This paper provides a loss analysis and comparison of high power semiconductor devices in 5MW Permanent Magnet Synchronous Generator (PMSG) Medium Voltage (MV) Wind Turbine Systems (WTSs). High power semiconductor devices of the presspack type IGCT, module type IGBT, presspack type IGBT, and presspack type IEGT of both 4.5kV and 6.5kV are considered in this paper. Benchmarking is performed based on the backtoback type 3level Neutral Point Clamped Voltage Source Converters (3LNPC VSCs) supplied from a grid voltage of 4160V. The feasible number of semiconductor devices in parallel is designed through a loss analysis considering both the conduction and switching losses under the operating conditions of 5MW PMSG wind turbines, particularly for application in offshore wind farms. This paper investigates the loss analysis and thermal performance of 5MW 3LNPC wind power inverters under the operating conditions of various power factors. The loss analysis and thermal analysis are confirmed through PLECS Blockset simulations with Matlab Simulink. The comparison results show that the presspack type IGCT has the highest efficiency including the snubber loss factor.
I. INTRODUCTION
In the multiMW wind turbine market, the maximum power rating of commercial wind turbines has been increased to more than 5MW with the aim of generating more power from wind power sites
[1]
. The power electronic converters in mediumvoltage level applications are generally realized as multilevel (ML) VSCs instead of 2LVSCs in order to improve the performance factors in terms of switch power losses, harmonic distortion, and common mode voltage/current
[2]
,
[3]
. In the family of multilevel (ML) inverters, the threelevel topology, called the Neutral Point Clamped (NPC) inverter, is one of the few topologies that have received a reasonable consensus in the high power community. These NPC inverters have also been implemented successfully in industrial applications for high power drives and wind turbines
[2]

[4]
.
In multiMW wind turbine systems, there are many different types of power converter topologies and highpower switching devices in use. In particular, the recent development of high power semiconductor technology has resulted in a wide variety of practical power devices. The benchmarking of these topologies and their optimal power switches is important for industry to select the most feasible solution in the product development of wind turbines. In general, a comparison of various power switches involves a great deal of engineering work considering many different aspects of device characteristics. Therefore, it is necessary to pick critical performance factors on which the comparison for devices is made so that the selection of the most feasible power device can be made with meaningful engineering work and insight.
This paper investigates the utilization of the four most feasible high power switching devices for 5MW PMSG WTSs; the presspack type IGCT, module type IGBT, presspack type IGBT, and presspack type IEGT. The power converter topology of 3LNPC VSCs is selected as the main platform for the device comparison. Power loss dissipated in the semiconductor devices is one of the most important performance factors in high power drives considering the total system efficiency and the requirements in terms of the cooling system. In addition, a power loss analysis gives an engineer insight into costeffective system designs
[5]
,
[6]
. This paper compares four different high power switching devices with respect to power loss dissipation. Furthermore, this paper investigates a transient thermal description of the IGCT platform in 3LNPC VSCs. 6.5kV and 4.5kV class devices are considered in the loss calculation of the 3LNPC VSCs. The loss distribution among several switching devices in the converter including the snubber circuit is also explained in this paper.
This paper is structured in four main sections. Section II describes the power semiconductor devices under comparison for a 5MW PMSG WTS. Section III discusses the electrothermal modeling of the power semiconductor devices. Section IV presents simulation results of 3LNPC VSCs in a 5MW PMSG MV WTS. Finally, a comparison of the losses in 3LNPC VSCs is shown in Section V.
II. POWER SEMICONDUCTOR DEVICES UNDER COMPARISON
A simplified schematic of the backtoback type 3LNPC VSCs for 5MW PMSG MV wind turbines with the generator connected to a grid is presented in
Fig. 1
[2]

[7]
. The semiconductor devices commonly used in high power converters are IGBTs (in a module or presspack package), presspack type IEGTs and IGCTs for 3LNPC converters. The recent technological development of 4.5 kV and 6.5 kV IGCTs, IGBTs, and IEGTs has been enabling a substantial improvement of MV converters in many aspects
[8]
,
[9]
. These four major types of semiconductor devices are considered in this paper. The major operating characteristics of the target semiconductor devices are summarized in
Table I
and are used in the loss analysis throughout this paper. The important characteristics for the loss calculation are the threshold voltage (
V_{TO}
), slope resistance (
R_{T}
) as a function of the collector/anode current, turnon energy (
E_{on}
), turnoff energy (
E_{off}
), diode reverse recovery energy (
E_{rr}
), maximum operating junction temperature (
T_{vj_max}
), and thermal resistance (
R_{th}
).
