Engel 2015
Engel 2015
Engel 2015
fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
1
Comparison of the
Modular Multilevel DC Converter and the
Dual-Active Bridge Converter for Power Conversion
in HVDC and MVDC Grids
Stefan P. Engel, Student Member, IEEE, Marco Stieneker, Student Member, IEEE,
Nils Soltau, Student Member, IEEE, Sedigheh Rabiee, Student Member, IEEE, Hanno Stagge, Member, IEEE,
and Rik W. De Doncker, Fellow, IEEE
(Special Issue on Modular Multilevel Converters, 2014)
Abstract—It is expected that in the near future the use of high- connections but also as multi-terminal dc networks [2]. This
voltage dc (HVDC) transmission and medium-voltage dc (MVDC) requires efficient dc-dc converters to control the power flows
distribution technology will expand. This development is driven in these networks.
by the growing share of electrical power generation by renewable
energy sources that are located far from load centers and the Additionally, the amount of renewable energy is still in-
increased use of distributed power generators in the distribution creasing steadily [3]. It has been shown, that large renewable
grid. Power converters that transfer the electric energy between energy sources such as windfarms [4], [5] and photovoltaic
voltage levels and control the power flow in dc grids will be key power plants [6], [7] operate more efficiently when they are
components in these systems. The recently presented Modular connected to a medium-voltage dc (MVDC) collector grid
Multilevel DC Converter (M2DC) and the three-phase Dual-
Active Bridge converter (DAB) are benchmarked for this task. instead of an ac collector grid. With the increasing amount
Three scenarios are examined: a 15-MW converter for power of renewables, power fluctuation in the ac grid is becoming
conversion from an HVDC grid to an MVDC grid of a university an issue as well [8]. Therefore, storage systems are essential
campus, a gigawatt converter for feeding the energy from an for leveling out the volatility of the supply [9], [10]. These
MVDC collector grid of a wind farm into the HVDC grid and storage systems, e.g. battery energy storage systems [11], [12],
a converter that acts as a power controller between two HVDC
grids with the same nominal voltage level. electrolyzers for power-to-gas technology [13] or pumped-
The operation and degrees of freedom of the M2DC are hydro storages [14] are medium-voltage dc sources or ac
investigated in detail aiming for an optimal design of this sources with variable frequency. Utilizing HVDC links, all
converter. The M2DC and the DAB converter are thoroughly these renewable energy sources and storages can be integrated
compared for the given scenarios in terms of efficiency, amount efficiently using a dc conversion from MVDC to HVDC.
of semiconductor devices and expense on capacitive storage and
magnetic components. Moreover, large medium-voltage industrial drives [15]–[17]
can be energized more cost effectively through an HVDC
Index Terms—Modular multilevel dc converter (M2DC), dual- link. This requires dc-dc converters to efficiently transform
active bridge (DAB) converter, dc-dc converter, HVDC, MVDC,
dc collector grid, costs. the electric power from the high-voltage to the the medium-
voltage level.
In such HVDC applications it is pursued to build power-
I. I NTRODUCTION electronic converters in a modular manner. A modular con-
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
2
SM SM
SM SM
SM SM
(a) (b)
Fig. 2. Submodules (SM) of Modular Multilevel Converters: (a) unipolar
Fig. 1. Unipolar push-pull M2DC [18] submodule, (b) bipolar submodule
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
3
B. Number of Submodules
Fig. 3. Bipolar push-pull M2DC A sufficiently high number of submodules Nsm,a are needed
in each converter arm a to generate the required arm voltage va
ranging from a minimum of Va,min to a maximum of Va,max .
By combination of two unipolar M2DC (Fig. 1), a bipolar In the following, subscript a ∈ {n1, p1, n2, p2, n, p}, while
version as depicted in Fig. 3 can be composed. n1, p1, n2, p2 refer to specific arms, whereas n and p refer to
any lower and upper arm, respectively.
A. Fundamental Equations Equations (4) to (6) yield the minimum and maximum
For a basic understanding of the push-pull M2DC converter, voltages of the lower and upper arms
the unipolar version (Fig. 1) is investigated. Ignoring losses, Vn,min = V2,min − V̂sec , Vn,max = V2,max + V̂sec , (10)
for given primary and secondary voltages V1 and V2 and Vp,min = V1,min − Vn,max , Vp,max = V1,max − Vn,min ,
nominal power PN , the primary and secondary current is (11)
PN PN
I1 = , I2 = . (1) where V1,min , V1,max , V2,min and V2,max are the specified
V1 V2 limits of the primary and the secondary voltage.
In steady-state operation, the average dc component of the It is assumed in the following that |Va,max | > |Va,min |.
voltages of the left and the right branch of the converter vp1 + Equations (10) and (11) show that this is fulfilled if V2,max >
vn1 and vp2 +vn2 , respectively, equals the primary-side voltage −V2,min and V1,max − V2,min > V2,max − V1,min , which is the
V1 . The same applies for the voltage of the lower arms, where case for the given scenarios.
the dc component is equal to the secondary-side voltage V2 . The number of submodules Nsm,a of a converter arm a can
Using power balancing principles, it is clear that the con- be estimated by the nominal voltage VC,0 of the submodule
verter has to carry the total input power to the output thereby capacitors and the maximum arm voltage Va,max that has to
realizing a voltage drop V1 − V2 . This is accomplished by be generated during a cycle of the secondary power loop:
a so-called secondary power loop that uses frequencies other
Va,max
than 0 Hz to transport energy [18]. The power that has to be Nsm,a = · ks , (12)
VC,0
transferred by this secondary power loop is
where ks is an additional safety factor in the range of 110 %
Psec = I1 · (V1 − V2 ). (2) to 120 % as a buffer for the control and for the energy
Assuming a power loop with sinusoidal current and voltage variations in the capacitors during operation of the converter.
at frequency fsec , the peak value of the current for a given If redundancy is required, the number of submodules increases
voltage of the secondary power loop V̂sec is further.
Bidirectional modules are needed only if a negative arm
Psec
Iˆsec = . (3) voltage is required. The number of required bidirectional
V̂sec · cos ϕsec submodules is
(l V m
ϕsec is the phase shift between voltage and current of the − Va,min · k if Va,min < 0
s
secondary power loop. The lowest current is achieved for Nsm,bi,a = C,0
(13)
ϕsec = 0. 0 else.
