## rectifier tx 00046621
TRANSCRIPT
-
8/10/2019 ## Rectifier TX 00046621
1/16
Parallel operation of bridge rectifiers without an
interbridge reactor
J.K. Hall, MSc(Eng), PhD, CEng, MIMechE,
FlEE
J.G. Kettleborough, MSc
A.B.M.J. Razak, MSc
Indexing terms: Power electronics
Abstract: Parallel connection of rectifier circuits is
needed for very high direct-current supplies such
as those for electrochemical plant. Appropriate
phase displacement of their AC supply voltages is
used to increase the pulse number, and an inter-
rectifier reactor, often known as an interphase
transformer (IPT), is connected on the DC side to
prevent circulating AC components between the
rectifiers and to allow them to operate indepen-
dently. At very high power levels, this component
is occasionally dispensed with on economic
grounds. The interactions between two six-pulse
bridge rectifiers operating without an IPT are
described and analysed comprehensively. The rec-
tifier, transformer winding and AC-system current
waveforms are computed, together with their har-
monic components, and some typical operating
characteristics are given. Computed results are
supported by practical measurements on a labor-
atory rectifier system. It is noted that removal of
the IPT will considerably influence the rectifier
transformer design.
List of symbols
C
=
connection matrix
E,
e
=
EMF
E d =
DC-load back EMF
I
i =
current
I d
= total direct current
I = direct current from bridge 1
I d ,
= direct current from bridge 2
L =
self or mutual inductance
N
=
number of turns
p = d /d t operator
R =
load resistance
r =
winding resistance
V =
voltage
Z
=impedance
a
= thyristor firing-delay angle
/3 = angle of cessation of thyristor condution
y
(discontinuous load current)
= angle at completion of interbridge commutation
Paper 6987B (P6), first received 27th January and in revised form 26th
July 1989
Dr. Hall and Mr. K ettleborough are with the Department of Electron ic
and Electrical Engineering, Loughborough University of Technology,
Loughb orough, Leicestershire, LEI
1
3TU, United Kingdom
Mr. Razak is w ith the U niversity of Bradford, Bradford, West Y ork-
shire, BD7 IDP, United K ingdom
I E E P R O C E E D I N G S , Vol. 137, P t .
B,
N o .
2,
M A R C H 1990
6
p
Y
=
angle at completion of interphase commutation
=
interphase commutation angle
( p = 6 - a)
=
angle at completion of overlapping of interphase
within a bridge
commutations, one in each bridge
1 Introduct ion
The increasing size of electrochemical and electrometal-
lurgical plant for manufacturing such products as chlo-
rine, aluminium and copper, requires AC-DC power
convertors capable of delivering direct currents of up to
100
kA, and sometimes higher. This necessitates multiple
paralleling of the semiconductor devices within the recti-
fier circuit and parallelling complete circuits.
It is usual to use as high a pulse number as is eco-
nomical to minimise both the ripple in the DC output
voltage and the harmonic content in the AC-supply
current waveform to the rectifier installation. For
example, 12-pulse operation is provided by supplying
two parallel six-pulse rectifiers with
30
displaced AC
voltages. A 24-pulse arrangement requires four six-pulse
rectifiers mutually phase displaced by 15 . A disadvan-
tage of a high pulse number is the increased size and cost
of the rectifier transformer arrangement to provide the
necessary phase displacement. In standard practice, the
transformer tank also encloses the centre-tapped induc-
tor, usually known as an interphase transformer (IPT),
connected between each pair of rectifier circuits, and
circuit pairs if appropriate, on the DC side. This device
prevents interaction between the rectifiers by sustaining
the instantaneous voltage difference in their outputs
caused by the phase displacement on the AC side. Nowa-
days, it is usually considered most economical to use a
12-pulse system.
The three possible six-pulse rectifier circuits for this
application are illustrated in Fig. 1. Each circuit may
have to provide direct current up to about
50
kA, and the
physical arrangements for liquid cooling the parallelled
diodes or thyristors often requires the use of a circuit
containing a common cathode connection, i.e. Fig. l a or
b. An advantage of the double three-phase star connec-
tion is that the semiconductor devices carry only half the
load current, but at the expense of an additional IPT.
Fig. 2 shows a typical electrochemical 48.5 kA rectifier
arrangement with parallelled devices, snubbers and
water-cooled busbars. Generally, parallelled circuits
of
the types in Fig. l a or b are used for higher currents at
voltages up to 400V. For higher voltages of about
lo00 V and lesser currents the three-phase bridge is
popular, owing to its considerably lower transformer
rating in relation to the DC output.
125
-
8/10/2019 ## Rectifier TX 00046621
2/16
For continuously variable control of the DC output
voltage, the choice lies between using thyristors, or diodes
D
D E
load
a b
A
load
C
Fig. 1
n
Six-phase, halfwave
b
Double three-phase star with IPT needed due to the
60
hase difference of the
two three-phase circuit
c
Three-p hase full-wave bridge
Some six-pulse rectifier configurations
with flux-reset transductors in the AC-supply lines to the
rectifier. The availability of diodes with higher ratings
than thyristors eases problems of device parallelling, but
has the added expense of magnetic components.
Fig. 2
126
Electrochemical plant 48.5 kA rectrJier
In this paper, the operation of two parallel three-phase
thyristor bridges with a
30
phase offset is considered.
The size and cost of an IPT connected between them in
electrochemical plant is such that, on economic grounds,
it has occasionally been omitted. As stated above, the
resulting penalty
is
the interaction between the bridge cir-
cuits and the consequent circulating AC flowing between
them. The ability to predict the transformer winding and
AC supply current waveforms and establish their harmo-
nic content with reasonable accuracy is obvious, and this
has provided the incentive for this investigation.
2
Operating modes
2.1 Circuit representation
Fig. 3 shows a schematic diagram of the parallel bridges
supplied by two separate transformer output windings
I
=I. .
I
I
%Firnary
I
three-phase
SUWlY
turns
l
''71
B1tertiary
I
load
bridges n parallel
Fig.
3
Three w inding rectifier transformers
Schematic diagram
of
the 12-pulse system
(the secondary and tertiary) with equal open-circuit volt-
ages, and with the AC voltage in bridge 2 lagging that of
bridge 1 by 30 . The number of turns per phase on the
star-connected secondary windings is
1/J3
times that on
the tertiary. The primary winding has a delta connection.
