Professional Documents
Culture Documents
tion of an RPC or SVC to a load may allow addition of substantial new REACTOR CAPACITORS LOAD
load to existing feeders or substations. A reactive-power compensator SvC
suited to industrial ratings (1.0-25 MVA) is described. It utilizes a force- I2 2
commutated current-source bridge to provide both leading and lagging O - 20 MVAr 20 MVAr 0-20 MVAr
LAG LEAD LAG
reactive power. The ability to operate both leading and lagging can reduce
by 2:1 the ratings of the RPC itself and the capacitors and magnetics (a)
associated with it. The characteristics of the power circuit, the means used
to control it, and the resulting dynamic performance is described. Speed
of response compares favorably to the thyristor-controlled reactors now
in common use as SVC's at higher MVA ratings. This RPC is suited to
compensate any balanced three-phase dynamic load. +10 MVA
FOR-CE 20
2 MVA
COMMUTATED
-"& |_1 -
LOAD
Y
RPC 10 VAr I
INTRODUCTION b
CAPA TORS .~~~~~~~~~~~~~~
10 MVAr LAG to 10 MVAr 0-20 MVAr
YNAMIC static var compensators are increasingly being 10 MVAr LEAD LEAD LAG
used to raise the amount of load that can be carried by a (b)
given utility power installation [1]. The compensators to be Fig. 1. Equivalent installations of static var compensation, TCR, and RPC.
discussed herein are usually located adjacent to a large cyclic (a) Compensator draws lagging current only. (b) Compensator draws
or intermittent load. They are controlled to cancel the effect of leading or lagging current.
the load on the voltage of the bus, so that even large cyclic
loads will not introduce voltage flicker into nearby sensitive illustrates this. The example load requires 0-20 MVA reactive
loads. Proper application of such a compensator can allow (MVAr) of varying lagging reactive power. The conventional
substantial expansion of an existing plant without the necessity method of compensation is shown in Fig. 1(a) using 20-MVAr
of increasing the rating of the utility feeders and transformers capacitors and a 20-MVAr static var compensator of the
supplying a plant. Dynamic compensation of reactive power thyristor-controlled reactor (TCR) type. This TCR can pro-
often improves the performance of the cyclic load it compen- duce only lagging vars. Thus the capacitor is rated to be able to
sates, provide the peak lagging demand of the load. The compensator
The new force-commutated compensator, called a reactive- is rated to cancel the capacitor bank when the load requires no
power compensator (RPC), is suited to cyclic loads of compensation. Thus both the compensator and the capacitor
moderate rating (1.0-25 MVA), which are balanced three- are rated 20 MVAr.
phase loads. These include motors, drives, rectifiers, and most Fig. l(b) shows the application of the force-commutated
industrial processes, with the specific exception of arc RPC to the same load. Since the compensator can either cancel
furnaces, which are not balanced loads. Dynamic performance the capacitor current or reinforce it, both the capacitor and the
of this RPC is similar to the fastest available SVC's. For loads RPC can be one-half the rating of those of the TCR approach
larger than 25 MVA, this RPC can be used as an interpolator in Fig. 1 (a).
between steps of switched capacitance to provide compensa- A second advantage of the symmetrical force-commutated
tion up to any total load. RPC is that when the load is at a typical nonpeak continuous
The important characteristic of this force-commutated RPC operating point (for example, requiring 10 MVAr or half the
is that it can provide either leading or lagging reactive power. available compensation), the capacitors approximately match
This allows it to be of approximately one-half the rating of the compensation demand and the RPC operates near zero
some other SVC approaches for a given application. Fig. 1 current. In the TCR approach shown in Fig. 1(a), the
compensator will be at 10 MVAr under these same conditions.
Paper IPCSD 86-24, approved by the Static Power Converter Committee of
The TCR will be at zero current only at the instants of peak 20-
the IEEE Industry Applications Society for presentation at the 1984 Industrial MVAr demand by the load. Thus the duty-cycle losses (and
and Commercial Power Systems Conference, Atlanta, GA, May 7-10. harmonic currents) drawn by the force-commutated compen-
Manuscript released for publication June 25, 1986. sator are minimized by its symmetrical operation capability.
