volume of low re
sistivity. The curve of electric field intensity S is discontinuous at r = b, but
those of J and V are continuous.
The resistance of the electrode to remote ground is
1 (p
K = 2n\ ai i", P2 -Pi
ohms
r\~-
(6)
which is less than that for uniform earth of resistivity p2 by the ratio
Pi
P2
p iV*
1-
P2>b
(7)
With the values used in Figure 6 (b - 10a and p2 = lOp, R'jRe = 0.19, and
for the ground current 1= 6.28 A the electrode potential is reduced from
100 to 19 V.
Again, a hemispherical electrode embedded in a hemispherical volume of
material of low resistivity is impractical. Lowering of the electrode resistance
and potential, however, can be obtained by locating electrodes in marshy
ground or in a trench back-filled with coke.
It must be emphasized that this effect is purely local and has no effect at
points beyond the area of low-resistivity material. It has an important use,
Buried electrode
Fig. 7. Geometry for calculating electric field of spherical electrode buried in uniform
earth.
electrode through a small insulated wire from a source above the surface. In
order to calculate the electric field around this electrode, we resort to an
imaginary device.
First, we imagine the medium (earth or water) to be extended above the
original surface until it fills infinite space. The present case then resembles the
first case considered (page 393), that of a hemispherical electrode at the
surface of a half-infinite conducting medium, except that now the current is
distributed through twice the solid angle of the earlier case; hence, for a
given current, the current density is now only half of its previous value at a
given distance r from the center of the electrode. Likewise the electric field
and potential are halved.
Second, we assume a second spherical electrode like the first one to be
placed directly above the latter at distance 2h, center to center, that is, at a
distance h above the original surface of the medium. The second electrode
injects into the medium a current equal to that injected by the first electrode.
400
9-2
GROUND RETURN
r.=
tr
volts
10'
Electrode
surface
10
Surface electrode
where r = \jx2 + h2 (see Figure 7). The electric field intensity at the surface
is horizontal and of magnitude
dV plx
=
dx 2 3
/t
(9)
Per metre
>
10"
at
pJ
0.0614pi
*max=
Buried
e lectroc e
max
-2
401
>/2A = 0.71A
(10)
(11)
but here the field is only 4/3 V 3 = 0.77 times the field of the surface electrode
at the same point. The maximum field intensity of the surface electrode,
however, occurs at its surface, where x = a; hence the maximum surface field
is reduced in the ratio
(12)
by burying. In our example ajh = 0.1, and the maximum field intensity is
10"
0.1
10
1
x, horizontal distance from electrode, m
\\
100
Fig. 8. Electric field intensity at surface of earth caused by spherical electrode of 1-m
radius buried with center 10 m below surface in uniform earth of 100-Q m resistivity and
conducting current of 6.28 A into the earth, compared with the field intensity caused by
hemispherical electrode at surface (Figure 4).
reduced by the ratio 3.85 x 10 ~3. Although the reduction in electric field
intensity due to burial of the electrode is only a local effect, it has an import
ant practical use. Burial to adequate depth is a good way of reducing hazardous
step voltages to safe values. The step voltage is the voltage between two feet of
a person or an animal.
In Figure 8, the curve for the buried electrode is asymptotic to two straight
lines, one with slope of +1, the other with slope of -2, intersecting at
x = h = 10 m, 6 1 V/m. The use of asymptotes for sketching curves of field
intensity versus distance from the electrode is another advantage of log-log
coordinates. The limiting,values of $ are found by substituting r3h3 =
402
9-2
GROUND RETURN
constant for small x (x<h) and r 3 s x3 for large x (x>h). Then & varies
as x1 for small x and as x-2 for large x.
Curves qualitatively like those in Figure 8 were obtained by measurements
made near Tracy, California, using as ground electrodes (a) the ground mat of
Tracy Substation of U.S.B.R. and (b) an electrode in a well 480 m deep.49
Resistance. The potential of the buried electrode is found by the addition
of the potential due to its own current field in an infinite medium and the
potential due to the image electrode. This sum, divided by the current of one
electrode, gives the resistance of the buried electrode to remote earth as
hmS
(13)
403
The cross-sectional areas of water and earth on the spherical shell are,
respectively,
Ax
= 2otr2
A2 2nr2
and
(14)
and the total current passing from the electrode into both media is
I
= J1A1+J2-2-2r2(/la + J2K)
t15)
Jz =i
whence
Pz2
Pz
Air
Wh6nce
Water,
pi
(18)
=2r\*LpM
9)
J2 = -17
; ctp2ipi)
2r (n +
Likewise
Earth, p2
I=2r2J1(+7yj
-
the land surface slope at an angle a to the horizontal. Assume, further, that
both the land mass and the sea bottom have equal, uniform resistivity p2 and
that the water has a much lower resistivity p x .Consider the part of land, water,
and air bounded by a sphere of radius r centered in the electrode. The water
inside the sphere is a spherical wedge, shaped like one segment of an orange;
the land and sea bottom constitute a hemisphere; and the air fills the re
mainder of the sphere. The current flow lines are again radial, and no current
crosses the plane boundary between any two of the three media. The current
densities in water and earth are inversely proportional to their respective
resistivities. The gradients and potentials are the same functions of r in both
media, the equipotential surfaces again being spherical shells. We now pro
ceed to find the gradient and potential as functions of r.
(20)
_y
2r(alp1+nlp2)
(22)
5R'e = 2a(ajpx
+ %lp2)
(23)
1
R'
(*M(P2IPi)
+1
Re
(24)
The potential and potential gradient at any point on the land are decreased
by the same ratio. Some values of the reciprocal of this ratio are given in
404
9-2
GROUND RETURN
10
20
50
100
200
500
1000
0.01
0.02
0.05
1.03
1.06
1.16
1.32
1.06
1.13
1.32
1.64
2.25
4.18
1.16
1.32
1.64
2.59
4.18
7.37
2000
5000
16.9
7.37
13.5
32.8
0.1
0.2
0.5
1.32
1.64
1.64
2.25
4.18
2.59
4.18
8.96
16.9
32.8
80.6
1.80
2.59
2.59
4.18
8.96
16.9
32.8
80.6
4.18
7.37
16.9
32.8
64.7
160.0
7.37
13.5
32.8
64.7
160.0
128.0
319.0
319.0
797.0
=
w2
Suppose that
Pa
Pz
(26)
The apparent resistivity of the earth, measured at any point on the surface
of the earth in the field caused by one or more electrodes, is,the resistivity of
a fictitious uniform earth which, for the same electrodes and currents, would
give the same field intensity as that measured at the point in question. Let
<?
2~P*1
2
%r
Apparent Resistivity
405
were recalculated for resistivity pa, the result would be S, the observed field
intensity. Thus Eq. (25) is consistent with the foregoing definition of apparent
resistivity and may be taken as an alternative definition of that quantity.
Equation (21) gave the electric field intensity at distance r from a shore
electrode. Suppose that this intensity were measured on the landward side of
this electrode, where the surface resistivity is p2 . The field intensity in a
uniform earth of resistivity p2 would be
u2
Thus, a shore electrode, and especially one at salt water, is seen to have a
considerable advantage over a land electrode in soil of average or even low
resistivity.
7u Ps
7
(25)
Since the field intensity in a uniform earth with given current injections is
Anisotropic
Earth67
Some parts of the crust of the earth consist of numerous thin horizontal
strata having different resistivities. Suppose that several kinds of such strata
occur in a cyclic order. The electrical properties of such strata in bulk can be
described approximately in terms of two resistivities :
the beds
the
beds
resistivity,
across
pt, transverse
=[ \m
) = coefficient of anistropy
(p
(27)
jplPt
(28)
406
9-2
GROUND RETURN
If, on the other hand, the strata were vertical instead of horizontal, making
the horizontal resistivity greater than the vertical, the apparent resistivity
would be
407
mm-
.V/AW.V
(29)
Pa=Pt
P2
< Pl
Two-layer Earth
The simplest assumption about the earth's structure over moderate dis
tances, after that of a uniform earth, is that of a two-layer earth or single
overburden. A horizontal layer of constant thickness h and uniform resistivity
Pi is assumed to lie on a " basement" rock of infinite depth having a different
uniform resistivity p2 , which may be either greater or less than px . Figure 10
Fig. 11. Current flow from electrode at surface of two-layer earth with lower layer
more conductive than upper layer; also the corresponding equipotentials.
pi
P2
> Pl
= PiJ =
Fig. 10. Current flow from electrode at surface of two-layer earth with lower layer more
resistive than upper layer; also the corresponding equipotentials.
pj
2nrh
(30)
Figure 12 compares the electric field intensity at the surface of such a twolayer earth with that at the surface of a uniform earth.
The potential difference between the electrode and a point on the surface
of the earth at distance r is
F= -
$dr
dr
27ih
-y2nh a
volts
(31)
potential of the electrode or of any point in its field with respect to remote
408
GROUND RETURN
10'
Y"
10
tf\\
2?
409
10"1
10
(b)
10*
1
10
Distance r, m
10'
Distance r, m
10'
(a)
Fig. 12. Comparison of electric field intensity at surface of the earth: (1) in two-layer
earth and (2) in uniform earth. The upper layer of the two-layer earth has resistivity
Pi = 100 n m thickness h= 10 m and cylindrical electrode of height 10 m and radius
1 m; the lower layer has p2 ~ co and infinite thickness. The uniform earth has p =100 Q mand hemispherical electrode of radius 1 m; I
= 6.28 A.
Upper layer
An electrode long enough to reach all the way through the upper layer, as
assumed above, is not generally practical. A more realistic electrode is one
at the surface of the earth, such as the hemispherical one assumed earlier in
this chapter; but such a one presents more mathematical difficulty.
One approach to the problem would be, as in the case of a buried electrode,
page 399, to seek a configuration of fictitious electrodes in a homogeneous
earth that would satisfy the boundary conditions. For the two-layer earth
with nonconducting lower layer, the boundary conditions are that no current
should cross either of two parallel, horizontal surfacesthe surface of the
earth and the surface separating the two layers. This condition is satisfied, as
Fig. 13. Fictitious array of electrodes in an infinite uniform medium that produces the
same field of current in the upper layer as the actual electrode (shown in black) when that
layer is bounded by insulating media.
these are equally spaced by 2h on a vertical line through the center of the
actual electrode, which is expanded from a hemisphere to a sphere to preserve
the symmetry. Each electrode conducts current 21 into the fictitious earth, /
being the current of the actual hemispherical electrode.
The potential at any point in the upper layer can be found as the algebraic
sum of components of potential due to the separate electrodes. Thus, at
410
9-2
GROUND RETURN
411
point P on the surface of the real earth at distance x from the center of the
real electrode, the potential is
... + >-.
r.emi+>+l
2n \x rt r2 +
r
=*v{l + 2xl})
where
V0lts
(34)
cos
i
nrt
/E
(33)
+ (2nh)2y12
r = x2
Upper
layer
nx2
2IK
2IK
p ,I(
00
+ 2x3
nx2I\ *
n= 1
z I
rnJ
(35)
The series in Eq. (35) converges but that in Eq. (33) does not.
The electric field intensity relative to that which would be produced in a
uniform earth of resistivity pt is
$
co
=1
+ 2x3
co
1+2
pT(wf
<36>
Fig. 14. Equivalence of current rays arriving at P from fictitious electrodes B and F
or from reflections of rays from real electrode O.
case is that of p2fpt = 0, or p2 = 0, that is, the lower layer a perfect conductor.
The pattern of current flow is similar to that sketched in Figure 15. The lines
are normal to the lower boundary. This pattern can be produced multiply by
reversing the current in alternate electrodes in Figure 13 or 14, that is,
by making K= - 1 instead of + 1 in Figure 14. It can be produced also by
negative reflections at the lower boundary of the upper layer. The arrowheads
on rays AP, CD, and DE would be reversed, so that the vector sums OA + PA,
OC + DC, and ED + EP become vertical instead of horizontal.
2/0
Fig. 15. Part of a fictitious array of electrodes similar to that of Figure 13 except that
the injected currents alternate in sign. The lines of current flow are normal to the boundary
plane midway between electrodes.
412
GROUND RETURN
&
=1
+ 2x3
Zt
co
jsn
=1
+2
100
jfn
[;1 + (2nhjx)1]3l2
(37)
10
K=~
(38)
P2 + P1
It obviously fits the three special cases already considered. Proof of the
correctness of K in the general case rests on the assumption of a transmitted
ray AG as an extension of incident ray OA in the same direction, and on
equating the sum of the normal (vertical) components of incident and
reflected rays of current density in medium 1 to the normal component of
transmitted current density in medium 2, and similarly equating the sum of
the tangential (horizontal) components of incident and reflected electric field
intensity in medium 1 to the tangential component of electric field intensity in
medium 2, and observing the vector relationships $x pxJx and
= p2J2
Note that K in Eq. (38) has the same form as the reflection coefficient of
electromagnetic waves at the end of a transmission line.
Results of calculation of $u as a function of hjx in log-log coordinates are
shown in Figure 16. There is observed a lack of symmetry between the upper
and lower halves of the figure. The upper half has an envelope of unit slope
for p2 = 00, as in Figure 12b; the lower half has a curved envelope of everincreasing slope.
Asymptotes. Intuitively, the apparent resistivity at short distances from the
electrode must be pls the resistivity of the upper layer; at long distances, it
must be p2 , the resistivity of the lower layer. Let these intuitive deductions be
proved mathematically. In Eq. (37), as x 0, SjSu = pjpx -* 1, or pa~*PiFor x ;> h, r -> x, and
~\+2
But
(39)
1~ K
K2<1
(40)
"
2
.e
c;
*1
> c
'(O
to
0.5
\v
E
. g
0.2
"D
Q>
0.1
\v_
0.1
0.05
Q>
II
0.02
0.01
0.01
vv
0.005
0.002
0.001
0.1
10
x _ distance from electrode
ft "thickness of upper layer
0.001
100
Fig. 16. Electric field intensity at the surface of a two-layer earth as a function of the
,
,
distance from the electrode.
414
9-2
GROUND RETURN
1(p2
2\pl
(41)
415
terized with good accuracy by two parameters representing transverse resistance per unit area : Tx = pxhx and T2 =p2h2 ohm-metres2. The dimensionless ratio T2jTx is used as a parameter in Figure 17, which shows the ratio
and hence
1 + 2 y Kn = 1 + - - 1 = Pi
Pi
<$u
Hence
Pi
Pa-Pi
Pi
(42)
(43)
(44)
Figure 16 shows that, with increasing distance from the electrode, the
apparent resistivity changes smoothly from the resistivity of the upper layer
to that of the lower layer. This transition occurs in a shorter distance (x = h
to 10/z) if the lower layer is more conductive than the upper layer than it does
if the contrary is true.
10
Fig. 17. Electric field in three-layer earth with highly resistant second layer:
and T2 = p22- (Adapted from Ref. 34 by permission.)
Tx = pjii
416
GROUND RETURN
electric field intensity at the surface caused by a given current into an electrode
at a given location. Geophysicists, however, use surveys of apparent re
sistivity made with varying spacing between current electrodes as part of
the evidence for deducing the geological structure.
The principal conclusion that we can draw from their work and apply to
ground-return dc circuits is that at distances up to several hundred kilometres
from the electrode the electric field gradient at the surface is greater than those
in a uniform earth of resistivity equal to that of the upper layer. At greater
distances the contrary is true, but at such distances the field intensities
attributable to the ground return current are already negligible.
