Power Electronic Converters in Wind Energy Systems - Considerations of Reliability and Strategies for Increasing Availability

ibexforeheadElectronics - Devices

Nov 24, 2013 (3 years and 6 months ago)


Power Electronic Converters in Wind Energy Systems -
Considerations of Reliability
and Strategies for Increasing Availability
Matthias Boettcher and Friedrich W.Fuchs
Kaiserstrasse 2
D-24143 Kiel,Germany
Phone:+49 (0) 431-880 6105
Fax:+49 (0) 431-880 6103
This work has been done in CEwind e.G.Competence Centre for Wind Energy of the Universities of
Schleswig-Holstein.It has been financed by the European Union and the State of Schleswig-Holstein.
<<Wind turbine>>,<<Power electronic converter>>,<<Reliability>>,<<Availability>>,
<<Condition monitoring>>,<<Prognosis>>,<<Fault tolerance>>.
This paper points out the need for increasing availability of the power electronic converter in offshore
wind turbines by means of statistical data.Therefor possible strategies such as condition monitoring,
prognosis,and fault tolerance are proposed and discussed by taking into consideration state of the art
and feasibility.Fault tolerant operation is evaluated to be the most efficient way to increase availability
of the converter.One fault tolerant converter topology is proposed.
Reliability is one of the key aspects regarding wind energy systems.The high availability of today’s on-
shore wind turbines is mainly due to periodical maintenance and quick repair in case of a fault.But wind
turbines are more frequently installed in areas with reduced accessibility,especially offshore.Hence
failures may cause turbine breakdowns for a longer period of time if they are not rectified quickly.As
a consequence the availability is reduced.The power electronic converter is a component of high sig-
nificance in this context because it has a relevant failure rate.Thus it can be expected that the influence
of the converter besides other components on the decrease of availability,when going from onshore to
offshore wind turbines,is eminent.This problemis described in different publications such as in [1],[2]
for instance,but only few propose possible strategies are proposed to overcome it.
In this paper strategies to increase the availability of the power electronic system such as condition
monitoring,prognosis,redundancy,and fault tolerance are discussed.In the subsequent chapter statistical
failure data are analized in order to show the need for increasing the availability of the power electronic
converter in offshore wind turbines.Afterwards different strategies to achieve this aim are presented.
Condition monitoring and prognosis applied to the power semiconductors and the DC link capacitors,
which turn out to be weak points of the converter,are outlined and fault tolerant operation is covered in
more detail.Finally a conclusion sums up the most important results.

speed concept
Simple Danish concept
Advanced Danish concept
Wind turbine population of WMEP

Simple Danish concept
Advanced Danish concept
speed concept
Wind turbine population in Germany:
Figure 1:Distribution of wind turbine concepts in the WMEP database (left) and distribution of wind turbine
concepts in Germany in 2010 (right) [3].
Reliability of Power Electronic Converters in Wind Energy Systems
Sources of Statistical Data
Reliability is one of the biggest concerns regarding wind turbines.For increasing the availability by
taking adequate measures identification and analysis of the existing problems are required.Therefor
failure statistics as detailed as possible are necessary.Different databases that are pursuing this target
are available.Furthermore there are different publications with investigations of these failure statistics.
In [1] and [4] data from Windstats [5] and Landwirtschaftskammer Schleswig-Holstein (LWK) [6] are
analysed.Windstats contains failure data of approximately 4000 onshore wind turbines in Germany and
approximately 1000 in Denmark over a period of 11 years.The merging of data for the entire population
of wind turbines is one significant disadvantage of Windstats since it does not allow to differentiate
important aspects such as power size and turbine concept.In contrast the database from LWK allows
differentiation,but includes failure reports of only approximately 350-650 wind turbines in Germany.
[7] studies two databases with statistics of Swedish wind turbines during 1997-2005,also with respect
to power size and operational year.Further sources with statistics of smaller coverage are listed in [3].