The author is with Drives Systems Operations, General Electric Company,
1501 Roanoke Boulevard, Salem, VA 24153. An additional advantage of the force-commutated RPC is
IEEE Log Number 8610534. that the usual problems associated with application of any
In lagging operation the bridge acts as a conventional 6- Fig. 2. Power circuit of force-commutated reactive-power compensator.
pulse phase-controlled rectifier without forced commutation,
gated near 900 phase retard. The dc voltage across the
inductor is near zero, and the dc current is regulated by LINE TO
NEUTRAL
controlling small deviations from the 900 gating angle, c. To VOLTAGE
the ac source, the fundamental component of current drawn is __ I - LINE
at zero power factor lagging (due to the 900 gating delay) and CURRENT
I\N\VN ~ ~ ~ ~ ~+ ~DC
shown in Fig. 3(a).
In leading operation, gating to the thyristors is ideally
advanced (rather than retarded) 90° from the a = 0 position.
The forced commutation, which is required when gating is VOLTAGE
earlier than a = 0°, is provided by the commutating
capacitors. The dc link voltage is near zero at a = 900,
leading as it is at a = 900 lagging, and the magnitude of dc (a)
current can be adjusted by small changes of gating angle. To LINE TO
the ac source, the fundamental component of current drawn is / \/ NEUTRAL
VOLTAGE
at zero power-factor leading, and of adjustable magnitude.
Ideal waveforms in this mode of operation are shown in Fig. _ LINE
CURRF.NT
3(b).
The foregoing description of the operation of the force-
commutated RPC is sufficient to understand its application,
but a number of second-order effects, which will be discussed,
must be considered to design or control the bridge.
Typical Installation
7/4/tV/1/1v ~~~~~~~DC VOLTAGE
Ld I BRIDGE LOAD
(b)
Fig. 4. Typical installations of 2.5-MVA RPC. (a) Compensating a low-
voltage load. (b) Compensating a high-voltage load.
\REGION \ // EIN3v/.
the time in lagging operation is that the commutating capaci-
tors can be viewed as connected directly across the ac source.
Thus they function as power-factor correction capacitors,
drawing a fixed leading load from the ac source. This opposes
the desired lagging load in much of this region and thus
LEADING
IDEAL
CHARACTERISTIC
reduces the effective rating of the bridge as a source of lagging
VARS
vars.
The effect of the commutating capacitors on the bridge in
ACTUAL lagging operation is evident in Fig. 8. This figure relates dc
current to reactive VA. In the leading mode of operation (left
-1.0
PER UNIT
side of Fig. 8) dc current is related to leading vars by the
Fig. 7. Static-control characteristic of force-commutated RPC.
classical relation for a three-phase bridge:
after the thyristors are gated. The delay between thyristor Q(vars)= Id Vs (1)
commutation and diode commutation is the time it takes to 1r
reverse the charge on the commutating capacitors. Since it is where
the diodes that connect to the ac power source rather than the Q reactive voltamperes,
thyristors (see Fig. 2), the phase angle of the current in the Id dc current,
source is determined by the diode commutation time rather Vs line-to-line source voltage (rms).
than the thyristor commutation time. Thus if power factor is to
be zero, diode commutation must take place near a = 900 or This relationship applies in region A-B of Fig. 8. The regions
a = 2700 so that zero real power will flow. The broken-line denoted by A, B, C, D on Fig. 8 correspond to those similarly
ideal characteristic on Fig. 7 represents the time of diode labeled on Fig. 7.
commutation while the solid line represents the gating time of At the point labeled B, C on Fig. 8, dc current has reached
the thyristors. The thyristor gating time always leads the diode the minimum value that may be force-commutated. The bridge
commutation time. The amount of this delay between gating reverts to natural commutation and goes into lagging opera-
the thyristor and commutating the diodes when in forced tion, but the output vars are still leading. The dc current must
commutation is approximately inversely proportional to cur- build to a substantial value in the lagging direction (I cap)
rent magnitude at the time of commutation. Thus at large before the net reactive VA reaches zero. The current I cap is
leading currents (near the bottom of Fig. 7), the capacitors the dc equivalent of the leading fundamental current drawn by
discharge quickly and the horizontal distance between the the commutating capacitors when the bridge is not in forced
solid-line actual thyristor gating angle and the broken-line commutation. The Appendix gives an expression for the
diode commutating angle is at a minimum. As current magnitude of the I cap.
