IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO.

5, MAY 2011 1877

Multiple-Module High-Gain High-Voltage

DC–DC Transformers for Offshore Wind
Energy Systems
Nicholas Denniston, Member, IEEE, Ahmed M. Massoud, Member, IEEE,
Shehab Ahmed, Member, IEEE, and Prasad N. Enjeti, Fellow, IEEE

Abstract—Renewable energy sources, such as offshore wind farms, but it requires power electronic converters to boost and
farms, require high voltage gains in order to interface with power control wind turbine outputs.
transmission networks. These conversions are normally made us- Conventional HVDC, a mature technology that has been in
ing bulky, complex, and costly transformers and high-voltage ac–
dc converters with unnecessary bidirectional power flow use for decades, uses ac transformers for voltage boosting and
capability. Multiple modules of single-switch single-inductor dc– a 12-pulse thyristor bridge for rectification and power flow
dc converters can reach high gains without transformers in control. This technology is robust and reliable, but it requires
these applications due to low semiconductor conduction loss in bulky, complex, and costly line-frequency transformers at each
high-power devices. This paper describes a new approach for end of the conversion (rectification and inversion). The inherent
high-gain high-voltage dc–dc converters using multiple modules of
single-switch single-inductor transformerless converters. Results bidirectional power flow capability is of less importance at the
for low-voltage experimental prototypes show gains of up to 29 p.u. offshore wind farm side, so simpler unidirectional converters
and demonstrate the potential of the approach as high-gain dc–dc are more attractive in these applications.
converters for offshore wind farms. This paper then demonstrates High-frequency pulse transformers with large turns ratios are
the viability of multiple-module converters compared to a con- difficult to design at high voltage and power levels. Problems
ventional high-voltage dc converter and a theoretical full-bridge
converter due to fewer devices and valves, comparable isolation include poor coupling, dielectric losses in insulation, and core
levels, and ease of interleaving for increased reliability. losses from nonsinusoidal excitation. The distributed capac-
itance of the winding turns can lower efficiency and slow
Index Terms—DC–DC transformers, high-voltage DC transmis-
sion, offshore wind energy. the pulse transitions [15]–[18]. Unidirectional high-gain dc–dc
converters can eliminate the transformers.
I. I NTRODUCTION Many high-gain dc–dc transformerless converters have been
reported in the literature, although most are limited to a few

R ESEARCH in harnessing and delivering electrical power

from renewable energy sources (RES) has skyrocketed as
political and economic concerns have threatened traditional fos-
kilowatts at best. Several authors have proposed the use of
coupled inductors to extend gain, assist in turning off rectifier
diodes, or both. A boost converter with a coupled inductor that
sil fuel supplies. Wind energy is the most mature RES, and more provides additional gain and regenerative snubbing is proposed
than 100 GW of capacity has been installed throughout the in [19]. A converter using a complicated coupled inductor to
world. Recent research has investigated grid connections [1]– provide additional gain and lossless snubbing is proposed in
[4], modeling and control [5]–[11], and condition monitoring [20]. A multiple-input converter for fuel cell vehicles that uses
to increase reliability [12]–[14]. Locations that are well suited coupled inductors for additional gain and alleviating reverse
for large-scale wind energy production, such as offshore wind recovery problems is proposed in [21]. The coupled inductors
farms, are often far from demand centers. Efficient transmission in these converters operate like transformers and face the same
of the generated power over these long distances requires design challenges.
boosting turbine output voltages to high voltage. High-voltage A series of high-gain dc–dc converters known as Luo con-
dc (HVDC) transmission appears promising for offshore wind verters is described in [22]–[24]. Starting with either an inverse-
SEPIC converter or a buck–boost (BB) converter with output
Manuscript received March 1, 2010; accepted May 14, 2010. Date of filter, sets of capacitors, inductors, and diodes are added to
publication June 21, 2010; date of current version April 13, 2011. This work reach high gains at lower duty cycles. The low duty cycles
was supported by a National Priorities Research Program grant from the Qatar
National Research Fund.
allow greater converter efficiency, even with additional series
N. Denniston and P. N. Enjeti are with Texas A&M University, College element losses, but both the component count and the converter
Station, TX 77843 USA (e-mail: [email protected]; [email protected]). order increase dramatically. The large number of components,
A. M. Massoud is with Qatar University, Doha, Qatar (e-mail:
[email protected]). multiple series semiconductor drops, and complicated small-
S. Ahmed is with Texas A&M University at Qatar, Doha, Qatar (e-mail: signal response appear to limit the use of Luo converters in
[email protected]). high-power applications.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. Research has turned recently to multiple modules of
Digital Object Identifier 10.1109/TIE.2010.2053340 transformer-isolated converters, although much of the work has

0278-0046/$26.00 © 2010 IEEE

1878 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 5, MAY 2011

Fig. 2. Definition of ripple factor RF .