Another project collects data for state-of-the-art wind turbines in a national German research program
with the aim of increasing availability [8].Finally a large database has been accumulated in the period
from1989 to 2006 within the scientific measurement and evaluation program(WMEP) [9].The database
of WMEP is used here as source of information for several reasons:
 It contains approximately 64000 reports of maintenance and repair of about 1500 wind turbines
during the aforesaid period of 17 years.Each turbine has been monitored during a period of at
least 10 years.Therewith the WMEP is the most comprehensive study of the long-termbehaviour
of wind turbines worldwide [9].Thus it delivers a high degree of representativeness.
 The WMEP database allows differentiation of aspects that are important for analysis with regard
to the power electronic converter.One of these important aspects is the wind turbine concept.
Fig.1 shows the distribution of concepts of those wind turbines that are comprised in the WMEP
database as well as the distribution of concepts of German wind turbines in 2010 in comparison
[3].Concerning the power electronic converter it is important to account for the rate of variable-
speed concepts.This is due to the fact that in the Danish concept,which is widely used in early
wind turbine systems,the generator is directly connected to the grid und thus no power electronic
converter is used [10].
 Based on the WMEP database a deeper reliability study of the electrical parts is available in [3],
whereas also a reliability analysis of the power electronic converter is carried out for the first time.
Failure Causes of Wind Energy Systems
Before going into detail regarding the power electronic converter this section treats reliability aspects of
the entire wind turbine with the aim to show the significance of the electrical components concerning
reliability.On the left side of Fig.2 the distribution of general failure causes of all wind turbines in the
WMEP database is depicted [9].Component failures emerge to be the most frequent failure causes with
more than 35 % followed by malfunctions of the control system with less than 25 %.Failures due to
external impacts such as breakdown of mains voltage or stormonly appear with a ratio of approximately
19 %in sum.


1MW ≤ P < 1.5MW
P ≥ 1.5MW
Figure 2:Distribution of failure causes of all wind turbines (left) and distribution of affected components of wind
turbines of two power groups (right) in the WMEP database during 1997-2005 [9].
The right side of Fig.2 shows the distribution of components that are affected by failures,whereas only
two power groups of turbines in the WMEP database are considered.It has to be mentioned that this
diagram is related to all failures,not only to component failures.But as observed from the left diagram
of Fig.2 this is the main part.The electrical system,that also the power electronic converter belongs to,
turns out to be the most frequently affected part of the wind turbine.A big difference concerning this
matter is obviously the power size of the turbine.In a power range between 1 MWand 1.5 MWthe share
of failures in the electrical system is only somewhat higher than 20 % whereas for power sizes greater
than 1.5 MWthe share is almost at 35 %.This big difference can be ascribed to two main reasons.At first
it can be stated that the reliability of the electrical system decreases with turbine size since the number
of electrical components increases.Secondly the turbine concept has to be taken into consideration.
In the power range from 1 MW to 1.5 MW the share of turbines in the database with Danish concept
is prevalent.In contrast variable-speed concepts are predominant in the database for wind turbines of
power size greater than 1.5 MW.Adiagramof affected components with a classification of turbines with
Danish concept and variable-speed concept would lead to similar results.Consequently the important
role of the power electronic converter in regard to reliability is revealed indirectly by the right diagram
in Fig.2 as power electronics is a major part of the electrical system.
Failure Causes of Power Electronic Converters in Wind Energy Systems
As already worked out in the previous section the power electronic converter is one of the weak points of
the electrical system in context with reliability issues.In [3] the share of the power electronic converter
in the electrical components that are affected by failures is amounted to approximately 15 %.But this
value is yielded by considering the entire population of wind turbines in the WMEP database.Regarding
state-of-the-art wind turbines,which mostly exhibit a power electronic converter,on the one side a much
higher value can be expected,on the other side reliability has been improved.