Backtoback type 3LNPC VSCs for 5MW PMSG MV wind turbines.
CIRCUIT PARAMETERS AND OPERATING CONDITIONS[11],[12],[13],[15],[18]
CIRCUIT PARAMETERS AND OPERATING CONDITIONS [11], [12], [13], [15], [18]
 A. PressPack Type IGCT for MV Converters
MV IGCT presspack devices are mainly used in high power industrial applications owing to their advantageous features such as presspack housing cases, a higher thermal/power cycling capability, and explosionfree failure mode
[5]
. Recently, a 10kV IGCT device has been introduced and its switching capability has been confirmed
[10]
. In this paper, a 6.5kV/3800A presspack type IGCT (ABB 5SHY42L6500) is considered for the loss analysis
[11]
. As for the antiparallel and neutral point clamped diode, a 6.0kV/1100A presspack type FRD (ABB 5SDF10H6004) is employed
[12]
.
 B. Module Type IGBT for MV Converters
Module type IGBTs are widely accepted in the market for the power range below 34 MW approximately. MV IGBTs having a blocking voltage of 6.5kV have been developed by several manufacturers and employed in many industrial applications. The continuous switching current capability of 6.5kV IGBT modules has reached around 750A. In this paper, a 6.5kV/750A module type IGBT (ABB 5SNA0750G650300) is considered for the loss analysis
[13]
.
 C. PressPack Type IGBT for MV Converters
The recently developed presspack type IGBT devices combine the advantages of IGBTs with those of presspack cases. Thus, presspack IGBTs have become a competitor for IGCTs in medium and high power industrial applications such as the MV drives for wind turbines
[14]
. In this paper, a 4.5kV/2400A presspack type IGBT (Westcode, IXYS T2400GB45E) is considered for the loss analysis
[15]
.
 D. PressPack Type IEGT for MV Converters
As a highvoltage and largecapacity power semiconductor device to replace conventional GTO thyristors, the Injection Enhanced Gate Transistor (IEGT) has been developed in recent years and applied in a practical use. This development was led by Toshiba and GE Companies. Highvoltage, largecapacity, and fullcontrolled power device of the IEGT type are intended to combine the advantages of IGBT devices and GTO devices. The presspack type 4.5kV IEGTs have been placed on the market, mainly for the use in medium voltage converters
[16]
,
[17]
. In this paper, a 4.5kV/2100A presspack IEGT (Toshiba ST2100GXH22A) is considered for the loss analysis
[18]
.
Target power semiconductor devices for MV wind turbines [11], [13], [15], [18].
III. ELECTROTHERMAL MODELING OF THE POWER SEMICONDUCTOR DEVICES
 A. Power Loss Modeling of Semiconductors Devices
The losses of power semiconductor device are approximated by analytical expressions in terms of voltage and current.
Fig. 3
shows a simplified loss estimation model for power semiconductor devices.
Simplified device switching waveforms and its power losses.
1) Conduction Losses:
The total semiconductor device loss
P_{t}
consists of the conduction loss
P_{cond}
and the switching loss
P_{switching}
[19]
[20]
;
The conduction loss of each power semiconductor depends on the instantaneous onstate voltage
v_{sw}
(
t
) and the instantaneous switching current
i
(
t
) passing through it. The forward onstate voltage of the power semiconductor device,
v_{sw}
(
t
) can be modeled by using a firstorder linear approximation comprised of a threshold voltage
v_{on}
and a series resistance
R_{on}
as follows;
The total conduction loss in the power semiconductors can be expressed as;
where
T
is the fundamental period of the converter
[21]
.
2) Switching Losses:
The switching loss of the power semiconductor device is determined by the total commutation time in which the device is turned on/off, and by the voltage
v(t)
and current
i(t)
across the device. The energy dissipated during commutations is made up of
E_{on}
,
E_{off}
, and
E_{rr}
, for the turnon, turnoff, and diode reverse recovery, respectively. This information is provided by the device manufacturer on their datasheet. The average switching power loss
P_{switching}
over a complete fundamental period
T
may be determined by summing all of the commutations of a device during a respective interval of time. The switching loss for the turnon, turnoff, and diode revers recovery can be expressed as;
Equation (4), (5), and (6) represent linear approximations of the actual switching loss for the turnon, turnoff, and diode reverse recovery based on the specific values (
E_{on(spec)}
,
E_{off(spec)}
, and
E_{rr(spec)}
) provided by the manufacturer. Although the switching loss can vary depending on the gate impedance, parasitic circuit elements, and snubber characteristics, this linear approximation gives fairly good accuracy particularly in the vicinity of the manufacturer’s test point (
V_{test}
and
I_{test}
) and snubber condition
[5]
.
