720 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO.
2, MARCH/APRIL 2012
Negative-Sequence Reactive-Power Control by a
PWM STATCOM Based on a Modular Multilevel
Cascade Converter (MMCC-SDBC)
Makoto Hagiwara, Member, IEEE, Ryo Maeda, and Hirofumi Akagi, Fellow, IEEE
Abstract—This paper presents the application of a modular
multilevel cascade converter based on single-delta bridge cells
(SDBCs) to a STATic synchronous COMpensator (STATCOM),
particularly for negative-sequence reactive-power control. The
SDBC is characterized by cascade connection of multiple single-
phase H-bridge (or full bridge) converter cells per leg, thus
facilitating flexible circuit design, low-voltage steps, and low-
electromagnetic-interference emissions. This paper designs,
constructs, and tests a 100-V 5-kVA pulsewidth-modulated
STATCOM based on the SDBC, with focus on the operating
principle and performance. Experimental results verify that it
can control not only positive-sequence reactive power but also
negative-sequence reactive power and low-frequency active power
intended for flicker compensation of arc furnaces.
Index Terms—Flicker compensation, multilevel converters,
negative-sequence reactive power.
I. I NTRODUCTION
T HE FAMILY of modular multilevel cascade converters
(MMCCs) is expected as one of the next-generation power
converters suitable for high-voltage or medium-voltage ap-
plications without line-frequency transformers [1]–[20]. From
power-circuit and converter-cell configurations, it can be clas-
sified into the following [1]:
1) single-star bridge cells (SSBCs);
2) single-delta bridge cells (SDBCs);
3) double-star chopper cells (DSCCs);
4) double-star bridge cells (DSBCs).
The term “bridge cell” is a single-phase H-bridge (or full
bridge) converter, and the “chopper cell” is a bidirectional chop-
per consisting of a dc capacitor and two insulated-gate bipolar
transistors. The SSBC is particularly suitable for a STATic syn-
chronous COMpensator (STATCOM) for voltage regulation
Manuscript received October 25, 2011; revised December 8, 2011;
December 17, 2011. Date of publication December 30, 2011; date of current
version March 21, 2012. Paper 2011-IPCC-569.R1, presented at the 2011 IEEE
Energy Conversion Congress and Exposition, Phoenix, AZ, September 17–22,
and approved for publication in the IEEE T RANSACTIONS ON I NDUSTRY
A PPLICATIONS by the Industrial Power Converter Committee of the IEEE
Industry Applications Society.
M. Hagiwara and H. Akagi are with the Department of Electrical and
Electronic Engineering, Tokyo Institute of Technology, Tokyo 152-8552, Japan
(e-mail: mhagi@akg.ee.titech.ac.jp; akagi@ee.titech.ac.jp).
R. Maeda was with the Department of Electrical and Electronic Engineering,
Tokyo Institute of Technology, Tokyo 152-8552, Japan. He is now with Tokyo
Electric Power Company, Inc., Tokyo 100-8560, Japan.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIA.2011.2182330
0093-9994/$26.00 © 2011 IEEE
Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.
[10], [12] and battery energy storage systems [13], [14].
However, without significantly increasing the converter-cell
count, the SSBC-based STATCOM cannot draw any negative-
sequence reactive power because it has no circulating current.
Flicker compensation of arc furnaces requires the control
of positive-sequence reactive power, negative-sequence reactive
power, and low-frequency active power at the same time [15]–
[17]. The SDBC, DSCC, and DSBC have the capability to
control negative-sequence reactive power because they have the
circulating current(s) that flow inside. The authors of this paper
have described the DSCC acting as a STATCOM with focus
on control and performance [19], [20]. Experimental results
obtained from a 200-V 10-kVA downscaled model have shown
that the STATCOM can control the rated negative-sequence
reactive power with a small amount of circulating current [19].
However, the converter-cell count required for the DSCC is four
times of that for the SSBC [1].
Attention has been paid to STATCOMs based on the SDBC
[7]–[9]. The SDBC seems to be a better choice than the DSCC
from a practical point of view because√the converter-cell count
required for the SDBC is only 1.7 (= 3) times of that for the
SSBC [1]. Peng and Wang presented a control method of an
SDBC-based STATCOM with stair-case modulation, in which
the amplitude of each cluster current is determined by offline
calculation based on phasor diagrams [7]. In other words,
no feedback loop is formed to control the current circulating
among the delta-connected clusters. Moreover, the authors in
[7] made no description of the term “circulating current.” It has
been pointed out in [8] and [9] that the STATCOM can draw
negative-sequence reactive power by adjusting the circulating
current. However, no experimental verification has been made
in the literature.
The aim of this paper is to provide experimental verification
of an SDBC-based pulsewidth-modulated (PWM) STATCOM
for negative-sequence reactive-power control. This paper pro-
poses a control method that is characterized by forming a feed-
back loop of the circulating current among the delta-connected
clusters, leading to stable dc-mean voltage control of all the
dc capacitors. This paper derives theoretical equations related
to ac-voltage fluctuations of each dc capacitor. It is followed
by experimental verification using a downscaled model rated at
100 V and 5 kVA. Experimental results verify that it can control
not only positive-sequence reactive power but also negative-
sequence reactive power and low-frequency active power at the
same time.
HAGIWARA et al.: REACTIVE-POWER CONTROL BY PWM STATCOM BASED ON MMCC (MMCC-SDBC)
Fig. 1. Typical system configuration with ac arc furnaces, passive filters, and
a flicker compensator.
Fig. 2. Simplified circuit of Fig. 1.
II. F LICKER C OMPENSATORS FOR A RC F URNACES
A. Configuration of Flicker Compensators
Fig. 1 shows a typical system configuration with ac arc fur-
naces, passive filters, and a flicker compensator. The multiple ac
arc furnaces in a range from 10 to 100 MVA are connected to
a high-voltage grid [18]. Arc furnaces draw positive-sequence
and negative-sequence reactive currents, low-frequency active
currents, and harmonic currents from the grid. The harmonic
currents can be suppressed by connecting the passive filters
tuned to such specific frequencies as shown in Fig. 1. The other
currents that cause the so-called “voltage flicker” should be
mitigated fully by the flicker compensator.
B. Configuration of Flicker Compensators
Fig. 2 shows a simplified circuit of Fig. 1 in which iS is the
supply current, iL is the load current, and i is the compensating
current. Note that p and q are the instantaneous active and re-
active powers in the three-phase circuit [22] at the ac terminals
of the compensator. Without the compensator, a voltage flicker
would appear on the point of common coupling (PCC) due to
the time-varying load current iL drawn by the arc furnace. The
flicker compensator mitigates the voltage flicker by controlling
both p and q. Generally, the controller of a flicker compensator
calculates power commands p∗ and q ∗ from the load current
detected.1 Then, the current commands of the compensating
currents are calculated from p∗ and q ∗ . Note that the three-
phase voltages at the PCC would be almost balanced and
sinusoidal waveforms when the compensator worked properly.
For example, the following flicker compensators have been put
1 How to calculate p∗ and q ∗ is beyond the scope of this paper.
Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.
721
Fig. 3. Flicker compensators with the SDBC-based STATCOM. (a) 22-kV/
33-kV system with a grid transformer. (b) 6.6-kV system with no grid
transformer.
into practical use to date: One is based on line commutation,
called as static var compensators [15]. The other is based on
self-commutation, called as a STATCOM [18].
C. SDBC-Based STATCOM for Flicker Compensation
Fig. 3 shows two kinds of flicker compensators with an
SDBC-based STATCOM. The SDBC is characterized by easily
increasing the voltage and current ratings without using line-
frequency converter transformers. Fig. 3(a) shows a 22-kV/
33-kV system with a grid transformer. Each cluster of the
SDBC is connected in delta configuration via a single coupled
inductor. The leakage inductance of the transformer works as
ac-link inductors between the grid and the SDBC. Fig. 3(b)
shows a 6.6-kV system with no grid transformer. Each cluster
of the SDBC is connected via three noncoupled inductors. The
SDBC can be connected directly to the 6.6-kV grid because the
noncoupled inductors work as ac-link inductors.
III. C IRCUIT C ONFIGURATION OF THE SDBC
A. Circuit Configuration
Fig. 4 shows the detailed circuit configuration of the 100-V
5-kVA STATCOM used in experiments. Each cluster of the
SDBC consists of cascade connection of three bridge cells
(i.e., single-phase full-bridge PWM converters), and the three
clusters are connected in delta configuration via a single cou-
pled inductor L. The SDBC is connected to a three-phase ac
mains of 100 V (line to line in rms) via a three-phase ac-link
inductor LS that corresponds to the leakage inductance of the
grid transformer in Fig. 3(a). Here, vuv , vvw , and vwu are the
cluster voltages, iuv , ivw , and iwu are the cluster currents, and
p and q are the instantaneous active and reactive powers at the
PCC. The following relations exist between the compensating
currents and the cluster currents2 :
iu = iuv − iwu iv = ivw − iuv iw = iwu − ivw . (1)
2 The compensating currents and the supply currents are the same in Fig. 4
because no load (arc furnace) is connected.
722 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO. 2, MARCH/APRIL 2012