decreases (proceeding upward and to the left on Fig. 7), In the lagging range of operation, dc current is approxi-
current decreases and commutation time stretches as the mately equal to the sum of the expected value given by (1) and
reciprocal of current. the constant-value I cap. Due to the larger dc current in the
This commutating time would stretch to infinity at zero dc lagging range, the rating of the RPC is not truly symmetrical
current, but that point is never reached. As current is reduced but slightly smaller in lagging than in leading operation.
from the maximum leading value, commutating time stretches
reciprocally until a point is reached where the current in one Waveforms
diode has still not commutated when the next thyristor in the The idealized waveforms of current and voltage were shown
given set of three is gated. This occurs when the commutating in Fig. 3. Actual internal waveforms are somewhat more
WALKER: REACTIVE POWER COMPENSATOR 1095
LAGGING
IDEAL
VVOLTAGE
AJ11LlIK .2Z
(a)
7- LAG \| |
LAG CURRENT
LEADING
IDEAL
.4I VOLTAGE
/ LEAD
(b)
LEAD CURRENT
LU
t
Fig. 9. Comparison of ideal and actual waveforms at terminals of RPC bridge. (a) Lagging. (b) Leading. (Actual waveforms at 750
kVAr on ± 1250-kVAr RPC bridge.)
complex. The currents and voltages at the terminals of the The actual waveforms in leading operation, as shown in the
installation in a real application are quite sinusoidal, as will be photograph of Fig. 9, are very much like the ideal in both
shown. The low harmonic content is maintained by the current and voltage. In lagging operation the commutating
filtering effect of the tuned power-factor correction capacitors capacitors, which are now redundant, cause the commutations
always included in the installation of an RPC. of current to be followed by a resonant backswing. The
Within the RPC, currents and voltages are as shown in Fig. capacitors are almost always drawing current, and this adds an
9. If the currents were pure 120° square waves, the voltage at ac component to the expected flat-topped current during the
the terminals of the bridge would be a shown in the idealized conduction of a phase. The polarity of the ripple is such as to
sketches. In lagging operation the voltage would show a series reduce the fundamental component of lagging current as was
of notches caused by the effect of current commutation on the previously noted. The effect of the commutating capacitors on
reactors (Ls) in series with each phase of the source (Fig. 2). the terminal voltage of the bridge is to soften the repeated
In leading operation these notches would be displaced 1800 in notches. Note that the voltage waveforms shown are on the
time phase so that they appear as spikes on the voltage RPC side of the inductor Ls. The spikes and notches appear
waveform. The spikes shown in the sketch of leading primarily across that inductance and do not appear on the
operation in Fig. 9 have the shape of quarter-cycle sinewaves source.
characteristic of the CSI commutation [3]. Fig. 10 shows the net current drawn by the RPC bridge
1096 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-22, NO. 6, NOVEMBER/DECEMBER 1986
(a)
(b)
Fig. 10. Current waveforms. (a) In capacitors and RPC bridge. (b) In RPC
bridge alone.
RPC CURRENT-4V P
P = POWER
Q = REACTIVE VA
Inductive Positive
Fig. 11. Overall control diagram.
alone and the total current drawn from the power source by the the point of common coupling between the load facility and the
RPC and the tuned capacitors in the power-factor correction power authority or utility.
bank. The effectiveness of the capacitors in absorbing the Any in-phase current drawn by the load will cause a voltage
harmonics of the bridge is evident. drop across Ru. Any lagging reactive load drawn will cause a
voltage drop across Xu. Any leading load drawn will cause a
System Controls voltage rise across Xu.
