You are on page 1of 12

3684 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 67, NO.

5, MAY 2020

DC Voltage Control Strategy of Three-Terminal


Medium-Voltage Power Electronic
Transformer-Based Soft Normally Open Points
Shaodi Ouyang , Member, IEEE, Jinjun Liu , Fellow, IEEE, Yue Yang ,
Xingxing Chen , Student Member, IEEE, Shuguang Song , Student Member, IEEE, and Hongda Wu

Abstract—The power electronic transformer (PET) or


solid-state transformer (SST) is a key technique in the
future distribution grid. One of PET’s future possible appli-
cation is the soft normally open points (SNOP) to handle
the power and voltage challenge brought by the renewable
energy sources. In this paper, a PET-based three-terminal
medium-voltage SNOP topology is proposed. The topology
consists of three cascade H-bridge (CHB) stages and a mul-
tiactive bridge (MAB) stage. The dc voltage control strategy
for the proposed PET-SNOP is studied. Two strategies –
the CHB+CHB+CHB strategy and CHB+MAB strategies
are developed from the existing strategies but they both
have some drawbacks. In order to overcome the drawbacks,
Fig. 1. Future distribution network with SNOP.
a novel dc voltage control strategy—CHB×3+MAB strategy
is proposed, which combines the advantages and avoids the
drawbacks of these two strategies, moreover, it also brings The SNOP can also offer additional functions to the distribution
more flexibility and robustness to the dc voltage control.
The effectiveness of the proposed PET-SNOP topology as grid like energy storage [4], static synchronous compensator [5],
well as the proposed CHB×3+MAB dc voltage control strat- improving the grid’s restoration performance during faults [6],
egy is verified by both simulation and experiment results. etc.
Index Terms—DC voltage control, power electronic trans-
From the view of power electronic topology, the technical
former, soft normally open point, three-terminal. requirement of multiterminal SNOP is the capability of multi-
terminal power transfer and electrical isolation. The power elec-
tronic transformer (PET) [7] or solid-state transformer (SST)
I. INTRODUCTION [8] is an attractive topology for SNOP as it can achieve isola-
HE renewable energy sources raises a lot of challenges to tion through a high-frequency isolation stage instead of several
T the distribution grid like voltage violation, feeder load un-
balancing, intermittent power flow, un-optimized operation, etc.
low-frequency transformers, bringing advantage in efficiency
and system size. In this paper, a three-terminal medium voltage
The feeder power conditioner as shown in Fig. 1, which is also (MV) PET-based SNOP is proposed, whose circuit is shown in
named “soft normally open points” (SNOP) [1] or “soft open Fig. 2. The system consists of three cascaded H-bridge (CHB)
points” (SOP) [2], is a recently proposed solution to these chal- stages and a multi-active bridge (MAB) [9] stage. Each MAB
lenges. The SNOP can provide additional controllable electrical converter connects three CHB modules from three CHB stages
paths among feeders allowing for flexible power flow adjust- and they together form a PET cell as shown in Fig. 3.
ment, which can effectively handle these challenges and in- The power structure of the proposed PET-SNOP features in
crease the penetration rate of renewable energy sources [1]–[3]. symmetrical multiterminal structure and modularized structure,
making it different from the existing three major types of CHB-
Manuscript received July 15, 2018; revised November 10, 2018, Febru- based PETs: the FREEDM/Traction PET with a common low
ary 16, 2019, and May 13, 2019; accepted May 29, 2019. Date of publi- voltage direct current bus (the LV stage is not modernized) [10],
cation June 19, 2019; date of current version January 3, 2020. This work
was supported by the National Key Research and Development Plan un- [11], the asymmetrical multiterminal modularized UNIFLEX-
der Grant 2018YFB0905800. (Corresponding author: Shaodi Ouyang.) PM PET [12], [13], and the two-terminal modularized PET [14],
The authors are with the State Key Lab of Electrical Insulation and [15]. The dc voltage control strategy of such a symmetrical
Power Equipment, School of Electrical Engineering, Xi’an Jiaotong Uni-
versity, Xi’an 710049, China (e-mail:, oysd1989@stu.xjtu.edu.cn; jjliu@ multiterminal modularized structure has not been studied yet
mail.xjtu.edu.cn; flyeryy@stu.xjtu.edu.cn; xingxingchen@stu.xjtu.edu. thus this paper undertakes this research.
cn; shuguang.song.chine@gmail.com; hongda@stu.xjtu.edu.cn). To achieve dc voltage control of modularized PET, two dc
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. voltage control strategies have been proposed in the existing
Digital Object Identifier 10.1109/TIE.2019.2922915 literatures: 1) the CHB+CHB strategy in [15] and [16] in which
0278-0046 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
OUYANG et al.: DC VOLTAGE CONTROL STRATEGY OF THREE-TERMINAL MEDIUM-VOLTAGE PET-BASED SNOPS 3685

Fig. 4. Circuit of the CHB converter.

Fig. 2. Circuit of the proposed three-terminal PET-SNOP.

Fig. 5. Inner current loop of a CHB converter.

Fig. 3. Circuit of a PET cell.