TABLE I
C IRCUIT PARAMETERS U SED IN E XPERIMENTS

Fig. 4. Circuit configuration of the 100-V 5-kVA PWM STATCOM used in

experiments.

Fig. 5. Coupled inductor used in Fig. 4.

Let the circulating current flowing inside the delta-connected

clusters be iZ . It is defined as
1
iZ = (iuv + ivw + iwu ). (2)
3

B. Configuration of the Coupled Inductor

Fig. 5 shows the detailed configuration of the coupled in-
ductor used in Fig. 4. It consists of three windings, each of
which has a turn number of N/3 along a common magnetic
core. The following relation between the magnetic flux φ and
the magnetomotive force is obtained from Fig. 5 and (2) as:
N iuv N ivw N iwu
Rφ = + + = N iZ
3 3 3
where R is the reluctance of the core. Hence, the coupled induc-
tor presents an inductance value of L (= N 2 /R) only to the
circulating current iZ and no inductance to the compensating
currents. Note that L is the inductance value when the three
windings are connected in series.
C. Circuit Parameters Used in Experiments
Fig. 6. Block diagram of dc-capacitor voltage control. (a) Cluster-balancing
control. (b) Circulating-current control. (c) Individual-balancing control.
set as VC∗ = 60 V. The dc capacitor was set as either C =
16.4 mF (H = 53 ms) or C = 0.9 F (H = 2.9 s). The ac-link
inductor and the coupled inductor were set as LS = 0.5 mH
(8%) and L = 2.3 mH (37%), respectively.3
(3)
IV. C ONTROL M ETHOD OF THE SDBC
Fig. 6 shows the block diagram of the dc-capacitor voltage
control. Voltage control of the nine floating dc capacitors in
Fig. 4 can be divided into the following:
1) cluster-balancing control;
2) circulating-current control;
3) individual-balancing control.