New England
Adirondacks
X)
10
03
TD
C
CO
2
Q)
<u
>>
irt
0)
1
.2
417
Southern Nevada
0.1
Central California
t3
0)
LU
0.01
0.01
0.1
1
10
Distance from electrode, km
100
1000
Fig. 18. Comparison of electric field measurements made in various parts of the United
States. Those in the Columbia River plateau, southern Nevada, and central California
were made in investigations for the Pacific Interties, 1963 to 1965. The rest were made by
the U.S. Geological Survey, 1961 to 1963. (Adapted from Ref. 66 by permission.)
metres
(46)
WAV
where f frequency of current, Hz
p = ground resistivity, Q m
p. ground permeability, H/m
For nonmagnetic earth it becomes
5 = 503
metres
(47)
= 2.96 x
106/
(48)
For example, at 300 Hz (the fifth harmonic of 60 Hz) and with p = 150
a- m, 5 355 m and RJl = 0.888 Q/km; at 3000 Hz, 5 = 112 m and
Rjl= 8.88 Q/km.
418
9-4
GROUND RETURN
Current flow lines are parallel to one another and to the metallic conductors
except near the ends of the transmission circuit, where these lines diverge
from, or converge to, the electrodes. Here is an end effect, in a zone having
dimensions of 23 to 33, which results in an additional component of resistance
associated with each electrode.
As frequency is decreased, the ground currents spread farther below and to
the sides of the metallic conductors, and the end effects occupy a greater
length of the transmission line as well as zones beyond the ends of the line.
At zero frequency (direct current), nothing is left but end effects. The re
sistance of the ground-return is merely the sum of the resistances associated
with each electrode. The current flow lines become independent of the route
of the metallic conductors. The current leaves the electrode in all directions
without respect to the direction of the metallic conductors.
Both alternating and direct currents follow ground paths offering the least
impedance. For direct current, the impedance consists only of resistance, and
the paths are determined by the relative resistivities of different parts of the
earth. Ac impedance has both resistive and reactive components, and the
EARff
419 ,/
higher the frequency, the more dominant the inductive reactance becomes.
Alternating current tends to follow paths of low reactance, which are paths
near the metallic conductors. The crowding of the ground current into a small
cross-sectional area, however, increases the resistance of the circuit.
In dc transmission we are concerned with both direct ground current and
harmonic current, the former widely dispersed and the latter following
closely the route of the transmission line, which, fortunately, decreases the
harmonic voltages induced in other circuits.
If the two electrodes of a ground-return dc line are far apart compared
with the size of the electrodes themselves, the electric field intensity anywhere
except at points nearly equidistant from the two electrodes is very nearly
equal to the intensity due to the nearer electrode. At equidistant points, the
intensity is twice that due to one electrode. Figure 19 shows, in unconven
tional log-log coordinates, the field intensity at the surface of the earth along
a straight line from one electrode to the other, 2 km away. One curve is for
a homogeneous earth; the other, for a thin conducting layer.
9-4 THE NATURAL CURRENT FIELD IN THE EARTH66 67
>
c-
A'
CJ
Cause
O)
LlJ
Currents called telluric currents are induced in the earth by variations of the
,-3
>"4
1
10
100
100
1000
Distance from nearer electrode, m
10
Fig. 19. Electric field intensity along straight line on surface of earth between the two
electrodes 2 km apart. Curve A is for uniform earth, p = 100 Q m. Curve B is for con
ductive layer 10 m thick of pL = 100 H m, below which p2 = <x>. Hemispherical electrodes
of 1-m diameter: 1= 6.28 A,
magnetic field of the earth. The variations of the magnetic field are believed to
be caused by fluctuations inthe stream of charged particles emitted by the sun.
Some particles have positive charge ;others, negative charge. As these particles.
approach the earth, the earth's magnetic field deflects them eastward or west
ward, depending on the sign of the charge. They form a ring of current around
the earth in the ecliptic plane. If this ring current were constant in magnitude,
it would contribute only a static component to the earth's magnetic field, the
major part of which is believed to be caused by currents inside the earth.
However, the density of the stream of charged particles intercepted by the
420
9-4
GROUND RETURN
earth, as it travels in its orbit varies with time. Therefore, the ring current and
the component of magnetic field due to it vary with time. Moreover, the meet
ing of the two streams of particles on the back side of the earth sets up turbu
lence which leads to additional oscillations in the mmf.
p(il m)
1000 100 10
1
1000
100
~NS\\k
\\
Frequency
(mHz)
10 to 40
60 to 100
120 to 420
25 to 100
10 to 17
2.4 to 8.3
100 -
Some of the periods are long enough to justify the assumption that the effects
of the natural currents are not unlike the effects of ground-return currents
from dc transmission.
These oscillations undergo amplitude modulation at several different
\\
10 -
100
10 -
10
\"
\ \
Wbvv
100 ...
Period
(sec)
\\ \\
421
10
-I -
periods :
10
0.1
100
Fractions of an hour.
Daily. The amplitude at midday is 10 to 20 times that at night.
28 days (period of rotation of the sun), amplitude ratio from 2 : 1 to 20 : 1.
11 yr (sunspot cycle).
H\J5pf
* One gamma = 10~s oersted 0.8 mA/m is the unit used by geopbysicists.
(49)
Fig. 20. Graph for estimating the probability that geomagnetic micropulsations of mag
netic field intensity Hwill exceed specified levels, together with scales for converting these
to the amplitudes of electric field intensity 3 induced in earth of resistivity 10, 100, or
1000 H m. (Adapted from Ref. 66 by permission.)
/5/Y/2
W
H\ )
(50)
422
GROUND RETURN
Percent of
Time during
which Natural
Field Exceeds
Field from
Ground Return
50
10
1
0.1
Central Valley
of California
Southern
Nevada
Oregon)
(Tracy)
(Hoover Dam)
(km)
(mi)
(km)
(mi)
(km)
(mi)
90
29
9.5
56
18
6
66
17
4.5
1.2
41
120
52
13
75
32
2.0
Columbia
River plateau
(The Dalles,
3.2
423
11
2.8
0.7
2.9
General
Electrolytic Corrosion
The most serious problem that may be aggravated by direct ground current
is that of electrolytic corrosion of buried or immersed metallic structures,
such as underground and submarine power and communication cables; gas,
424
GROUND RETURN
water, and oil pipes, ships' hulls, docks, tower footings, and other foundations. The phrase " may be aggravated " was carefully chosen. There would
still be a serious problem of electrolytic corrosion of buried orimmersed metal
lic structures even if no dc lines were built. First, there are natural ground
currents, already discussed in Section 9-4. Second, even if there were none,
there would still be local ground currents caused by emfs of contact between
different metals, between local spots of different amounts of impurities in the
metal, and between spots in contact with ground moisture (electrolyte) containing different ions or the same ions in different concentrations. One
expects a piece of unprotected iron buried in the ground to rust, at first
superficially and eventually, completely.
A superposed uniform current field may aggravate corrosion, but under
other circumstances it may prevent corrosion. The different lies in the direc
tion of the current.
Metallic ions are positive and move through an electrolyte in the direction
of the electric current. Where current leaves a metal object, such as an iron
pipe, metal ions go into the electrolyte, and some of them combine with neg
ative ions; for example, a typical reaction between an iron anode and soil is:
Fe+ +
+ 20H
-> Fe(OH)2
(51)
The formation of the Fe++ ion releases two electrons that may traverse the
pipe to some other point where current enters it. At this point the electrons
combine with positive ions; for example,
2e- + 2H+
->
H2
(52)
This reaction coats the pipe with a layer of hydrogen that protects it from
oxidation and other corrosion. Metal from which current leaves to enter an
electrolyte is an anode. Metal that current enters on leaving an electrolyte is a
cathode. Both anodes and cathodes are electrodes. Metallic anodes are
corroded; most cathodes are protected from corrosion. Corrosion does not
occur where current passes directly from one metal to another metal or solid
conductor, such as carbon, with which it is in good contact, for here conduc
tion takes place by electrons, not by ions.
Different parts of the same metal object, such as the iron pipe considered
above, may be anode and cathode; and the same part may be sometimes
anode and at other times cathode because of reversal of the current. Metal
removed by leaving current is not redeposited by entering current unless the
reversals of current occur rapidly. Alternating currents of commercial power
frequency cause only about 1% as much corrosion as does direct current of
equal rms value.
The basic law of electrolysis is Faraday's : the mass of material that leaves
the anode and goes into solution in the electrolyte is proportional to the
'
425
Kec\i dt
(53)
Kec =
where
YYl
(54)
vKF
Element and
Its Symbol
Usual
Atomic
Valence Weight
24.32
26.97
65.38
55.85
207.21
EMF*
(V)
0.126
0.454
0.0932
0.335
-0.762
0.339
-0.440
-0.126
0.289
1.22
1.04
3.87
0.0374
1.18
2.37
Lead (Pb)
Hydrogen (H)
Copper (Cu)
Copper (Cu)
63.57
+0.345
63.57
+0.522
1.008
1.67
(g/A h) (kg/A-yr)
(mg/C)
-2.34
Magnesium (Mg)
Aluminum (Al)
Zinc (Zn)
Iron (Fe)
3
2
1.074
0.0104
0.329
0.659
3.97
2.94
10.66
9.13
33.87
0.328
10.39
20.77
426
9-5
GROUND RETURN
more than the theoretical value. The ratio of actual loss to theoretical loss is
known as corrosion efficiency,15 anode efficiency, or coefficient of deteriora
tion.31 Tests have shown that, for ferrous materials buried in soils, the corro
sion efficiency at high current densities (0.5 to 50 A/m2) varies from 0.2 to
1.4; at moderate current densities and average soil moisture, from 0.5 to 1.1;
at low current densities (say, 1 to 10 mA/m2), the loss of mass closely approxi
mates the normal corrosion due to local galvanic action added to the theoreti
cal value due to stray current. The recommended practice in the calculation of
loss of anodic metal is to assume corrosion efficiency of 1.0.
Thus the rate of loss of anodic metal can be calculated, at least approxi
mately, from the current, time duration of its flow, and the electrochemical
equivalent. If the current density were uniform over a given part of the
anodic surface, the rate of decrease of thickness at that surface could also be
calculated. If, however, the current density were nonuniform, the corrosion
would likewise be nonuniform, and only the average decrease of thickness
could be readily calculated.
An iron cathode does not waste away, as does the anode, but a lead cathode
may deteriorate in certain soils if it has too great a negative potential with
respect to the soil.
If two metals are in contact with each other, only the one higher in the
electrochemical series corrodes and protects the other; for example, leadsheathed cable with steel armor or run in steel ducts is protected from corro
sion, but the steel is corroded.
Ceramic and plastic pipes do not corrode electrolytically.
The permissible density of the current leaving a buried metal object depends
on the desired useful life of the object, the thickness of metal that can be lost
without impairing the serviceability of the object, and the electrochemical
equivalent and density of the metal. For buried pipes and cable sheaths the
permissible maximum current densities due to stray currents from groundreturn transmission electrodes is usually taken as 1 to 10 mA/m2. For thicker
objects the permissible current density is greater.
Example
A certain buried telephone cable has a lead sheath 137 mils (3.48 x 10~3m)
thick with an outer diameter of 2.54 in (6.45 x 10 ~"2 m). What is the per
missible leakage current density for loss of 10% of the thickness of the sheath
in 40 years?50
The following characteristics of lead are known:
427
SOLUTION
428
GROUND RETURN
9-5
429
2= $0
(55)
Pi
where p2 is the resistivity of the sphere and px is that of the medium.
In estimating the maximum surface current density of nonspherical objects,
one must expect saliencies in the direction of the external field to have density
several times greater than the maximum surface current density of a sphere
at the same location.
Bare Pipe Crossing the Field of the Main Electrode. A bare metal pipe, or a
cable with bare metal sheath or armor, is assumed to run straight across the
field (Figure 21) and, therefore, to have a point P closest to the electrode Eand
at a perpendicular distance y from it. Any other point Q on the pipe distant x
from P is distant r = (x2 + y2)1/2 from E. It is assumed that the earth has
uniform resistivity p and that the field is not distorted by the presence of the
PipeThe latter assumption is strictly true only if the resistance per unit length
of the tubular pipe or sheath is equal to that of the cylindrical part of the dis-
-1.00
Fig. 21. Leakage-current density J on buried pipe relative to density Jo at a point on the
pipe nearest to current electrode and distant y from it as a function of distance x from
that point.
dV
A-"*'
(56)
dir*p
dx
(57)
d2V
dip
dx 2
dx
(53)
Tl
430
T5
9-5
GROUND RETURN
The potential V of the pipe is the same as that of the soil with which it is in
contact, and, for electrode current / amperes into the homogeneous earth,
this is
V-f
2nr
(59)
=r
where
k=Pi
d2V
, ,
dx
k
r
,,1
Radial Bare Pipe (Figure 22). This is a special case of the previous case,
j.
d2V
dr 2
* is
2k
TT
(66)
pi
(67)
Jl = 7i2D0tr2
JL =
nD
431
(60>
(62)
with greatest value (not a mathematical maximum) at the end of the pipe near
the electrode, where r = r0 . This density is twice that in the previous case for
equal distances from the electrode to the nearest point of the pipe.
JL = 2 1
3)
3( - )
(63)
Jm = Hemp
Pipe
ro
(64)
which is the maximum absolute value. The negative sign denotes current
entering the pipe.
Adoption of dimensionless ratios J' = JLIJL0 and x' = xjy leads to the
equation
2(*')2 ~ 1
[(x')2+l}5/2
(
}
According to the last equation, the current is leaving the pipe everywhere ;
and this appears impossible. If, in accordance with the assumptions, the pipe
were merely cut off or capped, the current would have to enter the butt end or
cap, not the cylindrical surface ; but in practice, although not in this rough
theory, current would enter the cylindrical surface in a short section of pipe
near the end. Perhaps a more practical case is one in which the pipe terminates
in a partly buried tank or makes a right-angled bend to go to a tank or
other equipment on the surface of the earth. Then most of the current would
enter the surface of the tank or of the vertical run of pipe.
432
9-5
GROUND RETURN
Example 2
Effect of Distortion of Field Caused by Pipe. Rusck40 has shown that the
expression for leakage-current density with field distortion neglected
Eq. (64) should be multiplied by the dimensionless correction factor,
plotted in Figure 23 and calculated as
my) =
433
(68)
SOLUTION .
where
JIn (0.7y/
\p
v Dh)J
metre-1
(69)
ymin =
S0 = Struve's function7
N0 = Bessel function of second kind for real arguments7
The graph shows that the approximate formula gives good results for large
separations between the electrode and the pipe and for low resistivity of the
medium but gives much too high a value of leakage current density for small
separations and high resistivity.
1
|
|
{ii?Dmz)
(a)
where p = 180 fi m
/ = 2.0 x 103 A
D = 6.45 x 10 ~2 m
m = 3.6 x 10-4 n/m
JL0= 2.9xl0-3A/m2
mi
This result is based on the approximate theory, which neglects the distortion
of the electric field caused by the presence of the cable. It rests also on the
assumption that the dc line operates at full load with ground return for the
whole 40 yr.
Example
0.01
0.1
(7i
180 In
10
ftFig. 23. Correction factor for density of leakage current of bare buried pipe. See Eqs.
(68) and (69). (Adapted from Ref. 40 by permission.)
x 3.6 x 10~4
\1/2
0.7 x 6.47 x 103 | x 6.47 x
= (8.02 x
a/6.45
lO"2 x
103
1/
(b)
From the graph in Figure 23, at fiy = 5.2, read F(fiy) = 0.85. The corrected
434
9-5
GROUND RETURN
value of the surface current density is 0.85 x 2.9 = 2.5 mA/m2. For 2.9
raA/m2, the closest distance could be reduced to (0.85)1/3 6.47 = 0.95 x 6.47
= 6.15 km. The error is only 5%, and the accuracy must be considered very
good in view of the uncertainty of other factors.
Example
distances.
Cathodic
Protection12'15'23'44
435
Forced drainage is the only method used for protecting long cables or
pipelines. The necessary emf is usually provided by rectifiers if ac power is
available. Wind-driven generators with floating storage batteries are used in
some areas where the wind blows almost constantly. Generators driven by
gas engines are used along some natural-gas pipelines. Rectifiers are mounted
on poles; the engine-driven generators, in cabins.