Fig. 6. Overall dc voltage control.


the dc voltages are controlled by both CHB stages, meanwhile
the DAB stage do not take responsibility for dc voltage con- VI show the experiment platform and the experiment results.
trol; and 2) the CHB+DAB strategy in [12], [17]–[20] in which Section VII concludes this paper.
the dc voltages are controlled by one CHB stage and the DAB The naming convention for the PET cell, H-bridge modules,
stage, meanwhile the other CHB do not take responsibility for dc voltages and currents is “terminal phase+number(X t pn ).” For
voltage control. These two strategies can be extended to the pro- example, CELL pn (p = A,B,C; n = 1, 2, . . . , N) consists of
posed three-terminal PET-SNOP as CHB+CHB+CHB strategy six H-bridges: CHB I pn , CHB I I pn , CHB I I I pn , MAB I pn ,
and CHB+MAB strategy. However, they both have drawbacks: MAB I I pn , and MAB I I I pn . The MAB converter is marked as
the CHB+CHB+CHB strategy suffers from the cell power mis- MAB pn . The dc voltages of this cell are marked as Vdc I pn ,
match owing to MAB parameter mismatch; the CHB+MAB Vdc I I pn , and Vdc I I I pn . The three high-frequency ac square
strategy can overcome the MAB parameter mismatch issue but voltages of the MAB are marked as vM AB I pn , vM AB I I pn , and
sacrifices the symmetrical control structure, which will compli- vM AB I I I pn . The cell A1 is taken as an example in Fig. 3.
cate three multiterminal application.
In order to overcome these drawbacks, a novel dc control
II. BASIC PRINCIPLE OF CHB AND MAB
strategy: CHB×3+MAB strategy, is proposed in this paper
which combines the advantages, meanwhile avoids the draw- The ac–dc stage of the proposed PET-SNOP is three-phase
backs of both strategies. The CHB×3+MAB strategy can also CHB, the circuit is shown in Fig. 4.
bring more flexibility and robustness to PET’s dc voltage control The general goal of CHB control is ac current regulation and
than the existing strategies. dc voltage regulation. A generally used ac control block diagram
This paper is organized as follows. Section II briefly in- of a CHB converter is shown in Fig. 5.
troduces basis of CHB and MAB. Section III introduces the The dc control of a CHB converter generally consists of three
CHB+CHB+CHB strategy, CHB+MAB strategy, and the pro- loops, an overall dc voltage loop controls its average (overall)
posed CHB×3+MAB strategy for the three-terminal PET- dc voltage by adjusting the d-axis current command as shown
SNOP. Section IV shows the simulation results. Sections V and in Fig. 6; the among-phase power control loop can be used to

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
3686 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 67, NO. 5, MAY 2020

Fig. 7. Among-phase dc voltage balancing control.

Fig. 10. CHB+CHB+CHB Strategy.




V V
PI = D Cπ IX 3DLC I I δI I (π − δI I )






V V
+ D C πIX D3 LC I I I δI I I (π − δI I I )
Fig. 8. In-phase dc voltage balancing control. ⎪




⎨ PI I = V D C I I V D C I (−δI I )(π + δI I )
πX3L
. (1)

⎪ V V
+ D C πI IX 3DLC I I I (δI I I − δI I )(π − δI I I + δI I )





⎪ V V

⎪ PI I I = D C πIXI I3 LD C I (−δI I I )(π + δI I I )



⎩ V V
+ D C πI IXI 3 LD C I I (δI I − δI I I )(π − δI I + δI I I )

III. DC VOLTAGE CONTROL STRATEGY FOR THE THREE


TERMINAL MV PET
Fig. 9. Equivalent circuit for the MAB.
In this paper, three dc voltage control strategies are dis-
cussed, among which the CHB+CHB+CHB strategy and the
CHB+MAB strategies are developed from existing strategies of
balance the dc voltage among phases by zero-sequence voltage
two-terminal PET, and the CHB×3+MAB strategy is proposed
injection [21], [22] or negative-sequence current injection [23]
in this paper.
as shown in Fig. 7. An inner-phase power control loop can be
used to balance all the dc voltages among the phase by adjusting
the amplitude of the modulation wave of the H-bridge converters A. CHB+CHB+CHB Strategy
within each phases as shown in Fig. 8. It could be seen that 3N The CHB+CHB+CHB strategy is developed from the
proportional-integral (PI) controllers (1 in Fig. 6, 2 in Fig. 7, CHB+CHB strategy in [15]. In this strategy, DC I is controlled
and 3N-3 in Fig. 8) are needed to achieve dc voltage control. by CHB stage I; DC II is controlled by CHB stage II; DC III
The isolation stage of the proposed PET-SNOP is formed is controlled by CHB strategy III; the MAB stage controls the
by MAB converters, whose circuit is shown in Fig. 3. The active power transferred between the three terminals; all the
Y-equivalent circuit for the MAB is shown in Fig. 9(a), where MABs share the same phase shift angle δI I and δI I I , as shown
three square-wave voltage sources are connected to a common in Fig. 10.
magnetizing inductor through three inductors [9], [24]. A poly- As all the CHBs need to achieve its own dc voltage control,
gon model could be obtained [9] and further simplified by ne- they all need to apply a complete dc voltage control including
glecting the magnetizing inductor [25] as shown in Fig. 9(b). the overall dc voltage control in Fig. 6, among-phase power
The terminal powers are determined by the phase shift differ- control in Fig. 7 and in-phase power control in Fig. 8. The three
ence among the square-wave voltages. The power between any CHBs require 9N PI controllers to achieve dc voltage control.
two terminals can be easily determined by a similar way to DAB As the MAB does not need to achieve dc voltage control,
and the overall terminal power can be derived as follows: it only needs to manage the transferring power among the

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
OUYANG et al.: DC VOLTAGE CONTROL STRATEGY OF THREE-TERMINAL MEDIUM-VOLTAGE PET-BASED SNOPS 3687

Fig. 11. Control block diagram of MAB under CHB+CHB+CHB


strategy.