Statistical data about failures in power electronic converters are very limited.In [3] the WMEP database
is analysed for the first time in terms of the subassemblies of the converter.The study turns out not to
be very representative due to an insufficient data and service acquisition [3].Nevertheless it could be
concluded that half of all failures in the converter can be led back to failures in semiconductors.But
it should also be mentioned that the study is based on data during 1997-2005 and reliability has been
improved since then.For getting further information about reliability of subassemblies in the power
electronic converter other sources,that are not specially related to wind energy,are helpful,although
power cycling of the subassemblies such as power semiconductors is different in other applications in
comparison to those in wind turbines.According to a survey based on 200 products from 80 companies
failures in the converter are 30 % due to capacitors,26 % due to PCBs,21 % due to semiconductors,
and 13 % due to solder [11].Another industry-based survey is given in [12].Here the power device
turns out to be the most fragile component of the converter with a share of 40 %followed by capacitors
with a share of 26 % and the gate drives with a share of 24 %.Finally it can be concluded that it is not
possible to give stiff representative values for the share of affected subassemblies in the power electronic
converter in wind turbines.Nevertheless it can clearly be stated that the power semiconductors as well

Support & housing
Electrical system
Mechanical brakes
Number of failures per year
Yaw system
Rotor hub
Control unit
Rotor blades
Drive train

Downtime per failure [days]
Figure 3:Number of failures per year (left) and downtime per failure of affected components of all wind turbines
in the WMEP database during 1997-2005 [9].
as the DC link capacitors are weak points in the converter and both together may be responsible for more
than half of all failures in power electronic converters,independently fromthe application.
Importance of Power Electronic Converter Availability in Offshore Applications
This section aims to show the significance of the power electronic converter availability in regard to
offshore wind turbines.Availability of a system describes the probability to find it in proper service
condition at any point of time [13].Mathematically it can be expressed in terms of mean time between
failures (MTBF) and mean time to repair (MTTR):
A =
The MTBF is the basic criterion for practical evaluation of systemreliability.It is the reciprocal value of
the failure rate of a systemor device and describes the average time of proper operation between failures.
The MTTR is the average time for eliminating a failure and putting the systemback in operation [13].
It can be directly concluded from the formula that a high availability is yielded by a long MTBF and
a short MTTR.The high technical availability of today’s onshore wind turbines of approximately 95-
99 % [3] is particularly a result of a frequent service and fast repair in case of a fault [7] and thus of
a comparatively short MTTR.But in contrast this short MTTR cannot be maintained for offshore wind
turbines due to a worse accessibility.The availability of today’s offshore wind turbines is not sufficient
and improvement is absolutely necessary [2].Since the MTTR is an adjusting parameter which is limited
in regard to offshore,this has to be accomplished by increasing the MTBF in order to get values of
availability in a range comparable to those of onshore turbines.This means that improvement of the
reliability is necessary.
The importance of the reliability of the power electronic system arises from Fig.3,where the number
of failures per year as well as the downtime per failure of affected components of all wind turbines in
the WMEP database are depicted [9].It should be noted that this diagramis almost completely based on
onshore wind turbines.It becomes clear that the electrical systemexhibits the biggest number of failures,
but entails only a comparatively short downtime per failure.This is a result of a fast service in case of a
fault.Thus a theoretical step fromonshore to offshore turbines leads to the conclusion that the electrical
system is the component that would be responsible for the biggest increase of downtime per failure.
Especially the power electronic converter is a key component of the electrical system in the context of
reliability,not only because of the share in the number of failures,as shown in previous sections as well,
but also because it is essential for maintaining operation of the turbine.