In some cases, the turn off loss of an IGBT device, unlike IGCT devices, is not linear. Therefore, the switching loss of an IGBT can be modeled as a 2
^{nd}
order polynomial curve having a parabolic shape like the following.
where
k_{0}
,
k_{1}
, and
k_{2}
are polynomial coefficients. However, even in the case of (7), the approximations of
P_{on}
and
P_{off}
, by (4)(6), have a quite high engineering accuracy.
 B. Loss Modeling of a Snubber Circuit for the IGCT Platform
Converters employing IGCTs need a
di/dt
limiting inductance to meet the required
di/dt
characteristics during switching on transients. This
di/dt
limiting inductor (
L_{i}
) usually necessitates an additional over voltage protection snubber or clamping circuitry, as shown in
Fig. 1
. This snubber circuitry dissipates additional power loss and gives a rise to an important loss factor which should be taken into consideration for a comprehensive loss comparison.
Fig. 4
presents an equivalent circuit of the
di/dt
limiting inductor, over voltage protection, and the particular IGCT being subject to switching transients in
Fig. 1
. At the instant of switching off, the stored magnetic energy in the
di/dt
limiting inductor due to the onstate current of the IGCT is given by;
OVP clamp and di/dt snubber circuit of IGCT in the upperhalf part of 3LNPC VSC.
This stored energy is mainly dissipated in the snubber resistor (
R_{Cl}
) or fed back to charge the dc link capacitor (
C_{DClink}
)
[22]
. In this paper, the total stored energy in the
di/dt
limiting inductor is considered to be snubber circuit power loss since the loss analysis is performed to compare the full functionality of different power semiconductor switching devices on an equal basis. Therefore, the clamp (snubber) circuit loss
P_{cl}
can be expressed as;
 C. Thermal Modeling of the Semiconductor Devices
The thermal performance of the power semiconductor devices is closely related to the reliability and cost of the whole power converter system
[23]
. The power loss modeling can be implemented based on the current and voltage values in the power devices. Afterwards, the thermal modeling is applied to the estimation of the loss dissipation. The electrothermal model deals with the analysis of both the electrical and thermal performances, which interact with each other by the power dissipation of the electronics devices
[24]
. The main purpose of the electrothermal model is to estimate the junction and case temperature for the 3LNPC VSCs. The electrothermal model is implemented for a presspack type IGCT (5SHY42L6500) and Diode (5SDF10H6004) from ABB.
Fig. 5
shows the electrothermal models involved in the model of the power device, power loss, and thermal model
[25]
. According to
Fig. 5
, the junction temperature is determined by the power losses of the power semiconductors. The steady state average junction temperature of each power semiconductor device can be expressed as follows according to
[26]
.
T_{j}
and
P_{t}
represent the junction temperature and the total semiconductor device loss, respectively
[24]
. The junction temperature is obtained as;
Electrothermal model structure for junction temperature of power semiconductor [25].
The total thermal resistance
R_{th}
consists of
R_{th(jc)}
,
R_{th(ch)}
, and
R_{th(ha)}
;
R_{th(jc)}
is the thermal resistance from the junctiontocase,
R_{th(ch)}
is the thermal resistance from the casetoheatsink,
R_{th(ha)}
is the thermal resistance from the heatsinktoambient.
In order to obtain the transient behavior of the junction temperature, the transient thermal impedance should be taken into consideration in the thermal modeling of the semiconductor devices. The transient thermal impedance is defined to be the transient response of the junction temperature against the unit step power loss. Therefore, the transient thermal impedance indirectly provides the information on the thermal resistance and capacitance of the Foster RC network model.
The curve for the transient thermal impedance
Z_{th(jc)}
can be fit into the series of the exponential term as (12).
The thermal impedance from the junction to the case
Z_{th(jc)}
is modeled as a fourlayer Foster RC network as depicted in
Fig. 6
. The case to the heatsink thermal impedance is modeled as a simple thermal resistor. Furthermore, the ambient temperature is set to 30℃. It is worth mentioning that a number of the four parameters are more than sufficient for a good estimation of the Foster Network. The parameters for a presspack type 6.5kV IGCT platform are summarized in
Table II
[11]
,
[12]
,
[27]
.