Fig. 11 shows the system controls for the RPC. This control It is possible to draw a controlled leading reactive load that
is suited to provide either 1) unity power factor, 2) constant will cause a voltage rise, compensating exactly the voltage
power factor other than unity, or 3) undisturbed voltage. In the drop due to real load. If such a compensating current is caused
undisturbed-voltage mode, the RPC does not attempt to to flow and no other reactive power is allowed to flow, then
correct voltage changes due to causes other than the compen- the voltages Vp will be undisturbed by the presence of the
sated load, but it does remove any disturbance to the voltage dynamic load.
source due to the changes in real or reactive power in the The obvious method of voltage control, that of sensing the
compensated load. The other two operation modes are voltage Vp and controlling the RPC to maintain it constant, is
simplifications of the undisturbed voltage mode. not desirable here. If the utility voltage were low due to causes
Undisturbed Voltage Mode: The control approach for outside the load facility, the entire leading capability of the
undisturbed voltage involves compensating the load's real RPC could be consumed in an attempt to hold up the utility
current with leading current, and compensating the load's voltage. This would prevent the RPC from performing its
reactive current to zero. Consider the power circuit shown in desired function: that of preventing dynamic swings (flicker)
Fig. 11. The elements Xu and Ru represent the impedance of on the source. Although this problem could be overcome with
the power system supplying the point at which voltage is to be a special control, there are other reasons to avoid the feedback
maintained undisturbed. This point, labeled Vp, is typically control of voltage. Both stability and speed of responses are
WALKER: REACTIVE POWER COMPENSATOR 1097
VARS VARS G
LEADING LAGGING
Fig. 12. RPC bridge controls.
enhanced by the method used, because the RPC regulator is account for the effect of the impedances Rp and Xp if they are
dynamically independent of other regulators and other loads significant.
on the power bus. It can maintain voltage variation below 1 It should be further noted that all the power calculators
percent with very rapid (10 pu/s) changes in the compensated shown on Fig. 11 sense instantaneous three-phase power, so
load. A control that sensed voltage would have to be slower by that their output responds instantly and has low ripple when
more than 10:1 to provide stability. loads are balanced.
To accomplish the undisturbed voltage mode, the current is Other Control Modes: All other control modes are
measured at IL on the power circuit of Fig. 11. This is the simplifications of the undisturbed voltage mode. For the
combined current of the dynamic load and the capacitors. IL constant-power-factor mode, the function Fv is made linear.
and VT are used as input to control blocks Cl and C2 to This calls for a net reactive load proportional to real load. For
calculate the load's real and reactive components PL and QL. the unity-power-factor mode the output Fv is made zero. The
Blocks Cl and C2 are implemented as shown in [5]. The real RPC vars are the opposite of the sum of the vars in the load
power component PL is input to a nonlinear function Fv. This and capacitors leaving no net reactive power in the source.
function calculates the exact amount of leading load needed to
compensate the existing real power. The output of the block RPC Controls
Fv is labeled "total var command" because no reactive power The RPC controls respond to the RPC var command. They
other than that needed to compensate the real power will be include a reactive-power calculator; block C3, which mea-
allowed to flow. To calculate the var command to the RPC sures the vars of the RPC bridge; and the RPC bridge controls,
bridge, the signal proportional to existing reactive VA in the which determine the gating angle of the bridge thyristors. A
load and capacitors QL is added to the total var command. The block diagram of the RPC controls is shown in Fig. 12. The
resulting summation is the RPC var command. If the RPC block C3 is repeated from Fig. 11. The concept of the bridge
perfectly follows this command, then the load's reactive controls is to provide a feed-forward nonlinear function that
power will be absorbed and the effects of its real power will be sets the gating angle a to approximately the correct value, and
compensated by leading power. a proportional plus integral regulator that trims the value of a
It should be noted that the function Fv is valid for both to the exact value needed. The feed-forward element is the
positive and negative real power since many of the loads to be bridge-model nonlinear network, block NL. It produces an
compensated are regenerative. This function must be individu- output characteristic that models the bridge-control character-
ally adjusted for the values of Xu and Ru in each installation. istic (Fig. 7). If it were driven directly by the RPC var
If these values are known from previous fault studies then they command, then it would be a classic feed-forward function,
can be preset to avoid field adjustment. The location of current providing the precise value of a needed for each level of the
transformers feeding the power calculators in Fig. 11 causes RPC var command. It is shown to be driven partly by this
the real power losses of the RPC bridge and reactor to be command and partly by the var feedback. The result is that in
uncompensated. These losses are small and can be neglected. any steady-state condition, the sum of the two inputs to
If it is desired to compensate these losses, the controls can be junction S2 is the same as the RPC var command, and the
modified slightly to do this. The function Fv may also have to output of the bridge model is the desired value of a. When the
1098 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-22, NO. 6, NOVEMBER/DECEMBER 1986
RPC var command changes suddenly and the var feedback has
not yet moved, the ratio of the two gains at the input to S2
defines the amount of transient forcing supplied by the bridge
AC Current
model. The gain K is chosen to make the dynamic response of
the RPC similar in the leading and lagging modes of operation.