Three arrangements of anodes are shown in Figure 24. The single anode is
7A
(a) Single anode
<<
jj
|j
Fig. 24. Three arrangements of anodes for cathodic protection of pipe. Multiple anodes, if
far apart, would have separate dc supplies.
|j
11
j
j\
j]
|:
jj
ij
|j
;
j
436
GROUND RETURN
437
Gil
Resistivity, fl-m
100
10
1
300
30
3
0.1
0.3
0.01
0.03
to be 1.0 V,
V=
+ 103 x
150
= +2.4 V
In x 104
Under such circumstances one or more of the rectifiers may well be of the
kind having their voltages automatically controlled so as to maintain the
desired pipe-to-soil voltages.
Calculations. The variation, with distance from current drain point, of
difference in pipe-to-soil voltage due .to cathodic protection, and of longi
tudinal current and leakage current can be calculated as for a leaky dc line.
On a very long pipe with one drainage point, these quantities all decrease
438
9-5
GROUND RETURN
439
Insulated joints
Track rails
Rheostat
AC Transmission
Many three-phase transmission lines terminate at both ends at substations
where there are grounding transformers usually transformer banks with
Y-connected windings and grounded neutral points. Zero-sequence direct
currents may flow in such lines. The magnitude of such currents depends on
the difference of potential of the ground mats at which these transformers are
located. These currents tend to saturate the magnetic cores unsymmetrically.
As such currents increase, the first indication is increase of the audible hum
from the affected transformers. This increase of hum is noticeable when the
neutral current per phase exceeds the normal alternating exciting current by a
ratio of 1.2 to 1.5 and is pronounced at a ratio of 2. At higher currents the
transformers may overheat because of increase of iron loss both in the core
and in the paths of stray flux (core clamps, tank, lid, etc.). Such currents, if
they exceed the relay setting, can also cause false tripping by overcurrent
ground relays of the plunger type.
Such troubles may be prevented by locating the dc ground electrode at
sufficient distance both from the converter station itself and from other sub
stations. Adequate separation of the electrode from substations is advisable
for another reason also for reducing electrolytic corrosion of the substation
structures and ground mat.
Excessive direct currents, however, occur infrequently in lines and trans
formers as a result of unusually great natural ground currents, especially
during the so-called magnetic storms. (See also Section 9-4.) The lines most
susceptible to such troubles run easterly and westerly in high northern
latitudes.
Series capacitors in a transmission line would block stray direct currents
whether from natural sources or from dc ground-return transmission. There
is no record to date of series capacitors having been installed for this purpose.
In some instances the transformer neutrals can be grounded through
resistors of a few ohms for reducing the direct neutral current.59
Railway Signals19'20
70
Battery
Relay coil
1 to 6 V
rail at the other end. Contacts of the relay may control block signals, inter
locking devices at railway crossings, junctions, and terminals, or indicators in
the dispatcher's office for centralized traffic control. If no train is in the sec
tion, the relay is energized and, if used for a block signal, puts the signal in the
"clear" or " proceed " aspect. If, however, the section is occupied by a train
or even by a single car, the wheels and axles form a short circuit, deenergizing
the relay and making the block signal indicate " stop." Note that this system
fails safe: if the battery is dead or if a rail is broken, the signal indicates
" stop" even if the section of track is not occupied.
During some ground-current tests made in Sweden about 1952, false
operation or conditions that could give false operation of signals were ob
served at distances as great as 20 km (12 mi) from an electrode conducting
170 A, and calculations indicated that an electrode current of 1000 A might
cause false operation of signals at distances up to 60 km (37 mi).19 Naturally,
this caused alarm. The track circuits used in Sweden at that time, however,
had only one insulated rail, the other rail being continuous. The continuous
rail readily picks up stray currents, which by their IR drops introduce
unwanted voltages into the track circuits. A stray current of 2.5 A or more
might cause misoperation.
In Great Britain, the United States, and probably elsewhere both rails are
sectionalized by insulating joints, making the track circuit relatively immune
from stray currents. Not only does a sectionalized rail pick up much less
ground current than a continuous one, but also, under normal conditions,
the currents in the two rails, the resistances of the rails, and the IR drops in
the two rails are approximately equal; therefore, the net voltage drop around
the track circuit caused by stray current is very small. The balance can be
upset by inequality of the leakage resistances of the two rails to the ties and
ballast or, even more, by a broken rail. Then, if the stray current were great
enough and in the right direction, it could prevent a relay already energized
from dropping out and thus cause a signal to give an unsafe false indication
of "clear."
440
GROUND RETURN
SERVICES
441
h
I
a
2izr r
Idh
2nr2
Lh
2n(h2
+ x 2)
(71)
Compass
Hdc cos X
(b)
Compass Error21,35,62
The magnetic field of direct current in a submarine cable can deflect ships'
magnetic compasses from their normal direction. This effect varies with the
current in the cable, the depth of the cable below the surface, the azimuth of
the cable, the horizontal distance of the ship from a position directly over the
cable, and the strength of the horizontal component of the earth's natural
magnetic field. It is greatest when the cable runs parallel to the undisturbed
direction of the compass that is, approximately north and south and when
the ship is directly over the cable. The compass error is
&=
tan-1 cosA
(70)
approximately
* H varies with position, especially with latitude. Inthe temperate zones it is
0.2 Oe= 16 A/m.
442
GROUND RETURN
the other, the cancellation of their fields is almost negligible. Instead, the
compass of a ship sailing over the two cables in succession shows almost the
full deflection due to one cable, first in one direction and then in the other.
Ground currents are so widely dispersed that the effect of a ground-return
circuit on a compass is essentially that of the cable or cables alone.
The question of compass error was raised by the British Admiralty in con
nection with the Cross Channel scheme because of the dense ship traffic in
the Strait of Dover. Preliminary tests were made in the Solent.62 Two ships
were used in the tests, one having a nonmetallic hull and a compass near the
water level and the other having a steel hull and a compass well above the
water level. Each ship made eight runs across a submarine cable conducting
100 or 200 A in direction nearly magnetic north-south, with sea return.
Compass errors ranging from 3 to 11 were both calculated and measured
with good agreement (median difference 0.5 = 0.009 rad). Compass errors
measured on the steel-hulled ship were only 0.5 to 0.6 of those measured on
the ship with nonmetallic hull, the difference being attributable to the shield
ing effect of the steel hull and to the greater height of the compass above
level.
The error of a ship's compass directly over a cable of the Cross Channel
scheme at 46-m (25-fathom) depth and conducting 800 A was calculated
as 7, which was deemed intolerably great.35 Consequently ground-return
operation was not permitted, and the two cables were laid only 3 m apart so
that their fields would effectively cancel each other.
In the Gotland scheme, in which one submarine cable carries 200 A with
sea return, the maximum compass error was calculated as 0.3, and this value
was confirmed by a measurement made on the ice in the winter of 1955. The
small deflection results from the direction of the cable, which does not deviate
by more than 11 from east-west, and from the smaller current. The fact that
the cable sheath and armor carry about 15% of the return current is probably
of little importance as regards compass error.
No problem of compass error has been reported on three other schemes
using submarine cable and sea return: Konti-Skan, crossing the Kattegat,
with its dense ship traffic; the New Zealand scheme, crossing Cook Strait;
and the Sardinian scheme, crossing the Tyrhennian Sea.
Although the compass error may be considerable directly over a dc cable,
it decreases rapidly with horizontal distance and, thus, is significant only in a
narrow strip of water along the route of the cable. Ships habitually plying a
nearby course would be aware of the error, while most other ships in the
strait would be sailing almost perpendicularly to the cable and thus would be
subject to significant cable error for only a very short time. Besides, nowadays
most big ships use gyro compasses normally with magnetic compasses only
water
as backup.
9-6
443
Because the design of land electrodes differs considerably from the design
of sea or shore electrodes, the two categories are discussed separately. First,
however, we discuss those considerations which are common to both
categories.
Requirements
1 Low resistance
2. Adequate current-carrying capacity
3. Low maintenance cost
4. Accessibility
5. Limitation of damage to other facilities
6. Safety for persons, livestock, and fish
444
9-7
GROUND RETURN
and in bipolar schemes in which either pole may be temporarily out of service,
the electrode at each end can function sometimes as anode and at other times
as cathode; therefore, both electrodes must be designed as anodes.
Accessibility. Electrodes must be accessible for inspection and main
tenance. Most electrodes are made of several separate units, arranged so that
the current to each unit can be measured and so that any unit can be discon
nected and, when necessary, replaced. Some shore electrodes have their
elements suspended in tubular wells. The electrode elements can be pulled up
for inspection and replacement.
Location. Electrodes should be located far enough away from other buried
or immersed metal structures so that electrolytic corrosion attributable to the
dc line is negligible in comparison with other causes of corrosion. In particu
lar, the electrodes should not be close to the terminal substation of the dc
line itself because of the hazards of corrosion and of transformer saturation.
Usually the electrodes are placed from 8 to 50 km (5 to 30 mi) from the
terminals and are connected to the converters through electrode lines lightly
insulated from ground. Depending on local circumstances, the electrode line
may be carried on the same towers as the main dc line, perhaps serving the
additional function of overhead shield wire, or on separate poles or towers,
or in cable, on a separate route. For avoiding corrosion of tower footings near
the electrode, these may be insulated from ground or wooden poles may be
used.
Safety. The maximum surface gradient near a land electrode should be
limited to about 5 V/m so that a walking or standing man cannot receive an
annoying electric shock. If the land is to be used for grazing, the same limit
ing gradient (5 V/m) should be acceptable. The maximum potential difference
between two feet is called the step voltage. This matter is discussed in more
detail in "Safety," page 455.
Attention must be given also to touch voltage between a person's feet and
such metal objects as fences. Fences enclosing an electrode site can be
designed to follow equipotential lines, at least approximately. Also, these
fences and others in the vicinity can be sectionalized by insulators, grounded
at many points, or made of nonconducting material.
In the vicinity of a sea or shore electrode, areas of high voltage gradient
must be enclosed or screened to exclude fish and unauthorized persons.
Limitation of voltage gradients near a dc electrode is easier than near an
ac electrode, because dc short-circuit currents are effectively limited to a
smaller multiple of rated current. On the other hand, those gradients caused
by load current may persist for a long time.
445
Duties
Three different kinds of duty for dc ground electrodes are listed below in
order of decreasing severity :
1. Continuous duty
2. Intermittent duty ground return used only during emergencies for a
limited time
3. No load current.
The requirements of the electrodes of different categories differ chiefly
General
The factors to be considered in the design of a land electrode are the choice
of site, the choice of electrode materials, the form or layout of the electrode,
the computation of its dimensions for the avoidance of overheating of the
soil, the computation of its resistance and time constant, and the assurance
of low step and touch voltages.
446
GROUND RETURN
9-7
447
quate area of land having soil of low resistivity and (b) far enough from the
converter station and other substations and from densely populated areas,
power plants, pipelines, railways, etc., so that problems of electrolytic cor
from a plain iron anode, and, besides, the cost of coke per unit mass is much
less than that of iron.
If carbon or graphite rods are used instead of iron rods for distributing
current to the coke, the loss of material from the anode by electrolysis is
almost entirely eliminated.
The method of construction is to dig a trench (Figure 28), to fill it with coke
Surface of ground
Choice of Site
The location of the electrode should be chosen (a) where there is an ade
It is assumed that the electrode is an anode, at least part of the time. The
principal problem is to find a cheap material that does not suffer excessive
electrolytic corrosion.
Although iron, especially scrap iron, is cheap, it does not last long enough
when carrying large direct currents (1000 A or more). The theoretical loss of
mass from an anode carrying 1000 A (time-average) is 9000 kg (9 tonnes) per
yr, which is clearly an excessive amount. The actual loss from an iron anode
buried in the soil may range from 0.2 to 1.4 times the theoretical loss, depend
ing on the nature of the soil, but even the lower figure is still excessive.
Loss of graphite is said to be about 0.6 that of iron and hence is also
excessive.
Experiments carried out in connection with the Kashira-Moscow dc line31
showed that if iron anodes, instead of being buried in direct contact with the
soil, were buried in a pit filled with crushed coke, the loss of iron was reduced
by a factor ranging from 68 to 21,000, depending on the time of year and,
presumably, on the moisture in the soil. Using the more optimistic factor, the
annual loss of iron at 1000 A dc is reduced to 0.4 kg, or, with the more
pessimistic factor, to 140 kg, which is an acceptable amount.
The explanation of the great reduction in corrosion of the iron is that
conduction from iron to coke is mostly electronic rather than ionic. If
moisture reaches the iron, however, a small part of the conduction from the
iron is ionic, resulting in some electrolytic corrosion. This supposition is
confirmed by observations made during the same experiments, that the lowest
rate of wastage of iron occurred when the ground was frozen, and the highest
rate when the soil was wettest.
Fromthe standpoint of loss of iron, ideal operation is with a current density
at the surface of the iron that keeps the surface dry. In the cited experiments,
this density was found to be 15 to 25 A/m2. The best coke to use has a low
sulphur content.
In an electrode so constructed, the coke becomes the true electrode, and
the role of the iron is merely to distribute current to the coke. There is some
wastage of coke, but less mass of coke is lost per kiloampere-year than of iron
Backfill
Distribution
conductor
Coke
Fig. 28. Cross section of land electrode made with fill of granulated coke in the bottom
of a trench: depth of burial h; sectional area ax b.
to half the final depth, to lay the distributing conductors on this layer of
coke, to put the remaining half of the coke on top of the lower half, and
finally to backfill the remaining space in the trench to the original grade. At
The layout of an electrode in plan view may take various forms, several of
which are illustrated in Figure 29.
The ring form, if located in uniform soil, gives the most uniform current
into the ground per unit length of trench and thus uses the materials to best
448
9-7
GROUND RETURN
Star
(a)
449
12
N(n)
0.53
1.45
3.42
5.5
10.0
Notation
Re = electrode-ground resistance, Q.
(b)
Fig. 29. Plan views of several forms of land electrode: (a) straight; (b) ring; (c) sixpointed star; (d) branched star.
advantage. It requires, however, a large tract of land with uniform soil and is
rather inflexible.
p ground resistivity, Q m
I= total length of conductors, m
D diameter of ring, m
b = yjdh, m
d = diameter of conductor, m
h = depth of burial of center of conductor, m
In x = natural logarithms of x = 2.303 log10 x
The straight line and the star have current densities from coke to soil that
increase at the ends of the arms. The length and shape of the arms can be
varied as required by local conditions.
In the star-shaped electrodes the decrease in electrode-to-ground conduc
tance for each additional arm decreases with an increasing number of arms.
Rusck40 suggested the six-armed star as a good compromise between con
ductance per unit area of site and conductance per unit volume of electrode
material. A six-pointed star of distorted form was used at Benmore.54'57
A ring was used at Rice Flats.78
A branched star appears to use the area of site efficiently.
Electrode-ground Resistance3,6'16111,18
Formulas are given below for the resistance, with respect to remote earth,
161,1
All these electrodes are assumed to be
of buried electrodes of several forms.
made of conductors having a constant circular cross section and buried to
uniform depth in soil of uniform resistivity.
RiY/a
coth ( ~
(75)
..
Straight Rod(Cylinder)
R.=
Rin = R,11
sAe)
when
nl
(72)
Ring (Torus)
4D
ll Tib
It
~b
(73)
account.
450
9-7
GROUND RETURN
Resistivity of Soil
Since the electrode-to-ground resistance and the voltage gradients near any
given electrode are both proportional to soil resistivity, low resistivity (high
conductivity) is important.
The conductivity of the soil is attributable almost entirely to the moisture
in it and to the minerals dissolved in this moisture. Therefore, an adequate
supply of moisture to the soil in the vicinity of the electrode must be assured.
Accordingly, land electrodes can be classified into two kinds irrigated
and unirrigated. The soil around an irrigated electrode is kept moist by
pumping water to it; the unirrigated electrode relies on the natural supply
of water. The land electrodes constructed to date have been unirrigated. Of
course, irrigation could be added later if it were found necessary.