Thermal model of power semiconductor devices with fourlayer Foster RC network.
PARAMETERS OF THE THERMAL IMPEDANCE FOR A IGCT PLATFORM
PARAMETERS OF THE THERMAL IMPEDANCE FOR A IGCT PLATFORM
IV. SIMULATION RESULTS OF THE 3LNPC VSCS IN A 5MW PMSG MV WTS
 A. MediumVoltage 3LNPC VSCs Topology
Fig. 7
shows the 3LNPC VSCs topology for the loss analysis of selected high power semiconductors in a 5MW PMSG MV WTS. Each leg of the VSC consists of two neutralpoint clamped diodes, four switches, and four antiparallel diodes. The DCbus voltage is split into threelevels by two series connected capacitors. The middle point of the two capacitors N can be defined as a neutral point. The output voltage
v_{AN}
has three states (
V_{dc}
/2, 0, and –
V_{dc}
/2) in each leg, which are produced by specific conduction paths depending on the output current direction and the output voltage polarity. Due to the relatively low switching frequency, a 2ndorder LCfilter system has been employed at the grid side of the converter to meet the harmonic constraints of the grid code, for example IEEE519
[28]
. In addition, a
di/dt
snubber is essential for all of the IGCT converters to achieve the required
di/dt
characteristics during switching on and to mitigate the reverse recovery stress on the fast diodes.
3LNPC VSCs for loss analysis in 5MW PMSG MV WTS.
 B. Power Loss Distribution in a 5MW PMSG WTS
The simulation is performed based on the parameters of a 5MW MV 3LNPC VSC as specified in
Table III
. Since the Grid Side Converter (GSC) is a 3LNPC type converter connected to an ac line of 4160V, the nominal dclink voltage is chosen to be 7kV. The switching frequency, i.e. the PWM carrier frequency, adopted for the GSC is set to 1020Hz. This switching frequency is selected to be 17 times the fundamental frequency. The selection of the switching frequency is done on the basis of compromising the switching loss and the harmonic content of the ac input current.
SIMULATION PARAMETERS OF 5MW MV 3LNPC VSCS
SIMULATION PARAMETERS OF 5MW MV 3LNPC VSCS
In
Fig. 8
, the ac input currents at the converter pole under the three different power factor conditions (0.9 leading condition, 1.0, and 0.9 lagging condition) are given with respect to the grid phase voltage. It is noted that the amplitude of the ac input current for the case of the 0.9 leading condition under inverter operating mode is at its largest among three conditions. This is due to the fact that the grid side LCfilter adds an additional leading power factor so that the input current at the converter pole requires an increased leading angle to generate the 0.9 leading power factor at the ac input, i.e. upstream of the grid side LCfilter.
Waveforms of ac input current at the converter pole under inverter operating mode (Positive current flowing into converter from grid).
Figs. 9
and
10
show waveforms of the switching current and voltage for the 6.5kV IGCTs in each phaseleg during one ac line period under the power factor of the 0.9 leading condition.
Waveforms of switching voltage and current in the upper side of each phaseleg under inverter operating mode (pf=0.9 leading condition).
Waveforms of switching voltage and current in the lower side of each phaseleg under inverter operating mode (pf=0.9 leading condition).
The converter operates under the inverter operating mode, i.e. power flows from the converter into the grid. In order to obtain the maximum semiconductor losses for the worst possible case, the simulation has been performed under the condition of a maximum ac input current; a line undervoltage of 90% and a power factor of the 0.9 leading condition. The switching voltage and current are sampled at the switching instant from the simulation waveform for each of the semiconductor devices. These sampled voltage and current values are then used to calculate the switching losses based on the specified loss values (
E_{on(spec)}
,
E_{off(spec)}
, and
E_{rr(spec)}
) given in the datasheet of the power semiconductors. The calculation of the switching loss is done in a linear manner as described in (4)  (6), assuming the typical gating impedances and snubber conditions suggested by the manufacturer.
 C. Transient Junction Temperature
The 3LNPC VSC thermal model is built in PLECS to obtain the junction temperature profiles of the IGCT platform for each power device.
Fig. 11
shows the main electrical and thermal circuit of the 3LNPC VSC.
Figs. 12

14
show the transient junction temperature of the one phaseleg IGCT platform under the three different power factor conditions (0.9 leading condition, 1.0, and 0.9 lagging condition) in a 5MW PMSG MV wind turbine.