The position of a conventional IPT is shown in broken
lines, although in this investigation it is omitted. The 12-
pulse DC output is applied to the load, which comprises
resistance, inductance and a back EMF, as is typical for
electrochemical cells.
There are two basic transformer arrangements which
may be used for supplying three-phase bridges in the
required manner. These use either
(a)
two separate three-phase transformers having the
same primary winding connection (delta) supplied in
parallel, each with a secondary (one in star and the other
in delta) connected to a bridge or
(b) a single three-phase three-winding transformer with
the star-connected secondary supplying one bridge and
the delta-connected tertiary the other, as depicted in Fig.
3.
The latter is more complicated to deal with analytically,
owing to the larger number of mutual inductances, and is
the arrangement adopted here.
The complete equivalent circuit used for analysis is
given in Fig.
4.
Each of the transformer windings is rep-
resented by a designated branch impedance Z comprising
a resistance and self-inductance. Also included are the
mutual inductances between the windings in each phase
and between the phases, e.g. L,, and L respectively.
Inductances are included for the IPT for completeness,
although these values are zero for the present investiga-
tion. A conducting thyristor is represented by an imped-
I E E P R O C E E D I N G S ,
V o l . 137,
P t . B , N o .
2,
M A R C H
199
-
8/10/2019 ## Rectifier TX 00046621
3/16
ance in series with a constant voltage source. A DC load. Higher operating modes occur during overload
nonconducting thyristor is assumed to have infinite and short-circuit
[2].
In this study, the simplest condition
impedance. The simple conventional three-winding trans- occurs for large thyristor firing-delay angles, when the
L13
L14j L25.L36
424
/
233 L23
1
L39. L17. L28
Fig.
4
z , , =
I ,
+
p L , ,
and similarly for each winding
Equivalent circuit showing impedances
former equivalent circuit is shown in Fig.
5[l].
Although
helpful in showing the important reactances for limiting
circulating currents, it was not considered sufficiently
comprehensive for the analysis adopted here. However, it
is used later to obtain approximate results for compari-
son.
If
the AC side inductances are assumed to be zero and
with no IPT, only the bridge which experiences the
highest instantaneous voltages in any 12-pulse ripple
period of 30 will conduct. Thus the normal conduction
interval of
120
for a thyristor in one bridge is broken up
by the conduction of thyristors in the other bridge. With
the AC-side inductances present, instantaneous transfer
of load current between the bridges (and between phases
within a bridge) is impossible, and commutation effects
occur in various degrees of complexity.
When defining the operating modes of single rectifier
circuits it is conventional to consider as Mode
1
the con-
dition where the load-current flow is continuous and the
commutation angles are small, i.e. less than a pulsewidth,
this being typical of normal operation with an inductive
l t e z
primary branch
secondary branch
Fig. 5
Conventional equivalent circuit of
a
three-winding transformer
(per phase)
I E E P R O C E E D I N G S , V o l . 137, P t . B, N o . 2, M A R C H 1990
load current is discontinuous and therefore no com-
mutations occur. This condition is adopted as Mode
1.
As the firing-delay angle is reduced, the load current
becomes continuous and both the operation and analysis
becomes increasingly complicated, with simultaneous
conduction of first the two bridges and then the rectifier
phases as the firing delay is reduced and current
increases. At each change
of
operation the mode number
rises.
The assumptions made for consideration of these oper-
ating modes are
(i) the AC supply voltages to the transformer are sinus-
oidal
(ii) the thyristor firing-delay angles are all equal
(iii) the transformer reactances presented to the two
bridges are equal.
Thus the bridges operate perfectly symmetrically and
with the required
30
phase displacement.
2.2 M o d e
1
: i scont inuous load current and
discont inuou s br idge currents
Fig. 6 shows the conduction patterns for the five oper-
ating modes considered here and covers 150 of the oper-
ating cycle. The numbers of the conducting thyristors are
defined in Fig.
7.
In Mode
1
the load, bridge and thyris-
tor currents flow in short pulses having a conduction
interval less than 30 . The bridges conduct alternately
into the load, there being one load-current loop present
at a time, say
i,
in Fig.
7.
Typical waveforms are shown in Fig. 8.The conduc-
tion interval is B
- a),
with the firing-delay angle
a
mea-
sured from the zero-voltage crossing of the appropriate
bridge-supply line voltage. This reference is more conve-
nient than the usual definition of
a = 0
when the incom-
127
-
8/10/2019 ## Rectifier TX 00046621
4/16
ing thyristor becomes forward biased, which is 75 later
in a 12-pulse system as shown in this Figure.
bridge 2 4 4 R f 6 Y ' $28:
d 30 60
90
120
150
0
5 I
I I
I I
v/3Y +4R/A/2B*3Y//1/1R+2B//1
V/lR
+ 2BA v2B'+3YYA
-I
2
1
2
Fig.6
Bridge 2 conduction lags bridge
1
conduction by 30
Interbridge commutations are shown
i /b
Interphase commutations are show n by i / p
Overlapping interphase commutations between bridges are shown by
i /pb
Interphase commutation overlap interference within each bridge is shown by
Conduction patterns or various operating modes
m2
2.3
M o d e 2: Cont inuous load c urrent , d iscont inuous
bridge currents and interbridge com mutation s
Mode 2 occurs when the thyristor firing delay is reduced
and the load current rises so that the conduction angle
increases to between 30 and 60 . The transition between
Modes 1 and 2 occurs at a critical firing-delay angle aCl.
w5B
R
Ed
In Mode 2 a thyristor in one bridge is turned on before
that in the alternate bridge has ceased to conduct. This
7 E O
-1-
12v
a
n
Fig.
8 Mo d e
I
waveforms
a = DC output voltage
b = load current
c = bridge
1
current
d = bridge
2
current
e = bridge 2 supply line current 1;
fl = angle of cessation current flow (measured from voltage zero)
initiates the transfer of current from one bridge to the
other, i.e. an interbridge commutation takes place, as
shown in Fig.
6.
Referring to the circuit of Fig.