The current of the secondary power loop is controlled Consequently, the number of unidirectional submodules is
by the voltage across the common-mode inductance of the
primary-side coupled inductors L1 . This inductance is chosen Nsm,uni,a = Nsm,a − Nsm,bi,a . (14)
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
4
C. Submodule Capacitance by weighting the conduction losses with the number of devices
in on-state:
The required capacitance in each submodule can be calcu-
lated to ZT
1 1
P cond,S1,a = · vth (ia− (t))
∆Ea NHB,a T
Csm,a = 2 · ∆v , (15) 0
Nsm,a VC,0 rel
· ia− (t) · N S1,on,a (t) dt, (25)
where Nsm,a is the number of cells in the according arm, ZT
VC,0 is the nominal cell voltage, ∆vrel is the specified relative 1 1
P cond,D1,a = · vf (ia+ (t))
voltage ripple at the capacitors and ∆Ea is the energy variation NHB,a T
0
over one cycle of the secondary power loop.
· ia+ (t) · N S1,on,a (t) dt. (26)
This equation is derived from the minimum and maximum
energy in the capacitors ∆Ea,min and ∆Ea,max and the corre- where vth (i) represents the on-state characteristic of the switch
sponding minimum and maximum capacitor voltages VC,min and vf (i) the forward voltage of the diode at the current i, with
and VC,max : (
ia if ia > 0
ia+ = , (27)
∆Ea = ∆Ea,max − ∆Ea,min (16) 0 else
Csm,a 2 2
(
−ia if ia < 0
= Nsm,a · VC,max − VC,min (17)
2 ia− = . (28)
∆vrel
0 else
with VC,max = VC,0 · 1 + (18)
2 For the estimation of the switching losses, the characteristics
∆vrel from the data sheets for the turn-on and turn-off energies
and VC,min = VC,0 · 1 − . (19) ES,on (v, i) and ES,off (v, i), respectively, and the reverse re-
2
covery energy of the diode ED,rr (v, i) for a given device
The energy variation ∆Ea is current i and voltage v is used. The voltage ripple of the
capacitors is neglected, that is a constant capacitor voltage
∆Ea = max (Ea (t)) − min (Ea (t)) (20) VC = VC,0 is assumed, resulting in
t∈[t0 ,t0 +T [ t∈[t0 ,t0 +T [
1 ZT
with T = , (21) f sw,a
fsec P sw,S1,a = ES,on (VC,0 , ia− (t))
T
where 0
tZ0 +t + ES,off (VC,0 , ia− (t)) dt, (29)
Ea (t) = (va (t) · ia (t)) dt + Ea,0 (22) ZT
f sw,a
t0 and P sw,D1,a = ED,rr (VC,0 , ia+ (t)) dt. (30)
T
0
for a given arm voltage va (t), arm current ia (t) and initial
energy Ea,0 . f sw,a is the average switching frequency of the semiconduc-
In the following, ∆vrel is set to 20 %, corresponding to a tors. If the arm is operated in fundamental switching frequency
voltage ripple of ±10 %. mode (i.e. the output waveform of the arms is a stepwise sine),
the average switching frequency of a half bridge is calculated
to
D. Semiconductors 2V̂sec
f sw,a = ksw · fsec · (31)
The conduction losses are estimated separately for each arm NHB,a VC,0
using the characteristics given in the data sheets. where ksw is an additional safety factor that accounts for addi-
For a given arm voltage and number of submodules, the tional switching action necessary for balancing of the capacitor
average number of switches S1 and S2 in on-state at a given voltages. ksw is set to 1.5 for the following investigations.
point in time t is The losses of S2 and D2 can be calculated accordingly.
va (t)
N S1,on,a (t) = + Nsm,bi,a (23)
VC,0 E. Inductors
and N S2,on,a (t) = NHB,a − N S1,on,a (t), (24) Two coupled inductors are used in the M2DC topology; one
on the primary side L1 and one on the secondary side L2 . In
respectively, where NHB,a = 2Nsm,bi,a + Nsm,uni,a is the the following, the design of these inductors is discussed and
number of half-bridges in the arm. the losses are estimated.
Assuming a homogeneous distribution of the losses among The primary-side inductor can be represented as a trans-
all half bridges of a converter arm, the average conduction former with its well-known equivalent circuit diagram con-
losses in a single switch and diode of this arm are calculated sisting of the magnetizing inductance L1,m and the leakage
0885-8993 (c) 2013 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.
\vsec \frac{\
i_{\rm
This article has been accepted for publication in a future issue sec} but has not been
of this journal, fully edited. Content may change prior to final publication. Citation information: DOI
\LhsII
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics -\vsec
\hat{=} 5
\Ll
inductances L1,σ1 and L1,σ2 (cf. Fig. 4). The primary cur-
rent I1 has to flow through the magnetizing inductance and, Fig. 5. Push-pull M2DC secondary-side inductor
additionally, a considerably high effective inductance for the
currents ip1 and ip2 , which is the leakage inductance of the
transformer, is required. Therefore, it is advantageous to split According to (3), for a given Psec , a small voltage in
the coupled inductors into three inductors L1,m , L1,σ1 and the secondary power loop results in high Iˆsec . Therefore, a
L1,σ2 with equivalent behavior as depicted in Fig. 4. The high V̂sec is favorable concerning the conduction losses in the
design of such inductors is straight forward [43]. semiconductors of the submodules. Furthermore, the size of
The secondary-side coupled inductor L2 is operated similar capacitive storage decreases with decreasing Iˆsec .
to an interphase transformer [44]. As illustrated in Fig. 5, Using solely unidirectional modules, i.e. Nsm,bi,a = 0 and,
the orientation of one winding is inverted. However, as the consequently, Va,min ≥ 0, it can be verified by (10) and (11)
primary and secondary phase voltages are 180° phase shifted, that V̂sec is limited by V̂sec < V2 and V̂sec < (V1 − V2 ).
the voltages imposed on the magnetic core are the same A higher V̂sec can be gained by using bidirectional mod-
as for an ordinary transformer. These voltages generate the ules. However, the use of bidirectional modules results in a
magnetization current im2 . An air gap in the transformer higher number of semiconductors. Additionally, the number
is not needed, as the magnetic fluxes generated by the dc of switches and diodes in the current path that produce losses
components of the transformer currents cancel out. Therefore increases. Consequently, increasing V̂sec above a certain value
the design of L2 can be conducted analogously to a standard reduces the efficiency of the converter.