Finally it can be ascertained that taking adequate measures in order to increase the availability of the
power electronic converter is reasonable,especially in offshore wind energy systems.But also for on-
shore wind turbines it might be sensible since power size increases and good accessibility is not always
ensured,even for onshore application.Moreover the power electronic converter only takes a small pro-
portion of the overall costs of a wind turbine on the one hand,but on the other hand it can cause high
financial losses in case of a fault,when operation of the turbine is interrupted.Fromthis point additional
measures would be justifiable.Furthermore increasing the availability of the converter would open new

failure rate
infancy period
 early failures &
extrinsic failures
useful life period
 extrinsic failures
wear-out period
 deterioration failures
& extrinsic failures
Figure 4:Bathtub Curve [14]:Hypothetical failure rate versus time.
possibilities to increase the overall availability of the turbine.For example direct drives of today do not
appear to have a lower failure rate than indirect drives,although they eliminate the gearbox.This is
because of a higher failure rate of the electronic subassemblies [1].But improvement of their availability
might lead to direct drive concepts with a higher overall availability.
Strategies for Increasing Availability of Power Electronic Converters
In the previous section the need for increasing the availability of the power electronic converter is
emerged.There are different strategies thinkable to achieve this aim.In this paper it is concentrated
on measures to handle reliability problems of the power semiconductors and the DC link capacitors
since they turn out to be the most fragile components of the power electronic converter as worked out be-
fore.But at first it is necessary to classify different kinds of failures in general because possible measures
are closely connected to the type of failure.
Classification of Failures
Different classifications of failures of electronic devices are reasonable.One prevalent approach is to
differentiate them in terms of their completed service life.The established bathtub curve [14] in Fig.4
drafts the three fractions of time with occurrence of the main failure types:
 The infancy period is characerized by a decreasing failure rate that starts at a comparable high
initial value.This is mainly due to early failures such as fabrication defects which only appear at
the beginning of service life.
 In the useful life period there is an ideally constant failure rate.This is due to extrinsic failures
which appear spontaneously and are independent from operation time and thus can also occur
in the infancy period and in the wear-out period.They are a consequence of overstress such as
overvoltage or overcurrent in an electronic device.
 The wear-out period is the final period in service life.Due to intrinsic causes such as deterioration
and fatigue in electronic components for instance intrinsic failures arise predominantly.
Survey of Basic Strategies for Increasing Availability
In the following different strategies in order to increase the availability are proposed and explained in
terms of the previously given kinds of failures.Since the strategies are valid in general the terms ’sys-
tem’ and ’component’ are used,whereas ’system’ can be perceived as power electronic converter and
’component’ as power semiconductor or DC link capacitor.
 Improvement of the reliability of a component:Naturally the availability of the component and
thus of the systemwould increase by improving the reliability.The number of early failures could
be reduced by an improved quality control.Deterioration failures could be retarded by extension
of their operating life.Also extrinsic failures could be reduced by increasing the resistance against
different kinds of overstress.
 Condition monitoring applied to the system:The health status of the system is measured in real-
time in order to detect potential overstress on the components.Protection of the component against
thermal overstress could be avoided by real-time measurement of the temperature for instance.
Thus condition monitoring applied to the system would help to decrease the number of extrinsic
failures in the components.
 Condition monitoring applied to the components:The condition of a component is measured in
real-time in such a way that appropriate measures can be taken if it drifts away from the healthy
condition [15].Dependent on the time between indications of a failure and its occurrence early
detection of failures either enables optimization of maintenance strategies or preventive measures,
such as operation stop or fault tolerant operation,before appearance of consequential damage.
Thus condition monitoring applied to the components is related to the number of deterioration
failures in the wear-out period of the components.
 Prognosis applied to the components:The current health of a system or a component is assessed
in order to predict the health at some point in the future [15].This enables the possibility to
predict the remaining useful life until the component reaches the wear-out period.This means that
a period of a higher failure probability due to deterioration failures is entered,but an exact point in
time for the occurrence of a failure cannot be predicted.Often condition monitoring of parameters
that record the stress on the components such as the temperature is needed additionally since the
operating life depends on power cycling.