Electrothermal model of each device for junction temperature in 3LNPC VSC.
Transient junction temperature of one phaseleg IGCT platform (pf=0.9 leading condition).
Transient junction temperature of one phaseleg IGCT platform (pf=1.0 condition).
Transient junction temperature of one phaseleg IGCT platform (pf=0.9 lagging condition).
In
Fig 12
, under the condition of a maximum ac input current, Q
_{1}
and Q
_{4}
are the most stressed device with a junction temperature of 120℃ and a temperature fluctuation of 13℃ amplitude. It can be seen that the mean temperature (111℃) of Q
_{1}
and Q
_{4}
is higher than the mean temperature (48℃) of Q
_{2}
and Q
_{3}
. In addition, the mean temperature (32℃) of D
_{1}
and D
_{4}
is higher than the mean temperature (30℃) of D
_{2}
and D
_{3}
in the 3LNPC VSCs. The mean temperatures (79℃) of NPD
_{5}
and NPD
_{6}
are similar to each other.
V. COMPARISON OF LOSSES
The total power losses (
P_{t}
) of the semiconductor devices in a 5MW PMSG MV wind turbine are summarized in
Fig. 15

17
using (1)(6) under the 0.9 leading power factor condition of the inverter operating mode. The average conduction loss of each device is calculated by integrating the power loss of each device for one cycle of line frequency as shown in
Fig. 8
. The average conduction loss factor is calculated as in (2) and (3). The switching on and off loss factors are calculated based on the device voltage and current values sampled from the simulation waveform, as shown in
Fig. 9
and
Fig. 10
. The total power loss (
P_{t}
) is then obtained by adding the conduction loss (
P_{cond}
) and switching loss (
P_{on}
and
P_{off}
).
Total power loss (P_{t}) distribution in four devices (Q_{1}  Q_{4}) of four different types of power semiconductors (Total power loss value for threephase).
Total power loss (P_{t}) distribution in six devices (D_{1}  D_{4}, NPD_{5}  NPD_{6}) of two different types of diodes; 5SDF10H6004, 5SNA0750G650300 (Total power loss value for threephase).
Total power loss (P_{t}) distribution regarding conduction, switching, and snubber losses in four devices (Q_{1}  Q_{4}) of four different types of power semiconductors (Total power loss value for threephase).
The total power losses of the four target semiconductor devices are compared including the snubber losses for the case of the IGCT. The power loss numbers in
Fig. 15

17
represent the values for the complete 3phase in a GSC of 5MW 3LNPC VSCs as depicted in
Fig. 7
. In general, the turnon losses of the antiparallel diode and neutralpoint clamp diode are very small, so they are ignored in this paper. For the case of a module type IGBT of 6.5kV/750A, it is necessary to employ two devices in parallel (n
_{p}
=2) to meet the converter operating specifications for an ac input current of 708A as shown in
Table III
. This parallel operation can be implemented either by paralleling devices or by paralleling two converter systems. In this paper, it is assumed that the current is shared equally between the two parallel devices of IGBT modules.
Figs. 15
and
16
show the total power loss (
P_{t}
) distribution in each device of the four different power semiconductors and the two different diodes using the power loss model (1)(6) for the GSC of 3LNPC VSCs. In
Fig. 15

16
, the switching devices show symmetrical power losses in one leg of the converter, i.e. the outer devices and the inner devices have almost same losses. The junction temperature of each switching device has been computed based on the thermal resistances given in
Table I
. The ambient temperature, i.e. the temperature of inlet cooling water, has been assumed to be 30 degrees. It is interesting to note that the inner antiparallel diodes (D
_{2}
and D
_{3}
) do not have switching off losses. This is due to the modulation characteristic of 3LNPC VSCs, that is Q
_{2}
and Q
_{3}
being left turn on at the commutation instant of D
_{2}
and D
_{3}
. In
Fig. 16
, a presspack diode of 5SDF10H6004 is commonly employed as antiparallel and neutralpoint clamp diode in the cases of presspack type IGCTs, presspack type IGBTs, and presspack type IEGTs, for the sake of a fair comparison. In the case of module type IGBTs, the antiparallel diode part integrated into the package of a 5SNA0750G650300 is also utilized as a neutralpoint clamp diode. As for neutralpoint clamp diodes of NPD
_{5}
and NPD
_{6}
, two antiparallel diodes of 5SNA0750G650300 are paralleled (n
_{p}
=2).