transformer [43]. fsec directly influences the size of the capacitors, magnetic
The value of L1,σ1 = L1,σ2 is selected in a way that the components and the losses of the semiconductors. From (15) to
voltage required to control the secondary current is less than (22), it can be derived that the required capacitance is inversely
5 %. Successively, L1,m is dimensioned such that the primary proportional to fsec . Furthermore, Ap is approximately in-
current ripple is below 5 % of the maximum primary current. versely proportional to fsec [43]. In these terms, it is favorable
To benchmark the performance of the M2DC, the copper to choose a high fsec . However, for a sufficiently high V̂sec
and core losses of the inductors are calculated for the given many or even all submodules are required to switch during one
scenarios. The Generalized Steinmetz Equations (GSE) are period of the secondary power loop frequency, resulting in an
used to calculate the core losses assuming a silicon steel increased average switching frequency (cf. Equation (31)) and
material with a thickness of 0.18 mm [45]. thus increased switching losses. Therefore, for a given design,
Beside the converter losses, the size of the passive compo- the switching losses are proportional to fsec .
nents is evaluated. For the magnetic devices, the area product
Ap is used as a figure of merit to compare the costs of IV. D ESIGN OF THE M ODULAR M ULTILEVEL DC
the magnetic components. Ap is defined as product of the C ONVERTER S YSTEMS
winding window area Aw and the cross section area Ac of For the given scenarios, the design of an appropriate Modu-
the magnetic core. It relates to the apparent power of the lar Multilevel DC Converter is carried out. The characteristics
magnetic components, i.e. to the size of a magnetic core and of the converters for all scenarios are summarized in Table II.
the size of the copper windings. As it is an important value for
the design of transformers and inductors, it is usually given Scenario A: Linkage between MVDC and HVDC Grids
in the data sheets of magnetic cores. To illustrate the area
product, a design of a conventional 50-Hz transformer with This scenario exhibits a high conversion ratio at low power.
an apparent power of 630 kVA has been carried out. For the HiPak 5SNA 0650J450300 IGBTs of ABB with a rating of
chosen example, the design leads to: 4500 V and 650 A are used for the upper arms, while StakPak
5SNA 1300K450300 IGBTs of ABB [46] rated for 4500 V and
Ap = Aw · Ac = 0.15 m2 · 0.05 m2 = 0.0075 m4 (32) 1300 A are employed in the lower arm. Due to the low current,
a parallel connection of IGBTs is not necessary. Hence, the
number of parallel semiconductors Npar,p and Npar,n in the
F. Design Optimization upper and lower arm, respectively, is equal to one.
The rating of the secondary power loop, namely the choice V̂sec has been set to 10 kV. Increasing V̂sec results in
of V̂sec and fsec , significantly influences the design and the higher switching losses and increased losses in the magnetics,
performance of the converter. whereas a lower V̂sec would result in a higher secondary power
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
6
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
7
primary DC AC
transformer voltages
0 AC DC
DC AC
AC DC
secondary
DC AC
AC DC
currents
phase
0
DC AC
AC DC
currents
DC AC
dc
AC DC
0
DC AC
angle in rad AC DC
A. Principle
loop can be implemented to keep the primary or secondary
Both bridges modulate a six-step voltage at the transformer
voltage constant [53].
terminals. The phase shift angle ϕ between the primary
and secondary transformer voltage vp,AC,rms and vs,AC,rms
results in a voltage difference across the leakage inductance C. Design of the Dual-Active Bridge Converter
of the transformer and hence in a transferred power PDAB .
According to [42], it is The DAB converter as the smallest unit of the DAB con-
verter system shown in Fig. 8 uses two series-connected 4.5 kV
Vp2 2 ϕ
· dϕ · ( −
PDAB = ), (33) semiconductor devices with a 100-FIT voltage of 2.6 kV to
ωLσ 3 2π gain the required blocking capability of 5 kV. The reactance
where Vp gives the primary dc voltage, Vs the secondary of the total leakage inductance Lσ = Lσ,p + Lσ,s limits the
dc voltage, ω the angular frequency of the ac link, Lσ the maximum power transfer. The maximum leakage inductance
total leakage inductance of the transformer and d = Vs/Vp
the voltage ratio. The waveforms of voltage and current are 1 Vp Vs
Lσ,max = (34)
exemplary shown in Fig. 7, where ϕ between the input and 12 fac PDAB
output voltage of the first phase is indicated.
can be derived by inserting ϕmax and the rated power PDAB
Since the ac voltage is modulated by operating the power
into (33). When the DAB converter is required to operate with
converters with fundamental frequency switching patterns, the
varying input and output voltages, Lσ should be optimized.
converter can be designed such that all devices are operating
The voltage imbalances (d 6= 1) increase the apparent power
in softswitching mode. Hence, the switching losses PDAB,sw
in the transformer. The calculation algorithms presented in [48]
are lower than the losses in hardswitching converters that
estimates the optimum leakage inductance Lσ,opt that leads to
use pulse-width modulation strategies such as the M2DC.
minimimum transformer current in all operation points.
Furthermore, the transformer’s core volume strongly depends
on the frequency of the applied voltage; a medium-frequency Hence, conduction losses of the semiconductors and of the
fac leads to a significant reduction of the transformer size and transformer windings are reduced leading to an improvement
weight in comparison with 50-Hz transformer in the same of the efficiency.
power range and with the decreased core size also the core The minimum required capacitance to limit ∆V operating at
losses decrease leading to a higher transformer efficiency. nominal power (ϕ = ϕnom ) is estimated with
Previous studies have shown that a 1000-Hz transformer, 2
Vp 2 ϕnom
designed for same reliability as standard 50-Hz transformers, C= (2 − d) 3dϕ nom ( − ) − ϕ nom .
has a tenfold increase of power density [45], [48]. 6 ω 2 Lσ ∆V 3 2π
(35)
Equation (35) takes the electric charge causing a voltage ripple
B. Control of the Dual-Active Bridge Converter ∆V across the capacitors of the DAB converter into account.
Fast control of the power PDAB is based on the instanta- The electric charge can be evaluated by integrating the dc
neous current control presented in [49]. Also, an outer control- current ripple plotted in Fig. 7.