extended. Because switching frequencies are limited in high-

power converters to reduce losses, device turn-off times in the
microsecond range (less than 1% of a 1-kHz period) still allow
for extreme duty cycles. Evaluating the gain range of the pro-
posed approach requires accurate modeling of its nonidealities,
such as device conduction losses and nonzero ripple current.
Fig. 1. Multiple-module converters. The principle of energy conversion [29] increases accuracy
in modeling dc–dc converters by removing the zero-ripple
focused on input-series–output-parallel connections for large assumption from the analysis. The actual conduction resistance
voltage step-down ratios [25]–[27]. For example, a 10 kV-to- carrying an rms current is replaced by a modified resistance
400 V step-down converter using multiple low-voltage dc–dc carrying an average current and dissipating the same power
forward converter modules is presented in [28]. The high- according to (1). The uppercase subscripts represent the equiva-
frequency isolation transformers in these approaches allow lent averaged element, while the lowercase subscripts represent
arbitrarily many series connections. the actual element. This equivalent averaged resistance (EAR)
This paper describes a new approach to high-voltage high- can be reflected between branches of an appropriate circuit
gain dc–dc converters using multiple modules of single-switch model based on the average branch current relationships
single-inductor transformerless converters. The use of high- 2 2
power semiconductors with low-per-unit conduction losses Irms rx = Iavg rX . (1)
allows operation well beyond the traditional limits of these
For the continuous-conduction-mode (CCM) current wave-
converters. The simple structure makes interleaving possible
form of Fig. 2, the rms and average currents are well known
for reduced loss and increased reliability. In this paper, two
as (2), where RF is the ripple factor (peak of ripple magnitude
modules are used to increase reliability and reduce control
divided by average value)
complexity. Equations for the gain and efficiency of cascade
and series configurations (Fig. 1) are derived and verified 
RF 2
via simulation. The sensitivity of the converters to changes Irms = Iavg 1 + . (2)
3
in inductor resistance is derived. Experimental results from
low-power prototypes that reach gains of up to 29 p.u. are The loss in a diode is due only to its average current and
presented. The multiple-module approach is compared to a its forward drop Vf if the diode resistance is neglected; this
conventional HVDC approach and to a theoretical transformer- assumption is acceptable as the diode resistance is normally
based dc–dc converter approach to demonstrate its advantages much smaller than the inductor dc resistance. In this case, diode
in high-voltage high-gain offshore wind applications. losses are unaffected by ripple factor, and the averaged voltage
drop VF = Vf .
The following assumptions are made in modeling the single-
II. T HEORY
switch single-inductor boost and BB converters.
A. Modeling Conduction Losses 1) Diode losses are modeled as a forward voltage VF .
The gain of the boosting single-switch converters (boost 2) Insulated-gate bipolar transistor (IGBT) losses are
and BB) is limited by parasitic losses that dominate at large modeled as a forward voltage VF .
duty cycles. The gains of practical low-power converters are 3) Inductor losses are modeled as resistance rl .
limited to less than 3 p.u. due to large-per-unit conduction 4) Switching losses are neglected.
losses from inductor dc resistances and diode forward volt- 5) Converters operate in CCM.
ages. The maximum duty cycles are further limited in high- 6) The per-unit system is used with a normalized input
frequency converters to allow the power switch time to turn off. voltage of 1 V and a normalized input power of 1 W.
High-power components show much smaller per-unit losses, The converters are modeled using the averaged transformer
and the range of duty cycles for high-efficiency operation is model, with all parasitic resistances and forward drops reflected
DENNISTON et al.: TRANSFORMERS FOR OFFSHORE WIND ENERGY SYSTEMS 1879

TABLE I
C ASCADE C ONVERTERS : M AXIMUM D UTY C YCLES
AND G AINS FOR 95% E FFICIENCY

Fig. 3. DC model of boost and BB converters.

into the diode branch, as shown in Fig. 3. The gain of each

converter is given by (3), with rL,pu given in (4)
    
 VO  V 1
  F,pu
 VIN  = M 1 − M (1 − D) rL,pu
1 + (1−D)
(3) Hereafter, the combined IGBT/diode term will be ignored. If
equal ripple factors are assumed, the η1 η2 term becomes η 2 for
2

 
rl RF 2 any combination of converters. Equation (6) can then be solved
rL,pu = 1+ . (4) for D as in
Ro 3