9,
thyris-
tor 1R and 6Y, say, are conducting the falling current i
in the outgoing bridge and 1R' and
6 Y
carrying the
rising current i in the incoming bridge. The interbridge
commutation is completed when
i
falls to zero, after
which
i
flows alone until the next pair of thyristors in the
sequence (1R and 2B) are fired. A point to note is that
during a bridge 1-bridge 2 commutation, devices in the
same phases are coming into conduction, e.g 1R' and 6 Y
take over conduction for 1R and
6Y.
During a bridge
2-bridge 1 commutation the third phase is involved, e.g.
1R and 2B take over from
1R
and 6Y'. Hence the trans-
former current paths are different in the two com-
mutations.
Fig. 10shows typical waveforms. The continuous load
current is made up of two alternating parts
r~
Fig.
7
Secondary and tertiary windings shown
128
Typical current loop paths o r Mo de I
Fig.
9
Secondary and tertiary windings shown
Typical current loop paths o r mode 2
I E E P R O C E E D I N G S , Vol . 137, P t .
E,
N o .
2,
M A R C H 1990
-
8/10/2019 ## Rectifier TX 00046621
5/16
( a )
the sum of the two bridge currents, one rising, the
other falling, during the interbridge commutation interval
Y
-
a)
( b )
a single bridge current similar to Mode 1.
75
a
n
3Y+4R 4R+5B
C
I
e
- in bridge2
Fig.
10 M o d e
2
waveforms
a
= DC output voltage
b
=
load current
c = bridge 1 current
d = bridge 2 current
e =
bridge 2 supply line current 1;
y = angle at completion
of
interbridge commutation
2.4
M o d e 3: Cont inuous load current , cont inuou s
bridg e currents and separated interphase
com mutat ions a l ternate ly in each br idge
A reduction of the firing-delay angle to a second critical
value a,* , defined by the bridge conduction intervals
reaching
60 ,
results in the whole of each 30 output
voltage pulsewidth being taken up by an interbridge
commutation. Operation moves into Mode 3 when
a
is
less than
aC2
and the conduction intervals within each
bridge overlap. Fig.
6
shows the conduction pattern.
Referring to Fig. 11,the sequence of events is defined
as follows
:
(i) let an interbridge commutation occur with bridge 2
incoming and bridge 1 outgoing; loop currents
ib
and i,,
respectively, flow in the bridges
7 7
50
i
0
I
I
I
Fig.
11
Secondary and tertiary windings shown
I E E P R O C E E D I N G S ,
Vol .
137, P t . B, N o . 2, M A R C H 1990
Typica l current loop paths fo r Mo de 3
(ii) before the interbridge commutation is completed,
the next thyristor is sequence 2B in the outgoing bridge is
fired and an interphase commutation commences,
6Y
current transferring to 2B with an additional loop
current
i,
flowing
(iii) on completion of the interphase commutation,
i,
has fallen to zero leaving an interbridge commutation
with the bridge roles reversed, i.e. the previous incoming
bridge now being the outgoing one and loop currents
i,
and
i,
remaining. The interphase commutation angle
< 30 .
Fig. 12 shows typical waveforms. Now the bridge cur-
rents are continuous, and the individual thyristors
75
a
n
b
C
0
I I
I
d
commutations
in bridge 2
-
Fig.12 M o d e 3 waveforms
a =
DC
output voltage
b
=
oad current
c = bridge 1 current
d
=
bridge 2 current
e
=
bridge 2 supply line current
ry
6 =
angle of completion
of
interphase commutation within a bridge
p
=
nterphase commutation angle
conduct for an angle of between 120 and
150 ,
as the
two pulses of Mode 2 have widened and merged. Since
commutations are taking place throughout the operating
cycle, the DC output voltage at no time follows the sinus-
oidal AC supply, but is at some intermediate value
between the appropriate phase voltages depending on the
commutation conditions.
2.5 M o d e
4:
Cont inuous load current , cont inuo us
br idge currents and in terphase commutat ions n
the br idges par t ly over lapping
With further reduction of firing delay and increased
current, the interphase commutation angles
p
in the
bridges increase to 30 at the third critical firing-delay
angle aC3.Operation then moves into Mode 4 f p
> 30,
with the individual bridge interphase commutations over-
lapping each other by an angle p
- 30) ,
as shown in
Fig.
6.
Again, a new loop current flows during this inter-
val, shown as id in Fig. 13. The stages of operation are as
follows:
(i) first, let an interphase commutation take place in
bridge 1 with current t ransferring from thyristor
6Y
to
2B;
i,
is rising and
i,
falling. Bridge 2 carries a single
current
ib
129
-
8/10/2019 ## Rectifier TX 00046621
6/16
(ii) before stage (i) is completed thyristor
2 B
in bridge
2 is fired, starting at
6Y'
to 2B' interphase commutation
with loop current
id
increasing and
i
decreasing
(iii) the interphase commutation is completed when i,
falls to zero, leaving the single current loop
i
in bridge
1
and the existing interphase commutat ion proceeding in
bridge
2.
R
L
Fig. 13
Secondary and tertiary windings
Typical current loop pathsfor M o d e
4
0
0
0
e
0
Fig.14 M o d e 4 w a v e f o r m
a =
DC output voltage
b = load current
c
= bridge
1
current
d = bridge
2
current
e
= bridge I supply line current
I
t
= angle
of
completion of coincidence of interphase com mutations between the
bridges
130
Typical waveforms are shown in Fig.
14.
The interphase
commutations overlap for an angle (I+
a) ,
and each
thyristor now conducts for an angle of between 150 and
180 .
An interesting feature is that the convetor can operate
with a thyristor firing-delay angle of less than 75
because the incoming thyristor becomes forward biased
earlier in the cycle than would occur with purely sinus-
oidal voltages applied to the bridges. The degree of such
firing 'advance' depends on the AC voltage distortion at
the bridge terminals and on the load current. This effect
was observed experimentally as results show later.
2.6 M o d e
5:
Cont inuous load current , cont inuous
br idge currents and comp lete over lapping of
br idge in terphase com mutat ions w i th o ver lap
interference in each b ridge
If the firing-delay angle is further reduced until the inter-
phase commutation in each bridge is
60,
both bridges
are permanently in a state of interphase commutation,
with three thyristors per bridge conducting. The critical
value
aC4
denotes the onset of Mode
5,
for which p is
greater than
60 .