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
8
VI. D ESIGN OF THE D UAL -ACTIVE B RIDGE C ONVERTER series to achieve the high transmission line voltage. Since the
S YSTEMS secondary-side voltage (V2 = ±10 kV) requires four bridges
The Dual-Active Bridge converter system is composed of in series connection, fifteen of these series-connected bridges
series and parallel connected DAB converters (cf. Fig. 8). are connected in parallel. The assumed wind farm power of
The modular design of each single DAB converter allows 1.21 GW generates a current I2 = 60.5 kA in the MVDC grid.
the easy adaption of power and voltage rating of the system. The HVDC transmission line is loaded with I1 = 4.03 kA.
If redundancy is required, additional DAB converter can be However, the semiconductors need a current capability of
connected to the system. Note that the amount of primary ip,AC,rms = is,AC,rms = 2.85 kA. The characteristics of ABB
bridges has to be the same as the amount of secondary bridges, 5SHY 55L4500 asymmetrical IGCT and ABB 5SDD 11D2800
because only the application of two-windings transformers diodes [46] are applied for loss calculations. Furthermore,
is assumed within the scope of this work. The transformer the influence of turn-off resonant snubbers are taken into
designs are summarized in Table III. In the following, different account for the (soft-switching) loss calculation. Capacitors
application scenarios are presented and discussed. Csn connected in parallel to the devices significantly reduce
turn-off losses by up to 70 % when the converter operates at
rated power [48]. Within this work, Csn = 3.5 µF is assumed.
Scenario A: Linkage between MVDC and HVDC Grids The dc link capacitance is 90 µF ensuring ∆V ≤ 500 V. The
The linkage between an HVDC transmission line and an optimum leakage inductance Lσ,opt is calculated to 45.5 µH.
urban MVDC distribution grid according to Scenario A re- Due to the thermal limits of the semiconductors, the switching
quires a converter system with a voltage transformation ratio frequency is limited to 1 kHz.
from V1 = ±150 kV to V2 = ±5 kV. The high voltage V1
can be realized by series-connection of 60 bridges. At the Scenario C: Grid Interconnection I
secondary side, the series-connection of two bridges is capable
In this system, 60 primary bridges and 60 secondary bridges
to connect the converter system to the 10 kV distribution
are connected in series to allow the connection of the converter
grid. Consequently, thirty of these series-connected bridges are
system to ±150 kV (V1 , V2 ) and a power flow of 1.21 GW. The
connected in parallel. Within this work, a bidirectional power
current rating of the semiconductor devices is the same as for
flow between these two grids of 15 MW is assumed. This leads
Scenario B. Hence, also in this scenario ABB 5SHY 55L4500
to a secondary current I2 of 1500 A and a comparatively small
IGCT modules and ABB 5SDD 11D2800 diodes [46] are
primary current I1 at the high-voltage side of 50 A. Since the
applied with Csn = 3.5 µF. In particular, this scenario takes
power is distributed equally among the DAB converters within
voltage variations of ±10 % of the rated voltage into account.
the converter system, each semiconductor is loaded with the
This results into an increased dc link capacitance of 143 µF
same current of ip,AC,rms = is,AC,rms = 36 A. For the design
that is required for a maximum voltage ripple at a single
of a DAB converter system for this application the character-
DAB converter of ±5 %. The leakage inductances and the
istics of ABB 5SNG 0150P450300 IGBT modules, rated at
switching frequency are equal to the values of Scenario B:
4.5 kV and 150 A per device [46], are taken into account. The
Lσ,opt = 45.5 µH and fsw = 1 kHz.
dc link capacitance connected to each primary and secondary
bridge is 1.12 µF limiting ∆V to 500 V (equivalent to a voltage
ripple of ±5 %). The optimized leakage inductance Lσ,opt in Scenario D: Grid Interconnection II
this application scenario is 3.7 mH. The DAB converters are This scenario investigates the linkage of two HVDC trans-
operated with fsw = 1 kHz. mission systems operating on different voltage level. The
secondary side that operates on ±300 kV requires 120 series-
Scenario B: Wind Farm Connection connected secondary bridges to achieve the grid voltage V2 .
The primary side connected to V1 = ±150 kV can be realized
The DAB converter system can be used for connecting with 60 input bridges in series. Further 60 series-connected
wind farms to the HVDC transmission line (V1 = ±150 kV). primary bridges are connected in parallel to energize all of
Equivalent to the linkage between urban MVDC grids and the 120 output bridges. The power flow is also in this scenario
HVDC lines, also in this scenario 60 bridges are connected in equal to 1.21 GW. Due to the doubled number of DAB, the
power rating of each converter is half compared to Scenario
B and C. Hence, the characteristic of ABB 5SHX 26L4520
TABLE III
M EDIUM - FREQUENCY TRANSFORMER DESIGN PARAMETERS reverse conducting IGCT [46] are applied with Csn = 1 µF.
Due to the insufficient current capability of the monolithic
Scenario A Scenario B & C Scenario D diodes, ABB 5SDF 11H4505 fast recovery diodes [46] have
S 0.3 MVA 24.3 MVA 12.2 MVA to be applied in parallel to the IGCTs.
J 13 A/mm2 5 A/mm2 5 A/mm2 Also this scenario takes voltage variations of ±10 % of the
Lσ 3700 µH 45.5 µH 91 µH rated voltage into account. The dc-link capacitance limiting
Number of turns 209 9 19
Turn ratio 1:1 1:1 1:1 the maximum voltage ripple to ±5 % is determined to be
fsw 1 kHz 1 kHz 1 kHz 72 µF. The medium-frequency transformer is designed with
Vcore 0.0146 m3 0.5638 m3 0.3838 m3 the to the optimum leakage inductance of Lσ,opt = 91 µH for
Ap 0.0002 m4 0.0022 m4 0.3101 m4
fsw = 1 kHz.