three terminals. As a feeder power balancer, the power com-


mand PI schedule , PI I schedule , and PI I I schedule must be as- Fig. 12. CHB+MAB strategy.
signed by the grid schedule using communication. According to
these power commands, the phase shift angle between the three
H-bridge converters among an MAB can be calculated in a nu-
merical way [25] from inversed (1). A power feedback loop can
be used to adjust the phase shift angle which decreases the power
error. Therefore, the MAB control block diagram is shown in
Fig. 11. Fig. 13. Control of MAB stage under CHB+MAB strategy.
Letting all the MAB share the same phase shift angle is based
on the assumption that all the MAB have precisely equal param-
eters. Unfortunately, in real HFT and uH-level inductor machin- the phase shift angles. The basic form of this control is formed
ing, it is difficult to make all MAB parameters precisely equal by two PI-based controllers [27], [28]. A power calculation-
[26]. Consequently, the transferred power of MAB as well as based feedforward [25] as well as a local-linearization-based
the PET cells will not be well balanced, causing negative effects decouple loop [29], [30] can be added to improve the dynamic
such as total harmonic distortion increase, unbalance compo- performance. In this paper, the basic form is used as shown in
nents to the grid, etc. Fig. 13.
To balance the MAB transferred power with different param- Owing to that, all the dc voltage control variables are zero,
eters, the powers of all MABs need to be sampled at the dc in both CHB II and CHB III, all the converters in a same phase
capacitors and then the phase shift angles need to be adjusted by can share a same voltage command as (2). As all the H-bridge
to balance the transferred powers of all MABs [16]. This control converters in same phase also shares a same current as (3), thus
strategy requires high-bandwidth MAB power sample system to the transferring power of H-bridge converters in a same phase
achieve precise measurement on MAB high-frequency currents, are also the same as (4)
moreover, additional 2N-2 PI controllers are needed to achieve
power balancing control which increases the system coding cost. u∗CHB t p1 = u∗CHB t p2 = · · · · · · = u∗CHB t pN (2)
iCHB t p · u∗CHB t p1 = iCHB t p · u∗CHB t p2 = ······
B. CHB+MAB Strategy
= iCHB t p · u∗CHB t pN (3)
This strategy is developed from CHB+DAB strategy for
UNIFLEX-PM and two-terminal modularized PET in [12], PCHB t p1 = PCHB t p2 = · · · · · · = PCHB t pN (4)
[17]–[20]. In the CHB+MAB strategy, DC I is controlled by
CHB I; DC II and DC III are controlled by the MAB stage; where p = A, B or C; t = II or III.
CHB II and CHB III does not take part in dc voltage control Therefore, the transferring powers of all PET cells are bal-
but only control the ac current, as shown in Fig. 12. anced and so the active power difference among cells is only the
As the CHB II and CHB III does not need to control the power loss difference. Comparably, in the CHB+CHB+CHB
dc voltage, the v0∗ II , v0∗ III , kA n I I , kB n I I , kC n I I , kA n I I I , strategy, the active power difference among cells is both the
kB n I I I and kC n I I I (n = 1, 2, . . . N) are all zero. Meanwhile, power loss difference and the transferring power difference
CHB I should apply a complete dc voltage control including the caused by the MAB parameters’ difference. Consequently, it
overall dc voltage control in Fig. 6, among-phase power control can be easily concluded that the CHB+MAB method has an
in Fig. 7 and in-phase power control in Fig. 8. obvious advantage in PET cell power balancing.
The MAB has to control the DC II and DC III. That means, The number of PI controllers of CHB+MAB strategy is also
for each MAB converter, one of its dc voltage is controlled by the fewer than the CHB+CHB+CHB strategy. As one of the CHB
CHB stage, meanwhile the other two dc voltages needs to be con- need 3N PI controllers and the MAB needs 2N PI controllers,
trolled by the MAB converter. This can be achieved by adjusting only 5N PI controllers are needed.

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
3688 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 67, NO. 5, MAY 2020

Fig. 14. Possible three-terminal PET operation scenario that excludes


a terminal.

However, the CHB+MAB strategy would suffer from the


asymmetrical control structure. It can be seen that the CHB
stages have different responsibilities in dc balancing control:
CHB I is “special” as it is fully responsible for dc balancing
control, while other CHBs are not responsible for dc balancing
control. In a symmetrical multiterminal application, any termi-
nal could be excluded from operation owing to the grid power
schedule or a grid fault as shown in Fig. 14. If the excluded termi-
nal happened to be the “special” terminal, the system would lose
the dc voltage control. Therefore, a symmetric control strategy
with no “special” CHB will be more suitable for multiterminal
application.

C. CHB×3+MAB Strategy
In this paper, a modified dc balancing control strategy that
combines the advantage of the aforementioned two strategies is
proposed. Similar to the CHB+CHB+CHB strategy, the pro-
posed balancing control strategy is symmetric in structure, all
Fig. 15. (a) Physical circuit for a single-phase three-terminal MV PET
the CHBs share the responsibility of dc balancing control so and (b) its equivalent circuit from the view of CHB×3+MAB strategy.
there is no “special” CHB. Similar to the CHB+MAB strategy,
the MAB phase shift angle is not obtained by power schedule,
but still by dc voltage loop, thus there is no unequal transferred and Vdc I I I A 1∼N as well as the MAB stage no longer exist;
power issue caused by the unequal MAB parameters, so the instead, they are replaced by a PET cell average dc voltage
MAB power sampling system as well as the MAB power bal- group Vdc Cell A 1∼N . The control of Vdc C ell I 1∼N is shared
ancing loop is not needed. by all the CHBs by a set of user-determined ratio kI , kI I , and
The key concepts of the proposed dc voltage strategy are as kI I I , thus there is no “special” CHB in this control strategy.
follows. The proposed dc voltage control includes four layers. One is
1) Each MAB balances the three individual dc voltages for the MAB stage: the in-cell dc voltage balancing control; the
within its PET cell. other three are for the CHB stage: the overall dc voltage control,
2) The CHBs do not operate on the individual dc voltages; the among-phase dc voltage balancing control, and the in-phase
instead, they balance the PET cell average dc voltages. dc voltage balancing control.
3) The responsibility of PET cell average dc voltage control The overall structure of dc voltage processing is shown in
is shared by all the CHBs. Fig. 16. The 9N dc voltages Vdc t pn are sampled, they are first
As all the four stages take part in the dc voltage control, it is transferred to the MAB stage to achieve in-cell dc voltage bal-
called CHB×3+MAB strategy. ancing, then the 3N cell average voltages Vdc Cell pn are obtained
The phase A PET is taken as an example to illustrate the and transferred to the CHB stages to achieve in-phase balancing,
control strategy. The physical circuit of the phase A PET among-phase balancing, and the overall voltage control.
is shown in Fig. 15(a): CHB I A is connecting Vdc I A 1∼N , In order to avoid the i∗d of CHBs as well as the phase shift
CHB I I A is connecting Vdc I I A 1∼N , and CHB I I I A is angles of MABs from being influenced by the 100 Hz compo-
connecting Vdc I I I A 1∼N ; the three dc capacitor groups are nent on the dc capacitors, a low-pass filter was added to the dc
connected by the MAB stage. From the viewpoint of the voltage sample process before obtaining Vdc t pn as [15].
CHB stages, every three dc voltages within a PET cell are 1) In-Cell DC Voltage Balancing Control (MAB): Com-
unified as one dc voltage. Therefore, the equivalent circuit is pared to the MAB control in the CHB+MAB strategy, the