In this mode, when the next (fourth) thyristor is gated,
this must necessarily be associated with the same phase
as that carrying the outgoing interphase commutation
current,
i,
in phase
Y
shown in Fig.
15.
As long as
i
flows in thyristor
6Y,
the newly gated thyristor
3Y
cannot conduct since its anode is negative with respect to
its cathode. A period of interphase overlap interference
follows until the outgoing thyristor
6Y
current
i,
falls to
zero. Then the incoming thyristor
3Y
commences to
conduct current i giving the start of a new interphase
Commutation. Hence all interphase commutations are
constrained to be
60
in Mode
5.
Such commutation
overlap interference is well known in single bridges [2].
Fig.
6
again gives the conduction pattern, with each thy-
ristor conducting for
180 .
For
typical values of AC-side inductances, interphase
commutations approaching
60
only occur under abnor-
mally high overload conditions and Mode 5 will not be
considered further.
Table 1 summarises the conduction conditions in the
various modes.
Fig.
15
Secondary and tertiary windings shown
Typical current
loop
paths
for
projected Mode 5
I E E P R O C E E D I N G S , Vol. 137, P t .
B,
N o . 2, M A R C H 1990
-
8/10/2019 ## Rectifier TX 00046621
7/16
Table 1 Showin g commutat ion angles and interact ion between br idges
~~
Mode Device Bridge Interphase Overlapping of Interphase commutation
conduction conduction commutation bridge interphase overlap interference
within a bridge, commutation within each bridge
angle p between bridges
1 2 x
0 -+
30 Independent -
2
2 ~ 3 0 + 6 0 - - -
B
- 0
pulses
Y
a
bridges
I
Ibridges
bridges
3 1 x
120
-+ 150 Continuous
0
-+
30
-
pulse overlapping
(p
=
6
U)
between
150 +180 Continuous 30 + 60 0 --+ 30
overlapping ( p )
( w -
a) =
(p
- 30 )
between
5
180 Continuous 60
-+ 90
30 0
-+
30
overlapping
(p)
Continuous
between
3
Outl ine of analysis
3.1 In t roduct ion
The analysis is based on Kron's tensor methods, modified
to assist in the handling of the device conduction patterns
[3],
and it incorporates all circuit elements without any
simplifying assumptions such as infinite DC-load induc-
tance. The description of the model is limited to an
outline of the technique, which is covered fully in Refer-
ence 4. The only assumptions made are those stated in
Section 2.1, which provide symmetrical operation of the
bridges. The equivalent circuit adopted is tha t of Fig.
4.
3.2 Method
The technique uses tensor analysis with the circuit mesh
currents defined as the system state variables. The con-
nection tensors, expressed in relation to the conducting
thyristors, are used to assemble the mesh state variable
equations.
The analysis involves an algorithm, which
(a) assembles automatically, for any thyristor conduc-
tion pattern, the closed-path or mesh (loop) equations in
state variable form
e15
_ -
e2
V21
e 2 1
2
I21
1%
--
Fig. 16 Primitive referenceframe
(b )
solves the equations using a numerical integration
method
(c)
determines the individual branch currents and volt-
ages from the mesh values.
Assembly of the mesh contour equations requires three
references frames
(i) the primitive
or
branch reference frame
(ii) the intermediate reference frame
(iii) the mesh reference frame.
3.3 Primitive reference frame
The primitive reference frame is concerned with individ-
ual branches of the network, as shown in Fig. 16. The
system-voltage equation in abbreviated matrix form is
v b Eb
=
Z b b I ,
(1)
The elements of the separate matrices for the transformer,
thyristors, IPT and load are given in Appendix 11.1,
(eqns.
9,
10 and 11). The main diagonal terms contain the
branch resistances and self-inductances defined in Fig.
4,
and the
off
diagonal terms contain mutual inductances. I t
should be noted that the laboratory transformers used in
this study were single-phase three-winding units and the
I E E P R O C E E D I N G S ,
V o l . 137,
P t . B , N o . 2, M A R C H 1990
131
-
8/10/2019 ## Rectifier TX 00046621
8/16
mutual inductance terms in the I,,,
, L14 L7,
groups of
Fig. 4are zero.
-
0 1
0 0
0 0
1 0
-1
0
O 0
1
0
0 1
1 0
1 1
-
3.4 intermediate reference frame
Fig. 17 shows the branches of the primitive reference
frame connected together to form the circuit defined by
the intermediate reference frame. Applying Kirchhoff
s
current law to the nodes defines a current transformation
between the branch and intermediate reference frame cur-
rents reflected into the secondary and tertiary windings,
to give the relationship
b
= (2)
The complete form of
Cbi
s given in eqn. 12 in Appendix
11.1. The intermediate branch currents are related by
E, +
Vi
= ZiiZi
(3)
Assuming power invariance, the following relationships
apply:
(4)
b = cbIzI
vi
=
CLi
v
E, = cL,E,
ziI = CL zbb
noting that matrix CL, is the transpose of matrix C,,
.
[ I
b
tion in which thyristors l R ,
6Y, R
and
6Y
are conduct-
ing. Three mesh currents
i, i,
and ib are required, as
shown in Fig. 9.The intermediate reference frame cur-
rents (Fig. 17)
Z
are related to the mesh currents
Z
by
the transformation tensor
C,,
as follows:
The complete master conduction transformation tensor
C,,, of which the above forms part, is given in eqn. 13 in
Appendix 11.1. It contains the twelve meshes which
describe all practical thyristor conduction patterns, with
positive mesh current defined such that thyristor extinc-
tion is indicated when the current becomes negative.
T
T I
10
I10
11
=Id
6 14
122-110-112=114 v2
11
3.5
Mesh reference frame
The mesh reference frame is concerned with the meshes
formed when individual thyristors are conducting. Kron's
'equation of constraint' technique [ S ,
6 ,
71 is used used,
whereby the currents of nonconducting devices are set to
zero. A new transformation tensor is defined, which
relates the intermediate reference frame to the mesh refer-
ence frame currents and introduces the changes in circuit
topology produced by switching. This new matrix
C,,
is
a reduction of
c b , .