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
9
300 100
Efficiency η in %
v in kV
DAB Scenario A
200
95
100
0
90 M2DC Scenario A
2
i in kA
0 85
0 0.2 0.4 0.6 0.8 1
−2
P/PN
2.2 Fig. 10. Efficiency of M2DC and DAB converter for Scenario A
v̄c in kV
2
100
1.8 M2DC Scenario C
99
Efficiency η in %
4
i in kA
98 DAB Scenario D
3
DAB Scenario B+C
2 97
M2DC Scenario D
0 0.5 1 1.5 2 0 0.5 1 1.5 2
96
t in ms t in ms M2DC Scenario B
95
Fig. 9. Voltage and current waveforms of the upper part of the bipolar M2DC 0 0.2 0.4 0.6 0.8 1
of Scenario D at nominal voltages and nominal power
P / PN
Fig. 11. Efficiency of M2DC and DAB converter for Scenario B, C and D
VII. S IMULATION R ESULTS AND C OMPARISON
In this section, the efficiencies of the two compared dc-dc
converter topologies are presented. The results take switching as shown in Fig. 13(b). In this scenario, the M2DC can
and conduction losses of the semiconductors into account compete with the Dual-Active Bridge only in low partial load
as well as winding and core losses of the transformers and conditions concerning the efficiency. However, the number of
inductors. installed semiconductors is significantly higher for the M2DC,
The semiconductor losses of the M2DC are estimated with as evident from Table IV.
the equations from Section III-D. To verify the operation Scenario B investigates a converter system with a power
principle and the analytical calculation of the semiconductor of 1.21 GW and primary voltage V1 of ±150 kV and a
losses of the M2DC, the scenarios have been built up in secondary voltage V2 of ±10 kV. With a value of 98.93 %
MATLAB/Simulink and PLECS [54]. In Fig. 9, the simulation at rated power, the efficiency of the Dual-Active Bridge
of the M2DC of Scenario D operated at nominal voltage and system (Fig. 11) is significantly higher compared to Scenario
nominal load is exemplary shown. PI controllers successfully A. The breakdown of losses in Fig. 12(b) shows increasing
control input current, output current and the energy balance conduction and switching losses with power, whereas the share
of the converter arms. The semiconductor losses of the DAB of the transformer losses becomes smaller. Due to the high
have also been simulated with PLECS. voltage transformation ratio, the M2DC has poor efficiency. Its
The improved generalized Steinmetz Equation (iGSE) [55] maximum efficiency of 96.5 % can be observed at partial load
is applied for core loss calculations. The used Steinmetz (206 MW). The efficiency at rated power is only 95.4 %. As
parameters can be found in [45]. with Scenario A, the major part of the losses are conduction
The efficiencies of the converter system designed for Sce- losses, see Fig. 13(b). Furthermore, the converter exhibits a
nario A are depicted in Fig. 10.
For the DAB converter, the diagram shows a slightly lower
efficiency for partial load compared to the M2DC. However, TABLE IV
C OMPARISON OF THE M2DC AND DAB CONVERTER FOR DIFFERENT
the DAB converter shows a better performance if the trans- SCENARIOS
ferred power is above 0.18 pu. At rated power, the efficiency
is about 96.7 %. The breakdown of the losses is given in Scenario Converter Ap Ecap Pswitch
Fig. 12(a). It can be seen that the magnetic losses (core A M2DC 0.06 m4 582 kJ 3.16 GVA
and winding losses) decrease with transferred power. Also DAB 0.012 m4 1.7 kJ 0.486 GVA
the influence of the switching losses becomes smaller. The B M2DC 25.1 m4 35 MJ 109 GVA
conduction losses do not increase significantly. DAB 0.132 m4 135 kJ 16.2 GVA
The losses of the M2DC are higher at this point leading to C M2DC 0.670 m4 1.55 MJ 51.3 GVA
a lower efficiency of 92.9 %. The high transformation ration DAB 0.132 m4 215 kJ 16.2 GVA
(V1 = ±150 kV and V2 = ±5 kV) results in a high secondary D M2DC 1.71 m4 1.75 MJ 78.5 GVA
DAB 0.312 m4 216 kJ 1.43 GVA
power loop current, leading to very high conduction losses,
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
10
10
Magnetics
8 Switching
1.5 Conduction 1.5
Relative losses in %
Relative losses in %
Relative losses in %
6
1 1
0.5 0.5
2
0 0 0
0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1
P / PN P / PN P / PN
(a) Scenario A (b) Scenario B & C (c) Scenario D
Fig. 12. Losses of the DAB converter system
12 12 12
Magnetics
10 10 Switching 10
Conduction
Relative losses in %
Relative losses in %
Relative losses in %
8 8 8
6 6 6
4 4 4
2 2 2
0 0 0
0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1
P / PN P / PN P / PN
(a) Scenario A (b) Scenario B (c) Scenario D
2 2 2
Relative losses in %
Relative losses in %
Relative losses in %
1 1 1
0 0 0
0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1
P / PN P / PN P / PN
(d) Scenario C with V1 = V2 (e) Scenario C, V1 = V2 , reduced V̂sec (f) Scenario C with max. unbalance
Fig. 13. Losses of the M2DC converter systems
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
11
TABLE V
Efficiency η in %
S CENARIO C: SYSTEM EFFICIENCY AT NOMINAL LOAD IN CASE OF
99.0 VOLTAGE VARIATIONS
98.5 V1 / V2
variable V̂sec 135 kV / 165 kV 150 kV / 150 kV 165 kV / 135 kV
98.0 const. V̂sec
M2DC 97.94 % 99.45 % 98.35 %
DAB 98.29 % 98.93 % 99.48 %
−30 −20 −10 0 10 20 30
∆V in kV TABLE VI
S CENARIO D: SYSTEM EFFICIENCY AT NOMINAL LOAD IN CASE OF
Fig. 14. Efficiency of the M2DC of Scenario C for voltage variations at VOLTAGE VARIATIONS
nominal power
V1 / V2
270 kV / 165 kV 300 kV / 150 kV 330 kV / 135 kV
high amount of semiconductors and a very large secondary-
side transformer. It can be stated that the DAB converter M2DC 97.01 % 96.46 % 95.91 %
DAB 99.29 % 99.04 % 98.62 %
system is superior in this application scenario.