 Redundancy and fault tolerance:The termfault tolerance embodies the feature of a systemto keep
up operation even in case of a failure.This is possible for example by extending the system with
redundancy so that in case of a failure the affected component can be isolated and the redundant
component takes over the operation.Fault tolerant measures do not aim to reduce the number
of component failures in the system as the previous mentioned strategies do.They rather take
them as a given event at some point in time and tolerate them in order to increase availability
of the system.This requires fault diagnosis for detection,localization,and identification of a
failure after its occurrence.In this case all kinds of failures that are shown in the bathtub curve
can be tolerated,provided that no consequential damage has appeared.This is a big advantage
in comparison to other strategies.If consequential damages are problematical in the application
and have to be avoided,it is also possible to connect fault tolerance with condition monitoring or
prognosis applied to the components,whereas only deterioration failures would be covered by this
In this paper some of these strategies are outlined or discussed in more detail in regard of the power
semiconductors and the DC link capacitors,whereas other strategies are skipped.Since reliability is a
key aspect of electronic devices development of new manufacturing techniques and materials is a big
field.Consideration of this issue concerning power semiconductors and DC link capacitors would go
beyond the scope of this paper,also condition monitoring of the power electronic converter.Condition
monitoring and prognosis applied to both kinds of components are big fields as well.The possibilities
are outlined by taking into consideration the state of the art and the feasibility.Redundancy and fault tol-
erance are proposed to be the most effective strategies to improve the availability of the power electronic
converter and are discussed in more detail.
Condition Monitoring and Prognosis Applied to DC Link Capacitors
There are two kinds of capacitors suitable for application as DC link capacitor in power electronic con-
verters.These are the aluminum electrolytic capacitor and the metallized polypropylene film capacitor.
The electrolytic capacitor is characterized by a high power density at a comparatively low price,but it
belongs to the weakest elements in a converter [16].In contrast the filmcapacitor offers a high reliability,
longer lifetimes and can withstand higher voltages and currents [17],but has a lower power density.
Electrolytic capacitors exhibit three aging indicators that can be detected offline.These are the capacity,
the equivalent series resistance and the dissipation factor [18].Condition monitoring can be applied by
real-time measurement of these parameters.So it is possible by definition of threshold values to assess
the starting point of critical health condition.Feasibility of implementing online detection depends from
the application [18].In regard to DC link capacitors in power electronic converters there are only few
strategies proposed in publications such as [19],[20],[21].Despite requirements such as standstill of the
generator or current injection for instance some of these strategies might be applicable to converters in
wind energy systems.But in parallel and series connection of capacitors in the DC link aging parameters
superpose each other and additional hardware to overcome this problemwould lead to increase of costs.
A cheaper but less reliable solution could be prognosis of the critical health condition which could be
based on prediction of thermal stress in the capacitors since temperature is the main aging accelerator.
Real-time measurement of the temperature could increase accuracy,but would also increase costs.
Although the filmcapacitor is very reliable due to its self-healing capability,it is not free of failures at all
[17].In contrast to electrolytic capacitors only the capacitance is an aging indicator,whereas strategies
for online detection can be the same.Besides others the main aging accelerator are current peaks and
thus the useful life period depends from the application [17].This has to be considered for application
of prognosis.
Condition Monitoring and Prognosis Applied to Power Semiconductors
In order to meet the demands that are made on wind turbines switches with turn-off ability are almost
mandatory in the power electronic converter [22].For power conversion in turbines with low voltage
drives mostly low-voltage insulated-gate bipolar transistor (IGBT) modules,typically 1700Vdevices for
690V rated voltage are used [23].Due to increase in power size of the turbines low-voltage converters
will no longer be convenient.Medium voltage converters based on three-level neutral point clamped
(NPC) using either IGBTs oder integrated gate-commutated thyristors (IGCT) are proposed to be appro-
priate and already used in practice partially [23].The IGBT can either be in module technology such
as the HiPak module family or in press-pack technology similar to the IGCT,whereas the press-pack
design is well suited for three-level converters [22].Reliability of press-pack devices is much higher
compared to power modules,since bond wires do not exist and double-sided cooling allows to lower the
thermal impedance and thus a better tolerance to thermal cycling is achieved [15].Power modules are
more susceptible in regard to reliability.Thus condition monitoring and prognosis is reasonable to aim
for.Both applied to power semiconductor devices is challenging and at a more embryonic stage,but this
situation will rapidly improve since progress due to a lot of research is made [15].In the following the
possibilities concerning IGBT modules are outlined.