The M in (3) is the ideal magnitude of the converter volt- η 1/2
age gain, which is 1/(1 − D) for the boost converter and Dmax,casc = 1 − rL,pu . (7)
1 − η 1/2
D/(1 − D) for the BB converter (D is the duty cycle). The
second and third terms of (3) give the efficiency of the converter This equation gives the maximum duty cycle that a cascade
and separate the losses in the semiconductors from the losses converter can operate at with a specified efficiency under a
in the inductor resistance. Note that the combined IGBT/diode given parasitic resistance. Note that the maximum duty cycles
loss term is independent of D in boost converters. In BB con- are equal for each converter type. The maximum duty cycle for
verters, the combined IGBT/diode term is inversely dependent an efficiency of 95% is calculated in Table I for sample parasitic
on D; however, for large gain (D ≈ 1), the losses are approxi- resistances and ripple factors. Small parasitic resistances are
mately independent of duty cycle. Note that the expression for needed to obtain high gain at high efficiency. Note that the
rL,pu is the same for both boost and BB converters because the cascade–boost/BB converter is called cascade hybrid.
relationship between the inductor current and the output (diode) The ripple factor assumption requires closer examination.
current is identical for both converters. Ripple factor equations are derived from inductor volt–seconds
in (8), where fs is the switching frequency, L is the inductance,
B. Cascade Converters P is the input power, and VIN is the input voltage. For D ≈ 1,
the expressions are virtually identical
The cascade configuration shown in Fig. 1(a) connects the V2
output of the first converter to the input of the second one. Each IN
D, (boost)
converter is rated for the full input power. To derive an equation RF = 2fVs LP 2 (8)
2
IN
2fs LP D , (buck−boost).
for the total gain, the losses are assumed equal on a per-unit
basis for each converter. The first stage has a lower absolute Physically, the second converter sees an input voltage that
2
per-unit output resistance (RO1pu = VO1 /P = M12 ) than the is M1 times the input voltage of the first converter, and the
second stage (RO2pu = VO2 /P = M1 M22 ), but the parasitic
2 2
magnitude of the allowable current ripple is smaller by the same
elements have equal per-unit magnitudes. The converters are factor. L2 must then be M12 times larger than L1 to maintain the
operated at the same duty cycle for ease of comparison. The same ripple factor. A compromise can made between increased
gain of the cascade converter is given by (5) and (6), where Mi RF and decreased inductor size, and the efficiency terms in (5)
refers to the ideal converter gain must be evaluated separately. Table I shows, however, that the
Mcasc = M1 M2 η1 η2 (5) effects of RF decrease as rL,pu decreases.
   Cascade converters offer potential for large gains due to
VF,pu 1 the multiplicative effect, but the efficiency losses compound
ηi = 1 − rL,pu . (6)
Mi (1 − D) 1 + (1−D) 2
quickly for the same reason. The entire input power is processed
twice, and the losses quickly become a limiting factor with
This equation neglects the loading effect of the parasitic large parasitic resistances. Interleaving can reduce parasitic
resistances of the second converter on the first converter, but resistances and inductor sizes while increasing reliability. The
from (6), it can be shown that, with low parasitic resistances, individual modules can allow larger RF ’s for decreased induc-
this effect is only significant for D ≈ 1 and results in low tance, or the inductance can be held constant to increase overall
efficiency. Practical converters are designed to avoid this con- efficiency. In either case, the physical inductor size is reduced
dition. The forward voltages are often negligible in high-power due to smaller current magnitudes.
devices; for example, the Eupec DD400S33K2C 3.3-kV 400A Cascade converters suffer from other practical issues. The
IGBT shows a worst case forward drop of 3.5 V. For a base first converter experiences intermediate voltage stresses and
of 1000 V, the per-unit forward voltage is only 0.0035 V. large current stresses, while the second converter experiences
1880 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 5, MAY 2011

TABLE II
S ERIES C ONVERTERS : M AXIMUM D UTY C YCLES
AND G AINS FOR 95% E FFICIENCY

Fig. 4. Series boost connection. The additional BB converter in (b) pro-

vides an additional series input source. (a) Direct connection. (b) Practical
connection.

large voltage stresses and small current stresses. This increases

the inventory requirements because of limited interchange-
ability between converters. Control is also difficult because The gain of the series converters is lower than the gain of
of the interaction between converters. Load variations at the cascade converters because of the additive effect, but they offer
output of the second converter will result in fluctuations in the many advantages. Each converter processes only half the input
inductor current and output voltage of the first converter, and power, so the total input power is only processed once. The
stability problems may result, unless the converter bandwidths gains in efficiency allow a larger range of duty cycle operation.
are properly chosen. According to the data in Tables I and II for rL = 0.005 p.u., the
series converters outperform the cascade converters, although
this advantage disappears with smaller parasitic resistances.
C. Series Converters Semiconductors in series converters experience only half
The series configuration of Fig. 1(b) connects the inputs of the total voltage stress, a decided advantage in high-voltage
the converters in parallel and the outputs in series. The gain and converters where several switches must be combined in series to
efficiency of the series converters are derived in (9) and (10) withstand the rated voltage. Semiconductors in both converters
using the assumptions of Section II-B have identical ratings, reducing the inventory requirements.
Each converter sees an identical input voltage and neutral
Mseries = (M1 + M2 )η (9) point, so the output voltage of each converter can be controlled
   independently. This allows a variety of control strategies that
VF,pu 1 are beyond the scope of this paper.
η = 1− rL,pu . (10)
Mi (1 − D) 1 + (1−D) 2

Mi represents the ideal converter gain. The factoring of (9) D. Reverse Recovery Effects
requires that each efficiency term is identical (approximately
The preceding analysis assumes that switching losses are
true for D ≈ 1 per Section II-B).
negligible due to low fs ; however, diode reverse recovery
Practically, only the series–hybrid (boost–BB) connection
causes loss, even at low frequency, and increases the turn-on
can be directly implemented. Fig. 4(a) shows a directly con-
stress of the power switch. Silicon carbide (SiC) diodes do
nected series boost converter. The bottom output is shorted to
not exhibit reverse recovery effects, and they offer improved
the neutral voltage via the return of the source. An extra BB
switching speeds and high-temperature operation [30]. SiC
converter can be cascaded before the bottom converter to create
blocking voltages have approached 10 kV although at a lower
an effective second input voltage (D = 0.5) in series with the
current density than that in Si devices [30], [31]. As SiC
actual source, as shown in Fig. 4(b). The efficiency loss of the
technology continues to improve, they can be used in multiple-
cascaded converter is minimal (assuming negligible losses from
module converters to mitigate reverse recovery effects and
forward drops) at D = 0.5. For example, the second term of
increase efficiency.
(11) evaluates to 99.5% for a 0.001-p.u. inductor resistance and
100% RF operated at D = 0.5.
Equation (10) (neglecting VF ) is solved for D to determine E. Sensitivity Analysis
the maximum duty cycle for a specified efficiency in (11). The
maximum duty cycle for 95% efficiency is calculated in Table II The sensitivity of the gain to the inductor resistance is
for different parasitic resistances and RF ’s. High efficiency can investigated here. The Bode sensitivity of Y (x1 , x2 , . . . , xn )
be achieved over a greater duty cycle range in series converters with respect to x1 is defined as (12) for small changes in x1