As the same bridges are used for the Dual-Active Bridge
in Scenario C, the components, the efficiencies and the distri- two converter topologies is almost 1 %.
bution of the losses is identical to Scenario B. The efficiency In Scenario D, the DAB converter system is still highly ef-
of the M2DC designed for this application scenario is signifi- ficient, although twice as many DAB are applied and the rated
cantly better compared to the other scenarios. Concerning the power is kept constant. The maximum efficiency of 99.15 % is
losses, this topology is competitive with the DAB converter obtained at half of the rated power. At full power, the efficiency
system. At rated power and equal voltages at the primary and of 99.05 % is only slightly lower. The distribution of the losses
secondary side, efficiencies exceeding 99.5 % can be observed. are shown in Fig. 12(c). The difference to the loss distribution
At unity conversion ratio V1 = V2 , the current only flows within Scenario B and C is caused by the different amount
through the upper arms, as I1 = I2 and Iˆsec = 0. Therefore, of DAB and by the selected semiconductors. However, the
the efficiency for this operating point is very high. If there is M2DC is not competitive with the DAB converter system due
a difference ∆V = V1 − V2 between primary and secondary- to the high voltage differences and the resulting high secondary
side voltage, the current of the secondary power loop builds power loop current. With a maximum efficiency of 96.46 %
up, resulting in increased losses in the lower submodules and near nominal power, it shows a very poor performance. The
a decreased efficiency. As shown in Fig. 14, the efficiency losses are dominated by the switching losses due to the high
is also dependent on the sign of ∆V . The efficiency can fsec , see Fig. 13(c). As shown in Table VI, the efficiency
be improved by dynamical adjustment of V̂sec dependent slightly increases if the voltage difference between primary
on the power that has to be transferred by the secondary and secondary side becomes smaller. Though, it is even worse
power loop, see Fig. 14. By lowering V̂sec , the magnetization if the voltage difference is at its maximum (330 kV / 135 kV).
losses in the secondary-side coupled inductor L2 are decreased The variation in efficiency of the DAB converter system
and the switching frequency of the semiconductors can be results from the different semiconductor characteristics. The
reduced. However, the conduction losses increase. For a given conduction performance of the IGCTs that carry the current
transferred power, there exists an optimum for V̂sec where in the primary bridge is more efficient than the conduction per-
the sum of magnetization, switching and conduction losses formance of the diodes carrying the current in the secondary
is minimal. bridge.
Exemplary, Fig. 13(d) and Fig. 13(e) show the loss dis-
tribution at nominal voltages and power for nominal V̂sec VIII. E CONOMIC E VALUATION
and reduced V̂sec , respectively. As expected, the losses in
the magnetic components (magnetization losses of the output In this section, based on the analysis of the topologies for
transformer) and the switching losses are lower when V̂sec is the different scenarios, an evaluation of the investment costs
reduced. is performed. According to current market prices, the costs of
the capacitors of the M2DC are approximately 150 C/kJ. In
However, the efficiency decreases if there is a voltage
case of the DAB converter in Scenario A, the required dc link
difference between primary and secondary side, see Fig. 14
capacitance is significantly smaller, leading to higher specific
and Fig. 13(f).
costs of 400 C/kJ. To determine the correlation between the
The efficiencies for a voltage variation of ±10 % are given window area product Ap and the costs of the magnetic
in Table V. If the voltages are balanced, the M2DC shows components, the market prices of different 50-Hz transformers
a significantly better performance than the DAB converter between 160 kVA and 3.15 MVA have been evaluated. The cost
system. At maximum unbalance, the efficiency of the DAB of the magnetic devices in the different converter systems is
converters system is higher than that of the M2DC. In case approximately
of an increase of the voltage at the input of 10 % and a 10 %
lower output voltage, the difference of the efficiency of the Kmag ≈ Ap · 723 000 C/m4 + Nmag · 4000 C, (36)
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
12
TABLE VII
E CONOMIC EVALUATION
Investment costs1 in kC
Scenario Converter Capacitors Magnetics Power Electronics Total
A M2DC 87.3 58.33 11 100 11 200
DAB 0.68 249 1700 1950
B M2DC 5250 18 200 382 000 405 000
DAB 54 335 56 700 57 100
C M2DC 232 478 179 000 180 000
DAB 86 335 56 700 57 100
D M2DC 263 1250 275 000 276 000
DAB 86.4 706 49 900 50 700
1 All figures are rounded to three significant digits
where Nmag is the number of magnetic devices in the system. carried out for different scenarios employing HVDC and
For the bipolar M2DC Nmag = 4 is assumed, whereas for MVDC grids, and the systems have been compared.
the DAB converter system Nmag = 60 and Nmag = 120, The DAB converter system features an efficiency between
respectively. 98.9 % and 99.2 % independent of the voltage transformation
Costs for semiconductors, wiring, cabinets and control are ratio.
estimated with 3.50 C per kVA installed switching power In general, the M2DC requires a much higher amount
Pswitch . of semiconductors. Consequently, compared to a converter
Table VII gives an overview of the cost distribution for system based on DAB converters, the investment costs for
the different converter systems and scenarios, Fig. 15 depicts an M2DC converter system are at least by a factor of three
the specific costs per nominal power. Obviously, the power higher. If the frequency of the secondary power loop can be
electronics (including semiconductors, wiring, cabinets and increased to 1 kHz, the M2DC requires a similar amount of
control) account for the major part of the costs. Although magnetic components compared to the soft-switching DAB
the M2DC topology requires a high number of capacitors, the converter system.
influence on the total cost is negligible. The cost part of the It has been shown that the M2DC is not suitable for high
voltage ratios, because in this case the circulating current in
A, M2DC
the converter becomes high, which results in a poor efficiency
of less than 95.5 %. Furthermore, the amount of required
Scenario and Converter
A, DAB
semiconductors is immense. If primary and secondary volt-
B, M2DC
age differ only by a few percent and galvanic isolation of
B, DAB
primary and secondary side is not required, then the M2DC
C, M2DC
Power Electronics is the superior topology. In this case, the maximum efficiency
C, DAB Magnetics exceeds 99.5 %.
D, M2DC Capacitors
D, DAB
ACKNOWLEDGMENT
0 200 400 600 800
The authors wish to acknowledge the financial support of
specific costs in EUR/kVA
E.ON ERC gGmbH foundation, Project No. 1, “High-power
Fig. 15. Specific costs per nominal power PN of the different converter DC-DC Converter”.
systems and scenarios
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
13
[6] L. Zhang, K. Sun, Y. Xing, L. Feng, and H. Ge, “A modular grid- [30] B. S. Riar, T. Geyer, and U. K. Madawala, “Model predictive direct
connected photovoltaic generation system based on dc bus,” IEEE current control of modular multi-level converters,” in IEEE International
Transactions on Power Electronics, vol. 26, no. 2, pp. 523–531, 2011. Conference on Industrial Technology (ICIT), 2013, 2013, pp. 582–587.