Table I summarizes the circuit parameters of the SDBC

3 The percent inductance of the coupled inductor, i.e., 37%, is calculated on
which will be used in experiments later on. Each carrier
a three-phase, 100-V, 5-kVA, and 50-Hz base. A coupled inductor is much
frequency of phase-shifted PWM for bridge cells was set as smaller in physical volume than an ac-link inductor when both have the same
fC = 2 kHz. The command of each dc-capacitor voltage was percent inductance.
Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.
HAGIWARA et al.: REACTIVE-POWER CONTROL BY PWM STATCOM BASED ON MMCC (MMCC-SDBC) 723

A. Cluster-Balancing Control
Fig. 6(a) shows the block diagram of the cluster-balancing
control. The voltage major loop forces the average voltage
of each cluster, namely, v̄Cu , v̄Cv , and v̄Cw , to follow the
average voltage of the three clusters v̄C where they are de-
fined as

1
1

3 3
v̄Cu = vCju v̄Cv = vCjv
3 j=1 3 j=1

1

3
v̄Cu + v̄Cv + v̄Cw
v̄Cw = vCjw v̄C = . (4)
3 j=1 3 Fig. 7. Block diagram of instantaneous active and reactive power controls,
and overall voltage control.
Here, v̄Cu , v̄Cv , v̄Cw , and v̄C are instantaneous values contain-
ing both ac and dc components. It is desirable to extract only the
dc components (i.e., (v̄Cu )dc , (v̄Cv )dc , (v̄Cw )dc , and (v̄C )dc )
because the existence of the ac components deteriorates the
controllability. The following methods can be utilized to extract
the dc components:
1) the method using a low-pass filter [10];
2) the method using a feedforward control;
3) the method using a moving-average filter of 100 Hz [9].
The last method is adopted in this paper. Note that sin(ωt +
π/6) in Fig. 6(a) is in phase with vuv . When (v̄C )dc >
(v̄Cu )dc , the product of vuv and iZ (= i∗Z ) forms positive
active power because iZ contains the same component as
vuv . As a result, an amount of active power flows into the Fig. 8. AC-voltage command of each bridge cell. (a) u-phase. (b) v-phase.
u-phase cluster, thus leading to increasing (v̄Cu )dc . On the (c) w-phase.
other hand, the product of vuv and iZ forms negative active
power when (v̄C )dc < (v̄Cu )dc , thus leading to decreasing Hence, the sum of the voltage commands is equal to zero.
(v̄Cu )dc . This means that no interference occurs between the individual-
balancing control and the circulating-current control.
B. Circulating-Current Control
Fig. 6(b) shows the block diagram of the circulating-current D. Active-Power, Reactive-Power, and Overall
control. The current minor loop forces iZ to follow its com- Voltage Controls
mand i∗Z , producing the voltage command vA ∗
that is common Fig. 7 shows the block diagram of the active-power, reactive-
to the three clusters. power, and overall voltage controls [10], in which p∗ and q ∗
represent the power commands of p and q at the PCC. The
C. Individual-Balancing Control dc component of q ∗ is adjusted to control positive-sequence
reactive power keeping the relation of p∗ = 0. On the other
Fig. 6(c) shows the block diagram of the individual- hand, a couple of second-order components (100 Hz) with the
balancing control. It forms an active power between the ac same amplitude but a phase difference of 90◦ are superimposed
voltage of each bridge cell and the corresponding cluster cur- on p∗ and q ∗ , respectively, to control negative-sequence reactive
∗ ∗ ∗
rent [21]. The voltage commands vBju , vBjv , and vBjw are power. A low-frequency component is superimposed on p∗ to
given by control active power, keeping the relation of q ∗ = 0. The line-
∗ ∗ ∗
∗ to-line voltage commands vuv , vvw , and vwu are determined
vBju = K4 (v̄Cu − vCju )iuv
by decoupled current control of the compensating currents. A
∗
vBjv = K4 (v̄Cv − vCjv )ivw voltage major loop intended for compensating the converter
loss is formed as shown in Fig. 7, which forces (v̄C )dc to follow
∗
vBjw = K4 (v̄Cw − vCjw )iwu . (5) its command vC ∗
.
Fig. 8 shows the ac-voltage command of each bridge cell.
The following equation is obtained from (4) and (5): The voltage command is normalized by each dc-capacitor volt-


3

3

3 age. Then, it is compared with a triangular waveform having a
∗ ∗ ∗
vBju + vBjv + vBjw = 0. (6) maximal value of 1 and a minimal value of −1 with a carrier
j=1 j=1 j=1 frequency of fC .
Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.
724 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO. 2, MARCH/APRIL 2012