The retention of moisture is affected by gravity, capillarity, hygroscopy,
evaporation, and electroosmosis. Gravity affects the movement of free water,
which is that in excess of 15 to 20% of the dry weight of the soil. Capillarity
can raise moisture to a level higher than gravity does by an amount that
increases as the sizes of interstices decreases. It is more effective in clay than
in sand. Hygroscopic water is that held by chemical attraction. Evaporation
depends on soil temperature, air humidity, and surface covering. Plants
increase the evaporation; paving retards it.
An electrode tends to dry the soil in two ways : first, by heating it, thereby
increasing evaporation, and, second, around an anode, by electroosmosis.
Electroosmosis moves water in the direction of the current. It has been used
11,42,56
but it is
to advantage in stabilizing the soil in cuts or excavations,
detrimental around an anode. The tendency of an electrode to dry the soil must
be offset by the effects of rainfall, gravity, and capillarity in restoring soil
moisture.
The current density at which electroosmosis begins to have a marked
detrimental effect depends on the character of the soil and on its supply of
water. An idea of the order of magnitude of this current density is provided
by two sources. Russian investigators31 found the critical current density to
be 9.8 A/m2 and recommended densities of 1.5 to 2.0 A/m2. The New
Zealand Electricity Department found that at the electrode site near Benmore54 the critical current density was 1.0 A/m2. An accurate value of critical
electroosmotic current density need not be known, because usually a lower
current density is imposed by the requirement of thermal stability, discussed
below.
Compaction of the soil by methods well known in the construction industry
aids in retention of moisture by decreasing evaporation and increasing
capillarity. The pressures usually applied are in the range of 200 to 300
lbf/in2 (105 to 1.5 x 105 N/m2).
451
equal temperature.
452
GROUND RETURN
9-7
The relation between potential and temperature is found from the prin
ciple of conservation of energy. As one goes farther from the electrode, the
flow of electric energy through the equipotential surfaces decreases and the
flow of heat energy increases correspondingly, the sum of the two being
constant and equal to the electric power leaving the electrode, which is
VeI watts.
On a given equipotential surface at potential V with respect to remote
earth, the electric power (rate of flow of electric energy) is VI watts. On the
same surface, the heat-energy flow density is the product of thermal conduct
ivity X and temperature gradient or X dBjdr watts per square metre, where 6
is the temperature and r is the distance along the field vector. The heat-energy
flow over the entire surface is
, , de
X \ dA
J 4
dr
(76)
watts
where A is the area (square metres) of the equipotential surface. The total
energy flow through the surface is equal to the electrode power, or
dr
dA
+ VI = VJ
(77)
watts
453
B-=wP
=
V2JpBe
(84)
This relation is independent of the shapes of the electrode and of the surface
of the ground. It does rest, however, on the assumption of uniformity of both
electrical and thermal resistivities.
Let typical numerical values be substituted in Eq. (84) (Rusck40):
Maximum soil temperature in summer in a temperate climate, 25C
Boiling point, 100C
Maximum permissible temperature rise, 9e = 100 25 = 75C
Thermal conductivity of soil, X = 1 W/m C.
By rearrangement
X I
dA = (Ve - V)I
(78)
watts
Now
/=| JdA = -\
PJa
P J a dr
dA
(79)
so that
"\A7rdA (V--V)l\Jr-dA
=
(80)
Ve = ReId = V2 x
1 x 75 x
Vp S 12Vp
volts
(85)
for example
if p = 10 Q m,
ifp = 100 a -m,
if p = 1000 ft m,
Ve = 38 V
Ve = 120 V
Ve = 380 V
Thus, in order to avoid overheating of the soil near the electrode under
continuous ground-return operation, the voltage of the electrode is restricted
to moderate values.
Let us suppose, further, that the ground current is' Id= 1.2 kA. Then the
electrode resistance and the power loss in the soil are as follows:
xJ=~{Ve~V)T
dr
dr p
Re (Q).
(81)
10
0.03
100
1000
0.10
0.31
I&Re (kW)
46
150
460
(82)
Ordinarily, an electrode site is chosen where the ground resistivity is not much
454
9-7
GROUND RETURN
/ R
(2
\1/2
volts
(85a)
The temperature does not rise exponentially; hence, the soil and electrode
have no true time constant. We may define their time constant T, however,
as the time required to reach the final steady-state temperature Bf if the tem
perature increased linearly at its initial rate.
Initially the soil and the electrode are at a uniform temperature: there is no
temperature gradient and hence no flow of heat. The electric energy converted
into heat in any element of volume serves to raise the temperature of the
element. Therefore, the initial rate of the rise of temperature of the element
is the rate of heat production divided by the heat capacitivity y. Heat capacity
is heat energy per unit increase of temperature, and heat capacitivity is
heat capacity per unit volume. It differs from specific heat, which is heat
capacity per unit mass, by the density of the substance.
The rate of heat production is the rate of dissipation of electric energy,
which, in unit volume, is pJ2 watt per cubic metre. The initial rate of rise of
temperature is
dd
pJ2
T
*
defined, is
T=
(86)
yQf
jj[
(87)
T=US) =2$j)
I'
l.
seconds
(88)
For the last equation to be true, J must be interpreted as the average current
density at the electrode surface. Actually, the initial rate of the rise of temperature varies from one part of the surface to another in proportion to the
square of the current density, but the flow of heat inside the electrode tends
to equalize the temperature.
The expression in parentheses has the dimension of length. For a spherical
electrode, it is the radius.6 For the electrodes of Figure 29, page 448, it is
about six to nine times the diameter d.
Typical values of soil parameters are
2 = 1 to 3 W/m C
y = 1.3 x 106 J/m3 C for sand and clay, corresponding to specific
heat 0.2 cal/g and specific gravity 1.6 g/cm3
p = 10 to 100 Q m
Land electrodes of the size used for high-current dc transmission have time
constants of the order of 100 days.
Safety4'10'22,25,27'29,34'38
I
I
f,
J
0r
455
greater than 100 Q m. With low soil resistivities, the surface area of the
electrode is determined by temperature rise, but with high resistivities it may
be determined by acceptable power loss.
A more recent analysis83 replaces Eq. (84) by the following:
Ve
f
I
I
|
I
|
f
1
1
|
456
GROUND RETURN
9-7
and absence of freezing, and (c) decreasing the resistance from electrode to
remote earth. Usually, the depth of burial required for safe surface gradient
is no greater than, or only a little greater than, the depth required for other
purposes.
Tolerable Body Currents. The current through the body is more significant
than the voltage across the body. Body current has the following effects,
according to the magnitude of the current. Listed in order of increasing
current, they are (a) perception, (b) annoyance, (c) discomfort and strong
muscular contractions, (d) cessation of breathing, (e) ventricular fibrillation
(irregular heart action), and (f) burns. Danger begins with the third item;
the victim, grasping a live conductor, is unable to let go of it. He can be
rescued by deneergizing the circuit or by pulling him from it. Cessation of
breathing can usually be remedied by artificial respiration. Ventricular
fibrillation is very dangerous. The only known remedy is a controlled countershock of short duration within a few minutes. The equipment, and skill for
administering such a countershock are almost never available.
Considerable experimentation has been done on let-go currents with
human subjects25 and on the threshold of ventricular fibrillation with
animals.4,34 Most of this work has been done with 60-Hz current, but enough
has been done with direct current to show that the dc let-go and fibrillation
currents are five times as great as the corresponding 60-Hz currents. Let us
assume that this same ratio applies to current values causing perception and
annoyance.
457
The theoretical 60-Hz rms threshood fibrillating current for normal men is
I
= 165/ 1 mA, where t is duration of shock in seconds. The corresponding
%/
dc value is
825/r mA.
Design Criteria. The body current of men and livestock should be below
the let-go level. Suitable criteria might be (a) the annoyance level for standing
or walking men and (b) half the danger (let-go) level for prone men or
standing animals, both to be with good contact to the soil (wet hands and
feet) at the worst place in the field with rated direct current in the electrode.
With a. fault on the dc line, the ground current might be several times its
rated value but only for a very short time (fraction of a second). Men or
animals could readily move a short distance from an uncomfortable place to a
comfortable one.
The body currents corresponding to these criteria are (a) 5 mA for a
standing man, (b) 22.5 mA for a prone man, and (c) for a standing animal,
22.5 mA multiplied by the ratio of its body weight to 70 kg.
Body and Contact Resistances. It is clear that the current shunted through
a body has no perceptible effect on the electrode current nor on any part of the
electric fieldof the anode except very close to the points of contact of the body
to ground. Therefore the current through the body is equal to the opencircuit voltage between the points of contact divided by the sum of the body
resistance and the two contact resistances in series.
Various authorities have suggested values of human-body resistance
ranging from 500 to 3000 Q. Obviously, it depends very much on the condi
tion of the skin. Dalziel22 measured body resistance to direct current from
one hand, wet with salt water and grasping a metal electrode, to both feet
in 3/4 in (2 cm) of salt water; the minimum value of nine subjects was 1135 Q.
With one arm immersed in salt water to the elbow and both legs in 10 in (25
cm) of salt water, resistances were 400 to 600 Q. For our purpose, a value of
1000 D is conservative, perhaps somewhat pessimistic.
Contact resistances between a bare foot and moist soil were found to be
equivalent to that between a circular metal plate of 16 cm diameter and the
soil.6,29 This value is 3ps, where ps is the resistivity, in ohm-metres, of the
surface soil.18 Thus the resistance of a standing or walking man (foot to
foot) is 1000 + 6ps Q. The resistance of a standing man grasping a metal
object with one hand is 1000 + 1.5pf, on the assumption that the resistance
of the contact between hand and metal object is negligible. This situation is
not likely to arise near a land electrode unless there are metal fence posts or
wires in the area of high voltage gradient. The resistance between the hands
and feet of a prone man is 1000 + 3ps, on the assumption that the contact
458
9-7
GROUND RETURN
459
resistances from hand to soil are equal to that from foot to soil. The resis
tance between front and hind feet of a standing or walking animal can be
calculated in like manner if the body and foot-to-soil resistances are known.
If we assume that the average density and resistivity of an animal body
is equal to those of the human body, the body resistance is inversely pro
portional to the weight. Then the body resistance of a horse or cow weigh
ing 1000 to 1400 lb (400 to 600 kg) is about (70/500)1000 = 140 H. The
hoof-to-soil contact resistance is inversely proportional to the contact area
and is about the same as that for a man's foot. Thus the resistance between
fore and hind feet of a cow or horse is about 140 + 3ps Q.
Table 6 shows data from which the tolerable step voltage and gradient are
computed according to the criteria adopted above. It is clear that, under these
criteria, the maximum permissible gradient is 5 -t- 0.03ps V/m, determined by
annoyance to a man walking with wet feet and 1-m steps near the electrode.
It is proportional to the ground current per unit length, ///, and to the resis
tivity and is inversely proportional to the depth of burial h to the center of
the cylinder. This method of calculation is accurate enough for electrodes of
finite length much greater than the diameter and for square or rectangular
cross section as well as round. It is adequate also for star or ring layouts.
The surface gradients near the ends of a cylinder of finite length or near
the ends of the arms of a star, however, are somewhat greater than that
calculated by Eq. (90).
Figure 30 shows how the potential and its gradient vary over a small range
of horizontal distance from the electrode. It is apparent that the gradient
varies considerably over the length of a step if the step is not much smaller
than the depth of burial. It would be more correct to compute the maximum
difference of potential between two points separated by the length of step than
to compute the maximum gradient and multiply it by the length of step.
The latter procedure, however, is simpler, errs on the pessimistic side, and is
just as accurate as are the tolerable values of body current.
Man
Walking
Prone
Cow or Horse
Standing
70
70
500
1000
1000
3ps
140
6ps
Annoyance
Pain
Pain
+ 0.03/>,
22
22 + 0.07ps
160
22 + 0.48ps
+ 0.03/>s
11 +0.035ps
11 +0.24p,
11.0
11
35
59
3Ps
Tolerable
current (mA)
Voltage (V)
Step (m)
5
5
Gradient (V/m)
General
Ps= 0
p., = 100 n m
Ps = 200 n m
5.0
8.0
11.0
14.5
18.0
pi
pi x plx
=
nrl r nrl
cos 6 =
(pl\
nrl
\ zr lix +h
It is easy to show that the maximum gradient occurs at x h and is
pl_
max = 77
2nlh
(89)
(90)
Rusck published a proposed design for a land electrode for 1200 A into a
soil of resistivity 100 fl m. He chose to use a coke fill with square cross
section 0.5 by 0.5 m in ditches 1.5 m deep and to arrange the ditches in the
form of a six-pointed star. The average depth of burial of the center of the
coke fill is then 1.25 m. He chose to distribute the current to the coke by a
round iron bar.
The electrode-ground resistance already determined by the heating limit
(table on page 453) is 0.1 O.
460
9-7
GROUND RETURN
1.5
461
r*i
1.0
0.5
electrode
-0.5
-1.0
1.5
-4
1
-2
-1
x/h
Fig. 30. Surface potential V and gradient in the vicinity of a long, cylindrical, hori
zontal buried electrode : unit of V is its value at xjh= 3 ; unit of is its maximum value.
The total length / of all the electrode arms is found by trial substitution for
/ into Eq. (74) for n = 6 to be 3500 m. Hence the length of each of the six
arms is 3500/6 = 585 m. The area of the interface between coke and soil is
3500 x (4 x 0.5) = 7000 m2, and the average current density at this surface is
1200/7000 = 0.17 A/m2, which is well below the permissible limit of 1 or
2 A/m2 determined by electroosmosis.
A current density of 0.5 A/mm2 was chosen for the iron distribution con
ductor. Since the current per arm is 1200/6 = 200 A, the required crosssectional area is 200/0.5 400 mm2 = 4 cm2, corresponding to that of a
square bar 2 by 2 cm. The resistance of such a conductor 585 m long is about
0. 19 2. Assuming the current to diminish linearly from 200 A at the input end
to zero at the extremity, the power loss in the iron conductors is 6 x 0.19
x (200)2/3 = 15 x 103 W = 15 kW. The power loss in the ground is
120 V x 1200 A = 1.44 x 10s W = 144 kW, and the sum of these losses is
15 + 144 = 169 kW. The power loss in the coke is probably much smaller
than that in the iron.
The volume of iron is (2 x 10-2)2 x 3500 = 1.4 m3, and, with a density of
8 tonnes/m3, its mass is 11.2 tonnes. If we assume the loss of iron by elec
trolysis to be 0.015 of the theoretical value, it would amount to 160 kg/yr.
On the assumption that the iron conductors remain serviceable until half of
This was the first land electrode built except those for the VolgogradDonbass line, about which littlehad been published. Its rated current is 1.2 kA.
The site was chosen in a valley situated 4.75 mi (7.6 km) south of Benmore
power plant and having a tertiary deposit of clay, silt, and sand. Tests made
on the soil at the site showed that its effective resistivity was 61.5 2 m, that
its thermal conductivity varied from 1 to 3W/m C, with an average value of
2 W/m C, that its heat capacitivity was of the order of 2 x 106 J/m3 C,
and that its maximum ambient temperature was 25G. The effective resistivity
was taken as that of a uniform earth, which would give the same resistance
from a large buried ring electrode to ground as it would if buried in the upper
layer, of resistivity 33 2 m and thickness 36.5 ft (11.1 m) resting on a lower
layer of 100 2 m and at least 500 ft (150 m) thick. Several test holes were
bored at depths down to 70 ft (21 m), and soil samples were analyzed for
grain sizes. Moisture content of the soil varied from 4 to 21% of dry weight,
with a mean value over the site of 8%.
A test electrode buried in a straight trench 50 ft (15 m) long and having the
same cross section as that proposed for the main electrode was located in
soil similar to that on the site of the main electrode. Its resistance, calculated
from Eq. (72) with p = 68 2 m, was 3.5 2. When the test electrode was loaded
with the same current per unit length as that proposed for the main electrode,
its resistance was initially 3.2 2 and dropped slowly to 2.8 2. The temperature
rise of the soil near the electrode was found to be about 60C. Time constant
of the test electrode was about four days. When the current density was raised
to 1 A/m2 into the soil, electroosmosis was indicated by the behavior of
measured resistance and temperature.