Fig. 17
provides the total loss (
P_{t}
) distribution regarding the conduction (
P_{cond}
), switching (
P_{on}
,
P_{off}
), and snubber losses (
P_{cl}
) in the four devices (Q
_{1}
, Q
_{2}
, Q
_{3}
, and Q
_{4}
) of four different types of power semiconductors using the power loss model (1)  (6). It is noted from the graph that the IGCT is subject to the largest switch turnoff loss among the four kinds of power devices. On the other hand, the IGCT has the smallest switch turnon loss in the group. The total power loss dissipated in the snubber circuitry for the IGCT converter, as shown in
Fig. 4
, is shown to be around 4.92kW based on (8) and (9) for a threephase GSC. Even with this additional power loss factor due to snubber circuitry, the switch turnon loss of the IGCT is relatively smaller than those of the other power devices as shown in
Fig. 17
.
The total losses (
P_{t}
) of the 3LNPC VSCs for 5MW PMSG MV wind turbines are described in
Fig. 18
for each device in a one phaseleg IGCT platform (
pf
=0.9 leading condition) using the power loss model (1)(6). It shows that the presspack type IGCT has the lowest loss value, 56kW (1.13%), among all four kinds of power semiconductor platforms including the snubber losses. On the other hand, the presspack type IEGT has the highest loss value at 77kW (1.54%). The module type IGBT exhibits a loss value of 58kW (1.18%) which is close to that of the presspack type IGCT. The presspack type IGBT has a loss value of 61kW (1.22%). The power loss of the module type IGBT corresponds to the case of two devices in parallel (n
_{p}
=2). In this paralleling of devices or converters, the equal sharing of current becomes an important design task. However, it may further complicate the control algorithm or mechanical concept when compared to single device approaches such as the IGCT, presspack IGBT, and presspack IEGT.
Total loss distribution regarding conduction, switching, and snubber circuit losses of four different types of power semiconductors (Total power loss value for threephase at 1020Hz switching frequency).
In order to investigate the influence of switching frequency on the loss characteristic, a loss analysis for a switching frequency of 1380Hz has been performed. It is noted from
Fig. 19
that the loss has been increased by approximately 0.4% for the cases of the presspack IGBT, module IGBT, and IGCT. It is interesting to note that the presspack IEGT has the largest increase of loss by approximately 1.0%.
Total loss distribution regarding conduction, switching, and snubber circuit losses of four different types of power semiconductors (Total power loss value for threephase at 1380Hz switching frequency).
Fig. 20
describes the total loss distribution at worst possible case under the inverter operating mode for the target power semiconductor switch of the IGCT type. Under the inverter operating mode, Q
_{1}
, Q
_{4}
, NPD
_{5}
, and NPD
_{6}
are subject to most of the power loss. As noted in
Fig. 8
, the amplitude of the ac input current at the converter pole changes as the power factor is varied from 0.9 leading to 0.9 lagging under the rated output power generation. A higher amplitude of the ac input current at the converter pole naturally results in higher power losses in the power semiconductor switches. In addition, a different power factor angle also changes the loss distribution pattern among the power semiconductor devices in each phase leg due to the commutation property in 3LNPC VSCs.
Total loss distribution in semiconductor devices of three phaseleg under inverter operating mode employing 6.5kV IGCT (pf=0.9 leading condition).
VI. CONCLUSION
In this paper, the loss analysis of a 6.5kV/3800A presspack type IGCT, 4.5kV/2400A presspack type IGBT, 4.5kV/2100A presspack type IEGT, and 6.5kV/750A module type IGBT for 5MW PMSG MV wind turbines employing a backtoback type 3LNPC VSC is presented. The transient thermal junction temperature of the IGCT platform is described in GSC for 5MW PMSG MV wind turbines. The switching loss is obtained under the assumption of typical gate impedance and snubber conditions provided by the semiconductor manufacturers. The switching frequency is set to 1020Hz, under a grid side input voltage of 4.16kV. The presspack type IGCT has been found to have the lowest device losses, i.e. the highest efficiency, among the four candidates with the additional snubber loss of the IGCT being taken into consideration. The module type IGBT requires that two devices be put in parallel in order to meet the given operating specifications of an ac input current of 708A. This paralleled devices structure may complicate the mechanical and cooling systems which are very critical functional elements in multiMW WTSs. Considering the difficulty of measuring the losses of high power converters in the MW range in the lab, the method of loss analysis proposed in this paper has practical usefulness since it can accurately estimate the losses of the high power semiconductor devices in 5MW PMSG MV wind turbines based on the semiconductor device datasheets. An experimental verification of these comparison results is in progress and the results will be reported in future publications.