The output of the bridge model is combined at junction SI (10 msec/div)
with the output of the PI regulator and clamp. The input to the
PI regulator is the error between the RPC var command and DC Current
the present value of actual bridge vars. The summation of the
output of blocks PI and NL provides the net oa command to the
firing circuit. Fig. 13. Response of RPC to step in a.
In operation the bridge model provides the large swings in a
necessary to provide control when operation is in a region
where a must change rapidly. The PI regulator provides the
primary control of a in operating regions such as the lagging
VAR Command
region, where only small changes are required in a. The PI (0.1 PU Step)
regulator also assures that there are no steady-state errors in
following the RPC var command.
(5 msec/div)
Control Dynamics VARS
(0.04 PU/div)
When a normal 6-pulse phase-controlled bridge circuit is
inductively loaded and controlled by a closed-loop current
regulator, the maximum crossover frequency of the regulator (a)
is fixed by the transport delay introduced by the discrete or
sampling nature of the power circuit. The maximum crossover
can be calculated by several means, but one commonly used is
to model the statistical transport delay, which is in the range VAR Command
(0.1 PU Step)
0-60° at line frequency as a fixed delay of 300, the average
value. If the 1800 phase shift allowed at the crossover
frequency of the regulator is apportioned as 900 to the (5 msec/div)
statistical delay and 900 to the dc inductor, then the maximum
crossover frequency is approximated as follows: VARS
(0.1 PU/div)
substantially lower than ideal. The values of the fifth and HARMONIC NUMBER
Dynamic Performance
Testing of the RPC system for dynamic response and
linearity was performed using a triangular-wave var reference
to simulate a dynamic load reversing as quickly as possible.
The amplitude of the reference was adjusted to cause the RPC
bridge to swing to its maximum limits. The frequency of the
triangle wave was increased until the vars produced by the
bridge could no longer track the reference. Data taken by this
method is shown in Fig. 18.. In the top traces the reference is at
low frequency, 1.6 Hz, demonstrating the linearity of the
RPC. Close examination of the response curve shows a slight
distortion of the response when crossing from the lagging to
the leading mode of operation. This is a dynamic effect not
present with a steady command.
In the lower traces of Fig. 18, the triangle-wave reference is
0 0.25 0.50 0.75 1.0 at 5.7 Hz. The performance has deteriorated so that the RPC
FULL
LEAD PER UNIT OF FULL VAR RANGE
FULL
LAG
does not quite reach-the peak leading vars commanded, and the
Fig. 16. Ideal harmonic currents of force-commutated RPC compared to disturbance in crossing from lagging to leading operation has
those of thyristor-controlled reactor (TCR) in rating range where both are become obvious. This is considered the linearity limit of the
6-pulse. RPC. It requires the bridge to go from full-rated lagging to
0.125 - full-rated leading in less than 0.1 s. The RPC will respond if
commanded to change up to five times this rate, as was shown
HARMON IC CURRENT
FULL REACTIVE CURRENT RANGE
in Figs. 13 and 14, but its response will not be linear during
0.10 the rapid change.