η δY /Y x1 δY
Dmax,series = 1 − rL,pu . (11) SxY1 = = . (12)
1−η δx1 /x1 Y δx1

Inductor size and ripple factor are not problems in series Neglecting diode voltage, gain G can be written generally
converters; both converters see the same input voltage, so the as (13), where Mi (D) is the ideal gain, rpu is the pu inductor
inductors are the same size. Table II confirms that the effects of resistance, and k = (1 + RF 2 /3) is the EAR factor from (4).
ripple factor decrease with small parasitic resistances. Exponent n is one for series converters and two or more for
DENNISTON et al.: TRANSFORMERS FOR OFFSHORE WIND ENERGY SYSTEMS 1881

TABLE III
S IMULATION C IRCUIT PARAMETERS

TABLE IV
C OMPARISON OF R ESULTS FOR η = 95% AND 50% RF

Fig. 5. Sensitivity of a cascade–boost converter at D = 0.9 and

rpu = 0.001.

cascade converters. From (12) and (13), the sensitivity can be

derived as in (14)
 n
1 Fig. 6(c) shows the gains of the three series converters;
G = Mi (D) (13)
krpu
1 + (1−D) note that the series–boost and series–BB converters use a
2
cascaded BB converter operated with a gain of −1, as shown
rpu δG nkrpu 1 in Fig. 4(b). In the simulation of the series–hybrid converter,
SrGpu = =−
. (14)
G δr (1 − D)2 1 + krpu 2 the two converters are operated at equal gains instead of equal
(1−D)
duty cycles. Fig. 6(c) and (d) shows that this configuration
The sensitivity therefore varies with rpu and D (assuming reaches the same gain as the series–boost converter with only
a constant RF ). Examining the two extremes of D gives the a slight loss of efficiency. The falloff in efficiency at large
limits of the sensitivity function for krpu  1 duty cycles is due to increased inductor losses from the larger
 BB duty cycle. The maximum gain for the series converters is
−nkrpu , D ≈ 0 approximately 30, or less than one-sixth of the cascade gain,
SrGpu = (15)
−n, D ≈ 1. but high efficiency is achieved over a larger duty cycle range.
As in the cascade converters, the series–BB converter slightly
A cascade boost converter was simulated at D = 0.9 with
underperforms in both gain and efficiency; however, it does
RF = 50% to verify (14). The inductor resistance was varied
offer source inversion. Duty cycles and gains at 95% efficiency
from a nominal value of 0.001 p.u., and the results are shown
are compared to the calculated values in Table IV and show
in Fig. 5. The simulated sensitivity corresponds closely to the
excellent agreement.
calculated results, particularly at small variations. At larger
The simulation results validate the analysis at large duty
variations from nominal, the assumption of small changes in
cycles and demonstrate that the multiple-module approach per-
rpu is violated, but the results are still very close to the predic-
forms well at low parasitic resistances.
tions. The analysis shows that multiple-module converters are
relatively insensitive to small variations in inductor resistance.
IV. E XPERIMENTAL R ESULTS
III. S IMULATION R ESULTS
A 500-W prototype of each converter was built to verify the
The six converters discussed in Section II were simulated theory and simulation results. The converters were designed
in SIMPLIS. The circuit parameters are listed in Table III. to boost a 15-V 500-W input to a maximum ideal output of
Inductor values were chosen to limit RF to less than 50% at 500 V. The series converters and the second cascade stages
maximum duty cycles and were held constant throughout the used Fairchild FGL60N100BNTD 1000-V–60-A IGBTs, and
simulation. the first cascade stages used ST Micro STGE200NB60S
The gain and efficiency plots for cascade converters are 600-V–150-A switches. All stages used IXYS DSEI-60 600-V–
shown in Fig. 6. Fig. 6(a) shows that all three cascade converters 60-A diodes. The inductors in the series converters and the first
have similar gains, particularly at large D. The peak gain is cascade stages were Hammond 500-μH–150-A inductors (rl =
over 200, but the efficiency [Fig. 6(b)] at this duty cycle is 1.8 mΩ), and the inductors in the second cascade stages were
low. The converters using at least one BB converter show a low Hammond 10-mH–50-A inductors (two) in series (rl = 46 mΩ
efficiency at low duty cycles due to the IGBT/diode drop, as in total).
discussed in Section II. Table IV compares the simulations to The prototypes had practical issues that affected perfor-
the calculations from Table II, and the agreement is excellent. mance. At an input voltage of 15 V, the forward voltages
1882 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 5, MAY 2011

Fig. 6. Simulation results for multiple-module converters. (a) Cascade gain (M ). (b) Cascade efficiency (η). (c) Series gain (M ). (d) Series efficiency (η).