Acknowledgements
This work was supported by the National ResearchFoundation of Korea (NRF) grant funded by the Koreagovernment (MSIP) (No. 20100028509) & (No.2014R1A2A1A11053678).
BIO
Kihyun Lee was born in the Republic of Korea, in 1982. He received his B.S. degree in Electronic Engineering from Chonbuk National University, Jeonju, Korea, in 2008, where he is presently working towards his M.S. degree in Electrical Engineering. From 2008 to 2012, he was a Research Engineer in the High Power Discrete Division of KODENSHI AUK Co., Korea. His current research interests include high power conversion systems for renewable energy sources and medium electric drive systems.
Yongsug Suh (M'90/SM'07) was born in Seoul, Korea. He received his B.S. and M.S. degrees from Yonsei University, Seoul, Korea, in 1991 and 1993, respectively, and his Ph.D. degree in Electrical Engineering from the University of Wisconsin, Madison, WI, USA, in 2004. From 1993 to 1998, he was an Application Engineer in the Power Semiconductor Division of Samsung Electronics Co., Korea. From 2004 to 2008, he was a Senior Engineer in the Power Electronics & Medium Voltage Drives Division of ABB, Turgi, Switzerland. Since 2008, he has been with the Department of Electrical Engineering, Chonbuk National University, Jeonju, Korea, where he is currently an Associate Professor. His current research interests include power conversion systems of high power for renewable energy sources and medium voltage electric drive systems.
Yongcheol Kang received his B.S., M.S., and Ph.D. degrees from Seoul National University, Korea, in 1991, 1993, and 1997, respectively. He has been with Chonbuk National University, Jeonju, Korea, since 1999. He is currently a Professor at the Department of Electrical Engineering, Chonbuk National University, and the director of the WeGAT Research Center supported the Ministry of Science, ICT, and Future Planning (MSIP), Korea. His current research interests include the development of new protection and control systems for wind power plants and wind energy integration technologies.
2009
World Market Update 2008 (Forecast 20092013)
BTM Consult ApS
Ringøkbing Denmark
Kouro S.
,
Malinowski M.
,
Gopakumar K.
,
Pou J.
,
Franquelo L. G.
,
Wu B.
,
Rodriguez J.
,
Perez M. A.
,
Leon J. I.
2010
“Recent advances and industrial applications of multilevel converters,”
IEEE Trans. Ind. Electron.
57
(8)
2553 
2580
DOI : 10.1109/TIE.2010.2049719
Rodriguez J.
,
Bernet S.
,
Steimer P. K.
,
Lizama I. E.
2010
“A survey on neutralpointclamped inverters,”
IEEE Trans. Ind. Electron.
57
(7)
2219 
2230
DOI : 10.1109/TIE.2009.2032430
Senturk O.
,
Helle L.
,
MunkNielsen S.
,
Rodriguez P.
,
Teodorescu R.
2012
“Power capability investigation based on electrothermal models of presspack IGBT threelevel NPC and ANPC VSCs for multiMW wind turbines,”
IEEE Trans. Power Electron.
27
(7)
3195 
3206
DOI : 10.1109/TPEL.2011.2182661
Suh Y. S.
,
Steinke J.
,
Steimer P.
2007
“Efficiency comparison of voltage source and current source drive system for medium voltage applications,”
IEEE Trans. Ind. Electron.
54
(5)
2521 
2531
DOI : 10.1109/TIE.2007.900352
Faulstich A.
,
Steinke J. K.
,
Wittwer F.
“Medium voltage converter for permanent magnet generators up to 5MW,”
in Proc. Eur. Conf. Power Electron. Appl.
2005
1 
9
Zuckerberger A.
,
Suter E.
,
Schaub C.
,
Klett A.
,
Steimer P.
“Design, simulation and realization of high power NPC converters equipped with IGCTs,”
in Proc. of the ThirtyThird IAS Annual Meeting
1998
Vol. 2
865 
872
Rodriguez J.
,
Bernet S.
,
Wu B.
,
Pontt J.
,
Kouro S.
2007
“Multilevel voltagesourceconverter topologies for industrial mediumvoltage drives,”
IEEE Trans. Ind. Electron.
54
(6)
2930 
2945
DOI : 10.1109/TIE.2007.907044
Bernet S.