The number of mesh currents needed for a particular
topology [7] is
M = B - N + S 5 )
where
M
is the minimum number
of
independent meshes,
B is the number of branches, N is the number of nodes
and S is the number of electrically isolated subnetworks
S=
1 here). As an example, consider the Mode
2
condi-
132
The transformation tensor C,, is dynamic, since its
elements vary with thyristor switching, and is assembled
automatically by the computer program. The matrix
equation relating the mesh currents and voltages is
E , + v = z,,z,
(7)
From Kirchhoffs voltage law
V ,
=
0.
the mesh and intermediate reference frames
Similar relationships to those of eqn. 4 apply between
(8)
=
C i m
1,
v = c:, vi
E , = C:, E,
z,, = CILZrrC,,
The mesh equation is solved numerically using a 4th-
order Runge-Kutta routine.
I E E P R O C E E D I N G S ,
Vol . 137,
P t .
B,
N o . 2,
M A R C H 1990
-
8/10/2019 ## Rectifier TX 00046621
9/16
4 AC side-current waveforms
This section presents computed results which are com-
pared with experimental results to validate the program,
and to confirm the waveshapes of the operating modes
described in Section 2.
The circuit parameters used in the computer program
are given in Tables
4,
5 and 6 in Appendix 11.2, obtained
from measurements on the laboratory convertor. Reason-
able balance of the voltages and inductances of the three
single-phase transformer units was achieved and averages
values used as data. The conventional three-branch per
phase equivalent circuit (Fig. 5) data is included in Table
7.
The importance of the values of the secondary and ter-
tiary branch impedances will be referred to later.
Figs. 18, 19 and 20 show for Modes 2, 3 and
4,
respec-
tively, predicted and experimental waveforms for the
bridge-supply line (transformer secondary phase) and
transformer tertiary-phase currents. In addition, the
transformer primary side currents are included for Mode
3. The correlation is quite acceptable in view of the
experimental equipment limitations and confirms the
waveforms given in Figs. 10, 12 and 14. It is interesting to
note that the firing-delay angle for Mode
4
operation was
60 ,
which is less than the minimum
75
expected, due to
the bridge AC terminal-voltage distortion described in
Section 2.5.
Table 2 gives the predicted and the extremes of mea-
sured values for the first six characteristic harmonics in
the bridge 1 supply line-current waveforms for the four
operating modes. Again correlation is generally accept-
able, although there is considerable deviation in a few
cases. This is due to slight differences between the com-
puted and measured conduction periods of the thyristors
caused by inaccuracy in the measured parameters, phase
*
O
a
-5-
5 t
oo
b.kd
1.b?o-l.h
time,s
x10-2
a
-5-
-15 I
O-O.;o bo
100 1z 5 0 time,s
x10-2
a
5
-lo[
15
b
-
1
1
1
6
'able
2:
Computed bridge
1
supply l ine-current harmonics
ixpressed as p.u. of the fund amental w i t h th e t w o extreme
neasured values fr om the t hree phases shown in br ackets
irin g-delay Harmonic order
ingle a
5th 7th 11th 13th 17th 19th
65
node 1
50
node 2
20
Aode 3
io
Aode 4
0.860
(0.863)
(0.795)
0.640
(0.630)
(0.059)
0.290
(0.260)
(0.240)
0.085
(0.1 02)
(0.095)
0.735
(0.685)
(0.614)
0.380
(0.460)
(0.405)
0.030
(0.022)
(0.021
0.049
(0.050)
(0.046)
0.430
(0.508)
(0.455)
0.009
(0.080)
(0.081)
0.069
(0.056)
(0.060)
0.036
(0.020)
(0.023)
0.280
(0.355)
(0.318)
0.056
(0.031
(0.030)
0.040
(0.034)
(0.036)
0.030
(0.024)
(0.021)
0.046
(0.173)
(0.168)
0.01 9
(0.066)
(0.73)
0.039
(0.047)
(0.037)
0.006
(0.020)
(0.017)
0.031
(0.069)
(0.082)
0.01 4
(0.046)
(0.041
0.026
(0.023)
(0.025)
0.008
(0.017)
(0.016)
imbalance and imprecise adjustment of the firing-delay
angles.
The characteristic harmonics present in each six-pulse
bridge-supply current waveform (5th, 7th, 17th, 19th etc.)
are absent in the transformer supply current for the
'perfect' 12-pulse convertor system with commutation
angle neglected. Here the computed characteristic and
uncharacteristic harmonics for the supply line current are
shown in Fig. 21a and
b
with perfect phase voltage,
impedance and firing balance assumed. The experimental
transformer impedances are slightly unbalanced, resulting
in the presence of additional uncharacteristic harmonic
levels which are, however, too low to be distinguishable.
5
System operating ch aracteristics
Various system operating characteristics are shown in
Figs. 22 to 26. The DC output-voltage regulation charac-
Fig. 18
Frequency
= 50 Hz
Computed and observed waveforms od e 2
a Bridge supply line current
a = 150
I = 0.75 A
R = 5 . 1 n
I =
0.75
A
L = 0.0174H
I
= 1.5
A
E,
=
24 V
V,
= 32.4
V
IEE PROCEEDINGS, Vol.
137,
P t . E ,
No. 2,
M A R C H
1990
b Tertiary w inding current
Oscillogram scale
Horizontal ms/div
Vertical A/div
133
-
8/10/2019 ## Rectifier TX 00046621
10/16
teristic with increasing load current (Fig.
22)
is obtained
by reducing the load resistance for three set values of
firing delay, to give operation in Modes
2 ,
3 and 4.The
knee in the lower characteristic may be attributed to the
load conduction becoming continuous and interbridge
commutation effects commencing. The voltage drops are
mainly due to commutation and amount to 20% or less
for the two higher modes.
The increase in thyristor conduction interval with
reduction of the firing-delay angle for a set load imped-
15-
10
5-
-10
-15-
15
tirne,s
i=
-5 x10-2
-
-;Lx-hm+l~o.2
irne,s
-
-lot1
5
tirne,s
x10-2
-lo[
15
b
C
time,s
2 -5
-101
-1 5L
Fig.