Fig. 7. Experimental results: Cascade converters. The maximum gain is Fig. 8. Experimental results: Series converters. The maximum gain is 22 p.u.
29 p.u. (boost). (hybrid).

of the semiconductors were in the range of 0.08–0.12 p.u. uses a conventional HVDC converter with an ac transformer
instead of the 0.0015 p.u. assumed in the simulations. This and a 12-pulse thyristor bridge. Approach B uses a theoretical
corresponds to a 12% diode/IGBT loss for a boost converter high-power full-bridge converter with a 1-kHz transformer.
and up to a 24% loss in a BB converter operated at D = 0.5. Approach C uses a cascade–boost converter. Approach D uses
The efficiency of the cascade converters was superior to that a series–hybrid converter. The converters are modeled with the
of the series converters because the larger second-stage input following assumptions.
voltage resulted in lower diode losses for that converter. The 1) Switching losses are neglected (fS = 1 kHz).
series boost and series BB converters suffered worst because 2) No snubbers or rate limiters are used.
all three converters saw a low input voltage. Efficiencies ranged 3) Leakage inductance is neglected; this would be a severe
from 71% to 87%. problem in approaches A and B due to the large turns
To properly evaluate the prototypes in the face of large-per- ratios and high-voltage insulation, so this underestimates
unit diode drops, simulations were rerun for each converter the negatives of these approaches.
using the actual component values. The new simulation results 4) All devices are ideal.
are shown as dotted lines in Figs. 7 and 8, and the experimental 5) All currents are ripple free.
results are shown as data points. The experimental results 6) Balancing networks are neglected.
agree closely with the actual-value simulations. The results also Each switch or diode is made up of several series-connected
closely match the analysis of (5), (6), (9) and (10). For example, devices to withstand the rated voltage. A string of such devices
(5) and (6) give an output voltage of 433 V for a cascade boost is referred to as a valve, and the individual devices are referred
operated at D = 0.825; the experimental output voltage was to as switches or diodes. External balancing components are
435 V. required to evenly distribute the valve voltage (steady state and
transient) across the devices; the design of these networks is
discussed elsewhere in the literature and is out of the scope
V. E VALUATION OF A PPROACH
of this paper. The approaches are compared based on the
Now, the multiple-module converter approach is evaluated following: 1) total devices; 2) device ratings; and 3) voltage
against other HVDC approaches. The input source is a 1-MW isolation levels for gate drivers. Semiconductor ratings are
wind turbine with an output of 1 kVLL,rms . This voltage will taken from the Eupec catalog. The approaches are discussed
be boosted to 132 kV for HVDC transmission. Approach A hereinafter and summarized in Table V.
DENNISTON et al.: TRANSFORMERS FOR OFFSHORE WIND ENERGY SYSTEMS 1883

Fig. 11. Approach C: Multiple-module cascade–boost converter.

Fig. 9. Approach A: Conventional HVDC converter.

Fig. 12. Approach D: Multiple-module series–hybrid converter.

pulse transformers with large turns ratios are difficult to design

in high-voltage high-power applications. A full discussion of
these issues is out of the scope of this paper, so for purposes of
comparison, it is assumed that an acceptable transformer can be
designed. Although the switch valves must only withstand the
rectifier output, the diode valves must withstand the reflected
rectifier output voltage. The isolation level required is 1.35 kV.
Device ratings are shown in Fig. 10.
Fig. 10. Approach B: Theoretical full-bridge converter.

A. Conventional HVDC C. Multiple-Module Cascade–Boost Converter

The conventional HVDC approach is shown in Fig. 9. This The cascade–boost configuration is shown in Fig. 11. The
approach requires an ac line-frequency transformer with a large wind turbine output is rectified and input to the first boost
turns ratio and two secondaries (Y and Δ) for the 12-pulse converter. The cascade configuration must boost the 1.35 kV
rectifier. The maximum output voltage of the SCR bridge is rectifier output to 132 kV (98×). Each boost converter is
1.35 times the input line voltage, so each transformer must designed to provide a gain of 9.9 at a duty cycle of 0.899. The
provide a gain of 50 for a total gain of 135. Each SCR bridge switch and diode valve voltage stresses in a boost converter
sees half the input voltage, so SCRs must be rated for one- are equal to the output voltage of each stage, and the average
half of the input sinusoidal peak voltage. The SCR bridges diode valve current equals the output current of each stage. The
are controlled to produce 132 kV. The isolation level depends average switch valve currents are equal to D times the stage
on the choice of ground point; grounding the midpoint of the input currents. The peak values of both switch and diode valve
two rectifiers requires isolation of ±66 kV, while grounding currents are determined by the inductor current. The maximum
the return of the bottom rectifier requires isolation of 132 kV. isolation voltage is given by the valves in the output converter
Device ratings are shown in Fig. 9. (132 kV). Ratings are shown in Fig. 11.