 
√ 7π
vwu = 2V sin ωt − (8)
6
where V is the rms value. The compensating currents iu , iv ,
and iw are expressed as
√
iu = − 2I sin(ωt + φ)
 
√ 4π
iv = − 2I sin ωt + φ −
3
 
√ 2π
iw = − 2I sin ωt + φ − (9)
3
where I is the rms value and φ is an initial phase. Applying the
d−q transformation into (9) produces id and iq as
√
id = 3I cos(2ωt + φ)
√
iq = 3I sin(2ωt + φ). (10)

The three-phase instantaneous active and reactive powers at

the PCC p and q under the balanced and sinusoidal ac-mains
voltages are expressed as follows:
√
p = VS id = 3VS I cos(2ωt + φ)
√
q = VS iq = 3VS I sin(2ωt + φ). (11)

The cluster currents and the circulating current are expressed as

follows (see the Appendix):
 
Fig. 9. Experimental waveforms when the rated negative-sequence reactive 2 π  π
power of 5 kvar was controlled with a condition of C = 16.4 mF (H =
iuv = 2 I sin φ − sin ωt −
3 3 3
53 ms). 
2
ivw = 2 I sin φ sin ωt
V. T HEORETICAL A NALYSIS W HEN N EGATIVE -S EQUENCE 3
R EACTIVE P OWER IS C ONTROLLED     
2 2π 2π
iwu = 2 I sin φ − sin ωt −
A. Cluster Currents and Circulating Current 3 3 3

The supply or ac-mains line-to-neutral voltages in Fig. 4 are 2
given by iZ = I cos(ωt − φ). (12)
3
   
2 2 2π The following conclusions are obtained from (12).
vSu = VS sin ωt vSv = VS sin ωt −
3 3 3 1) The initial phase φ determines either the capacitive or
   inductive operation of each cluster, along with the am-
2 4π
vSw = VS sin ωt − . (7) plitude of each cluster current. √
3 3 2) The maximal rms value of the cluster current is 2I/ 3.
The cluster voltages vuv , vvw , and vwu are not completely
balanced when a negative-sequence reactive power is con- B. DC-Capacitor Voltage
trolled. This means that their rms value and phase difference are
An operating condition of φ = 5π/6 is considered where
different from 100 V and 120◦ , respectively. The experimental
the ac-voltage fluctuation of vC1u and the amplitude of iuv
results in Fig. 9 show that the difference in the rms value is less
are maximal. The cluster current iuv can be expressed from
than 13% with respect to 100 V, and the phase difference is less
(12) as
than 12% with respect to 120◦ , both of which are not so large. 
To make the theoretical analysis simple and easy, the cluster 2  π
voltages are assumed to be balanced and sinusoidal as iuv = 2 I sin ωt − . (13)
3 3
√  π
vuv = 2V sin ωt + The dc-capacitor voltage vC1u can be obtained as follows:
6 √
  π
√ π 3VS I
vvw = 2V sin ωt − vC1u = VC + cos 2ωt + (14)
2 9ωCVC 3
Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.
HAGIWARA et al.: REACTIVE-POWER CONTROL BY PWM STATCOM BASED ON MMCC (MMCC-SDBC) 725

where VC is the dc component of vC1u (= VC∗ ). Equation (14)

indicates that a second-order frequency component of 100 Hz
is superimposed on VC .

VI. AC-VOLTAGE F LUCTUATION W HEN ACTIVE

P OWER IS C ONTROLLED
The instantaneous active and reactive powers in the three-
phase circuit p and q are given to the STATCOM as follows:
√
p= 3VS I sin(ω0 t + φ0 ) q=0 (15)

where ω0 /2π is a frequency and φ0 is an initial phase. The

following relation exists in Fig. 4 when the effects of the ac-
link inductors and the coupled inductor are ignored:

p = vuv iuv + vvw ivw + vwu iwu . (16)

The instantaneous active power in the u-phase cluster yields

d(vC1u + vC2u + vC3u ) dv̄Cu

vuv iuv  CVC = 3CVC .
dt dt
(17)
The similar relationships come into existence in the v-phase and
w-phase clusters as

dv̄Cv
vvw ivw  3CVC Fig. 10. Experimental waveforms when a positive-sequence capacitive-
dt
reactive power of 1.7 kvar, a negative-sequence reactive power of 1.7 kvar, and a
dv̄Cw 10-Hz active power of 1.7 kW were simultaneously controlled with a condition
vwu iwu  3CVC . (18) of C = 16.4 mF (H = 53 ms).
dt

The following equation is obtained from (4), (15)–(18) as: A. Negative-Sequence Reactive-Power Control
√ The instantaneous active and reactive power commands in
p = 3VS I sin(ω0 t + φ0 ) the three-phase circuit p∗ and q ∗ are given by.4
d(v̄Cu + v̄Cv + v̄Cw ) dv̄C  
 3CVC = 9CVC . (19) 5π
dt dt p∗ = 5 cos 2ωt + [kW]
6
Equations (15) and (19) produce  
∗ 5π
q = 5 sin 2ωt + [kvai] (21)
√ 6
3VS I
v̄C = VC − cos(ω0 t + φ0 ). (20)
9ω0 CVC where the initial phase from (9)–(11) was set as φ = 5π/6 so
that the amplitude of iuv has its maximal value. Equation (21)
Hence, a low-frequency component of ω0 /2π appears on v̄C . and Fig. 7 give i∗d and i∗q as follows:
Note that the low-frequency components included in the nine  
dc-capacitor voltages are identical in amplitude and phase. 5π
i∗d = 50 cos 2ωt + [A]
6
 