The failure mechanisms of power modules can be either associated with the chip or the packaging,
whereas the former is irrelevant to condition monitoring since they are mainly extrinsic due to overstress
and cannot be predicted [15].The main driving force of package related failures is the mismatch in
the coefficients of thermal expansion of the different materials used in the module.Besides others two
main failure mechanisms result from this,namely bond-wire liftoff and solder fatigue [15].The most
promising approaches to discover these mechanisms are based on detecting the collector-emitter satu-
ration voltage V
and the internal thermal resistance R
.An increase of V
could be used as an
indicator for bond-wire liftoff,whereas an increase of R
could be interpreted as solder fatigue [15].
The main challenge of practical realization is the detection of these parameters with a sufficient accu-
racy,either model based or by measurement with sensors.Also correct interpretation of the parameters is
challenging due to interactions between the failure mechanisms and due to the influence of temperature.
Furthermore it has to be considered that the permanent progress of module technology influences the
failure mechanisms and thus the possibilities concerning condition monitoring.
Prediction of the remaining useful life of electronic components such as IGBTs is one of the biggest
challenges of prognosis technique [24].For lifetime estimation of IGBT modules there are many dif-
ferent approaches based on analytical models and experimental results [25].But despite incorporation
of module design,thermomechanical characteristics,knowledge of failure mechanisms,the influence
of power cycling and experimental results for parameterization the accuracy of these models is not yet
Redundancy and Fault Tolerance
Wind turbines of today grow in power size,especially concerning offshore applications,and there is a
trend towards using synchronous generators with fully rated power converters [26],also due to reliability
issues.Operation in low voltage is often handled by several two-level converters in parallel in order to
reach the power size.The rated output voltage of usually 690V entails high currents and hence increase
of size and costs for cable and filtering.Furthermore because of the high voltage drop in the cables the
power converter has to be installed close to the generator in the nacelle which requires a large fraction
of the available space and increases the total weight of the nacelle [23].For offshore applications the
power density of these lowvoltage converters is not convenient since weight and space are of paramount
importance [23].Operation on medium voltage appears to be very attractive since it leads to higher
power density and lower currents and thus reduction of size and costs for cable and filtering as well as
for voltage step-up equipment required for connection at the point of common coupling [26].Medium
voltage operation can be achieved by using multi-level converters.The three-level back-to-back converter




Figure 5:Three-level NPC converter (left) and proposed fault tolerant converter topology (right).
seems to be the most appropriate concept for this purpose [26],[23].The following considerations of
redundancy and fault tolerance are mainly related to this converter topology.
There are different possibilities to accomplish redundancy in a power electronic converter,for single
components,for half-brigdes,or for entire converter modules for instance.Regarding single compo-
nents,such as power semiconductor devices and DC link capacitors,connection in parallel or in series is
possible.Series connection of components could remedy short-circuit failures of the device whereas par-
allel connection could resolve open-circuit failures.Implementation of both per component would lead
to dramatically improvement of availability of the converter [27],but also to heavy increase of costs.If
press-pack technology is used for the power semiconductors,the advantage of going into short-circuit in
case of a failure allows to use series connection only and thus to reduce the costs.The more prevalent
way to provide redundancy arises fromthe low voltage concept of connecting several converter modules
in parallel in order to cope with the power size of the turbine [23].In case of a failure the affected
converter module can be isolated and operation can be maintained with a part of rated power or,if the
modules are oversized,still with full rated power [28].Another advantage is the possibility of harmonics
elemination by synchronizing and shifting the PWM-switching [28].Concerning medium voltage con-
verters based on three-level NPCs parallel connection of converter modules increases the availability in
the same way.But due to the higher number of switching devices in a converter module the probability
of failure occurrence and thus of required module deactivation is higher.Furthermore each deactivation
means a disuse of a higher number of components in comparison to a two-level converter.