modified as shown in Fig. 15(b): Vdc I A 1∼N , Vdc I I A 1∼N , difference in CHB×3+MAB strategy is to replace the Vdc

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
OUYANG et al.: DC VOLTAGE CONTROL STRATEGY OF THREE-TERMINAL MEDIUM-VOLTAGE PET-BASED SNOPS 3689

Fig. 20. In-phase dc voltage balancing control of CHB×3+MAB


strategy.

Fig. 16. DC voltage signal processing.

Fig. 17. Control of MAB stage under CHB×3+MAB Strategy.

Fig. 21. Flowchart of CHB stages’ control of CHB×3+MAB Strategy.

Fig. 18. Overall dc voltage control of CHB×3+MAB strategy. dc voltages to the average dc voltage. The zero sequence voltage
is calculated according to the current of all three terminals.
4) In-Phase DC Voltage Balancing Control (CHB): The
in-phase dc voltage balancing loop is shown in Fig. 20. By
comparing the PET cell average dc voltages to its phase average
dc voltage, trimming coefficients kCELL pn (p = A, B, C, n =
1, . . . , N ) are generated to adjust the input/output active power
of the PET cells within phase p.
In the proposed strategy, there would be no “special” CHB,
thus the overall dc voltage control, the among-phase dc voltage
Fig. 19. Among-phase dc voltage balancing control of CHB×3+MAB balancing control as well as the in-phase dc voltage balancing
strategy. control are shared by all the CHBs. That means the i∗d , v0∗ ,
and k Cell pn are distributed to all the three CHB stages simul-
by Vdc I as shown in Fig. 17. This loop will only achieve taneously through distribution parameters kI , kI I , and kI I I .
Vdc I = Vdc I I = Vdc I I I but not maintaining them to any spe- The distribution parameters of 1) overall loop, 2) among-phase
cific value. loop, 3) and the in-phase loop can be set differently. The overall
As the in-cell dc voltage balancing is done by the MAB, control flowchart is shown in Fig. 21.
the CHB stages do not operate on the individual dc voltages. The number of PI controllers is also 5N, where 2N belongs
The CHBs only need to control the cell average dc voltages to MAB stage, meanwhile 3N belongs to the dc voltage control.
Vdc Cell pn . Compared to the CHB+MAB strategy, the added procedure is
2) Overall DC Voltage Control (CHB): This control loop the cell average dc derivation and the distribution parameters.
maintains the overall dc voltages of the whole PET as shown in The existence of kI , kI I , and kI I I can make the
Fig. 18. CHB×3+MAB strategy very flexible as they can be set
3) Among-Phase DC Voltage Balancing Control (CHB): according to the operating states of the PET-SNOP. Also,
The among-phase dc voltage balancing loop is shown in Fig. 19: the CHB×3+MAB strategy is also more robust than the
The among-phase power compensation commands of the whole CHB+MAB strategy. When facing a terminal loss in Fig. 14,
PET system, PA and PB , are obtained by comparing phase it can automatically transfer the dc voltage control of the lost

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
3690 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 67, NO. 5, MAY 2020

TABLE I
PET PARAMETERS

Fig. 23. Simulation results of CHB×3+MAB strategy.

Fig. 24. Simulation results of CHB+MAB strategy.

Fig. 22. Simulation results of CHB+CHB+CHB strategy.

terminal to the other terminals to ensure the PET operation,


which cannot be achieved by CHB+MAB strategy as the dc
capacitors of the lost terminal will be continuously charging or
discharging and finally causing the system collapse.

IV. SIMULATION RESULTS


A PET simulation model is built in MATLAB/Simulink to
verify the proposed topology as well as the three proposed con-
trol strategies. The circuit is as Figs. 2 and 3 and the parameters
Fig. 25. Simulation results of CHB×3+MAB strategy.
are shown in Table I.
The first simulation is built to verify the power sharing ad-
vantage of CHB×3+MAB over the CHB+CHB+CHB strat- The second simulation is built to verify the robust advan-
egy. Within each phase, the MAB inductors of the two cells tage of CHB×3+MAB over the CHB+MAB strategy. In the
are set 10% different from each other. The simulation results of CHB+MAB topology, the CHB I is selected as the “special”
CHB+CHB+CHB strategy are shown in Fig. 22 and those of CHB to control the dc voltage. In the CHB×3+MAB strategy,
CHB×3+MAB strategy are shown in Fig. 23. the dc voltage control is set to be shared by all the three CHBs
In the CHB+CHB+CHB strategy, owing to different MAB with kI : kI I : kI I I = 1/3 : 1/3 : 1/3.
inductors among cells, the power of PET cells are also different Before 1.0 s, the system is normally running; at 1.0 s, the
thus the H-bridge converters within each CHB phase would terminal I is cut off from the grid. The simulation results are
generate different ac voltages as shown in Fig. 22. shown in Figs. 24 and 25.
In the CHB×3+MAB strategy, the MAB phase shift com- In the CHB+MAB strategy, the original power(current) com-
mand is given by voltage closed-loop, thus the power of PET mand is PI∗ = 330 kW + PLoss , PI∗I = −110 kW and PI∗I I =
cells are still balanced (regardless of the power loss differ- −220 kW, while in the new state, the power(current) command
ence) despite having different MAB inductors. The H-bridge becomes PI∗ = ∞(saturated), PI∗I = −110 kW and PI∗I I =
converters within each CHB phase would generate a same ac −220 kW. As the CHB I actual input current becomes zero, the
voltage as shown in Fig. 23. The power sharing advantage of dc group I continues to loss power and eventually becomes zero.
CHB×3+MAB strategy over the CHB+CHB+CHB strategy As dc group I drops, the dc group II and dc group III are also
is verified. unable to maintain at 2000 V, instead, it drops to around 1500 V