5π
VII. E XPERIMENT i∗q = 50 sin 2ωt + [A] (22)
6
Figs. 9–12 show the experimental waveforms obtained from
the 100-V 5-kVA downscaled model. All the experimental where the dc component of i∗d coming from the overall volt-
waveforms were taken in a personal computer (PC) through age control is excluded from (22). Applying the inverse d−q
the YOKOGAWA WE7000 PC-based data acquisition system
with different sampling frequencies. Figs. 9, 10, and 12 had a
sampling frequency of 100 kHz, and Fig. 11 had a sampling 4 The unit volt ampere imaginary (“vai”) is introduced to the instantaneous
frequency of 20 kHz. reactive power, according to [24]
Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.
726 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO. 2, MARCH/APRIL 2012

 
5π
i∗w = 41 sin ωt − [A]. (23)
6

It is obvious from (23) that i∗u leads i∗w by 120◦ and i∗v by 240◦ ,
thus resulting in drawing the negative-sequence reactive power
from the ac mains.
Fig. 9 shows the experimental waveforms when the rated
negative-sequence reactive power of 5 kvar was controlled. The
cluster voltage vuv is a seven-level PWM waveform with a
voltage step of 60 V (= VC∗ ), containing much less harmonic
voltages as well as much less common-mode voltages than
traditional two-level voltage-source PWM converters. Since the
carrier frequency of each chopper cell is 2 kHz, the equivalent
switching frequency of the cluster is 12 kHz ( 2 kHz × 6).
The compensating currents iu , iv , and iw agree well with
their current commands given by (23). The waveform of iu
can be considered as a sinusoidal waveform with a funda-
mental component of 50 Hz. The total harmonic distortion
(THD) value of iu is as low as 3.3%.5 The amplitudes of the
compensating currents are slightly different (i.e., Iu = 42 A,
Iv = 35 A, and Iw = 39 A) due to the converter loss.
Fig. 11. Experimental waveforms when a 1-Hz active power of 5 kW was The u-phase cluster acts as an inductor because iuv lags vuv
controlled with a condition of C = 0.9 F (H = 2.9 s).
by 90◦ . On the other hand, the v- and w-phase clusters act as a
capacitor because ivw leads vvw by 90◦ and iwu leads vwu by
90◦ . The amplitudes of vuv , vvw , and vwu are different because
their operating modes are different. The circulating current
defined as (2) flows inside the delta-connected clusters to draw
the negative-sequence reactive power. As a result, the rms value
of iuv is as large as 34 A. Since the coupled inductor presents
the inductance as large as 37% to the circulating current, the
harmonic currents included in iZ are negligible.
The dc-capacitor voltages vC1u , vC1v , and vC1w contain both
dc and ac components, in which the voltage control regulates
the dc component at 60 V. The ac component consists of a
second-order (100 Hz) frequency component as shown in (14).
Fig. 9 shows that the peak-to-peak voltage fluctuation in vC1u
was observed as 3.6 V, which agrees well with the theoretical
value of 3.6 V obtained from (14).

B. Active-Power and Reactive-Power Controls

Flicker compensation of arc furnaces requires to control
positive-sequence reactive power, negative-sequence reactive
power, and low-frequency active power at the same time. As
an example, p∗ and q ∗ are given by
 
5π
p∗ = 1.7 sin(ωt/5) + 1.7 cos 2ωt + [kW]
6
 
Fig. 12. Experimental waveforms when a negative-sequence reactive power 5π
was increased from 2.5 to 5 kvar in 20 ms, kept constant for 20 ms, and q ∗ = 1.7 sin 2ωt + − 1.7 [kvai]. (24)
decreased from 5 to 2.5 kvar in 20 ms with a condition of C = 16.4 mF (H = 6
53 ms).
Fig. 10 shows the experimental waveforms when a positive-
sequence capacitive-reactive power of 1.7 kvar, a negative-
transformation into (22) produces i∗u , i∗v , and i∗w as sequence reactive power of 1.7 kvar, and a 10-Hz active power
 π of 1.7 kW were simultaneously controlled with a condition of
i∗u = 41 sin ωt − [A]
6
 π 5 The harmonic currents were counted up to the 40th component according to
i∗v = 41 sin ωt + [A] the Japanese harmonic guidelines.
2
Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.
HAGIWARA et al.: REACTIVE-POWER CONTROL BY PWM STATCOM BASED ON MMCC (MMCC-SDBC) 727