This leads to the idea to equip each converter module itself with additional circuitry for fault tolerant
operation.Different fault tolerant topologies based on the three-level NPC are proposed in literature
that vary in regard to important aspects such as costs and effort,type and number of tolerated failures,
modulation index in fault operation,and dimensioning of the power devices [29],[30].In fault tolerant
NPC topologies with three legs either only half modulation index is possible after a fault or the power
semiconductors have to be dimensioned to withstand full DC link voltage [29].The former case disables
grid-connected applications and in the latter the three-level NPC would lose one of its main advantages.
Topologies with four legs are better suited since they allow operation with full modulation index after
a fault without oversizing the semiconductors.One topology is proposed in Fig.5 on the right side,
whereas on the left side the three-level NPC is depicted.In the proposed topology,which is similar
to that proposed in [30],the basic NPC is extended by an additional flying capacitor leg.In normal
operation mode,where S
and S
are switched off and S
and S
are switched on,the fourth leg is able
to provide a stiff neutral point voltage.This is a big advantage since the residual NPC does not have
to balance the DC link voltage,which is an important concern in regard to NPCs,and thus modulation
techniques can be optimized in terms of minimizing harmonics or switching losses for instance [30].
In case of a failure the affected leg can be substituted by the fourth leg.When the fault diagnostic
system has detected a failure in phase x,either short-circuit or open-circuit,the reconfiguration process
starts with switching off the power semiconductors and proceeds with switching off contactor R
switching on R
.Afterwards S
and S
are switched on in order to provide the neutral point voltage and
fault operation mode can start.In this mode the converter works almost like a NPC,whereas it must be
avoided that switches S
and S
are in on-state at the same time.
Finally it can be stated that this fault tolerant topology nearly provides the same degree of redundancy
compared to parallel connection of two NPCconverter modules.But in contrast a fewer number of power
semiconductor devices is required,thus size and costs are less,and additional advantages in normal op-
eration mode are achieved.So by parallel connection of fault tolerant converter modules the number
of modules could be halved,whereas the degree of redundancy and thus the availability could be main-
tained.Achallenge is the process of fault diagnosis and reconfiguration with avoidance of consequential
damage in the converter.In regard to the affected leg consequential damage can be tolerated since this
leg is completely substituted.Probably the best possible way to overcome the problem of short time
between failure occurrence and heavy damage is combining fault tolerant operation with early detection
of faults by means of condition monitoring.
The need for increasing the availability of the power electronic system in offshore wind turbines is
pointed out by means of statistical failure data from the WMEP database.Power semiconductors and
DC link capacitors turn out to be the weakest points in the converter.Condition monitoring,prognosis,
redundancy and fault tolerant operation are discussed as possible strategies in order to achieve a higher
availability.In regard to the power semiconductors condition monitoring and prognosis are only related
to certain failures in the wear-out period,whereas state of the art methods are still unreliable and inaccu-
rate.Redundancy and fault tolerance turn out to be more effective strategies,since all kinds of failures
are covered.A fault tolerant converter topology based on the three-level NPC is proposed to be suited in
wind energy systems with advantages in comparison to parallel connection of converter modules.Con-
cerning the DC link capacitors condition monitoring appears to be the most effective strategy,although
improvement of methods is necessary,especially regarding parallel and series connection of capacitors.