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
OUYANG et al.: DC VOLTAGE CONTROL STRATEGY OF THREE-TERMINAL MEDIUM-VOLTAGE PET-BASED SNOPS 3691

TABLE III
HIGH-FREQUENCY MAGNETIC COMPONENT DESIGN

Fig. 26. Experiment platform of 36 H-bridge three-terminal PET.

TABLE II
PET PARAMETERS

and turns into the diode-rectifier state as shown in Fig. 24. Ob- Fig. 27. (a) iA of three terminals before disturbance. (b) iA of three
viously the system collapsed after the terminal I is cut off. terminals in the new steady state. (c) id of three terminals.
In the CHB×3+MAB strategy, as the dc voltage control is
shared by all the CHB stages, the dc voltages can still be con-
trolled by the other CHB stages. The original power(current) The experiment comparison between the proposed
command is PI∗ = 330 kW + 1/3PLoss , PI∗I = −110 kW + CHB×3+MAB strategy and the other two are shown in 6.3.
1/3PLoss , and PI∗I I = −220 kW + 1/3PLoss , while in the new
state, the power(current) command becomes PI∗ = 495 kW + A. Module Power Disturbance
1/2PLoss , PI∗I = 55 kW + 1/2PLoss , and PI∗I I = −55 kW + At normal state, the input current of three terminals are set as
1/2PLoss , the system can still be balanced thus all the dc volt- 4, 8, and –12 A which means absorbing 1.32 kW from termi-
ages can still maintain at 2000 V as shown in Fig. 25. Obviously nal I, 2.64 kW from terminal II and delivering 3.96 kW to
the system can maintain running without collapsed. The ro- terminal III. The dc voltage control is distributed to
bust advantage of CHB×3+MAB strategy over the CHB+MAB all three CHBs by kI  1 :kII 
1 :kIII 1 = 1/3:1/3:1/3,
strategy is verified. kI  2 : kII 
2 III 
:k 2 = 6/11:3/11:2/11, k I 3 II 
:k 3 :kIII 
3 =
1/3:1/3:1/3. The output of overall dc voltage loop is  i∗d =
V. LOW VOLTAGE EXPERIMENT PLATFORM 3.3 A to compensate for the SST power loss. The  i∗d is evenly
A downscaled experiment platform of the three-terminal PET distributed to three CHBs thus the commands are adjusted to 5.1,
described in Figs. 2 and 3 was built, as shown in Fig. 26. The 9.1, and –10.9 A. The phase A currents are shown in Fig. 27(a).
high-frequency magnetic components were constructed of fer- A 75 Ω resistor was added to the dc capacitor I B1, which
rite and Litz wire. The three coils of HFT were equal in position. brings a 133 W power consumption rise on that capacitor. This
The external inductors were 65 μH each. The PET parameters action causes power interrupt to inner Cell B1, also causes
and magnetic component design are shown in Tables II and III power unequal among three phases, and power unequal within
individually. the cells of phase B. The A-phase currents in new steady state
The system efficiency under BTB operation is ࣈ75%, the and the d-axis currents during the transient process are shown in
CHB stage loss is ࣈ7.5%, meanwhile the MAB stage loss is Fig. 27. To compensate for the increased power consumption,
ࣈ10%, which is a reasonable efficiency at this low-voltage level. the power-in terminal has to increase the input power, mean-
while the power-out terminal has to decrease the output power.
Theoretically, a 133 W power consumption rise will cause a
VI. EXPERIMENT RESULTS 0.54 A of  i∗d rise considering the system efficiency. The ac-
Two sets of experiment are done to verify the proposed tual  id is 0.57 A according to Fig. 27(a) and (b).
CHB×3+MAB strategy itself. 6.1: module power disturbance: Fig. 28(a) shows the in-cell dc voltage balancing process,
an additional power loss is added to one of the modules to check which is balanced after a dynamic process. Fig. 28(b) and (c)
the effectiveness of the CHB×3+MAB strategy; 6.2: unsched- shows the MAB ac voltages and its magnetizing voltage. Be-
uled terminal loss: terminal II is cut from the system to check fore the module power disturbance, the phase shift angle is
the robustness of the proposed control strategy. δI –I I I = 7.96 μs and δI I –I I I = 9.7 μs; in the new steady state,

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
3692 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 67, NO. 5, MAY 2020

Fig. 28. (a) In-cell dc voltage balancing. (b) MAB ac voltages before
disturbance. (c) MAB ac voltages under new steady state.

Fig. 31. (a) iA of three terminals. (b) i∗d of three terminals.

Fig. 32. (a) In-cell dc voltage balancing. (b) MAB ac voltages in new
Fig. 29. (a) Among-phase dc voltage balancing. (b) Zero sequence steady state. (c) MAB ac currents in new steady state.
voltage of all terminals.