To relate this data to the small signal transient response of
the bridge alone, as shown in Fig. 14, it is noted that the RPC
. 07 5 0 2.5 MVAr 4RATING <25 MVAr
regulator had a crossover frequency of 140 rad/s. It would be
expected to follow a sinusoidal reference at frequencies less
than 140 rad/s (22 Hz). To follow a triangle reference with
0.05 significant third and fifth harmonic content, the limit in
reference frequency might be expected to be 140/5 = 28 rad/s
= 4.5 Hz. This is consistent with the large signal perform-
0.025 ance.
Maintaining Voltage Undisturbed
The RPC was tested when configured in the undisturbed
0 .25 0.5 0.75 1.0
FULL
voltage mode by compensating a load consisting of a large 6-
FULL
LEAD PER UNIT OF FULL VAR RANGE LAG pulse motor drive driven through a rapid speed reversal with a
Fig. 17. Ideal harmonic components of force-commutated RPC compared to mechanical load which was primarily inertial. Figs. 19 and 20
those of thyristor-controlled reactor (TCR) in rating range where TCR is 6-
pulse and RPC is 12-pulse. show this test of the reactive power compensator, which
compensates a thyristor-controlled motor drive. The RPC is
(curved lines) except near the center of the operating range. As controlled in the undisturbed voltage mode, and the scales are
previously discussed, the RPC bridge tends to operate in pu VA on the base of the RPC bridge rating. Fig. 19 shows the
center of its range much of the time. conditions with the RPC turned off and power factor capaci-
For ratings greater than 2.5 MVAr, the RPC will be 12- tors disconnected. The load demand for real and reactive
pulse. The TCR may also be 12-pulse but this is not common power is shown in traces 3 and 4. The resulting swing in
in TCR ratings below 25 MVAr. Thus in the primary rating source voltage is shown in trace 5, with a maximum dip of 5.5-
range of the RPC, between 2.5 MVAr and 25 MVAr, the RPC percent. The var command shown in trace 1 was the signal
will have 12 or more pulses and the TCR will be 6-pulse. provided by the system regulator to the RPC bridge. It always
Harmonic comparisons in this case are shown in Fig. 17. The calls for leading compensation because there are no capacitors
RPC harmonic currents (straight lines) are smaller than those connected. The RPC bridge is prevented from responding.
of the TCR (curved lines) because the fifth and seventh Note that the var command saturates at the rating of the RPC
harmonics of the RPC are almost eliminated by the 12-pulse bridge.
connection, while those of the TCR remain. It should be Fig. 20 shows the same conditions with the RPC bridge
emphasized that the harmonic content of either the TCR or active and the associated power-factor correction capacitors
RPC can be reduced to acceptable levels by properly tuning connected. The load demand in traces 3 and 4 is the same as
the power-factor correction capacitors. The data in Figs. 15- Fig. 19. The var command is shifted toward lagging by the
17 is without these capacitors. steady-state effect of the capacitors, bringing it into the range
WALKER: REACTIVE POWER COMPENSATOR
LEAI1 .
rrace I
I
.
!
1
J:1 i
; :-
- t
;--::.;'' i' "t ,:
:: S7 ,
s
LAG 1. O
I k L/ A
li ldi 7 . 5, :fi: :: , \ r X 1gx ii
R.PC
LEAD1.0
0 Frace 2
LAG 1. 0
LOAD
LAG 1.0 VARS
rrace 3
LEA -. ''''-'' ,.,,
L A1.O0
O_
Trace 4
- . osec 5 .76
Hz VAR COMMAND
RPC BRIDGE VARS
4 . ; _; ;ffDLAG- :;--: - ::--
I
r t. - .-. - j 1 ! -- .: ~-::: _ _
NO LOAD Trace 5
-S%
Fig. 20. Test of reactive-power compensator. RPC and capacitors on.