2000
“Recent developments of high power converters for industry and traction applications,”
IEEE Trans. Power Electron.
15
(6)
1102 
1117
DOI : 10.1109/63.892825
Bernet S.
,
Carroll E.
,
Streit P.
,
Apeldoorn O.
,
Steimer P.
,
Tschirley S.
2005
“10 kV IGCTs,”
Industry Application Magazine
11
(2)
53 
61
DOI : 10.1109/MIA.2005.1405827
“Asymmetric Integrated GateCommutated Thyristor 5SHY 42L6500”, Datasheet, Doc. No. 5SYA124503 Dec. 12
ABB Switzerland Ltd.
Online:
“Fast Recovery Diode 5SDF 10H6004”, Datasheet, Doc. No. 5SYA110903 January. 10
ABB Switzerland Ltd.
Online:
“IGBT Moduled 5SNA 0750G650300”, Datasheet, Doc. No. 5SYA 160002 042012
ABB Switzerland Ltd.
Online:
Alvarez
,
Filsecker R.
,
Bernet S.
“Characterization of a new 4.5 kV press pack SPT+ IGBT for medium voltage converters,”
in Proc. of the 1st IEEE Energy Conversion Congress and Exposition, ECCE 2009
2009
3954 
3962
“Insulated Gate BiPolar Transistor, Type T2400GB45E”, Datasheet, T2400GB45E Nov. 2011
IXYS
Online:
Kon
,
Nakayama K.
,
Yanagisawa S.
,
Miwa J.
,
Uetake Y.
“The 4500V750A planar gate press pack IEGT,”
in Proc. ISPSD’98
1988
81 
84
Ichikawa K.
,
Tsukakoshi M.
,
Nakajima R.
“Higher efficiency threelevel inverter employing IEGTs,”
in Proc. 19th Annu. IEEE APEC
2004
Vol. 3
1663 
1668
“Toshiba Silicon NChannel IEGT ST1200GXH24A”, Datasheet 20090907
Toshiba Ltd
Online:
Alemi P.
,
Lee D. C.
2014
“Comparative analysis of power losses for threelevel Ttype and NPC PWM inverters,”
Transactions of Korean Institute of Power Electronics(KIPE)
19
(2)
173 
183
DOI : 10.6113/TKPE.2014.19.2.173
Kim M. K.
,
Woo D. G.
,
Lee B. K.
,
Kim N. J.
,
Kim J. S.
2014
“Loss analysis of power conversion equipment for efficiency improvement,”
Transactions of Korean Institute of Power Electronics(KIPE)
19
(1)
80 
90
DOI : 10.6113/TKPE.2014.19.1.80
Clotea L.
,
Forcos A.
“Power losses evaluation of two and threelevel NPC inverters considering drive applications,”
in Proc. OPTIM
2012
929 
934
Buschendorf M.
,
Weber J.
,
Bernet S.
“Comparison of IGCT and IGBT for the use in the modular multilevel converter for HVDC applications,”
in Proc. 9th Int. MultiConf. SSD
2012
1 
6
Zhou D.
,
Balabjerg F.
,
Lau M.
,
Tonnes M.
2013
“Thermal cycling overview of multimegawatt twolevel wind power converter at full grid code operation,”
IEEJ Trans. Ind. Appl.
2
(4)
173 
182
Ma K.
,
Blaabjerg F.
2012
“Thermal optimized modulation methods of threelevel neutralpointclamped inverter for 10MW MW wind turbines under lowvoltage ride through,”
IET Power Electron.
5
(6)
920 
927
DOI : 10.1049/ietpel.2011.0446
Sintamarean C.
,
Blaabjerg F.
,
Wang H.
“A novel electrothermal model for wide bandgap WBGsemiconductor based devices,”
IEEE European Conference on Power Electronics and Applications
EPE  ECCE Europe
2013
2007
ABB Application Note: ‘Applying IGBTs’
Wang H.
,
Khambadkone A. M.
,
Yu X.
2010
“Control of parallel connected power converters for low voltage microgrid – Part II: Dynamic electrothermal modeling,”
IEEE Trans. Power Electron.
25
(12)
2971 
2980
DOI : 10.1109/TPEL.2010.2087394
Rockhill A.
,
Liserre M.
,
Teodorescu R.
,
Rodriguez P.
2011
“Gridfilter design for a multi megawatt mediumvoltage voltagesource inverter,”
IEEE Trans. Ind. Electron.
58
(4)
1205 
1217
DOI : 10.1109/TIE.2010.2087293