19
Frequency = 50 Hz
Computed and observed waveforms ode 3
a
Bridge supply line current
a
=
120
R = 5.1 R
L = 0.0174 H
E ,
= 24 V
I = 4.78 A
I
= 5.46 A
I
= 10.24 A
V,
=
76.8 V
b Tertiary winding current
c Supply line current
d Primary phase current
Oscillogram scale:
Horizontal ms/div
Vertical A/div
134
I E E P R O C E E D I N G S , V o l . 137, P t . B,
N o .
2, M A R C H 199
-
8/10/2019 ## Rectifier TX 00046621
11/16
ance is evident in Fig. 23, as is the sharp jump from 60
to
120
at the Mode 2-Mode 3 change when the two 60
current pulses/cyc le merge.
-
0 .6 -
2
0 4 -
u
U
E
5
c
time,s
U
-10
L.
10-2
-201
-30L
t ime,s
J -10 x10-2
- O I
30
Fig. 20
Frequency = 50 Hz
Computed and observed waveforms
~
Mode 4
a Bridge supply line current
a = W I = 12.17 A
R
=
5.1 R I = 12.25 A
L = 0.0174 H
I
= 24.42 A
V, = 150.1 V
= 24 V
The increase of interbridge commutation angle from
zero 30
is shown in Fig. 24, and the bridge interphase
b
60 80 100 120 140 160 180
firing delay angle, degree
a
b
Tertiary winding current
Oscillogram scale:
Horizontal
~
2 ms/div
Vertical
~
10 Aidiv
Fig.
21
a
Characteristic a 1 lth component b A 5th component
b Uncharacteristic x 13th component x 7th component
Computed AC-supply line-current harmonics
17th component
firing delay angle,
degrees
b
Load data:
R = 5.1 R
L
=
0.0174
H
E , = O V
IEE PROC EEDINGS , Vol . 137,
P t .
B , N o .
2,
M A R C H 1990
135
-
8/10/2019 ## Rectifier TX 00046621
12/16
commutation angles within the bridges and the overlap-
ping of these between the bridges are given in Figs. 25
and 26, respectively. The critical firing-delay angles at
which mode changes occur are well defined by these
characteristics. A load EM F of 0 V was chosen for these
simulations, to allow the highest safe load current of 30A
to flow.
It is evident that differences exist in the computed
values of thyristor conduction and commutation angles
for the two bridges, which is surprising since balanced
(averaged) voltage and impedance values were used as
data. The differences are due to the fact that, for a per-
fectly symmetrical transformer, the resistance and self-
inductance values of the delta output winding should be
three times those of the star output winding and,
assuming the same coupling coefficient, the mutual
inductances between primary and delta output winding
should be
,/3
times those between the primary and the
star output winding. The adjustment of the experimental
transformer inductance and resistance balance to these
relationships was not perfect, with the deviation of the
resistance values being greatest, as can be seen by com-
paring the relative values in Table 5 (Appendix
11.2).
Computed results with suitably adjusted resistances
values show identical operation for the two bridges.
160-
140
v1
;
20-
ul
U
-
9
100-
t
.e
8 0 -
-
C
U
D
.-
-
60-
4
20-
6
With an interphase transformer present between the
bridges on the
DC
side, they operate without interaction
or specifically with very little circulating current between
them, depending on the value of the IPT inductance and
the firing-delay angle
a.
The circulating current is highest
Comparison wi th parallel bridges having an
IPT
-
-
140
>
*
3
a
3 100-
80
60-
4
0
10
20 30
40
0
load current, A
Fig.
22 Load characteristic
A computed a = 60 Load data:
x
experimental a = 60 L = 0.0174
H
computed a = 75 E , = O V
experimental
a = 75
R
=
variable
A computed a
=
120
experimental a
=
120
136
at larger firing-delay angles. Industrial installations nor-
mally operate in Mode
3
without an IPT, so the condi-
mode
1
two
pulsesof current
?-
9 0
$0 i o
lb 1o
i o + l k O lb
firing delay angle degrees
Fig.
23
A bridge
1
Load data:
x
bridge 2 ]computed R = 5.1 R
L
=
0.0174 H
bridge
1
-experimental
E , = O V
Thy risto r conduction interval characteristic
Ln
U
-
2
C
C
.-
._
c
c
i
E
8
firing del ay angle, degrees
Fig.
24
Interbridge commutation characteristic ode 2
Load data:
: Lzgi ] omputed
bridge 1 -experimental L = 0.0174 H
R = 5.1 R
E , = O V
delay
I E E P R O C E E D I N G S , Vol . 137, P t . B ,
N o .
2, M A R C H 1990
-
8/10/2019 ## Rectifier TX 00046621
13/16
tions shown in Fig. 19with
U =
120 will be adopted for
consideration with an IPT present, each half winding of
p
ritical firing angle uc3
firing delay angle, degrees
Interphase commutation characteristic ode 3
ig.
25
: ]
omputed
bridge
1
-
xperimental
L = 0.0174
H
Load data:
R
=
5.1
R
E , = O V
;a
critical firing delay
75 80
firing delay angle, degrees
Fig. 20
Overlapping interphase commutation characteristic - o d e
4
Load data:
: iz
]computed
R = 5 . 1 R
bridge 1 -experimental L = 0.0174
H
E , = O V
I E E P R O C E E D I N G S , Vol .
137,
P t . B , N o .
2,
M A R C H
1990
which will be assumed to have a self-inductance of
200 mH.
Fig. 27 shows the waveform of the bridge supply
current calculated by conventional methods with the rec-
tifier transformer commutating inductances obtained
from the averaged values quoted in Table 7. The com-
mutation angle is
5.9 .
The relatively small components
of the load-current ripple 0.24 A peak) and circulating
current
(0.13
A peak) are not shown.
Fig.
27 Calculated bridge supply current waveform with IPT present
Typical values of the first six harmonic components of
this waveform are given in Table
3.
Table 3: Predicted bridge-supply l ine-current harmonics as
a fraction of t he fundamental wit h IPT present
Firing delay- Harmonic order
angle a
5th 7th 11th 13th 17th 19th
120
0.198 0.141 0.086 0.071 0.050 0.045
It is evident from comparing these harmonic levels with
those of Table
2
for
U =
120
that omission of the IPT
results in considerably reduced values for all orders
except the 5th, which increases by about
50%.
Compari-
son of the waveforms (Figs.
19a
and 27) clearly demon-
strates the heavy circulating-current component with no
IPT, resulting in this substantially higher 5th harmonic.