Because it was difficult to fit a ring electrode onto the site, a six-pointed
star was adopted. As shown in Figure 31, the arms are not straight or of
equal lengths. When rock was encountered on one arm, this arm was ter
minated and another arm was extended. The junction point of the star was
placed in an area of lower than average soil resistivity.
The maximum resistance for thermal stability was computed by Eq. (84) as
Re
2pX9
T
136
= = 0.1132
1200
462
9-7
GROUND RETURN
463
Life Inspection after four years' operation, during which the electrodes
conducted 7.4 x 106 Ah as cathode and 1.6 x 106 Ah as anode, showed
evenly distributed pitting with very little reduction in diameter.
Design of Rice Flats Electrode78
Electrode line
I
0
0.5
'
1.0 km
Fig. 31. Plan of Benmore ground electrode as built. Ground surface equipotential lines
in percent of electrode voltage to remote earth. (Adapted from Refs. 57 and 81 by per
mission of the I.E.E.E.)
Use of Eq. (74) with n = 6 showed that each arm must have a length of at
least 400 yd (365 m) if the cross section of the coke were 20 by 20 in (51 by
51 cm) and if the depth of burial were 5 ft (1.5 m).
The area of the coke-soil interface was computed as 4500 m2, and the
corresponding current density as 0.27 A/m2. This is well below the critical
density (1 A/m2) found for electroosmotic drying.
The time constant was computed from Eq. (88) as 100 days.
Because the electrode would occupy an area of 200 acres (80 hectares),
fencing it would be expensive. It was decided to bury the electrode deeply
enough so that surface voltage gradients would be safe and to permit the land
to be used for grazing. A step voltage of 20 V in 5 ft (13 V/ml was taken as
safe for animals. The depth of burial of 5 ft (1.5 m) was foundao be adequate.
The maximum surface gradient, computed from Eq. (90) yvith ps = 33. 12 m
and with end effects neglected, is (33 x 1200)/(2je x 365 -x 6 x 1.5) = 2 V/m.
The calculated resistance of the electrode was 0. 11 II; the measured resis
tance varied between 0.29 and 0.32 12. The discrepancy between these two
values may be attributable to departure from the intended star configuration
and to a greater effect of the lower, high-resistivity layer on the effective
resistivity. The highest measured voltage gradient was 6 V in 4 ft (5 V/m).54
The cost of the electrode, including the cost of investigation, was about
15,600 ($44,000).
The chosen location is on the route of the dc line 6.6 mi (10.6 km) south of
Celilo, the southern terminal of the dc Pacific intertie. It is reasonably
distant from other plants that might be affected by corrosion, being 18 mi
(29 km) from one major gas pipeline and 35 mi (56 km) from another, and
7.5 mi (12 km) from the city of The Dalles, Oregon, with its underground gas
and water distribution systems.
The soil is a fine wind-deposited silt (loess) up to 50 ft (15 m) thick, resting
on basalt. Annual precipitation is only 16 in (42 cm), but the fine texture of the
soil permits it to retain moisture. Measurements of soil moisture made at
various points during the course of a year show that it varies from 5 to 14%
of the dry weight. Tests made with small electrodes over 18 months showed
that satisfactory operation could be achieved with this rather dry soil if the
current density at the electrode-soil interface were kept somewhat below
1 A/m2. Therefore, the expense of providing irrigation or of adopting a
moister site 20 mi (32 km) farther away was judged to be unjustified. The
chosen site is used for dry wheat farming.
Measurements of soil resistivity on the site showed 70 12 m at 5-ft (1.5-m)
depth, down to 36 Q m at 40 to 50 ft (12 to 15 m), up again to 50 2 m at
300 ft (92 m) and on up to 100 12 m at 700 to 1000 ft (210 to 300 m). The
design value was taken as 70 12 m. The thermal conductivity of the soil was
measured as 0.87 to 1.13 W/m C. Soil temperature at depths of 3.5 to 4.5 ft
(1.1 to 1.4 m) was measured at 15C or less in May, 1964, and as 16C in
August, 1965. The design value was taken as 25C, giving an allowable rise
of 75C. The specific heat of the soil was assumed to be 0.19 cal/g.
The following requirements were set as a basis for design :
Ve = V2 x
75 x 1.0 x 70 = 102 V
464
9-8
GROUND RETURN
465
2nTh
70 x 1800 .
= 4 Vm
2it x 1IOOtt x 1.5
= 3.04 x
General
Most of the dc transmission schemes built before 1970 have sea or shore
electrodes at one or both ends. These schemes include Gotland, KontiSkan, New Zealand, Sardinia, Vancouver Island, and the Pacific NorthwestSouthwest Intertie.
For the purpose of this discussion a shore electrode is defined as either an
electrode on the seashore near the sea or one in seawater near the shore, as in
a cove or lagoon. Only one true sea electrode has been built to date: this is in
the Pacific Ocean 6000 ft (1.8 km) off the shore near Santa Monica, Cali
fornia.
Advantages. The principal advantages of a sea or shore electrode over a
land electrode are shown by the analysis in Section 9-2, page 402.
=6-5*10-W
The current is distributed to the coke in the trench through a series of 1067
high-silicon cast-iron cylinders each 5 ft (1.53 m) long by 1.5 in (3.8 cm)
diameter. Groups of five cylinders were fabricated into strings with a 5-ft
insulated cable between adjacent cylinders. A 5-ft flexible lead at each end of
each string was connected to No. 4/0 AWG (108 mm2) copper distribution
cable laid in the same trench in a large closed circle fed at three equally
spaced points through buried insulated cables radiating from a point inside
the circle where the electrode line terminates at a steel tower, which also
carries the main dc line.
Cost. The total cost of the Rice Flats electrode, including surveys, tests,
design, land easements, materials, labor, and overhead, was about $300,000.
Problems. Three problems found with sea or shore electrodes but not with
land electrodes are the following:
1. Additional corrosiveness of seawater, especially at the anode.
2. The necessity for the protection of fish.
3. The protection of the electrode installation from mechanical damage
caused by waves, ice, and driftwood during storms, especially at high tides.
466
9-8
GROUND RETURN
gradient as low as 0.17 V/in" (6.7 V/m). Elder and Whitney79 state: "Some
observers note that it is the head-to-tail voltage rather than gradient that
affects fish. Voltages in the order of 1 to 5 V have been observed to make a
fish unconscious." Quinn and Howard58 state that a gradient of 2.5 V/m
causes discomfort to a man in the water, but that 1.25 V/m can be regarded as
safe for man and for large fish. Areas of high voltage gradient can be enclosed
by screens or perforated tubes or boxes, or they can be separated from the
main sea by a rock breakwater.
Protection against wave action may be obtained by covering the electrodes
with large rocks, placing them in a cove separated from the main sea by a
breakwater, or putting them on shore at a safe distance from the sea. Another
method is to place the electrodes on the sea bottom where the water is so deep
that wave forces are weak.
After the pioneering Gotland scheme, which began operation in 1954
with 200-A sea return, no more dc sea or shore electrodes were built until
1965, and these were for much greater currents (1.0 and 1.2 kA).
This scheme had a rated current of 1000 A and was initially monopolar
with sea return. It was decided that the current in the sea cables should always
flow westward, from Sweden to Denmark. Consequently, the electrode on
the Danish side is always the anode and that on the Swedish side always the
cathode. The cathode is very simple, consisting of about 300 m of bare
copper cable of 600 mm2 cross section laid in a ring on the sea bed in 10 to
15 m of water 3 km offshore. The anode, however, posed some interesting
design problems.
The site was chosen on the shore of the Kattegat about 8 km south of the
cable terminal at Stensntes. There a short, lightly insulated branch was built
from the main dc line to the coast at a fairly isolated spot 1 km northeast of
the village of Sora, remote from urban areas, harbors, cables, etc. The site
comprises 1.5 acres (0.61 hectare) of marshy flat land with coast line of 137 m
and width of 46 m. The sea is very shallow far from the land. At low tide the
water recedes as much as 400 m. It was decided to put the electrode on the
shore rather than in the sea because of the serious risk of damage from ice on
the sea in hard winters.
The subsoil consists of 1.4 m of sand, with blue clay underneath to a depth
of at least 44 m.
The first plan considered was to excavate a basin and suspend the anode
elements in the water of this basin from wooden frames. Because of the wide
areas of shallow water, the natural supply of new seawater to such a pool
would be inadequate. Indeed, the flow of underground fresh water into the
467
pool would make the salinity of the water in the pool considerably smaller
than that of seawater. Pumping of seawater into the pool would be necessary.
Pumping being necessary in any case, a cheaper alternative plan was
chosen. This plan was to suspend each anode element in a separate small well.
Each well would be lined with a pipe of material that would conduct current
but would keep the soil from filling the well.
Tests were made to find suitable materials for the anode element and the
well casing, both of whichwould have to be practically immune from deteriora
tion caused by the products of electrolysis of salt water (principally chlorine
and oxygen gases).
The anode elements are impregnated graphite rods 2.44 m long and
0.100 m in diameter and weigh 38 kg, including 5 m of cable. Tests showed
that when such elements were operated at 50 A for 2 months in pumped
seawater, the loss of material was only 5 g per 1000 Ah. Without pumping,
however, in hot water with increased concentration of chlorine and oxygen,
the rate of loss was 10 times as great, or 50 g per 1000 Ah.
For well casings hard polyvinyl chloride (PVC) pipes were chosen; PVC
pipes had been used by well drillers for some years for the sand filters placed
at the bottom of a well and from which the water is pumped out. These pipes
were available with an inside diameter of 0.230 m, wall thickness of 0.010 m,
and length of 6 m, with the number and size of slits required for passage of
current and of water. The slits cover 16% of the wall in a length of 2.8 m,
the total area of slits per pipe being 0.35 m2.
The suitability of anode-element and casing materials was ascertained by
two test wells using a somewhat smaller anode (1.50 by 0.075 m) and a
shorter casing (3 m). The test anodes could carry 50 to 60 A with water flow
as little as 0.2 liter per sec per well with very little temperature rise and with
no problems. The measuredresistance was 0.8 Q for one anode. Without water
flow, the temperature in the test wells rose 40C, and the corrosion of the
graphite anodes was increased tenfold, as already noted; but the PVC pipe
showed only a slight discoloration but no loss of mechanical strength in 8 to
12 months.
The resistance of the full-scale anode was calculated by proportion to be
0.5 Q per well. It was decided to use 25 wells, spaced 5 m apart in a straight
line. If there were no interaction between elements, the resultant resistance
would be 0.5/25 = 0.02 O. The interaction, however, was expected to increase
the resistance by a factor 1.6, making it 0.032 O.
The number of anode elements was chosen to give an average current of
40 A per well. It was expected, however, that the current of the elements at or
near the ends of the row would be greater than that of the other elements.
This tendency was partially offset by using closer spacings between wells
near the ends of the row. The spacing between elements 1 and 2, likewise
468
9-8
GROUND RETURN
between 24 and 25, was made 3 m ; and that between 2 and 3 and 23 and 24,
4 m. Another factor tending to equalize the currents was the greater length
of the cables from a central point to the end elements.
The 25 wells were dug in a row along the shore about 20 m from normal
high-water line. A steel tube of 16-in (41 cm) diameter was drilled 5 m into
the ground. Then the 10-in (25.4-cm) PVC pipe was inserted, the steel tube
was withdrawn, and the space between the PVC pipe and the 16-in hole was
filled with filter sand.
A section of one well is shown in Figure 32, and a plan of the site, in
Figure 33.
Junction box
Hff
PVC-insulated cable'
25
mm2
(X
'
mm2 ACSR
25
(normal)
==<>*rMain pipe
Sand
salt water
'.'Outflow!
supply
ditch >
cm diam li t
-Graphite electrode
Current-permeable zone
Road
! 13
,-iU
Electrode wells
Pump
Symbols
Dam
..........
...................
.,.'i
10
....
10
20
30
11
1ni
40
ir
1 1
50
Metres
Fig. 33. Site plan of the Danish shore electrode of the Konti-Skan link. (From Ref. 65
by permission.)
469
Road
.
Meters
Fig. 32. Longitudinal section of an electrode well for the Konti-Skan link. (From Ref. 65
by permission.)
The PVC tube projects above the surface of the ground and carries a
junction box where the neoprene-insulated suspension cable is connected to a
PVC-insulated 25-mm2 copper cable that runs underground to the guyed
terminal tower of the electrode line. On that tower is a copper busbar 1 m
long to which are connected the 910-mm2 ACSR electrode-line conductor
and 25 insulated cables that are connected to the individual anode elements.
The lengths of these cables vary from 18 to 78 m.
'{
|
|
|i
470
GROUND RETURN
The highest surface voltage gradient is 2 V/m near the outermost wells. In
the sea at normal shore line, the greatest gradient is 0.3 V/m, which is too
small to attract fish. The area south of the electrode station is marshy land
used for grazing. The maximum gradient in this area is 0.2 V/m, which has
no visible effect on the cattle. Gradients were measured out to distances of
about 1 km for estimating the densities of currents leaving hypothetical
buried pipes and cables. At 600 m, currents entering such facilities would be
30 to 150 mA/m2, depending on wall thickness, diameter, and kind of metal,
and in the more extended anodic zone the calculated current densities are
only 5 to 20 mA/m2. There are, in fact, no such uninsulated pipes or cables
near the electrode station, except some short water pipes for domestic supply.
The electrode station is surrounded by a fence that approximately follows
equipotential lines. The fence is built of plastic-covered wire mesh. It is
divided into 10-m sections insulated one from another but not grounded.
The electrode station is inspected at intervals of 1 or 2 weeks. Occasionally
it has been necessary to remove sand or mud from the bottoms of some of
the wells. The electrode, when drawn out, has a maximum voltage of 15 V
to adjacent ground. The use of rubber gloves, though not necessary because
of shock, is advisable because the electrodes are slimy and dirty. It has
never been necessary to interrupt transmission for performing maintenance
on the electrode station, although the pump failed and had to be replaced.
The savings in investment, including capitalized losses, of using ground
return on Konti-Skan instead of metallic returns, has been estimated at
16.5 x 106 Danish Kronor (about U.S. $2.2 x 106).
Shore Electrode on North Island, New Zealand55,60
Site. A site was chosen on the west coast of the North Island 15.8 mi
(25.4 km) from the dc terminal substation at Haywards, 8 mi (13 km) from
the dc cable terminal at Oteranga Bay, and far enough from the city of
Wellington and its harbor. The site was named Te Hikowhenua, which means
in Maori "electricity of or in the earth." A preliminary calculation showed
that the leakage current density in. the armoring of the dc cables would be not
more than 4.7 mA/m2 when the electrode was carrying 1200 A (the rated
current of the dc link). This is less than the value (10 mA/m2) deemed
acceptable.
The site is a narrow strip of beach covered with a mixture of sand, shingle,
and large boulders some up to 6 ft (2 m) across to a depth of about 10 ft
(3 m) over bedrock, which also forms the sea bottom. The designers decided
to bury the electrodes in the beach and to protect them with the large boulders.
The seabed sloped about 1.5 (0.026 rad). The resistivity of the seawater
was measured as 0.23 fl-mat 15C, and the design value was taken as 0.25.
9-8
471
The resistivity of beach material saturated with seawater ranged from 0.3
ra for 4-in (10-cm) stones.
m for fine sand to 0.9
As in the design of the Danish electrode, one of the first questions was that
of suitable materials for the electrode units, for insulation of cables attached
to the electrode units, and for porous enclosures around the electrode units.
Tests were made on various materials.