Fig. 33. (a) Among-phase dc voltage balancing. (b) Zero sequence


Fig. 30. (a) In-phase dc voltage balancing. (b) Power trimming coeffi-
voltage of all terminals.
cients of phase B.

the phase shift angle is δI –I I I = 6.8 μs and δI I –I I I = 9.36 μs strategy reconfiguration, which is the same with the simula-
which indicates higher power delivered to terminal I. tion results and verifies the robustness of the proposed control
Fig. 29 shows the among-phase dc voltage balancing. strategy.
Fig. 29(a) shows the three phase dc voltage of terminal I, which Fig. 32(a) shows the in-cell dc voltage balancing process. The
achieves balancing after a dynamic process. Fig. 29(b) shows dc voltage corresponding to the terminal II has a voltage drop
the injected zero sequence voltage. As the power interrupt on due to the input power loss, but achieves balancing again after a
B1 brought additional power unbalance among three phases, the dynamic process. The MAB voltage and current of new steady
required zero-sequence voltage becomes larger. state is shown in Fig. 32(b) and (c). It is seen that the HFAC
Fig. 30 shows the in-phase dc voltage balancing. Fig. 30(a) voltage of terminal II becomes in phase with the magnetizing
shows the Vdc I B 1 and Vdc I B 2 , Fig. 30(b) shows the in- voltage meanwhile the HF current changes to zero.
phase power trimming coefficient. As the power interrupt on Fig. 33 shows the among-phase dc voltage balancing.
B1 brought power unbalance within the phase B, a relatively Fig. 33(a) shows the three phase dc voltage of terminal I.
big power trimming coefficient is needed. Fig. 33(b) shows the injected zero sequence voltage. In the
Figs. 27–30 verify the effectiveness of the proposed dc voltage new steady state, the zero sequence voltage amplitude is in-
control strategy to handle the cell power interrupt. creased compared the to normal case, which is because the zero
sequence voltage in terminal II losses its effect in the among-
B. Unscheduled Terminal Loss
phase voltage balancing, thus the original zero sequence power
The terminal II is cut from the source to test the robustness in terminal II has to be taken by the other two terminals so their
of the proposed control strategy. With a power-in terminal loss, zero-sequence voltage has to increase.
the input power of terminal I is increased, meanwhile the output Fig. 34 shows the in-phase dc voltage balancing. Fig. 34(a)
power of terminal III is decreased and finally achieves a new shows the Vdc I B 1 and Vdc I B 2 , Fig. 34(b) shows the in-phase
steady state, which is shown in Fig. 31(a) and (b). This is au- power trimming coefficient. It is seen that after a dynamic
tomatically achieved by the overall dc balancing control loop process, the in-phase dc voltage balancing is achieved. The
in Figs. 18 and 21 without using any fault detection or control trimming coefficient is also small as before.

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
OUYANG et al.: DC VOLTAGE CONTROL STRATEGY OF THREE-TERMINAL MEDIUM-VOLTAGE PET-BASED SNOPS 3693

Fig. 34. (a) In-phase dc voltage balancing. (b) Power trimming coeffi-
cients of phase B.

Fig. 36. (a) In-cell dc voltage. (b) Phase A current. (c) d-axis reference
current. (d) d-axis current under terminal II loss of CHB+MAB strategy.

Fig. 35. (a) Modulation wave and trimming coefficient of CHB I A of


CHB+CHB+CHB strategy. (b) Modulation wave of CHB I A and trim- and the results in Section VI-B, the robustness advantage of
ming coefficient of phase A of CHB×3+MAB strategy. CHB×3+MAB strategy is verified.
The comparison results from experiment are the same as those
Figs. 31–34 verify the robustness of the proposed in the simulation section.
CHB×3+MAB strategy. The system can operate under an un-
scheduled terminal lose without modifying the control strategy. VII. CONCLUSION

C. Comparison With the Other Two Strategies In this paper, a PET-based three-terminal SNOP topology was
proposed. It consisted of CHB converters and MAB converters
In order to verify the power sharing advantage of to interface three MV feeders meanwhile achieve isolation.
CHB×3+MAB over the CHB+CHB+CHB strategy, the in- The dc voltage control strategy for the proposed PET-SNOP
ductors of one MAB in phase A are adjusted by adding addi- was studied. Two strategies were derived from the existing
tional small inductors, thus the PET cells of phase A would strategies: CHB+CHB+CHB strategy, which suffers from un-
transfer different amount of active power under the same phase balanced cell power; and CHB+MAB strategy, which suffers
shift angle when apply the CHB+CHB+CHB strategy. As a from the asymmetric structure. In order to overcome these draw-
comparison, the phase shift angles of CHB×3+MAB strategy backs, a novel strategy: CHB×3+MAB strategy, was proposed
is derived by voltage closed-loop, thus the MAB inductor mis- in this paper. The CHB×3+MAB strategy can achieve cell
match would not cause power mismatch between PET cells of power balancing as well as having a symmetric control struc-
phase A. The steady-state results are shown in Fig. 35, it shows ture, which avoids the drawbacks, meanwhile combines the ad-
that the modules are transferring different amount of power us- vantages of the two strategies. Moreover, the CHB×3+MAB
ing the CHB+CHB+CHB strategy as the modulation waves are strategy can increase the flexibility and robustness of the dc
different in amplitude, meanwhile transferring a same amount of voltage control.
power using CHB×3+MAB strategy as the modulation waves The effectiveness of the proposed PET-SNOP topology as
are overlapped. The power sharing advantage of CHB×3+MAB well as the proposed CHB×3+MAB dc voltage control strategy
strategy over the CHB+CHB+CHB strategy is verified. was verified by both simulation and experiment results.
In order to verify the robust advantage of CHB×3+MAB over The proposed circuit topology as well as the dc voltage control
the CHB+MAB strategy, the system is running in CHB+MAB strategy can be extended to applications with more terminals.
strategy with CHB II assigned as the special CHB which con-
trols the dc voltage. When CHB II is cut from the system cre-
ating the same scenario in Section VI-B, the response is shown REFERENCES
in Fig. 36. As the terminals I and III do not have the dc voltage [1] J. M. Bloemink and T. C. Green, “Increasing distributed generation pene-
control, they still attempt to maintain the scheduled current (ref- tration using soft normally-open points,” in Proc. IEEE PES Gen. Meeting,
2010, pp. 1–8.
erence current do not change), thus the whole system is running [2] P. Li et al., “Coordinated control method of voltage and reactive power
with negative active power. Terminal II dc voltage drops zero; for active distribution networks based on soft open point,” IEEE Trans.
terminal I and III dc voltage drops to a clamping voltage and the Sustain. Energy, vol. 8, no. 4, pp. 1430–1442, Oct. 2017.
[3] J. M. Bloemink and T. C. Green, “Effects of power electronic compen-
output currents are distorted. From the view of the whole sys- sation on distribution network thermal and voltage violations,” in Proc.
tem, it is definitely running out of control. Comparing Fig. 36 IEEE Power Energy Soc. Gen. Meeting, 2013, pp. 1–5.