C = 16.4 mF (H = 53 ms). Since three different operating
modes are intermixed, the amplitude and phase of the currents
are changing dynamically as shown in Fig. 10. The amplitude of
iZ in Fig. 10 is one-third of that in Fig. 9 because the negative-
sequence reactive power in Fig. 10 is reduced to one-third of
that in Fig. 9.
The dc-capacitor voltages contain two frequency compo-
nents. One is a second-order frequency component of 100 Hz
resulting from the negative-sequence reactive power. The other
is a low-frequency component of 10 Hz resulting from the
active power. The three clusters have 100-Hz components with
different amplitudes but the same phase difference as 120◦ ,
whereas the three clusters have the common 10-Hz component
with the same amplitude. Fig. 10 indicates that the amplitude of
the 10-Hz component is much larger than those of the 100-Hz
components because it is inversely proportional to ω or ω0 as
shown in (14) and (20). Hence, larger capacitors are required
to control low-frequency active power. Fig. 10 suggests that the
10-Hz peak-to-peak voltage fluctuation included in vC1u was
observed as 6 V, which agrees well with the theoretical value of
5.9 V obtained from (20).
capacitor is required to decrease the voltage difference during
the transient period.
VIII. C ONCLUSION
This paper has discussed a PWM STATCOM using
an MMCC-SDBC, with focus on operating principle and
performance. The experimental results obtained from the
100-V 5-kVA downscaled model have led to the following
conclusions.
1) Low-voltage steps at the ac terminals of each cluster make
a significant contribution to reducing the THD values of
the compensating currents.
2) The SDBC has a capability to control negative-sequence
reactive power with the help of the circulating current
among the delta-connected clusters.
3) Positive-sequence reactive power, negative-sequence re-
active power, and low-frequency active power can be
controlled simultaneously.
These conclusions suggest that the SDBC is applicable to a
STATCOM for flicker compensation of arc furnaces.

C. Low-Frequency Active-Power Control A PPENDIX

∗
To control a 1-Hz instantaneous active power of 5 kW, p and A. Derivation of (12)
q ∗ are given by
Equations (1) and (2) give the cluster currents as
p∗ = 5 sin (ωt/50) [kW] iu − iv
iuv = + iZ
q∗ = 0 [kvai]. (25) 3
iv − iw
Fig. 11 shows the experimental waveforms in which each of ivw = + iZ
3
the dc capacitors was replaced with C = 0.9 F (H = 2.9 s).
Both compensating and cluster currents form 1-Hz envelopes as iw − iu
iwu = + iZ . (26)
shown in 3
√Fig. 11, in which the amplitude of the cluster currents
was 1/ 3 of that of the compensating currents. Carefully Substituting (9) into (26) yields
looking into Fig. 11 reveals that the amplitudes of both cluster  
and compensating currents when p∗ > 0 are larger than that of 2 π
iuv = − I sin ωt + φ − + iZ
the currents when p∗ < 0 due to the converter loss. The relation 3 6
of iZ = 0 always exists because no negative-sequence reactive  
2 π
power was controlled. The dc-capacitor voltages vC1u , vC1v , ivw = − I sin ωt + φ + + iZ
and vC1w contain both 1- and 100-Hz components in which the 3 2
  
former is much larger than the latter. Fig. 11 shows that the 2 5π
1-Hz peak-to-peak voltage fluctuation is 3.3 V which agrees iwu = − I sin ωt + φ − + iZ . (27)
3 6
well with the theoretical value of 3.3 V obtained from (20).
Each dc-capacitor voltage lags p∗ by 90◦ as obtained from (15) The circulating current iZ can be expressed as follows:
and (20). √
iZ = 2IZ sin(ωt + φZ ) (28)

D. Transient-State Performance
Fig. 12 shows the experimental waveforms when the
negative-sequence reactive power was increased from 2.5 to
5 kvar in 20 ms, kept constant for 20 ms, and decreased from
5 to 2.5 kvar in 20 ms with a condition of C = 16.4 mF (H =
53 ms). It is clear from Fig. 12 that the STATCOM can achieve
fast negative-sequence reactive-power control without delay
time. However, the waveforms of vC1u , vC1v , and vC1w show
that a maximal voltage difference of 12 V (20%) occurs during
the transient state. To increase the capacitance of each dc
Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.
where IZ is the rms value and φZ is the initial phase.
The instantaneous active power formed between the voltage
and the current in each cluster is represented as
puv = vuv · iuv
pvw = vvw · ivw
pwu = vwu · iwu . (29)
Let the dc component of puv , pvw , and pwu be (puv )dc , (pvw )dc ,
and (pwu )dc , respectively. The following equation should be
728 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO. 2, MARCH/APRIL 2012