Impregnated graphite, used by a pulp and paper manufacturer for elec
trolytic production of chlorine, was chosen for the electrode elements. A test
cell with a graphite electrode in a steel drum filled with seawater was used to
test insulating materials. The free chlorine content was kept at 1000 parts per
million higher than the concentration expected in the electrode cells
to serve for accelerated life tests. The following insulating materials were
found to be suitable : Hypalon, high-density polythene, and polytetrafluorethylene. Suitable sealing and insulating compounds were found to be phenolic
and epoxy resins. Epoxy resin mixed with a small amount of powdered
graphite provided good adhesion to graphite and to most plastics.
For cell enclosures, hard, dense, cast-concrete pipe, made with pozzolan
sulphate-resisting cement, and drilled with 3/16-in (5-mm) holes to produce
an effective porosity of 6%, was selected. Calcium chloride was used as a
strength accelerator.
Design. The following dimensions were chosen for the electrode units and
their enclosures :
Re
2id
p! In
+ p2 In + p3 In + 271T3
(91)
472
___
GROUND RETURN
9-8
Electrode line
terminal poles
Switch house
.it cell
,
Junction pit
4' G.E.W. pipes set
in concrete
58'- 7"
/ foi M fig|
jH-
f 9'-0"
12 -6
9'- 8"
=srl4'-0"
datum-VX-
0'
"""
-
-5.7' lowest low-water level
Low-water _
,wr~
1""
120'-0" approximately
mar
(a)
665'- 0'
Switch house
)" concrete pipes
Approximate L-. ;
_
high-water s
Jine_-_L__.
Junction pits
T
V T TT
-ir-6-(y-6-i4-<5-6-4-i-
"""""IJmits of trench
in rock
Fence
Approximate
low-water
line
(b)
Fig. 34. (o) Section and (b) site plan of shore electrode of the New Zealand scheme.
(From Ref. 55, N.Z. Engineering, by permission.)
473
action may be estimated from Figure 3.2, page 79, in Ref. 16III, which indi
cates that 24 elements are required. The number chosen was 25, and the
spacing between adjacent elements was 25 ft (7.6 m), except for the last two
spaces at each end, which were reduced to 20 and 15 ft (6.1 and 4.6 m).
The distance between centers of the end elements is thus 570 ft (173 m).
A plan of the site is given in Figure 34b.
Construction. Beach material was pushed toward the sea to form a coffer
dam 10 ft (3 m) high. A trench was excavated in the bedrock, and two fullsized test cells were constructed. A small concrete pad for each cell was cast
on the floor of the excavation, and the perforated concrete pipes were set on
the pads. The trench was filled to the top of these pipes with large boulders.
Concrete slabs 18 in (46 cm) thick and 18 ft (5.5 m) in outer diameter were
cast around the top of the pipes. Their purposes were to prevent sideward
movement of the pipes and support inspection pipes of 4-ft (1.2-m) diameter.
After the latter pipes were placed, the rest of the trench was filled with beach
material. Graphite electrode elements were hung in the cells, and their leads
were brought out to two temporary junction pits. Tests were run with 100 A
for 12 h between the beach electrodes in parallel and secondary electrodes
placed in the sea. The resistance was measured continuously and was found
to vary 10%with the tide. The highest voltage gradient measured was 6.2 V
in 3 ft (6.8 V/m). The current division between the two cells was never worse
than 45 A : 55 A.
On the basis of these tests, it was decided to proceed with construction of
the remaining 23 cells. Construction proceeded from July to November,
1964. A severe storm with 9-ft waves breached the coffer dam and flooded
250 ft (76 m) of open trench, but this did not seriously delay the construction.
Leads from each cell were brought to a concrete-block switch house,
where the electrode line terminated at copper busbars. An isolating switch
was provided for each two cells.
At the end of the job, the beach was restored to its normal level.
Final Tests. The measured resistance varied from 0.23 to 0.30 2, according
to the tide. This value is 1.5 to 2.0 times the design value; The maximum
voltage gradient was 7 V in 3 ft (7.7 V/m) and occurred on the seaward side
of the trench line. This indicates that most of the current flowed in the upper
layers of shingle. The two outer cells carry about 30% more current than do
the adjacent cells.
Cost. The cost of the electrode, including investigation, was approximately
57,000 ($160,000).
Life. After 22 months' service with 6 x 10~6 Ah usage as anode, the elec
trode elements were found reduced in volume by about 25 %, partly due to
corrosion by sea water. Additional useful life of 10 to 15 years was estimated.
1
474
9-8
GROUND RETURN
-This electrode serves the southern terminus (Sylmar) of the Pacific North
west-Southwest dc intertie. It is the first major dc electrode to be placed in
the open ocean rather than on a beach or in a sheltered cove. Its rated current
is 1.8 kA. It can serve intermittently as either anode or cathode.
Because the entire Los Angeles area is underlain with many pipelines and
other substructures, it was difficult to find a suitable site for a land electrode
within a reasonable distance of the converter station at Sylmar without
causing serious problems of electrolytic corrosion. It was, therefore, decided
to place the electrode in the Pacific Ocean, in spite of the fairly long distance
(30 mi or 47 km) from Sylmar, so as to reduce greatly current densities on the
land in that area.
The site chosen is 6000 ft (1.8 km) offshore in Santa Monica Bay. This is
not only the closest part of the ocean to Sylmar but it is also at a considerable
distance from Los Angeles harbor, from the steam power plants on the
ocean shore, and from oil pipelines in the ocean. Another factor in the choice
of the site is that the electrode line could be run much of the way from
Sylmar (22 mi or 35 km) to the shore on towers of existing ac transmission
lines. This part of the electrode line consists of two 1272-mcm (646 mm2)
ACSR conductors, taking the place of the usual overhead ground wires and
insulated by two suspension insulators. The remaining 8 mi (13 km) to the
shore is by two 1250-mcm copper-conductor lead-sheathed 15-kV cables run
in duct banks. These terminate at a bus in a vault near the beach, where six
underwater conductors to the electrode also terminate. The resistance of the
electrode line is about 1.1
The electrode consists of a linear array (Figure 35) of 24 elements resting
on the bottom of the ocean, which is at least 40 ft (12 m) deep. At this depth
the effect of wave motion is expected to be very weak. The spacing between
elements is 25 ft (7.6 m) except near the ends, where the spacing is reduced to
23, 20, 16 ft (7.0, 6.1, 4.9 m). The overall length of the array between centers
of the end elements is 543 ft (166 m). Each element has two horizontal metal
rods, one above the other, in a concrete box. The lower rod is about 1.5 ft
(46 cm), and the upper rodis 3.5 ft (107 cm) above the ocean floor.
The rods are made of Durichlor 51, a high-silicon iron alloy (14.5% silicon,
4.5% chromium, 0.9% carbon, and 0.65% manganese) manufactured by the
Duriron Company for use as anodes in salt water. Each rod is 5 ft (1.52m) in
length and 3 in (76 mm) in diameter. The estimated life of these rods is 85
yr under an operating cycle of 54 A, each polarity half of the time, constantly
except for 12 h/yr at 1800 A.
Each pair of rods is enclosed in a rectangular box (Figure 36) 11 by 7 by
5.17 ft high (3.4 by 2.1 by 1.6m)made of unreinforced concrete and weighing
l6'-0"
25'- 0" \
-20'- 0"
1
V /
\,
17 at 25'-0" = 425'- 0 \
20'- 0"N
23'-0'\
543'- 0"
,23'-(r
2A/
shields s.
'*|
475
,16'- 0"
25' 0"
a-n-THfi-
( Unlighted
buoy
Untighted J
buoy
Top section
Center
sections
Lead wire
Bottom section
Fig. 36. Detail of electrode rods and electrode shield, Los Angeles ocean electrode.
(From Ref. 79 by permission of authors.)
ssppr
476
PROBLEMS
GROUND RETURN
9 tons (8.2 tonnes). Each box is made in four sections, separated by horizontal
tongue-and-groove joints. The box has in its bottom, top, and sides a total of
45 screened circular holes of 13.6-in (34.6-cm) diameter, having a total area
of 0.377 m2. The rods are suspended from polypropylene rope slings attached
to the top section of the box. The boxes serve (a) to prevent access of fish to
regions of high voltage gradient, (b) to protect the rods from underwater
turbulence and from the movement of silt, and (c) to decrease electrode-toground resistance by providing separation of the rods from the ocean bottom.
Maximum calculated gradient outside the box is 2.5 V/m in the first few
centimeters from the box.
The continuous rating of the electrode is 1410 A, set by thermal limitation
of the underwater cable. The electrode can carry 1800 A for 14 h without
exceeding this thermal limitation.
Two three-conductor 300-mcm (152 mm2) 5-kV copper cables connect
the electrode to the vault on the shore. Each conductor is spliced to
four electrode elements. The leads from each element are 2/0 AWG
(67.5 mm2) copper with polyethylene insulation.
The calculated resistance of the electrode to earth is of the order of 0.01
to 0.02 2. The electrode can operate satisfactorily with less than 0.02 2 resis
tance even if one of the three-conductor cables should be cut; for example, by
a dragging anchor of a small boat. Large ships are not expected to anchor
in this area.
PROBLEMS
477
with that during monopolar operation (a) with ground return and (b)
with metallic return using the other conductor. Voltage to ground is
400 kV at the sending end.
3. Derive an equation for the conductance per unit length from a buried
pipe to a parallel pipe of different size. Use the method of images.
4. Prove that, in the current field of a shore electrode (Section 9-2, page
402), the apparent resistivity measured at the surface of the sea is equal
to that measured on land.
5. Find an expression for the apparent resistivity of the sea when a shore
electrode is used and the land and sea bottom have infinite resistivity.
The assumptions made in the text should be used.
6. Prove that the expression for reflection coefficient in Eq. (38) is correct.
Find the relationship between the angle of arrival and the angle of
refraction of a ray of current passing from a medium of resistivity pt
through a plane boundary to a medium of resistivity pz . The arriving
ray is defined here as the resultant of the incident and reflected rays
in medium 1.
7. Check Rusck's calculation on rate of loss of iron at JL = 10 mA/m2
(page 427).
8. Compute the minimum acceptable spacing between rectifiers and the
current and voltage to be furnished by each rectifier in order to keep the
potential of a long, insulated pipeline between the limits of 0.85 and
1.20 V. Assume that the soil is chemically uniform and has a uniform
resistivity of 100 2 m. Neglect foreign currents. The following data
pertain to the pipe:
Outer diameter, 3.0 ft (0.915 m)
Wall thickness, 0.438 in (1.11 cm)
Resistivity of pipe, 2.1 x 10 ~7 2 m
Coating, felt and asphalt, leakage resistance of unit area, 6 x 104
2 ft2 (5.6 x 103 2 m2)
Depth of burial (top of pipe), 3 ft (0.9 m)
9. Plot curves of pipe potential, longitudinal current, and leakage current
per unit of length foT the insulated pipe in Problem 8.
10. Repeat Problem 8 for an uncoated pipe otherwise the same.
11. Find the decrease of conductance from electrode to ground for each arm
added to a star-shaped electrode.Assume each arm to be a cylinder 500 m
long and 0.5 m in diameter. Assume soil resistivity to be 100 2 m.
12. Calculate the power loss in the coke of the electrode designed by
Rusck (Section 9-7, page 459).
478
GROUND RETURN
BIBLIOGRAPHY
479
16. Earth Conduction Effects in Transmission Systems, by Erling D. Sunde, Van Nostrand,
New York, 1949.
Chapter II, "Earth Resistivity Testing and Analysis," pp. 38-63.
Chapter III, "Resistance of Grounding Arrangements," pp. 66-97.
Chapter VI, "D-C Earth Conduction and Corrosion Protection," pp. 177-214.
17. . Transient Performance of Electric Power Systems, by Reinhold Riidenberg,
McGraw-Hill, 1950, especially Chapters 24 and 25, pp. 310-339. Step voltage,
multiple electrodes, buried electrode, method of images, resistances to ground, and
heating of ground.
18. "Les Bases Generale de la Technique des Mises a la Terre dans les Installations
filectriques (General Technical Principles of Grounding Electric . Plant)," by P.-G.
.
Laurent, Revue Generale d"Electricity, 7th series, Vol. 1, No. 7, pp. 368-402, July,
1951. In French. Excellent general treatise on principles and methods of grounding,
including purposes and characteristics of operating and safety grounds, electrical
properties of soils, ground electrodes, hazards to life, and heating of the soil. English
translation of all except part VI, "Heating of Soil," published as Appendix IV of
Ref. 39.
BIBLIOGRAPHY
1. A Method of Measuring Earth Resistivity, by F. Wenner, Report 258, Bulletin of the
U.S. Bureau of Standards, Vol. 12, No. 3, pp. 469-482, Feb. 3, 1916.
2. Erdstrome (Earth Currents), by Franz Ollendorff, Julius Springer, Berlin, 1928. In
German. Reprinted by University Microfilms, Ann Arbor, Mich., 1962. 260 pp.
3. " Calculation of Resistances to Ground," by H. B. Dwight, Elec. Eng., Vol. 55, pp.
1319-1328, December 1936.
4. " Effect of Electric Shock on the Heart," by L. P. Ferris, B. G. King, P. W. Spence, and
H. B. Williams," A.I.E.E. Trans., Vol. 55, pp. 498-515, 1263, May 1936. Tests at
Columbia University on the effect of shock on heart action of anesthetized animals
(guinea pig, rabbit, cat, dog, pig, sheep, and calf). Current causing fibrillation increases
with body weight and heart weight. For calf (70 kg) 3-sec 60-Hz minimum fibrillation
current is 0.21 to 0.47 A; for sheep (56 kg), 0.16 to 0.39 A; sheep, 3-sec dc, 0.75 to
2.1 A.
5. " Practical Aspects of Earthing," by E. Fawssett, H. W. Grimmitt, G. F. Shotter, and
-H. G. Taylor, I.E.E. Jour., Vol. 87, pp. 357-390, October 1940. Disc., pp. 390-400.
Comprehensive article.
6. "Grounding Principles and Practice: I, Fundamental Considerations on Ground
Current," by Reinhold Riidenberg, Elec. Eng., Vol. 64, pp. 1-13, January 1945.
General principles of ground currents. Formulas for resistance of various forms of
ground electrode. Republished as Appendix III of Ref. 39.
7. Tables of Functions with Formulae and Curves, by Eugene Jahnke and Fritz Emde,
4th ed., 1945, Dover Publications, New York. Struve function So, p. 218; N0, p. 190.
8. "The Experimental Sending of D.C. through the Earth in Sweden," by R. Lundholm,
C.I.G.R.E., 1946, Report 134. 12 pp. Four tests in which direct current was circulated
between pairs of earth or sea electrodes and currents were measured in railroad rails
and in sheaths of telephone cables.
9. "Soil and Water Resistivity in Finland," by E. K. Saraoja, C.I.G.R.E., 1946, Report
201, 4 pp. Average measured resistivities of soil and water, including values for frozen
19. "Return Current through the Earth for High-voltage D.C. Transmission," by
R. Lundholm, Direct Current, Vol. 1, pp. 79-86, March 1953. Experiments in Sweden
with direct current circulated through ac transmission lines. Effects on railway signals
and currents in rails and cable sheaths.
20. "Direct Current Track Circuits for Railways," by D. G. Shipp, Direct Current,
Vol. 1, pp. 118-121, June 1953.
21. "The Magnetic Compass Error Caused by D.C. Single-core Sea Cables," by F. Busemann,E.R.A. report B/T 116, May 1953. Also in Direct Current, Vol. 1, pp. 208-213.
"A Study of the Hazards of Impulse Currents," by Charles F. Dalziel, A.I.E.E. Trans.,
Vol. 72, Part 3, pp. 1032-1041, October 1953. Disc., pp. 1041-1043. Direct-current
resistance of body from hand to both feet in 3/4 in of saltwater is 1135 Ci, minimum
of nine subjects.
22.
10. "Dangerous Electric Currents," by Charles F. Dalziel, A.I.E.E. Trans., Vol. 65, pp.