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
3694 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 67, NO. 5, MAY 2020

[4] J. M. Bloemink and T. C. Green, “Increasing photovoltaic penetration with [27] C. Gu, Z. Zheng, and Y. Li, “A power electronic transformer (PET)
local energy storage and soft normally-open points,” in Proc. IEEE Power with multiport bidirectional resonant DC-DC converters for electric trac-
Energy Soc. Gen. Meeting, 2011, pp. 1–8. tion applications,” in Proc. IEEE Transp. Electrific. Conf. Expo, 2015,
[5] P. Li et al., “Optimal operation of soft open points in active distribution pp. 1–6.
networks under three-phase unbalanced conditions,” IEEE Trans. Smart [28] Z. Zheng, Z. Gao, C. Gu, L. Xu, K. Wang, and Y. Li, “Stability and voltage
Grid, vol. 10, no. 1, pp. 380–391, Jan. 2019. balance control of a modular converter with multiwinding high-frequency
[6] A. Aithal et al., “Impact of soft open point on feeder automation,” in transformer,” IEEE Trans. Power Electron., vol. 29, no. 8, pp. 4183–4194,
Proc. IEEE Int. Energy Conf., 2016, pp. 1–6. Aug. 2014.
[7] J. Feng et al., “Power electronic transformer-based railway traction sys- [29] Y. Yu et al., “Power flow control of a triple active bridge DC-DC con-
tems: Challenges and opportunities,” IEEE J. Emerg. Sel. Topics Power verter using GaN power devices for a low-voltage DC power distribution
Electron., vol. 5, no. 3, pp. 1237–1253, Sep. 2017. system[C]//,” in Proc. Future Energy Electron. Conf. ECCE Asia, 2017,
[8] X. She, A. Q. Huang, and R. Burgos, “Review of solid-state trans- pp. 772–777.
former technologies and their application in power distribution systems,” [30] S. Nakagawa, J. Arai, R. Kasashima, K. Nishimoto, Y. Kado, and K. Wada,
IEEE J. Emerg. Sel. Topics Power Electron., vol. 1, no. 3, pp. 186–198, “Dynamic performance of triple-active bridge converter rated at 400 V,
Sep. 2013. 10 kW, and 20 kHz,” in Proc. IEEE 3rd Int. Future Energy Electron. Conf.
[9] S. Falcones, R. Ayyanar, and X. Mao, “A DC–DC multiport-converter- ECCE Asia, 2017, pp. 1090–1094.
based solid-state transformer integrating distributed generation and stor-
age,” IEEE Trans. Power Electron., vol. 28, no. 5, pp. 2192–2203,
May 2013. Shaodi Ouyang (S’14–M’19) was born in Hubei,
[10] C. Zhao et al., “Power electronic traction transformer—Medium voltage China. He received the B.S. degree from the
prototype,” IEEE Trans. Ind. Electron., vol. 61, no. 7, pp. 3257–3268, Huazhong University of Science and Technol-
Jul. 2014. ogy, Wuhan, China, in 2011, and the Ph.D. de-
[11] X. She, X. Yu, F. Wang, and A. Q. Huang, “Design and Demonstration gree from Xi’an Jiaotong University, Xi’an, China,
of a 3.6-kV–120-V/10-kVA solid-state transformer for smart grid appli- in 2018, both in electrical engineering.
cation,” IEEE Trans. Power Electron., vol. 29, no. 8, pp. 3982–3996, In 2019, he joined Xi’an Jiaotong Uni-
Aug. 2014. versity, as an Assistant Research Associate.
[12] S. Bifaretti et al., “Advanced power electronic conversion and control His research interests include power electronic
system for universal and flexible power management,” IEEE Trans. Smart transformer, multilevel converters, and dc–dc
Grid, vol. 2, no. 2, pp. 231–243, Jun. 2011. converters.
[13] S. Pipolo et al., “Reactive power control strategies for UNIFLEX-PM
Converter,” in Proc. 42nd Annu. Conf. IEEE Ind. Electron. Soc., 2016,
pp. 3570–3575. Jinjun Liu (M’97–SM’10–F’19) received the
[14] D. Wang et al., “A 10-kV/400-V 500-kVA electronic power trans- B.S. and Ph.D. degrees in electrical engineer-
former,” IEEE Trans. Ind. Electron., vol. 63, no. 11, pp. 6653–6663, ing from Xi’an Jiaotong University (XJTU), Xi’an,
Nov. 2016. China, in 1992 and 1997, respectively.
[15] H. Akagi and R. Kitada, “Control and design of a modular multilevel cas- He then joined the XJTU Electrical Engineer-
cade BTB system using bidirectional isolated DC/DC converters,” IEEE ing School as a faculty. From late 1999 to early
Trans. Power Electron., vol. 26, no. 9, pp. 2457–2464, Sep. 2011. 2002, he was with the Center for Power Elec-
[16] B. Zhao, Q. Song, J. Li, and W. Liu, “A modular multilevel DC-link tronics Systems, Virginia Polytechnic Institute
front-to-front DC solid-state transformer based on high-frequency dual and State University, Blacksburg, VA, USA, as
active phase shift for HVDC grid integration,” IEEE Trans. Ind. Electron., a Visiting Scholar. In late 2002, he was pro-
vol. 64, no. 11, pp. 8919–8927, Nov. 2017. moted to a Full Professor and then the Head
[17] L. Tarisciotti et al., “Multiobjective modulated model predictive control of the Power Electronics and Renewable Energy Center, XJTU, which
for a multilevel solid-state transformer,” IEEE Trans. Ind. Appl., vol. 51, now comprises 17 Faculty Members and more than 100 graduate stu-
no. 5, pp. 4051–4060, Sep./Oct. 2015. dents and carries one of the leading power electronics programs in
[18] S. Bifaretti et al., “Power flow control through a multi-level H-bridge China. From 2005 to early 2010, he served as an Associate Dean of
based power converter for universal and flexible power management in Electrical Engineering School, XJTU, and from 2009 to early 2015,
future electrical grids,” in Proc. 13th Int. Power Electron. Motion Control the Dean for Undergraduate Education with XJTU. He is currently
Conf., 2008, pp. 1771–1778. an XJTU Distinguished Professor of Power Electronics, sponsored
[19] S. Ouyang et al., “A single phase power electronic transformer considering by Chang Jiang Scholars Program of Chinese Ministry of Education.
harmonic compensation in Scott traction system,” in Proc. 9th Int. Conf. He has coauthored three books (including one textbook), published
Power Electron. ECCE Asia, 2015, pp. 2620–2627. more than 400 technical papers in peer-reviewed journals and con-
[20] X. Wang et al., “Control and experiment of an H-bridge-based three- ference proceedings, holds nearly 50 invention patents (China/U.S.),
phase three-stage modular power electronic transformer,” IEEE Trans. and delivered for many times plenary keynote speeches and tuto-
Power Electron., vol. 31, no. 3, pp. 2002–2011, Mar. 2015. rials at IEEE conferences or China national conferences in power
[21] L. Maharjan et al., “State-of-charge (SOC)-balancing control of a battery electronics area. His research interests include power quality control
energy storage system based on a cascade PWM converter,” IEEE Trans. and utility applications of power electronics, microgrids for sustain-
Power Electron., vol. 24, no. 6, pp. 1628–1636, Jun. 2009. able energy and distributed generation, and more/all electronic power
[22] L. Wang, D. Zhang, Y. Wang, B. Wu, and H. S. Athab, “Power and systems.
voltage balance control of a novel three-phase solid-state transformer Dr. Liu was the recipient of governmental awards for eight times at na-
using multilevel cascaded H-bridge inverters for microgrid applications,” tional level or provincial/ministerial level for scientific research/teaching
IEEE Trans. Power Electron., vol. 31, no. 4, pp. 3289–3301, Apr. 2016. achievements, Delta Scholar Award in 2006, Chang Jiang Scholar Award
[23] J. Tian et al., “Analysis and control of electronic power transformer with and Outstanding Sci-Tech Worker of the Nation Award in 2014, and
star-configuration under unbalanced conditions,” IET Elect. Power Appl., IEEE Transactions on Power Electronics Prize Paper Award in 2016.
vol. 9, no. 5, pp. 358–369, 2015. He served as the IEEE Power Electronics Society Region ten Liaison
[24] A. Sankala et al., “Modular double-cascade converter for high-power and then China Liaison for ten years, an Associate Editor for the IEEE
medium-voltage drives,” IET Power Electron., vol. 8, no. 9, pp. 1661– TRANSACTIONS ON POWER ELECTRONICS for 12 years, and from starting
1669, 2015. 2015, the Vice President for membership of IEEE PELS. He is on the
[25] C. Gu, Z. Zheng, L. Xu, K. Wang, and Y. Li, “Modeling and control Board of China Electrotechnical Society (CES) and was elected the Vice
of a multiport power electronic transformer (PET) for electric traction President of the CES Power Electronics Society in 2013. Since 2013,
applications,” IEEE Trans. Power Electron., vol. 31, no. 2, pp. 915–927, he has been the Vice President for International Affairs, China Power
Feb. 2016. Supply Society (CPSS) and since 2016, the inaugural Editor-in-Chief of
[26] T. Zhao et al., “Voltage and power balance control for a cascaded multi- CPSS Transactions on Power Electronics and Applications. Since 2013,
level solid state transformer,” in Proc. Appl. Power Electron. Conf. Expo., he has been serving as the Vice Chair of the Chinese National Steering
2010, pp. 761–767. Committee for College Electric Power Engineering Programs.