B. Full-Bridge Converter D. Multiple-Module Series–Hybrid Converter

The full-bridge circuit is shown in Fig. 10. A six-pulse bridge The series–hybrid converter is shown in Fig. 12. Because of
is used to rectify the wind turbine output. The remaining 98× the lower gain of this converter, an ac transformer with a 4×
of gain is provided by a 1-kHz transformer with a turns ratio turns ratio is added before the rectifier. In a practical converter,
of 150× and a switch valve duty cycle of 0.33. High-frequency this transformer would be much simpler than the ac transformer
1884 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 5, MAY 2011

TABLE V
C OMPARISON OF HVDC A PPROACHES : D EVICE R ATINGS , D EVICE C OUNTS , AND I SOLATION L EVELS

in approach A due to the small turns ratio, and the leakage proach is superior to other approaches due to the ease of
inductance would be much smaller. The remaining 33× of gain interleaving modules. If a single interleaved module fails,
must come from the rectifier (1.35) and the series converter the converter can still function at a reduced power level.
(24.4). The converters are designed for equal gains of 12.2 The circuits of approaches A and B cannot be easily in-
and equal input powers of 500 kW. The boost duty cycle is terleaved, so the failure of a single valve forces the entire
0.918, and the BB duty cycle is 0.924. Each converter operates converter offline. Based on these comparisons, the multiple-
from equal input voltages and draws equal input currents. module converter approach shows clear advantages over both
All semiconductors will be rated per the slightly larger BB conventional thyristor-based approach and transformer-based
requirements. The BB valves must withstand the sum of half approach.
the output voltage and the input voltage. The BB switch valves
carry the average value of the input current. The diode valves VI. C ONCLUSION
carry the average output current. The peak current is given by
the inductor current. The isolation level required is determined This paper has shown that multiple modules of single-switch
by the choice of ground point; similar to approach A, the single-inductor converters can reach high gains at high effi-
required isolation will be either ±66 or 132 kV. ciencies without transformers in offshore wind applications.
Continued advances in high-voltage SiC technology are ex-
pected to alleviate diode reverse recovery effects inherent to
E. Discussion
CCM operation. The multiple-module approach is general;
Table V shows that the device count in the multiple-module although the basic single-switch single-inductor configurations
converters (approaches C and D) is much smaller than that were presented, other nonisolated configurations, such as quasi-
of approaches A and B. A smaller device count simplifies resonant converters, can be combined similarly.
implementation by reducing the balancing network components Experimental results on low-power prototypes showed gains
used with series-connected devices in a valve. Fewer snubbers up to 22.5 p.u. for series converters and up to 29 p.u. for cascade
and rate limiters are required as well. Smaller device count converters. The results show that multiple-module gains are
means lower cost, fewer failure points, and higher reliability. increasing at duty cycles well beyond those where conventional
The voltage levels of the valve devices are important because boost converters normally peak. Efficiencies in the prototypes
devices built to withstand higher voltages are both more expen- were harmed mainly by IGBT and diode drops, which will be
sive and lossier due to doping requirements. The conventional much less significant at high power and voltage levels. The
HVDC approach requires (264) 6.5-kV devices, while the full- duty-cycle-dependent losses in the inductor resistance were not
bridge approach requires (250) 6.5 kV. The cascade approach significant in the prototypes, and inductor parasitics should
uses only (94) 6.5 kV devices, and the series approach uses only be even smaller at high power and voltage levels. It seems
(104) 6.5 kV devices. highly likely that performance will improve at higher power and
The isolation level is similar for all but the full-bridge voltage levels.
converter due to the need to drive the topmost IGBT or SCR Conceptual comparisons of multiple-module converters to
in a valve, but fewer drive circuits are needed in the multiple conventional HVDC systems show that the multiple-module
module converters of C and D. While the choice of ground point approach is superior in terms of device count, device rating, and
can reduce the isolation requirements in A and D, the cascade reliability. Voltage and current stresses within a valve can be
converter must provide the full 132 kV of isolation. balanced using the same techniques discussed in the literature.
Perhaps the most important concern for converters used The multiple-module approach is therefore well suited for high-
in power transmission is reliability. The multiple-module ap- power high-gain offshore wind applications.
DENNISTON et al.: TRANSFORMERS FOR OFFSHORE WIND ENERGY SYSTEMS 1885

R EFERENCES [23] F. L. Luo, “Double-output Luo converters, an advanced voltage-lift