7 Implications for transformer design
It is evident that, from the current-harmonic viewpoint,
omission of the IP T is beneficial except for the 5th har-
monic. With greater firing-delay angles, this can be
increased by more than the
50
demonstrated here, 75%
being observed with transformers having low-leakage
reactance.
The 5th and 7th harmonic components flowing in the
star and delta output windings of the transformer are in
antiphase, and
so
therefore are the associated flux com-
ponents. These are at
90 to
the preferred direction
of
the
main fundamental leakage flux relative to the winding
conductors. The much lower 7th harmonic component
gives no difficulty, but the increased 5th harmonic flux
can cause severe localised eddy-current heating in the
windings.
The transformer design must therefore be adapted, if
the IPT is omitted,
so
that the circulating current, and
thus the 5th harmonic winding current components, are
10%
a b
Fig.
28
three-winding equivalent circuit
B WithIPT
b Without IPT
Typ ical percentage reactance values for the conventional
137
-
8/10/2019 ## Rectifier TX 00046621
14/16
restricted. This takes the form of reconfiguring the wind-
ings to give relatively high leakage between the two
output windings on each phase. The change in leakage
flux distribution is well illustrated by the different reac-
tance values in the conventional three-winding equivalent
circuit. Fig.
28
illustrates percentage reactances which are
typical for an industrial transformer supplying parallel
bridge rectifiers. For
a,
each transformer phase may have
a single primary winding and multiple flat coils in alter-
nate layered sandwich form for the two output windings
to give low leakage between the latter. For
b ,
a primary
winding in two parallel-connected halves may be used,
with separate single output windings positioned end to
end for increased flux leakage between them.
8 Conclusions
Four operating modes for two 6-pulse bridge rectifiers
connected in parallel, without a DC side interbridge
reactor (IPT), have been described and analysed. A fifth
mode is postulated. The interbridge and interphase com-
mutation effects are defined, and the AC side current
waveforms and rectifier characteristics are predicted, by
the mathematical model, and verified experimentally
using a laboratory converter. Good correlation is evident
between computed and measured results, thereby con-
firming the analysis and computer program. Small dis-
crepancies are explained by imperfections in the balance
and symmetry of the laboratory equipment. Large-scale
industrial plant is expected to be much better in this
respect.
A rigorous modelling technique using tensors has been
adopted, which has the following features
:
( a )
all circuit impedance components are included and
have finite values
( b )
he DC load has all the typical elements of electro-
chemical plant ; resistance, inductance and back EM F
(c)
the analysis has been developed from first prin-
ciples
:
no separately derived equivalent circuit is used
(d)
the mathematical model includes all mutual induc-
tance terms, including those for a three-phase trans-
former and an IPT, and is therefore complete, although
three single-phase transformers and no IPT were used
here
(e )
perfect balance between the three phases and sym-
metrical firing of the thyristors are assumed.
The requirement for transformer da ta which include self-
and mutual-inductance values and not p.u. or percentage
winding leakage reactances is unconventional in terms of
industrial practice, where the three-branch equivalent
circuit per phase is more usual. The latter, however, is
not sufficiently comprehensive, since it neglects the inter-
phase coupling of a three-phase transformer unit.
The bridge-supply current waveform for typical oper-
ating conditions is predicted by conventional methods for
the case with an IPT included to compare the waveform
harmonic contents.
It is demonstrated that the absence of an IPT need not
be a handicap to the successful operation of two parallel
bridge rectifiers. However, it is necessary to modify the
rectifier transformer design to provide high-leakage reac-
tance between the two output windings to limit the inter-
bridge circulating current and hence the troublesome 5th
harmonic fluxes in the transformer. In simple terms, it
can be said that these high-leakage reactances act like an
IPT in this respect.
0
0
r +PL33
0
0 r4 +p L 4 4
0 0
0
PL47
0 0
PL36 0
PL39 0
9 Acknowledgments
Acknowledgments are made to the Department of Elec-
tronic and Electrical Engineering and the Computer
Centre of Loughborough University of Technology for
the provision of facilities, and to GEC TDPL, Stafford,
and Mr G. E. Snazell in particular for financial support,
helpful discussions and permission to reproduce Fig.
2.
10 References
1 LANGLOIS-BERTHELOT, R.
:
Transformers and generators for
power systems (Macdonald, London, 1960)
2 RAMAMOORTY, M.: An introduction to thyristors and their
applications (Macmillan Press Ltd., London, 1978)
3 KETTLE BOROUG H, J.G., SMITH, I.R., and FANTHO ME, B .A.:
Simulation of a transformer/rectifier unit for aircraft power supply
systems,
IEE P roc . B,
1982,
129,
(6), pp. 323-329
4 RAZAK, A.B.M.J.: Parallel operation of bridge rectifiers without an
interbridge reactor. MSc T hesis, Loughborough University of Tech-
nology, 1985
5
KRON, G .: Tensors
for
circuits (Dove r, 1959)
6 WILLIAMS,
S.,
and S MIT H, I.R .: SCR bridge converter computa-
tion using tensor methods, IEEE
Trans.,
1976, C-25 pp. 1-6
7 HAP P, H .H.: Diakoptics and networks (Academic Press, 1971)
11 Appendix
1 1 . 1
Matrices develop ed in the analysis
(a ) Primitive reference fram e:
the following are the
primitive reference frame matrices containing the branch
equations for the network components of the transformer
(eqn.
9),
thyristor branches (eqn.
10)
and IPT (if present)
with load (eqn. 11)
b)
Transformation matrix
C,,
for primitive to interme-
diate reference frame:
this matrix (eqn. 12) is used in eqn.
4
for transforming the primitive branch matrices into the
complete network (intermediate) frame
( c ) Intermediate branch to mesh transformation matrix
C,,,,:This matrix is used in eqn.
8
to transform the
network frame into the final mesh frame for solution and
it incorporates the sequential switching of the thyristors
0 0
PL17 0
0
0
0
0
0
PL36
r6 PL66
PL69
0
0
0
0
0
0
PL47
r7 +PL77
138
(9)
I E E P R O C E E D I N G S , Vol. 137,Pt. B, No .