satisfied: [8] K. Fujii, U. Schwarzer, and R. W. De Doncker, “Comparison of

hard-switched multi-level inverter topologies for STATCOM by loss-
(puv )dc = (pvw )dc = (pwu )dc = 0. (30) implemented simulation and cost estimation,” in Conf. Rec. IEEE PESC,
2005, pp. 340–346.
[9] K. Fujii, R. W. De Doncker, and S. Konishi, “A novel dc-link voltage
Substituting (8), (27)–(29) into (30) yields control of PWM-switched cascade cell multi-level inverter applied to
STATCOM,” in Conf. Rec. IEEE IAS Annu. Meeting, 2005, pp. 961–967.
I
IZ = √ [10] H. Akagi, S. Inoue, and T. Yoshii, “Control and performance of a trans-
3 formerless cascade PWM STATCOM with star configuration,” IEEE
Trans. Ind. Appl., vol. 43, no. 4, pp. 1041–1049, Jul./Aug. 2007.
π [11] J. A. Barrena, L. Marroyo, M. A. Rodriguez, and J. R. Torrealday, “Indi-
φZ = − φ + . (31)
2 vidual voltage balancing strategy for PWM cascaded H-Bridge converter-
based STATCOM,” IEEE Trans. Ind. Electron., vol. 55, no. 1, pp. 21–29,
Substituting (31) into (27) and (28) yields (12). Jan. 2008.
[12] B. Gultekin, C. O. Gerçek, T. Atalýk, M. Deniz, N. Biçer, M. Ermi,
N. Köse, C. Ermi, E. Koç, I. Çadirci, A. Açik, Y. Akkaya, H. Toygar, and
S. Bideci, “Design and implementation of a 154 kV, ±50 MVAr trans-
B. Derivation of (14) mission STATCOM based on 21-Level cascaded multilevel converter,” in
Conf. Rec. IEEE ECCE, 2010, pp. 3936–3948.
Concerning one of the u-phase bridge cells in Fig. 4, the ac- [13] L. Maharjan, S. Inoue, and H. Akagi, “A transformerless energy stor-
∗
voltage command v1u is obtained from Fig. 8 as age system based on a cascade multilevel PWM converter with star
∗
configuration,” IEEE Trans. Ind. Appl., vol. 44, no. 5, pp. 1621–1630,
∗ ∗ ∗ vuv Sep./Oct. 2008.
v1u = vA + vB1u + . (32)
3 [14] L. Maharjan, S. Inoue, and H. Akagi, “SOC (state-of-charge)-balancing
control of a battery energy storage system based on a cascade PWM
The first and second terms of the right hand are negligi- converter,” IEEE Trans. Power Electron., vol. 24, no. 6, pp. 1628–1636,
∗ ∗ Jun. 2009.
ble, compared to the third term (vA = vB1u  0).6 Therefore, [15] C. Schauder, “STATCOM for compensation of large electric arc
∗
v1u (= v1u ) is approximated as furnace installations,” in Conf. Rec. IEEE PES Summer Meeting, 1999,
pp. 1109–1112.
∗ vuv
v1u = v1u  [16] A. G. Cerrada, P. G. Gonzalez, R. Collantes, T. Gomez, and J. Anzola,
3 “Comparison of thyristor-controlled reactors and voltage-source inverters
√  π
for compensation of flicker caused by arc furnaces,” IEEE Trans. Power
2V Del., vol. 15, no. 4, pp. 1225–1231, Oct. 2000.
= sin ωt + (33)
3 6 [17] C. Han, Z. Yang, B. Chen, A. Q. Huang, B. Zhang, M. R. Ingram, and
A. Edris, “Evaluation of cascade-multilevel-converter-based STATCOM
∗
where the relation of vuv = vuv is applied. The following for arc furnace flicker mitigation,” IEEE Trans. Ind. Appl., vol. 43, no. 2,
pp. 748–755, Mar./Apr. 2007.
relationship exists in Fig. 4 as: [18] K. Usuki, F. Aoyama, and M. Hanamatsu, “Development of SVC control
for suppressing voltage fluctuations,” in Conf. Rec. IEEE ICPE, 2011,
dvC1u dvC1u
v1u · iuv = vC1u · C  CVC (34) pp. 2073–2080.
dt dt [19] M. Hagiwara, R. Maeda, and H. Akagi, “Negative-sequence reactive-
power control by the modular multilevel cascade converter based on
where a reasonable approximation of vC1u = VC is applied.7 double-star chopper-cells (MMCC-DSCC),” in Conf. Rec. IEEE ECCE,
Substituting (13) and (33) into (34) produces (14) with a 2010, pp. 3949–3954.
[20] M. Hagiwara, R. Maeda, and H. Akagi, “Control and analysis of the
relation of V = VS . modular multilevel cascade converter based on double-star chopper-cells
(MMCC-DSCC),” IEEE Trans. Power Electron., vol. 26, no. 6, pp. 1649–
1658, Jun. 2011.
R EFERENCES [21] M. Hagiwara and H. Akagi, “Control and experiment of pulsewidth-
[1] H. Akagi, “Classification, terminology, and application of the modular modulated modular multilevel converters,” IEEE Trans. Power Electron.,
multilevel cascade converter (MMCC),” IEEE Trans. Power Electron., vol. 24, no. 7, pp. 1737–1746, Jul. 2009.
vol. 26, no. 11, pp. 3119–3130, Nov. 2011. [22] H. Akagi, Y. Kanazawa, and A. Nabae, “Instantaneous reactive power
[2] J. S. Lai and F. Z. Peng, “Multilevel converters-A new breed of power compensators comprising switching devices without energy storage com-
converters,” IEEE Trans. Ind. Appl., vol. 32, no. 3, pp. 509–517, ponents,” IEEE Trans. Ind. Appl., vol. IA-20, no. 3, pp. 625–630,
May/Jun. 1996. May/Jun. 1984.
[3] F. Z. Peng and J. S. Lai, “Dynamic performance and control of a static [23] H. Fujita, S. Tominaga, and H. Akagi, “Analysis and design of a
var generator using multilevel inverters,” IEEE Trans. Ind. Appl., vol. 33, dc voltage-controlled static var compensator using quad-series voltage-
no. 3, pp. 748–755, May/Jun. 1998. source inverters,” IEEE Trans. Ind. Appl., vol. 32, no. 4, pp. 970–977,
[4] Y. Liang and C. O. Nwankpa, “A new type of STATCOM based on Jul./Aug. 1996.
cascading voltage-source inverters with phase-shifted unipolar SPWM,” [24] H. Akagi, E. H. Watanabe, and M. Aredes, Instantaneous Power Theory
IEEE Trans. Ind. Appl., vol. 35, no. 5, pp. 1118–1123, Sep./Oct. 1999. and Applications to Power Conditioning. Piscataway, NJ: IEEE Press,
[5] S. Sirisukprasert, A. Q. Huang, and J. S. Lai, “Modeling, analysis and con- 2007.
trol of cascaded-multilevel converter-based STATCOM,” in Proc. Power
Eng. Soc. Gen. Meeting, 2003, pp. 13–17.
[6] C. K. Lee, J. S. K. Leung, S. Y. R. Hui, and H. S. H. Chung, “Circuit-level
comparison of STATCOM technologies,” IEEE Trans. Power Electron., Makoto Hagiwara (M’06) was born in Tokyo,
vol. 18, no. 4, pp. 1084–1092, Jul. 2003. Japan, in 1979. He received the B.S., M.S., and
[7] F. Z. Peng and J. Wang, “A universal STATCOM with delta- Ph.D. degrees in electrical engineering from Tokyo
connected Cascade multilevel inverter,” in Conf. Rec. IEEE PESC, 2004, Institute of Technology, Tokyo, in 2001, 2003, and
pp. 3529–3533. 2006, respectively.
Since April 2006, he has been an Assistant Pro-
6 The values of v ∗ and v ∗ fessor in the Department of Electrical and Elec-
A B1u are as small as 1% of the rms value of tronic Engineering, Tokyo Institute of Technology.
∗ /3 (= 33 V) under the steady-state condition.
vuv
7 The ac components are as small as 5% of the dc component of 60 V His research interests are modular multilevel cascade
converters for utility applications.
from Fig. 9.
Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.
HAGIWARA et al.: REACTIVE-POWER CONTROL BY PWM STATCOM BASED ON MMCC (MMCC-SDBC) 729