+VC
Source
Voltage
tlIt2 -t
tc It3 t
- -VC
I
Tl Thyristor Ti Current
0
fE
T2
.L
--E ~ .~~
Thyristor T2 Current ~~~~~~~~~~
iZ_ ~ ~~~~
,
Dl Diode DI Current
i 1
D2 Diode D2 Current
I~~~~~~~~~~~~~~~~~~~~~atforcecommutat.1---
Fig. A- TIming
12/r
Q= VS2wSC (A. 10)
/37r 7r
2 -t-13)
This expression is valid in region A-B of Fig. 7, from
maximum leading current to oa = 150°. Fig. A-3 shows a plot Fig. A-3. Calculated static-control characteristic.
of this expression.
Constant Current, Region B-C, Fig. 7: The preceding
1104 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. IA-22, NO. 6, NOVEMBER/DECEMBER 1986
analysis showed that the capacitor voltage conditions for each given as in (A.8) by
commutation in a group of three thyristors are set up by the
previous commutation. This can be true only if the commuta- I cap=- Ic. (A. 15)
tion is complete (diode current has fully transferred) before the
next thyristor in the commutating group is gated. Thus the Substituting (A. 14) into (A. 15) yields the value for I cap in
preceding analysis is valid until commutation takes 120°, or a Fig. 8:
= 2700 - 1200 = 1500. If the thyristor is gated earlier than
1500 (but still not early enough for natural commutation) then
the current will not drop below that necessary to discharge the I cap= 3
3
VswsC. (A. 16)
capacitor. Thus current is constant over the region of a from
approximately 90° to 1500 (region B-C on Fig. 7). The offset between the ideal value of Id and the actual Id in
This minimum current for 60-Hz operation is analogous to a Fig. 8 is not exactly constant but decreases slightly as lagging
maximum frequency for light-load operation when a current vars are increased. This is because the fundamental voltage
source inverter is used as a motor drive as discussed in [3], across the deltas of capacitors is reduced by the fundamental
[4]. The value of minimum current is that which corresponds voltage drop across the inductances Ls. When Vs in expres-
to a commutating time of 120°. Commutating time in this case sion (A. 16) is corrected for this drop, the correct value of Id
is tl + t2, until the current is zero in the diode. The value of can be calculated at any value of lagging vars.
current I min is found by setting a = 1500 in (A.7) and
substituting t2 for t3: ACKNOWLEDGMENT
This circuit configuration was suggested by Mr. W. K.
Boice (General Electric Company; retired). The assistance of
I min= VswsC (A. 11) Mr. T. J. Kolb in evaluating and refining the RPC is greatly
2wr/3- (l+) COS - appreciated.
REFERENCES
Similarly the reactive VA in the constant-currerit region are [1] T. 1. E. Miller, Reactive Power Control in Electric Systems. New
York: John Wiley & Sons, 1982.
given by reference to (A. 10): [2] Laszlo Gyugyi, "Reactive power generation and control by thyristor
circuits," IEEE Trans. Ind. Appl., vol. IA-15 no. 5, pp. 521-532,
Sept./Oct. 1979.
12/xr [31 W. Lienau, "Commutation modes of a current source inverter,"
Q= Vs2wsC (A. 12)
presented at Second IFAC Symp. on Control in Power Electronics and
2ir/3- (l+2) os"fCi Electrical Drives, pp. 219-229, Dusseldorf, W. Germany, 1977.
[4] T. A. Lipo and L. H. Walker, "Design and control techniques for
extending high frequency operation of a CSI induction motor drive,"
IEEE Trans. Ind. Appl., vol. IA-19, no. 5, pp. 744-753, Sept./Oct.
Note on Fig. A-3 that the vars given by (A. 10) at a = 1500 1983.
are lower than the minimum given by (A. 12). Since diode [5] H. A. Kagi, Y. Kanazawa, and A. Nabae, "Instantaneous reactive
power compensators comprising switching devices without energy
i
nextit.guilshed
current lasts slightly after ca = 270 in (A. 10), the iext thyristor
begins to conduct before the diode is completely c extinguished.
storage components," in Conf. Record of IEEE/IAS Ann. Meeting,
Mexico City, 1983, pp. 825-830.
Thus the transition from (A. 10) characteristi c to (A.12)
characteristic begins at slightly higher than 15 00.
a