2,
M A R C H 1990
-
8/10/2019 ## Rectifier TX 00046621
15/16
r10 10
+PLlO 10
_ _
r20
20
+PLZO 20
r 2 1 2 1
PL2121 - -
r 1 2
12
+
PL12 12
...
'10 -
'1 1
I 2
'1 3
4
115
'1 6
1
7
'1
8
'1
9
1 20
1 2 1
-
...
..
..
...
...
...
r 2 2
22
+ P L 2 2 22 PL 2 2 2 3 0
P L 2 2
23 r23 23 + PL23 23 0
0 0
4
5
6 7
8
10 12 16 18 22 23 24
1
- -
0 0
- -
7 0 0
0 0 0 0 0 0
Nl
l
2
0 --
0
0 - -7 0 0 0 0 0 0 0
Nl
l
3
0 0 - - 3
3
0 0
0 0 0 0 0
NI NI
Nl
4
1 0
0
0
0
0 0
0
0 0 0 0
5 0
1
0 0
0
0
0
0
0 0 0 0
6 0
0 1
0
0
0
0
0 0 0 0 0
7
0
0 0 1
0
0
0 0 0 0 0 0
8 0
0 0
0
1 0
0
0
0 0 0 0
9
0
0 0
-1
-1 0
0 0
0 0 0 0
10
0
0
0 0 0
1 0
0 0 0 0 0
c,,= 11
-1 0
1 0 0
1 0
0 0 0 0 0
12
0 0
0 0 0 0
1 0
0 0 0 0
13
1 -1
0 0 0 0
1 0
0 0 0 0
14
0
0
0
0
0 - 1 - 1
0
0 1 0 0
15
0
1
-1 0
0 - 1 - 1
0
0 1 0 0
16
0
0
0
0 0 0 0
1 0 0 0 0
17
0 0
0 -1
0
0 0
1 0 0 0 0
18
0 0
0 0 0 0 0 0
1 0 0 0
19
0
0 0
0 -1 0 0 0
1 0 0
20
0
0 0
0 0 0
0 - 1 - 1 0 1 0
21
0
0 0
1
1 0 0 - 1 - 1 0 1 0
22
0
0 0
0 0 0 0 0
0 1 0 0
23 0
0 0
0
0 0 0 0 0 0 1 0
24
0
0 0
0
0 0 0 0
0 0 0 1
I E E P R O C E E D I N G S ,
Vol. 137,
P t .
B ,
N o . 2, M A R C H
1990
139
-
8/10/2019 ## Rectifier TX 00046621
16/16
Conduction path
10 10 12 12
14
14 16 16 18 18
20
20
thyristors
13 15 15 11 11 13 19 21 21 17 17 19
loop +
a b
c d
e
f g h i j k
Branch
number
4 1 0 0 - 1 0 0
0 0 0
0 0 0
5 0 0 1 0 0 - 1 0 0 0 0 0 0
6 0 - 1 0 0 1
0 0 0 0
0 0 0
7
0 0 0 0 0 0
1 1 0 - 1 - 1
0
8 0
0 0
0 0
0 - 1 0 1 1 0 - 1
c,,
=
10 1 1 0 0 0
0
0 0 0 0 0 0
12
0
0 1 1 0 0 0 0 0 0 0 0
16 0 0 0
0 0 . 0
1 1 0 0 0
0
18
0 0 0
0 0 0
0 0 1 1 0
0
22 1 1 1
1 1 1 0 0 0
0
0 0
23
0
0 0
0 0 0
1 1 1 1 1 1
24 1 1 1
1 1 1 1 1 1 1 1 1
Note: Other branches
are
omitted
here
as their effect
i s
included
in
the electrical connections transformation matrix.
(13)
1 1.2Experimental equipment data
(a) Transformer: Measured values of transformer
open-circuit voltages, turns ratios, resistances and induc-
tances, together with other circuit data are as below
Table 4: Measured transformer open-circuit voltages and
turns ratios
Primary winding (delta) line voltages
Secondary winding (star) line voltages
50 H z
R -Y
Y - B
Average
R -Y
B -R
Average
B R
Tertiary winding (delta) line voltages
Y - B
Turns ratios
NJN
Turns ratios
N , / N ,
240.0
V
129.5
V
129.8
V
129.0
V
129.4 V
129.0 V
129.8
V
129.5 V
129.4 V
0.539
0.31
1
Table 5 : Average measured transfor mer d ata used (Fig. 4)
Resistances, Q Self-inductances, H Mutual inductances, H
r ,
=r,=r ,=0.708 L,, L2,=L3,= 1.66 L,,= L,,=L,,=0.901
r 4 = r 5
= r ,
=0.289 L,,= L,,
=
L,, =0.5 06 L,,=L,,=L,, =0 .523
r 7 = r , = r 9
=0.307
L,,
=
L,,
=
L,, =0.279
,,= L,,
=
L,, =0.170
Table 6 : Measured thyristor branch and load data used (Fig.
4)
Thyristor model Load
No
IPT
e thv =
E
=
24.0 V
or
0 V
Z,,,, =
0
L?hv
=
L
=
0.01 74 H Z,,,,
=
0
R,,, =
0.1
Q
R
=
5.1
Zn,,
=
0
Table
7:
Measured transformer equivalent circuit param-
eters referred to the rectif ier side (Fig.
5)
Branch Transformer
A
Transformer B Transformer C
parameters
r , ,
-0.266
-0.207 -0.237
L,,
mH -3.075
-3.024 -3.1 77
r,,
Q
1.096
0.962 0.895
L,,
mH 8.321
8.270 8.203
r,,
0.994
0.895 0.949
L,,
mH
8.407
8.270 8.270
Erratum
AL ZAHAWI, B.A.T.,
JONES,
B.L., and DRURY,
W . :
Analysis and simulation of static Kramer drive under
steady-state conditions,
ZEE Proc. B,
1989, 136,
(6),
pp.
In this paper, the following correction should be made to
the text:
On page 285, eqn. 23 should read:
T = p{i,[-LL,i,
-
? M ( i , c ) / 2 ]
+
i , [ L , i ,
+
M i ,
-
M ( i ,
+
i c ) / 2 ] }
The results and conclusions of the paper are in no way
affected by this correction.
28 1-29 1
140
I E E P R O C E E D I N G S , Vol .
137,
P t . B , N o . 2, M A R C H 199