technique,” in Proc. Inst. Elect. Eng.—Elect. Power Appl., Nov. 2000,
[1] R. Takahashi, H. Kinoshita, T. Murata, J. Tamura, M. Sugimasa,
vol. 147, no. 6, pp. 469–485.
A. Komura, M. Futami, M. Ichinose, and K. Ide, “Output power smooth-
[24] F. L. Luo, “Positive output Luo converters: Voltage lift technique,” in
ing and hydrogen production by using variable speed wind generators,”
Proc. Inst. Elect. Eng.—Elect. Power Appl., Jul. 1999, vol. 146, no. 4,
IEEE Trans. Ind. Electron., vol. 57, no. 2, pp. 485–493, Feb. 2010.
pp. 415–432.
[2] S. Nishikata and F. Tatsuta, “A new interconnecting method for wind
[25] R. Giri, V. Choudhary, R. Ayyanar, and N. Mohan, “Common-duty-ratio
turbine/generators in a wind farm and basic performances of the inte-
control of input-series connected modular dc–dc converters with active
grated system,” IEEE Trans. Ind. Electron., vol. 57, no. 2, pp. 468–475,
input voltage and load-current sharing,” IEEE Trans. Ind. Appl., vol. 42,
Feb. 2010.
no. 4, pp. 1101–1111, Jul./Aug. 2006.
[3] J. Lopez, E. Gubia, E. Olea, J. Ruiz, and L. Marroyo, “Ride through
[26] R. Ayyanar, R. Giri, and N. Mohan, “Active input-voltage and load-current
of wind turbines with doubly fed induction generator under symmetrical
sharing in input-series and output-parallel connected modular DC–DC
voltage dips,” IEEE Trans. Ind. Electron., vol. 56, no. 10, pp. 4246–4254,
converters using dynamic input-voltage reference scheme,” IEEE Trans.
Oct. 2009.
Power Electron., vol. 19, no. 6, pp. 1462–1473, Nov. 2004.
[4] R. Cardenas, R. Pena, G. Tobar, J. Clare, P. Wheeler, and G. Asher, “Sta-
[27] J. W. Kimball, J. T. Mossoba, and P. T. Krein, “A stabilizing, high-
bility analysis of a wind energy conversion system based on a doubly fed
performance controller for input series–output parallel converters,” IEEE
induction generator fed by a matrix converter,” IEEE Trans. Ind. Electron.,
Trans. Power Electron., vol. 23, no. 3, pp. 1416–1427, May 2008.
vol. 56, no. 10, pp. 4194–4206, Oct. 2009.
[28] V. Vorperian, “Synthesis of medium voltage dc-to-dc converters from low-
[5] A. Timbus, M. Larsson, and C. Yuen, “Active management of distrib-
voltage, high-frequency PWM switching converters,” IEEE Trans. Power
uted energy resources using standardized communications and modern
Electron., vol. 22, no. 5, pp. 1619–1635, Sep. 2007.
information technologies,” IEEE Trans. Ind. Electron., vol. 56, no. 10,
[29] D. Czarkowski and M. K. Kazimierczuk, “Energy-conservation approach
pp. 4029–4037, Oct. 2009.
to modeling PWM DC–DC converters,” IEEE Trans. Aerosp. Electron.
[6] B. C. Rabelo, W. Hofmann, J. L. da Silva, R. G. de Oliveira, and
Syst., vol. 29, no. 3, pp. 1059–1063, Jul. 1993.
S. R. Silva, “Reactive power control design in doubly fed induction gen-
[30] R. Singh, J. A. Cooper, Jr., M. R. Melloch, T. P. Chow, and J. W. Palmour,
erators for wind turbines,” IEEE Trans. Ind. Electron., vol. 56, no. 10,
“SiC power Schottky and PiN diodes,” IEEE Trans. Electron Devices,
pp. 4154–4162, Oct. 2009.
vol. 49, no. 4, pp. 665–672, Apr. 2002.
[7] R. Cardenas, R. Pena, P. Wheeler, J. Clare, and G. Asher, “Control of
[31] R. S. Howell, S. Buchoff, S. Van Campen, T. R. McNutt, H. Hearne,
the reactive power supplied by a WECS based on an induction generator
A. Ezis, M. E. Sherwin, R. C. Clarke, and R. Singh, “Comparisons of
fed by a matrix converter,” IEEE Trans. Ind. Electron., vol. 56, no. 2,
design and yield for large-area 10-kV 4H-SiC DMOSFETs,” IEEE Trans.
pp. 429–438, Feb. 2009.
Electron Devices, vol. 55, no. 8, pp. 1816–1823, Aug. 2008.
[8] A. Bouscayrol, X. Guillaud, P. Delarue, and B. Lemaire-Semail, “Ener-
getic macroscopic representation and inversion-based control illustrated
on a wind-energy-conversion system using hardware-in-the-loop simu-
lation,” IEEE Trans. Ind. Electron., vol. 56, no. 12, pp. 4826–4835,
Dec. 2009.
[9] B. Beltran, T. Ahmed-Ali, and M. Benbouzid, “High-order sliding-mode
control of variable-speed wind turbines,” IEEE Trans. Ind. Electron.,
vol. 56, no. 9, pp. 3314–3321, Sep. 2009.
[10] J. Arbi, M. J. B. Ghorbal, I. Slama-Belkhodja, and L. Charaabi, “Direct
virtual torque control for doubly fed induction generator grid connection,”
IEEE Trans. Ind. Electron., vol. 56, no. 10, pp. 4163–4173, Oct. 2009.
[11] L. Wei, G. Joos, and J. Belanger, “Real-time simulation of a wind turbine
generator coupled with a battery supercapacitor energy storage system,” Nicholas Denniston (M’09) received the B.Sc. and
IEEE Trans. Ind. Electron., vol. 57, no. 4, pp. 1137–1145, Apr. 2010. M.Sc. degrees from the Department of Electrical
[12] A. Stefani, A. Bellini, and F. Filippetti, “Diagnosis of induction ma- Engineering, Texas A&M University, College Sta-
chines’ rotor faults in time-varying conditions,” IEEE Trans. Ind. tion, in 2001 and 2010, respectively.
Electron., vol. 56, no. 11, pp. 4548–4556, Nov. 2009. He is currently with Texas A&M University. His
[13] K. Rothenhagen and F. W. Fuchs, “Doubly fed induction generator model- research interests include high-gain dc–dc convert-
based sensor fault detection and control loop reconfiguration,” IEEE ers, converters for renewable energy applications,
Trans. Ind. Electron., vol. 56, no. 10, pp. 4229–4238, Oct. 2009. and fuel cell modeling.
[14] Y. Wenxian, P. J. Tavner, C. J. Crabtree, and M. Wilkinson, “Cost-effective
condition monitoring for wind turbines,” IEEE Trans. Ind. Electron.,
vol. 57, no. 1, pp. 263–271, Jan. 2010.
[15] L. Haoyu, B. Xianglin, and W. Jianqiang, “Study on modeling of high
frequency power pulse transformer,” in Proc. WAC, 2008, pp. 1–5.
[16] M. Borage, K. V. Nagesh, M. S. Bhatia, and S. Tiwari, “Design of
LCL-T resonant converter including the effect of transformer winding
capacitance,” IEEE Trans. Ind. Electron., vol. 56, no. 5, pp. 1420–1427,
May 2009.
[17] J. A. Martin-Ramos, J. Diaz, A. M. Pernia, J. M. Lopera, and F. Nuno,
“Dynamic and steady-state models for the PRC-LCC resonant topology
with a capacitor as output filter,” IEEE Trans. Ind. Electron., vol. 54, no. 4,
pp. 2262–2275, Aug. 2007.
[18] J. C. Fothergill, P. W. Devine, and P. W. Lefley, “A novel prototype design Ahmed M. Massoud (M’10) received the B.Sc.
for a transformer for high voltage, high frequency, high power use,” IEEE (first-class honors) and M.Sc. degrees from The
Trans. Power Del., vol. 16, no. 1, pp. 89–98, Jan. 2001. Faculty of Engineering, Alexandria University,
[19] W. Rong-Jong and D. Rou-Yong, “High step-up converter with coupled- Alexandria, Egypt, in 1997 and 2000, respectively,
inductor,” IEEE Trans. Power Electron., vol. 20, no. 5, pp. 1025–1035, and the Ph.D. degree in electrical engineering from
Sep. 2005. the Department of Computing and Electrical Engi-
[20] L. Wuhua and H. Xiangning, “An interleaved winding-coupled boost con- neering, Heriot–Watt University, Edinburgh, U.K.,
verter with passive lossless clamp circuits,” IEEE Trans. Power Electron., in 2004.
vol. 22, no. 4, pp. 1499–1507, Jul. 2007. He was a Research Fellow at Strathclyde Univer-
[21] R. J. Wai, C. Y. Lin, L. W. Liu, and Y. R. Chang, “High-efficiency single- sity, Glasgow, U.K., from 2005 to 2008 and at Texas
stage bidirectional converter with multi-input power sources,” Elect. A&M University at Qatar, Doha, Qatar, from 2008
Power Appl., IET, vol. 1, no. 5, pp. 763–777, Sep. 2007. to 2009. He is currently an Assistant Professor with Qatar University, Doha.
[22] F. L. Luo and H. Ye, Advanced DC/DC Converters. Boca Raton, FL: His research interests include power quality, active power filtering, distributed
CRC, 2004. generation, and multilevel converters.
1886 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 5, MAY 2011