579-585, August-September 1946. Disc., pp. 1123-1124.
11. "Applications of Electro-osmosis to Soil Drainage," by A. H. D. Marwick and A. F.
Dobson, Engineering, Vol. 163, No. 4230, pp. 121-123, Feb. 7, 1947.
12. " Cathodic Protection of Buried Metallic Structures," by R. de Brouwer, C.I.G.R.E.,
1948, Report 205.
13. "D.C. Transmission with Return Current through Earth," by C. E. Soderbaum,
I. Beckius, M. Bockman, and R. Lundholm, C.I.G.R.E., 1948, Report 401, 24 pp.
14. "Comparative Properties of Grounding Electrodes," by Reinhold Riidenberg, Elec.
World, Vol. 129, No. 5, p. 72, Jan. 31, 1948. Brief comparison of surface, volume,
current capacity, and gradient of electrodes of various shapes having equal ground
25. "Let-go Currents and Voltages," by Charles F. Dalziel and F. P. Massaglia, A.I.E.E.
Trans., Vol. 75, Part 2, pp. 49-55, May 1956. Disc., pp. 55-56. 60-Hz let-go currents
for 99.5% of 134 men, 9 mA; of 28 women, 6 mA. For dc exactly five times as great.
Mean dc release current for 28 men was 74 mA; value for 99.5% of these men, 0.82
x 74 mA. Safe voltages for wet hands and feet, 10.2 V 60 Hz, 51 V dc.
26.
27. "Electric Shock Hazard Analysis," by K. S. Geiges, A.I.E.E. Trans., Vol. 75, pp.
1329-1331, February 1957. Disc., p. 1331. Perception current, 0.1 to 0.2 mA; muscular
contraction, 6 to 20 mA; fibrillation of heart, 100 mA; these are for ac. Tolerable dc
somewhat higher.
resistance.
15. The Corrosion Handbook, edited by Herbert H. Uhlig and sponsored by the Electro
chemical Society, Wiley, New York, 1948, 1188 pp.
T|
i
480
GROUND RETURN
28. "Telephone Interference and Other Effects Caused by the Gotland H.V. D.C. Transmission," by G. A. Pettersson, L. Ahlgren, and H. Forsell, C.I.G.R.E., 1958, Report
324. Corrosion and compass variations are discussed.
29. "Hazards of Electric Shock at Stations during Fault and Method of Reduction,"
by A. Elek, Ontario Hydro Research News, Vol. 10, No. 1, pp. 1-6, January-March
1958. Includes use of crushed stone on ground surface and grounding of fences
around station.
30. "Corrosion Problems Arising from High Voltage D.C. Transmission with Earth
Return," by J. H. Gosden, Appendix V to C.I.G.R.E. Report 417, 1960, pp. 24-28.
Also in Direct Current, Vol. 4, pp. 159-161, September 1959. Disc., p. 161.
31. "Investigation of Some Problems Concerning the Earth-Return H.V.D.C. System,
Kashira-Moscow Line" (in Russian), by S. A. Bazhenov, V. P. Pimenov, and M. R.
Sonin, Elek. Stantsii, Vol. 30, No. 9, pp. 54-59, September 1959.
32. "EarthReturn Effects," by A. Stalewski, Direct Current, Vol. 4, pp. 155-158, Sep
tember 1959. Disc., p. 161. Report to C.I.G.R.E. Study Committee 10, June 1959.
33. "Examples of Geoelectric Surveys," by L. S. Palmer, l.E.E. Proc., Vol. 106A, pp.
231-241, June 1959. Disc., pp. 241-244.
34. "Alternating-Current Shocks of Varying Parameters Affecting the Heart," by W. B.
Kouwenhoven, G. G. Knickerbocker, R. W. Chestnut, W. R. Milnor, and D. J. Sass,
A.I.E.E. Trans., Vol. 78, Part 1, pp. 163-167, May 1959. Disc., pp. 167-169. Experiments on anesthetized dogs with 60-Hz shocks of various durations from 1/120 to
5 sec. Defibrillation by countershock.
35. "Use of Earth and/or Sea Return," Chapter 9 of High Voltage Direct Current Power
Transmission, by Colin Adamson and N. G. Hingorani, Garraway, London, 1960,
pp. 133-142.
36. Recommendations for the Protection of Underground Cables against Corrosion,
C.C.I.T.T., New Delhi, 1960.
37. " Soil Thermal Characteristics in Relation to Underground Power Cables," A.I.E.E.
Committee Report, A.I.E.E. Trans., Vol. 79, Part 3, pp. 792-844, December 1960.
Disc., pp. 844-856. Following is a list of the subdivisions of the report and the respec
tive authors:
'
BIBLIOGRAPHY
481
J
j
j
j
(
|
|
\
I
j
j
4
;
49. " Investigations of Ground Current Phenomena Associated with HighVoltage Direct
Current Transmission," by Fred E. Cornwell, ibid., pp. 17-22. Also in I.E.E.E. Trans.
on P.A.&S., Vol. 85, pp. 254-258, March 1966. Disc., pp. 258-260. Also presented at
I.E.E.E. WESCON, August 1964. Tests in Sacramento Valley of California on poten
tial gradients, transformer neutral currents, and pipe-to-soil potentials. One current
electrode was in a deep well.
50. " Corrosion Aspects of HVDC on Buried Telephone Plant," by L. E. Fiorito, ibid.,
pp. 23-30. Also in I.E.E.E. Trans, on P.A.&S., Vol. 87, pp. 35-40, January 1968.
Formulas are derived and illustrated for leakage current on a cable sheath near
current electrode and for minimum separation of cable from electrode for acceptable
rate of corrosion.
51. "Electrodes for EHV D-C Transmission in a Metropolitan Area," by S. G. Pann,
A. A. Galindo, and G. R. Elder, ibid., pp. 31-36. Calculations and tests in the Los
Angeles area indicated that a sea electrode should be used.
52. "H.V.D.C. Transmission: Earth Electrode Arrangements," by Anders Kohler,
Direct Current, Vol. 10, pp. 18-24, February 1965.
53. The Influence of H.V.D.C. Transmission on General Utility Operations, a report prepared for the Oregon Public Utility Commissioner by Cornell, Howland, Hayes, and
Merryfield, Corvallis, Oregon, 1965.
54. "The Benmore Land Electrode," by D. G. Dell, New Zealand Engineering, Vol. 20,
pp. 165-175, May 15, 1965.
55. " The North Island Sea Electrode," by D. G. Dell, ibid., pp. 213-222, June 1965.
56. "Practical Applications of Electro-osmosis," by J. K. Gladwell, New Zealand
Engineering, Vol. 20, No. 2, pp. 66-72, 1965.
:
j
J
j
ij
I
j
I
482
57.
" The Benmore Land Electrode for the Benmore-Haywards H.V.D.C. Transmission
Scheme," by D. G. Dell, I.E.E. Conference Publication 22, HVDC Transmission,
Manchester, Sept. 19-23, 1966, Part 1, Paper No. 82, pp. 415-418.
58. "Earth Electrodes," by P. Quinn and P. E. Howard, ibid., Paper No. 83, pp. 419-421.
Proposed designs of electrode in mud flat and of one in an arid region near a reservoir.
59. "Some Basic Concepts of HVDC Ground Return Currents," by E. V. Lindseth and
F. E. Cornwell, ibid., Paper No. 84, pp. 422-426. Potential gradients around Tracy
and Mead; transformer neutral currents; transients in tests with pulses.
60. "The North Island Sea Electrode for the Benmore-Haywards HVDC Transmission
Scheme," by D. G. Dell, ibid., Paper No. 85, pp. 427-430.
61. "Design of the Sea Electrode System, Sardinia-Italian Mainland 200 kV Scheme,"
by D. S. Thorp and D. MacGregor, ibid., Paper No, 86, pp. 431-434.
64.
65.
66.
Cranmer, ibid., Paper No. 87, pp. 435-438. Experiments in the Solent between England
and the Isle of Wight.
"A Study of the Electric Field in the Mediterranean Sea," by F. M. Cahen and
M. H. Macaire, ibid., Paper No. 88, pp. 439-441. Model and digital-computer studies
of field associated with Sardinia-Italy link.
Discussions of foregoing papers by R. F. Stevens, F. E. Cornwell, H. R. Gunn, and
D. MacGregor, ibid., Part 2, pp. 111-114.
"The Anodic Earth Electrode for the Konti-Skan H.V.D.C. Link," by E. Anderson
and N. R. Neilson, Direct Current, Vol. 11, pp. 54-56 and 58-63, May 1966. Shore
electrode in Denmark.
Statistical Study of Electric Fields from Earth-Return Tests in the Western States and
Comparison with Natural Electric Fields, by George V. Keller, report prepared for
71.
BIBLIOGRAPHY
GROUND RETURN
72. "The Effects of High Voltage Direct Current Transmission on Buried Metallic Pipelines," by John Van Bladeren, Paper No. X, ibid.
73. " Corrosive Effects of HVDC Transmission," by Fred E. Cornwell and T. E. Backstrom, Materials Protection, Vol. 5, pp. 8-13, February 1966. Disc., pp. 13-14.
f
]
j
:
\
\
{
\
|
I
?
j
j
|
=
j
|
j
f
1
I
\
j
{
|
|
f
ji
|
|
I
I
I
|
483
75. Ground Current Behavior as Related to Buried Pipelines, by Ira C. Bechtold, Richard
H. Jahns, Robert V. Langmuir, and William R. Smythe, a report prepared for Pacific
Lighting Service and Supply Co., Southern California Gas Co., and Southern Counties
Gas Co., Apr. 15, 1967.
76. " Electrophysical Study of the Soil for the North End DC Grounding Electrode of the
Northwest-Southwest Intertie," by A. L. Kinyon, l.E.E.E. Publication 31C46, pp.
52-64, January 1967.
77. "Statistical Study of Electric Fields from Earth-return Tests in the Western States
and Comparison with Natural Electric Fields," by George V. Keller, l.E.E.E. Publi
cation 31C46, pp. 35-51, January 1967. Also in I.E.E.E. Trans, on P.A.&S., Vol. 87,
pp. 1050-1057, April 1968. Similar to Ref. 66.
78. " The Rice Flats DC Ground Electrode for the Pacific Northwest-Southwest HVDC
Intertie," by Carleton L. Waugh, Proc. Western Water and Power Symposium, Los
Angeles, Apr. 8-9, 1968, pp. D-53-60.
79. "The Los Angeles HVDC Ocean Electrode," by G. R. Elder and D. B. Whitneyibid., pp. D-61-68.
80. "Major Features of the Vancouver Island 260-kV HVDC Submarine Link," by
H. M. Ellis and W. Chin, Proc. American Power Conf, Vol. 30, pp. 1017-1034,
April 1968; copyright, 1969.
81. "A Survey of Corrosion Aspects Related to the Operation of Electrodes for HVDC
Ground Return," by Ingvar Lid6n and Heine M&rtensson, a report to C.I.G.R.E.
Study Committee 36 (Interference), Volgograd, 1969.
82. Appendix of " Electric Power Transmission by H.V.D.C. Submarine Cable across
the Adriatic Sea from Yugoslavia to Southern Italy," by M. Visentini, A. Asta, and
F. Trimani, C.I.G.R.E., 1958, Report 210, pp. 13-22. Compass error.
83. " Thermal Analysis of Earthing Rods at the Ends ofD. C. Lines," by K.R.Rakhimov,
Electric Technology U.S.S.R., Vol. 4, pp. 36-43, July 1970. Translated from
Elektrichestro, No. 10, pp. 23-26, October 1969.
485
III ! IV
III
Appendix A
(Al)
IS2 COS 0
>~L0t
Fig. Al. Positive half-waves of (a) alternating line current and (6) alternating line-toneutral voltage of a six-pulse converter.
Seg
ment
+<- u
AS 60
Equation
Limits
I a - 60 < 9 < 5 - 60
II 5 60 < 9 < a
+ 60
+ 60
(A2)
ia = Id = /s2(cos a cos 5)
im = 7s2[cos (0 - 60) - cos 5]
(A3)
i, = /s2[cos
nI
(A4)
0m]2d9j
Hi
LVJa-eO9
AS +60
Ax +60
J (5-60
484
2dd+
te/j2 = Is22
J a+60
(A6)
\-U/2
hn2 dQ J
/J
(A7)
(A8)
By the substitution 9' -6 + 60, dB' dO, which displaces the origin hori
zontally, this becomes
W2
For any wave such as this in which the two successive half cycles are alike
except for sign, each 2n in Eq. (A6) may be replaced by x. For the wave in
question, the integral may be divided into segments, thus :
n / AS-60 d0
a 60
(A5)
=0
~I
s
i2
f (cos
j
Isi2[*' cos2
422 cos2
(cos2 a
/s22
(<5 a)
<5 a
+-
9' sin 26
2 cos a sin B + +
a 2 cos a(sin <5
sin 25 sin 2a
4
sin a)
(A9)
<5
a =u
and
<5 4- a = X
(A10)
486
B) cos %{A + B)
cos2 A
(A11)
we obtain
>r
= hi u cos
idi
a+~
2 cos a
2 sin ~ cos
A*
f*<5+60
2n
2n
d9 = Id2(a-b+)=Id2{-u)
~3
3
5-60
(A13)
= IS22
Sn
* sin u
Id2
/i
(A14)
c.
(A15)
=
cos a
T2n
-J
u{2
|_ 3
/I
Ad
(A16)
(A17)
/ A
(A22)
21
d
2n
1T
7=7d{f[l-(2,H)]}1/2
where
(A21)
The whole sum is obtained by adding Eq. (A13) for segment II. It is
This expression can be put in terms of 7d instead of Is2 by use of Eq. (A3),
becoming
cos A
= 7s22( -
2s22jw +
cos - sin - = 2
sin 2b sin 2a
and
Cd
hi2i 2 +
sin -
- cos -
cos2 - = (cos 2 + 1)
a+60
(A20)
But
Cd
/A1A.
(A19)
Segment III
hi"
(A18)
Ad
7s2
+ 1)
(A12)
I
'a +60
%hll
= (cos 2A
,
cos 2a + cos 2b + 2
-cos2 a + cos2 b = -
Segment II
nla2=ld2
487
'
f k
(A24)
(A25)
'
\
488
+
+
+
cos 2 cos u
2k
3
+ cos u)
+ cos u)
1.00
0.99
0.98
0.97
0.96
0.95
a>
0.94
0.93
a or 7
0.92
0.91
20
30
15
30s to 90
40
489
Hence
/0=WI
= Vl F(A,u)
l0
(A27)
(A28)
Hence
F{A,u) =
Reference
1. Tables of Integrals and other Mathematical Data, by H. B. Dwight, Macmillan, 1947.
(A26)
491
ji
Appendix B
V2
/*<5-60
-*s2
J a 60
j- =
'a +60
6-60
'<5+60
pa+60
2 cos a/ 6d9
1
J a 60
Refer to Figure A1 and Eqs. (Al) to (A5) for the wave shapes of voltage
and current during a half cycle with overlap angle less than 60.
Any periodic quantity may be represented by a Fourier series. The complex
form of such a series is
(*<5+60
<5-60
2 cos b/9d9
'560
if Dh/hB
h= <x>
a 60
(Bl)
(*5+60
+
where
and
Dfi =
f*
F(9)/-h9d0
(B2)
(B3)
Substitution of 2 cos x
The crest value and phase of the hth harmonic wave are given by 2Dh . The
limits of the integral need not be as shown but may span any range of 2n rad.