Ryo Maeda was born in Yokohama, Japan, in 1987. Hirofumi Akagi (M’87–SM’94–F’96) was born in
He received the B.S. degree in electrical engineering Okayama, Japan, in 1951. He received the B.S. de-
from Tokyo Metropolitan University, Tokyo, Japan, gree in electrical engineering from Nagoya Institute
in 2009, and the M.S. degree in electrical engineering of Technology, Nagoya, Japan, in 1974, and the M.S.
from Tokyo Institute of Technology, Tokyo, in 2011. and Ph.D. degrees in electrical engineering from
Since April 2011, he has been with Tokyo Electric Tokyo Institute of Technology, Tokyo, Japan, in 1976
Power Company, Inc., Tokyo. and 1979, respectively.
In 1979, he joined Nagaoka University of Tech-
nology, Nagaoka, Japan, as an Assistant Profes-
sor and then became an Associate Professor in the
Department of Electrical Engineering. In 1987, he
was a Visiting Scientist at the Massachusetts Institute of Technology (MIT),
Cambridge, for ten months. From 1991 to 1999, he was a Professor in the
Department of Electrical Engineering, Okayama University, Okayama. From
March to August of 1996, he was a Visiting Professor at the University of
Wisconsin, Madison, and then at MIT. Since January 2000, he has been a
Professor in the Department of Electrical and Electronic Engineering, Tokyo In-
stitute of Technology. His research interests include power conversion systems,
motor drives, active and passive electromagnetic interference (EMI) filters,
high-frequency resonant inverters for induction heating and corona discharge
treatment processes, and utility applications of power electronics such as active
filters, self-commutated back-to-back systems, and flexible ac transmission
system devices. He has authored or coauthored more than 100 IEEE T RANS -
ACTIONS papers and two invited papers published in P ROCEEDINGS OF THE
IEEE in 2001 and 2004. The total citation index for all his papers in Google
Scholar is more than 15 000. He has made numerous presentations as a keynote
or invited speaker internationally.
Dr. Akagi served as the President of the IEEE Power Electronics Society
for 2007–2008. He was elected as a Distinguished Lecturer of the IEEE
Power Electronics and IEEE Industry Applications Societies for 1998–1999.
He was a recipient of five IEEE T RANSACTIONS Prize Paper Awards and nine
IEEE Conference Prize Paper Awards. He was also the recipient of the 2001
IEEE William E. Newell Power Electronics Award, the 2004 IEEE Industry
Applications Society Outstanding Achievement Award, and the 2008 IEEE
Richard H. Kaufmann Technical Field Award.

Authorized licensed use limited to: Florida State University. Downloaded on February 23,2024 at 17:34:46 UTC from IEEE Xplore. Restrictions apply.