Shehab Ahmed (M’07) received the B.Sc. degree Prasad N. Enjeti (M’85–SM’88–F’00) received the
in electrical engineering from Alexandria University, B.E. degree in electrical engineering from Osmania
Alexandria, Egypt, in 1999 and the M.Sc. and Ph.D. University, Hyderabad, India, in 1980, the M.Tech.
degrees from the Department of Electrical Engineer- degree in electrical engineering from the Indian
ing, Texas A&M University, College Station, in 2000 Institute of Technology Kanpur, Kanpur, India, in
and 2007, respectively. 1982, and the Ph.D. degree in electrical engineering
He was with Schlumberger Technology Corpo- from Concordia University, Montreal, QC, Canada,
ration, working on downhole mechatronic systems in 1988.
from 2001 to 2007. He is currently an Assistant Pro- In 1988, he joined the Department of Electrical
fessor with Texas A&M University at Qatar, Doha, Engineering, Texas A&M University, College Sta-
Qatar. His research interests include mechatronics, tion, as an Assistant Professor, where he was pro-
solid-state power conversion, and electric machines and drives. moted as Associate Professor in 1994, became a Full Professor in 1998, and
is the Lead Developer of the Power Electronics/Power Quality and Fuel Cell
Power Conditioning Laboratories. He is actively involved in many projects
with industries while engaged in teaching, research, and consulting in the
areas of power electronics, motor drives, power quality, and clean power utility
interface issues. He is the holder of four U.S. patents and has licensed two
new technologies to the industry so far. His current research interests include
advanced converters for power supplies and motor drives; power quality issues;
active power filter development; utility interface issues; advancing switching
power supply designs and solutions to complex power management issues in the
context of analog- and mixed-signal applications; exploring alternative designs
to meet the demands of high slew rate load currents at low output voltages;
power-conditioning systems for fuel cell, wind, and solar energy systems;
and design of high-temperature power conversion systems with wideband-gap
semiconductor devices.
Dr. Enjeti is a Registered Professional Engineer in Texas. He was the recip-
ient of the select title “Class of 2001 Texas A&M University Faculty Fellow”
Award for demonstrating achievement of excellence in research, scholarship,
and leadership in the field. He was also the recipient of the IEEE–Industry
Applications Society (IAS) Second and Third Best Paper Awards in 1993, 1998,
1999, 2001, and 1996, respectively; the Second Best IEEE–IA Transactions
Paper published in midyear 1994 to midyear 1995; and the IEEE–IAS Mag-
azine Prize Article Award in 1996. In 2000, he was elected as Fellow by the
IEEE Fellows Committee for his “contributions to solutions of utility interface
problems in power electronic systems and harmonic mitigation.” He directed a
team of students to design and build a low-cost fuel cell inverter for residential
applications, which won the 2001 Future Energy Challenge Award Grand Prize
from the Department of Energy.