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.
OUYANG et al.: DC VOLTAGE CONTROL STRATEGY OF THREE-TERMINAL MEDIUM-VOLTAGE PET-BASED SNOPS 3695

Yue Yang was born in Shandong, China, in Shuguang Song (S’17) was born in Shandong,
1994. He received the B.S. degree in electri- China. He received the B.S. degree in elec-
cal engineering in 2017 from Xi’an Jiaotong trical engineering in 2014 from Xi’an Jiaotong
University, Xi’an, China, where he is currently University, Xi’an, China, where he is currently
working toward the Ph.D. degree in electrical working toward the Ph.D. degree in electrical
engineering. engineering.
His research interests include power His research interests include modular multi-
electronic transformer and reliability of power level converters, power quality, high voltage di-
electronic systems. rect current, and motor drive.

Xingxing Chen (S’17) was born in Zhe- Hongda Wu was born in Fujian, China, in 1994.
jiang, China. He received the B.S. degree in He received the B.S. degree in electrical engi-
electronic and information engineering from neering in 2016 from Xi’an Jiaotong University,
Southwest Jiaotong University, Chengdu, China, Xi’an, China, where he is currently working to-
in 2015. He is currently working toward the Ph.D. ward the M.S. degree in electrical engineering.
degree in electrical engineering with Xi’an Jiao- His research interests include modular multi-
tong University, Xi’an, China. level converter and space vector modulation.
His research interests include high voltage di-
rect current and modular multilevel converters.

Authorized licensed use limited to: Cardiff University. Downloaded on April 26,2023 at 10:07:29 UTC from IEEE Xplore. Restrictions apply.

You might also like