[7] H. A. B. Siddique, S. M. Ali, and R. W. De Doncker, “Dc collector grid [31] K. Ilves, S. Norrga, L. Harnefors, and H.-P. Nee, “On energy storage
configurations for large photovoltaic parks,” in EPE’13 ECCE Europe requirements in modular multilevel converters,” IEEE Transactions on
15th Conference on Power Electronics and Applications, 2013. Power Electronics, vol. 29, no. 1, pp. 77–88, 2014.
[8] M. Bragard, N. Soltau, S. Thomas, and R. W. De Doncker, “The balance [32] F. Deng and Z. Chen, “A control method for voltage balancing in mod-
of renewable sources and user demands in grids: Power electronics for ular multilevel converters,” IEEE Transactions on Power Electronics,
modular battery energy storage systems,” IEEE Transactions on Power vol. 29, no. 1, pp. 66–76, Jan 2014.
Electronics, vol. 25, no. 12, pp. 3049–3056, 2010. [33] S. P. Engel and R. W. De Doncker, “Control of the modular multi-level
[9] K. Yang and A. Walid, “Outage-storage tradeoff in frequency regulation converter for minimized cell capacitance,” in Proceedings of the 2011-
for smart grid with renewables,” IEEE Transactions on Smart Grid, 14th European Conference on Power Electronics and Applications (EPE
vol. 4, no. 1, pp. 245–252, 2013. 2011), 2011, pp. 1–10.
[10] Q. Schiermeier, “Renewable power: Germany’s energy gamble,” Nature, [34] R. Picas, J. Pou, S. Ceballos, J. Zaragoza, G. Konstantinou, and V. G.
vol. 496, pp. 156–158, 2013. Agelidis, “Optimal injection of harmonics in circulating currents of
[11] R. Wagner, “Large lead-acid batteries for frequency regulation, load modular multilevel converters for capacitor voltage ripple minimization,”
levelling and solar power applications,” Journal of Power Sources, in IEEE ECCE Asia Downunder (ECCE Asia), 2013, 2013, pp. 318–324.
vol. 67, no. 12, pp. 163–172, 1997, proceedings of the Fifth European [35] S. Kenzelmann, A. Rufer, D. Dujic, F. Canales, and Y. R. De Novaes,
Lead Battery Conference. “A versatile dc/dc converter based on modular multilevel converter for
[12] J. McDowall, “High power batteries for utilities - the world’s most energy collection and distribution,” in IET Conference on Renewable
powerful battery and other developments,” in IEEE Power Engineering Power Generation (RPG 2011), 2011, pp. 1–6.
Society General Meeting, 2004, 2004, pp. 2034–2037 Vol.2. [36] S. Kenzelmann, “Modular dc/dc converter for dc distribution and col-
[13] E.ON Ruhrgas, “E.ON inaugurates power-to-gas unit in Falkenhagen lection networks,” Ph.D. dissertation, École Polytechnique Fédérale de
in eastern Germany,” Press Release, Aug. 2013. [Online]. Avail- Lausanne, Swiss, 2012.
able: http://www.eon.com/en/media/news/press-releases/2013/8/28/eon- [37] A. Bendre, I. Wallace, G. A. Luckjiff, S. Norris, R. W. Gascoigne,
inaugurates-power-to-gas-unit-in-falkenhagen-in-eastern-germany.html D. Divan, and R. M. Cuzner, “Design considerations for a soft-switched
[14] R. Magureanu, M. Albu, A.-M. Dumitrescu, and M. Priboianu, “A modular 2.4-mva medium-voltage drive,” IEEE Transactions on Industry
practical solution for grid connected dispersed generation from renew- Applications, vol. 38, no. 5, pp. 1400–1411, 2002.
able sources: Dc connection,” in International Symposium on Power [38] A. J. Watson, H. Q. S. Dang, P. W. Wheeler, J. C. Clare, G. Mondal,
Electronics, Electrical Drives, Automation and Motion (SPEEDAM), A. R. Rufer, S. Kenzelmann, and Y. De Novaes, “A novel multilevel
2006, 2006, pp. 1228–1231. converter structure integrated into power systems and its performance
[15] H. Akagi, “Large static converters for industry and utility applications,” evaluation,” in 13th European Conference on Power Electronics and
Proceedings of the IEEE, vol. 89, no. 6, pp. 976–983, 2001. Applications EPE ’09, 2009, 2009, pp. 1–10.
[16] M. E. Baran and N. R. Mahajan, “Dc distribution for industrial systems: [39] H. Fan and H. Li, “High-frequency transformer isolated bidirectional dc-
opportunities and challenges,” IEEE Transactions on Industry Applica- dc converter modules with high efficiency over wide load range for 20
tions, vol. 39, no. 6, pp. 1596–1601, 2003. kva solid-state transformer,” IEEE Transactions on Power Electronics,
vol. 26, no. 12, pp. 3599–3608, 2011.
[17] S. Kouro, J. Rodriguez, B. Wu, S. Bernet, and M. Perez, “Powering the
[40] D. Montesinos-Miracle, M. Massot-Campos, J. Bergas-Jane,
future of industry: High-power adjustable speed drive topologies,” IEEE
S. Galceran-Arellano, and A. Rufer, “Design and control of a
Industry Applications Magazine, vol. 18, no. 4, pp. 26–39, 2012.
modular multilevel dc/dc converter for regenerative applications,” IEEE
[18] J. A. Ferreira, “The multilevel modular dc converter,” IEEE Transactions
Transactions on Power Electronics, vol. 28, no. 8, pp. 3970–3979,
on Power Electronics, vol. 28, no. 10, pp. 4460–4465, 2013.
2013.