If F(9) = F{9 + 7i), as is true for the wave in question, and if h is odd, the
limits may span a range of n rad, in which case the coefficient of the integral is
doubled. Thus the crest value and phase of the fundamental current (h = 1)
are given by
fa+ 120
60)/ 9d9
=/x / x gives
(*5+60
(/S - 9 -/-6 - 9) d9
(/a-9-/-a~9)d9~
60
5-60
5-60
'5+60
(/60 / 20 60) d9 +
(J 6Q / 20 + 60)
a+60
a- 60
/a bO d9 = /a b9
(B4)
where It is the complex rms value of fundamental current and i{9) is the
instantaneous current. The phase of the current is expressed as advance with
respect to the line-to-neutral voltage. The integral is subdivided into segments
I, II, and IIIin Eqs. (A2) to (A5). Multiplication of each side of Eq. (B4) by
7t//s2 and substitution of these equations give
490
2 cos (9
a+60
(*a+60
a
2 cos (9 + 60)/9d9
gives
'
\a+60
)
f j/a 9 -j/-a - 0/a
\
60
\<5+60
/
- ( j/b - 9 -j/-b - 9 )
/<'a+60
\
\5 60
+ 60
20 + 60
20 - 60 )
+ ( 0/ 60 - 0.5.//0/60 - 0.5//
/a+60
'/a 60 \ J--
jrm
492
p' _|_
= tV-2
25 ~72m)
4tt -"7"
'
. /~2<x ~/~2
cos a cos <5
V3EII* = 3ELNi;
s*n
wn 2
cos
+7"
2
sin X sin u
P' = - 2
u cos X sin u
--
per unit
Q' =
(B13)
{B14)
(B15)
(B12)
(B6)
(Bll)
sin 2a sin 25 + 2u
'
, / 2a / 2<5 /2m
ll=
7"d
-Q'
(B5>
Now, in the first per-unit system, the base alternating current is Is2 6/71, and
hence the per-unit fundamental alternating current is
!i_
--
1,2
cos 2a cos 25
(B9)
The real part of Eq. (B10) gives the per-unit active power,
/}
ELNgives
Since base power in the first per-unit system is Vd0Is2 , the per-unit complex
power is
/2a /2<5 +Jjlu
P' -}- jQ = 4r
(B10)
4
yMJ--/-+i2u
p+jQ =
493
(B8)
s' =
(B16)
Displacement Angle. The angle 4> by which the fundamental current lags
behind the line-to-neutral voltage is given by
tan
r
Q 2u + sin 2a sin 2d
= u csc u csc X cot X
P = cos
2a -cos 2<5
-
(B17)
}
v
SB
loop. We then put 5 = n y, and hence cos 6 = cos y, and multiply the
resulting equation by 1, obtaining:
Appendix C
(C3)
-Vt= Vd0 cos y-(Rc~ R,)Id + Farc
In Eqs. (CI) and (C3), Vd is negative for inversion (however, Vd0 is always
positive). The sign convention for Vd is sometimes reversed.
Equations
j
if
(CI)*
(C2)
495
The direct current passes through two phases of the ac circuit in series
(Y connection assumed) except during commutation, when it passes through
three phases in series-parallel. The resistance of the path, then, varies from
2Rp at u = 0 to 1.5Z?P at u = 60, where Rp is the resistance per phase. For
u = 0 we can readily verify that the same copper loss can be calculated either
from dc quantities as
Pca = 2RpId2
(C3)
(C4)
or from ac quantities as
,1
INDEX
INDEX
limitations, 19
A.E.G., 7
169, 170
Anisotropic earth, 405-406
Anisotropy, coefficient of, 405
Anode, 50, 424 {see also Electrode)
for cathodic protection, 435
Anode dampers, 260-269
Anode efficiency, 426
Anode reactor, 261, 262
Anodes in multiple {see Valve, multianode)
Apparent power of converter, 493
Arcback, 67, 157, 199-220, 241, 247
analysis, uncontrolled, 211-219
with blocking and bypassing, 207-210
caused by rapid commutation, 266-269
causes, 206-207, 266-269
consequences, 207
consequential, 232-233
crest currents, 216, 218-219
cure, 219
definition, 198
detection, 219
in inverter, 219
valves in series, 220
Arcback counter, 219
Arc chopping {see Quenching)
Arc drops, 50, 494-495
Arc quenching {see Quenching)
90,91,92,116
single-phase, 54
three-phase, 58, 71-123
Bridge inverter, analysis, 105-112, 203, 204,
219, 222-228, 230-231
Bridge rectifier, auto phase control, 104
complete characteristics of, 103
equivalent circuit, 88
wave forms, 76, 82, 86, 87, 101, 208,
217, 232
Bridges in parallel, 114
Bridges in series {see Cascade)
Brown Boveri, 7
Bundle conductors, 26
Bypassing, of bridges, 259
use of main valves for, 205-206
Bypass switch, 199
Bypass valve, 199-206
auxiliaries, 205
current divider, 205
damper, 261
discharge of dc line through, 244-247
grid-pulse supply, 205
transfer of current to and from, 201-204
in inverter, 203-205
in rectifier, 201-202, 204
497
Capacitors (continued)
cost, 345
heating from harmonics, 324
overvoltage from harmonic resonance, 324
protection of, 385-386
rating of, 369, 373, 383
reactive power of, 324
series, 438 {see also Compensation of .ac
lines)
498
INDEX
INDEX
Control (continued)
power, 31, 180
stability of, 174-179
tap changer, 179-180
Converter, apparent power of, 493
bridge circuit, single-phase, 54
bridge circuit, three-phase, 58, 71-146
development of, 4
Graetz circuit (see Converter, bridge circuit,
three-phase)
Marx, 4
misoperation, 198-233
['
j
multibridge, 115-123
power vectors, 79
pulse number, 61
single-phase, 51-55
bridge, 54
center-tap, 52
full-wave,52, 54
half-wave, 51
six-pulse, 58-65
displacement angle, 493
effective alternating current, 484-489
fundamental current, 490-492
power as function of converter angles,
492-493
rms alternating current, 61, 484-489
undesirability, 332, 372
twelve-pulse, 67, 115, 332, 370
voltage drop due to resistance, 494
voltage drop in arcs, 494-495
Y-Y -interphase connection, 62, 63
Converter angles, 91, 105-106
notation for, 105-106, 110
Converter chart 1, 129-138
Converter chart 2, 138-146
error in reactive power by approximate
method, 145
Converter charts, 129-146
loci of constant active power, 137,
138, 140
loci of constant alternating current, 138
loci of constant apparent power, 137, 138,
139,140
loci of constant converter angles, 130-135,
139-143
loci of constant direct current, 129, 137,
140, 143, 144
loci of constant direct voltage, 129, 137,
140, 143, 144
*j
i
416,418
natural, 419-422
compared to that of dc transmission,
422-423
near electrode, 393-417
499
270-271
Dc current transformer, 187-191
Dc distribution, vestiges of, 2
Dc generators, Edison bipolar, 1
in Thury system, 3
limitations of, 2
series wound, 1, 3
Dc line, telephone noise from, 339-341
Dc link (see also Eel River, English Channel,
Gotland, Kingsnorth, Konti-Skan,
Nelson River, New Zealand, Pacific
Northwest-Southwest, Sardinia,
Vancouver Island, Volgograd-Donbass)
bipolar, 11, 391-392
constitution of, 9
construction in stages, 392
definition, 9
homopolar, 11, 392
kinds, 11
'//
x
monopolar, 11, 391,
39
reliability, 392
table, 13
Dc motors, limitations of, 2
Dc power measurement, 191
Dc reactor, 51, 235-247, 315
as buffer of overvoltage surge, 288
effect on crest discharge current of line,
246-247
effect on ripple in direct current, 240-244
WW****
500
INDEX
Dc reactor (continued)
effect on selectivity between dc line faults
failures, 236-240
purposes, 235-236
Dc smoothing reactor (see Dc reactor)
Dc transmission, advantages, 19, 32
applications, 32
British Columbia (see Vancouver Island
dc link)
Charlottenberg-Moabit, 7
constant-current, 7, 156
constant-voltage, 156
control (see Control)
economics, 33
Elbe-Marienfelde, 7
English Channel (see English Channel
dc link)
future of, 35, 39-41
Gotland.(see Gotland dc link)
history, 1, 37
Kingsnorth (see Kingsnorth
dc link)
Konti-Skan (see Konti-Skan dc link)
limitations, 19, 32
losses, 156
Manitobe (see Nelson River dc link)
Mechanicville-Schenectady, 6
Mellerud-Trollhattan, 8
Moscow-Kashira, 8
Moutiers-Lyons, 3
New Zealand (see New Zealand dc link)
Pacific NW-SW Intertie (see Pacific NW-SW
dc link)
Sardinia (see Sardinia dc link)
status during ascendancy of ac, 2
Sweden, 7, 8 (see also Gotland dc link)
Thury, 3
U.S.S.R., 8, 14-15
Vancouver Island (see Vancouver Island
dc link)
Volgograd-Donbass (see Volgograd-Donbass
dc link)
Wettingen-Zurich, 7
Dc transmission link, 3 (see also Dc link)
"Dc year," 15
Dead band of tap-changer control, 1.79
INDEX
Diametrical connection, 64
Direct current, average (see Average direct
current)
discontinuous, 240, 283
ripple in, 240-244
Direct Current Institute, Leningrad, 8
Proceedings, 37 (ref. A3)
Direct voltage, average (see Average direct
voltage)
drops due to arcs and resistance, 494-495
measurement of, 191
Disconnecting switches, 199
Displacement, transmission by, 26, 27
Displacement angle, 493
Displacement factor, 79, 91
Distortion of alternating voltages of multibridge converter, 118-119
Distortion of wave shape, measures of,
325-331
Disturbing current, 331
Disturbing voltage, 331
Donets Basin (see Volgograd-Donbass dc link)
Dorsey (see Nelson River dc link)
Double commutation failure, 226
Double overlap, in multibridge converter,
120-121
in single bridge, 93
Durichlor, 474
279-280
501
f1
|
|
|
5
1
j
1
502
INDEX
INDEX
Ground electrode (continued)
safety, 444, 455-459
Ground return, 11, 30, 391-476
advantages, 391-392
compatibility with other services, 423-443
equivalent depth of, 335, 336
problems of, 423-443
Ground-return circuits, coupling between,
334,335
telephone, 335
Ground wires, effect on balance, 337
Half-wave rectifier, 5 1
Half-wave transmission, 27
Harmonic filters (see Filter)
Harmonic impedance of generator, 324
Harmonic instability, 319
Harmonic voltage, minimization of, 357
Harmonics, 31, 115, 295-326
ac characteristic, 297-313, 354
at no overlap, 297-304
at overlap, 305-313
effect of finite smoothing reactor, 3 15 ,
317-318
equations, 305-307
graphs, 308-313
with zero and infinite smoothing
reactor, 317
anticipated (see Harmonics, characteristic)
characteristic, 295-318
direct-voltage characteristic, equations,
307-313
graphs, 314-316
even, 320-321
noncharacteristic (see Harmonics,
uncharacteristic)
nontheoretical (see Harmonics, uncharac
teristic)
orders of, 295, 296,297
Harmonics (continued)
table, 98, 99
overlap less than 60, equations, 80-83
table, 84
503
Lamm, Uno, 6
Land electrode, 445-464
Benmore, 461-463
construction, 447
cost, 462, 464
form, 447448
general, 445459
irrigated, 450
layout, 447448
life, 461463
location, 446, 461, 463
materials, 446447
resistance, 448449, 459, 461, 462, 464
Rice Flats, 463464
ring, 448, 449, 464
Rusck's design, 459461
safe voltage, 453
'
v.;v:v"
~"X,:
504
INDEX
soil, 450,451,455,461,463
star, 448, 449, 459
straight, 448
dc, 287-290
for protection of filter inductor, 385
locations in converter station, 290
multiple-stack, 289, 290
on electrode line, 386
Lightning protection, 285-286
Line dampers, 175, 270-272
Line-drop compensator, 179
Link, transmission, 9 (see also Dc link)
Livestock, 455458,470
Loads in harmonic studies, 349-350
Longitudinal voltage, 330, 331, 335
Loss of load on rectifier, overvoltage caused
by, 287
Lydd (see also English Channel dc
INDEX
link)
Magma, 393
Magnetic storms, 420, 438
Magnetization curve, 187
Malfunctions of valves, 198-199, 206-220,
227-233
Mantle, 393
Marx,E., 4
Materials, resistivities 0394
263, 264,265
ov voltage across valve, 248, 251, 253
461,470
criterion on harmonics, 347
Noise voltage, calculation of, 338
Ocean electrode (see Sea electrode)
Oil-blast circuit breaker, 282
Oscillations on dc line, 270-272
Overhead lines, 10, 13, 19-30, 34, 237,
275, 285,349
Overlap, 73, 74, 80
double (see Double overlap)
greater than 60, 92-103
insufficient, 241, 284, 286
less than 60, 26-92
Overlap angle, 80 (see also Overlap)
Overshoot, of current during dc line fault,
276-279
of dc line voltage, 276
of rectifier voltage, 283, 286
of voltage across valve, 248, 253, 254
Overvoltage protection, 282-291
of converter transformer, 287
principles of, 282
summary of, 291
Overvoltages, external, 283
internal, 283-284
505
dc, 313-314
Phase-shift circuits, 171
Piecewise-linear circuit, 72
Pipe, cathodic protection of, 434-438
leakage conductance of, 436
I
li
/;
;
factor) .
Power limit of ac lines, cables, 25
overhead, 22, 23
Power reversal, 153-154
Power vectors of converter, 79
Protection, 235-291
j;
l\
jj
of capacitors, 385-386
of converter transformers, 287
of filters, 385-386
Psophometrically weighted voltage, 331
Psophometric EMF, 338
Psophometric voltage, 331
calculation of, 338
Psophometric weighting, 327, 330, 331
table, 330
Pulse number, 61, 295, 296, 318, 332, 372
Pulse transformer, 151, 196
\\
j i
ii
j |
|j
'
88,89
Peterson, Harold A., 276
Phadke, Arun G., 276
Phase-angle meter, 367
j
i
j I
j I
i j
i j
506
INDEX
349-352
Sardinia dc link, 13, 16,45
criterion on harmonics, 346
Saturable reactor, 187
Screened rod gap, 284
Screening, 336
Screening factor, 336
Sea electrode (see also Shore electrode)
general, 465-466
life, 474
Los Angeles, 474-476
Sea return (see Ground return)
Sea-return circuits (see Ground-return
circuits)
Series arrangements of valves or bridges,
112-115
Danish, 466-470
efficiency, 404
general, 465-466
life, 473
materials, 465, 467, 471 , 474
mud flats, 476
North Island, 470-473
problems, 465
resistance, 469, 473, 476
voltage gradient, 470, 473, 476
Short circuit, monopolar, on bipolar dc
line, 283
on rectifier, crest currents, 222
uncontrolled, 221-222
with blocking and bypassing, 221
phase to ground, 275, 283
Short-circuit current, 29, 276-279
Shunt reactors, 10
Siemens, 7
Silicon controlled rectifier (see Thyristor)
Single-phase converters, 51
Single-tuned filter, 355-363
Skin depth, 335,417
Smoothing reactor (see Dc reactor)
Soil, capillarity, 450
compaction, 450
heating, 451-454
moisture, 450, 461, 463
parameters (electrical and thermal), 455,
461,463
resistivity, 450, 45 1, 461, 463
specific heat, 454, 455, 463
temperature, 451, 463
thermal conductivity, 455, 461, 463
S0ra, 466
Spark gap, 284, 285
Stability, of ac transmission, 17, 25
of control, 174-179
Stalingrad (see Volgograd-Donbass dc link)
-507-
Transductor, 187
Transfer function of current regulator, 166
Transformer, converter, 51, 53, 54, 56, 59,
63-68
tertiary winding, 343, 381
dc, 36
favoring ac transmission, 2
interphase, 62, 63
overvoltage protection, 287
representation in harmonic studies, 349
saturation caused by direct current,
56,438
Transformer connections, 53, 54, 56, 59,
61-65,68,332
effect on phase of ac harmonics, 301-303
effect on wave shape of line current of
converter, 298
for increased pulse number, 332
Valve, 5
bypass (see Bypass valve)
148-149
Fleming, 5
malfunctions of, 198-199, 206-220,
227-233
mercury arc, 5
508
INDEX