[19] G. J. Stigler, “The economies of scale,” Journal of Law and Economics, [41] G. Kish, M. Ranjram, and P. Lehn, “A modular multilevel dc/dc
vol. 1, pp. 54–71, 1958. converter with fault blocking capability for hvdc interconnects,” IEEE
[20] L. R. Christensen and W. H. Greene, “Economies of scale in u.s. electric Transactions on Power Electronics, vol. PP, no. 99, 2013.
power generation,” Journal of Political Economy, vol. 84, no. 4, pp. [42] R. W. De Doncker, D. M. Divan, and M. H. Kheraluwala, “A three-
655–676, Aug. 1976. phase soft-switched high-power-density dc/dc converter for high-power
[21] M. Stieneker and R. W. De Doncker, “System efficiency estimation of applications,” IEEE Transactions on Industry Applications, vol. 27,
redundant cascaded-cell converters in applications with high-power bat- no. 1, pp. 63–73, Jan./Feb. 1991.
tery energy storage systems,” in International Conference on Renewable [43] W. G. Hurley and W. H. Woelfle, Transformers and Inductors for Power
Energy Research and Applications (ICRERA), 2012, 2012, pp. 1–6. Electronics - Theory, Design and Applications. John Wiley & Sons Ltd.,
[22] R. Withanage and N. Shammas, “Series connection of insulated gate 2013.
bipolar transistors (igbts),” IEEE Transactions on Power Electronics, [44] A. J. Forsyth and G. Calderon-Lopez, “Sampled-data analysis of the
vol. 27, no. 4, pp. 2204–2212, 2012. dual-interleaved boost converter with interphase transformer,” IEEE
[23] R. Marquardt, “Stromrichterschaltungen mit verteilten Energiespeich- Transactions on Power Electronics, vol. 27, no. 3, pp. 1338–1346, 2012.
ern,” German Patent DE10 103 031, 2002. [45] N. Soltau, D. Eggers, K. Hameyer, and R. W. D. Doncker, “Iron losses
[24] A. Lesnicar, J. Hildinger, and R. Marquardt, “Modulares Strom- in a medium-frequency transformer operated in a high-power dc-dc
richterkonzept für Netzkupplungsanwendungen bei hohen Spannungen converter,” IEEE Transactions on Magnetics, vol. 50, no. 2, 2013.
[modular converter concept for grid interconnection at high voltage [46] ABB. (2013, September) Semiconductor data sheets. Documentation.
levels],” in ETG-Fachtagung, 2002 (in German). ABB. [Online]. Available: http://www.abb.com/semiconductors
[25] A. Lesnicar and R. Marquardt, “A new modular voltage source inverter [47] D. Segaran, D. G. Holmes, and B. P. McGrath, “Comparative analysis of
topology,” in International Conference and Exposition on Electrical and single and three-phase dual active bridge bidirectional dc-dc converters,”
Power Engineering (EPE), 2003, 2003. in Australasian Universities Power Engineering Conference (AUPEC)
[26] R. Marquardt and A. Lesnicar, “New concept for high voltage - modular 2008, 2008, pp. 1–6.
multilevel converter,” in IEEE Power Electronics Specialists Conference [48] R. U. Lenke, “A contribution to the design of isolated dc-dc converters
(PESC), 2004, 2004. for utility appliactions,” Ph.D. dissertation, RWTH Aachen University,
[27] H.-P. Nee and L. Angquist, “Perspectives on power electronics and grid Germany, 2012, ISBN: 978-3-942789-05-9.
solutions for offshore wind farms,” Elforsk, Tech. Rep., Nov. 2010. [49] S. P. Engel, N. Soltau, H. Stagge, and R. W. De Doncker, “Dynamic
[Online]. Available: http://www.elforsk.se/Rapporter/?download= and balanced control of three-phase high-power dual-active bridge dc-
report&rid=10 96 dc converters in dc-grid applications,” IEEE Transactions on Power
[28] L. Angquist, A. Antonopoulos, D. Siemaszko, K. Ilves, M. Vasiladiotis, Electronics, vol. 28, no. 4, pp. 1880–1889, 2013.
and H.-P. Nee, “Open-loop control of modular multilevel converters [50] H. van Hoek, M. Neubert, and R. W. De Doncker, “Enhanced modulation
using estimation of stored energy,” IEEE Transactions on Industry strategy for a three-phase dual active bridge - boosting efficiency of an
Applications, vol. 47, no. 6, pp. 2516–2524, 2011. electric vehicle converter,” IEEE Transactions on Power Electronics,
[29] H. Barnklau, A. Gensior, and J. Rudolph, “A model-based control vol. 28, no. 12, pp. 5499–5507, 2013.
scheme for modular multilevel converters,” IEEE Transactions on In- [51] A. K. Tripathi, K. Hatua, H. Mirzaee, and S. Bhattacharya, “A three-
dustrial Electronics, vol. 60, no. 12, pp. 5359–5375, 2013. phase three winding topology for dual active bridge and its d-q mode
0885-8993 (c) 2013 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2014.2310656, IEEE Transactions on Power Electronics
14
control,” in Applied Power Electronics Conference and Exposition Sedigheh Rabiee (S’13) finished her B.Sc. degree
(APEC), 2012 Twenty-Seventh Annual IEEE, 2012, pp. 1368–1372. in electrical engineering and information technology
[52] D. Boillat, S. Roy, A. Tripathi, and S. Bhattacharya, “Design considera- from Shiraz University, Shiraz, Iran, in 2008. She
tions of a three phase dual active bridge based on reactive power flow,” received her M.Sc. degree in electrical power en-
in Energy Conversion Congress and Exposition (ECCE), 2012 IEEE, gineering from Technical University of Darmstadt,
2012, pp. 424–430. Darmstadt, Germany, in 2012. Her master thesis
[53] N. Soltau, H. A. B. Siddique, and R. W. De Doncker, “Comprehensive deals with the implementation of modular multi-
modeling and control strategies for a three-phase dual-active bridge,” in level converters in a multi-terminal HVDC system.
International Conference on Renewable Energy Research and Applica- Since February 2013, she has been with the Institute
tions (ICRERA), 2012, 2012, pp. 1–6. for Power Generation and Storage Systems (PGS),
[54] Plexim Inc., “PLECS – piece-wise linear electrical circuit simulation.” E.ON Energy Research Center, RWTH Aachen Uni-
[Online]. Available: http://www.plexim.com versity, as a Research Associate. Her research interests include application of
[55] K. Venkatachalam, C. R. Sullivan, T. Abdallah, and H. Tacca, “Accurate power electronic devices in medium-voltage dc grids.
prediction of ferrite core loss with nonsinusoidal waveforms using
only steinmetz parameters,” in IEEE Workshop on Computers in Power
Electronics, 2002, 2002, pp. 36–41.
0885-8993 (c) 2013 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.