[1] P.J.Tavner,G.J.W.van Bussel,and F.Spinato,“Machine and Converter Reliabilities in Wind Turbines,”
IET International Conference on Power Electronics,Machines and Drives,pp.127–130,March 2006.
[2] G.J.W.van Bussel,“Offshore Wind Energy,the Reliability Dilemma,” World Wind Energy Conference,
Berlin,Germany,July 2002.
[3] P.Lyding,S.Faulstich,B.Hahn,and P.Tavner,“Reliability of the Electrical Parts of Wind Energy Systems -
a Statistical Evaluation of Practical Experiences,” EPE Wind Energy Chapter,Staffordshire,UK,April 2010.
[4] F.Spinato,P.J.Tavner,G.J.W.van Bussel,and E.Koutoulakos,“Reliability of Wind Turbine Subassem-
blies,” IET Renewable Power Generation,vol.3,no.4,pp.387–401,December 2009.
[5] Windstats,available:http://www.windstats.com.
[6] Landwirtschaftskammer,Schleswig-Holstein,Germany,available:http://lwksh.de/cms/index.php?id=2878.
[7] J.Ribrant and L.M.Bertling,“Survey of Failures in Wind Power Systems with Focus on Swedish Wind
Power Plants during 1997-2005,” IEEE Transactions on Energy Conversion,vol.22,no.1,pp.167–173,
March 2007.
[8] “Erh¨ohung der Verf¨ugbarkeit von Windenergieanlagen (EVW),” German Federal for the Environment,Na-
ture Conversation and Nuclear Safety,available:http://www.evw-wind.de/.
[9] S.Faulstich,M.Durstewitz,B.Hahn,K.Knorr,and K.Rohrig,“Windenergie Report Deutschland 2008,”
Institut f¨ur Solare Energieversorgungstechnik (ISET),Kassel,2008.
[10] Z.Chen,J.M.Guerrero,and F.Blaabjerg,“A Review of the State of the Art of Power Electronics for Wind
Turbines,” IEEE Transactions on Power Electronics,vol.24,no.8,pp.1859–1875,August 2009.
[11] E.Wolfgang,“Examples for Failures in Power Electronics Systems,” ECPE Tutorial ”Reliability of Power
Electronic Systems”,April 2007.
[12] S.Yang,A.Bryant,P.Mawby,D.Xiang,L.Ran,and P.Tavner,“An Industry-Based Survey of Reliability
in Power Electronic Converters,” IEEE Energy Conversion Congress and Exposition ECCE,pp.3151–3157,
September 2009.
[13] P.Wikstr¨om,L.A.Terens,and H.Kobi,“Reliability,Availability,and Maintainability of High-Power
Variable-Speed Drive Systems,” IEEE Transactions on Industry Applications,vol.36,no.1,pp.231–241,
January/February 2000.
[14] G.-A.Klutke,P.C.Kiessler,and M.A.Wortman,“ACritical Look at the Bathtub Curve,” IEEE Transactions
on Reliability,vol.52,no.1,pp.125–129,March 2003.
[15] S.Yang,D.Xiang,A.Bryant,P.Mawby,L.Ran,and P.Tavner,“Condition Monitoring for Device Reliability
in Power Electronic Converters:A Review,” IEEE Transactions on Power Electronics,vol.25,no.11,pp.
2734–2752,November 2010.
[16] K.Abdennadher,P.Venet,G.Rojat,J.-M.R´etif,and C.Rosset,“A Real-Time Predictive-Maintenance Sys-
tem of Aluminum Electrolytic Capacitors Used in Uninterrupted Power Supplies,” IEEE Transactions on
Industry Applications,vol.46,no.4,pp.1644–1652,July/August 2010.
[17] G.M.Buiatti,J.A.Mart´ın-Ramos,A.M.R.Amaral,P.Dworakowski,and A.J.M.Cardoso,“Condition
Monitoring of Metallized Polypropylene Film Capacitors in Railway Power Trains,” IEEE Transactions on
Instrumentation and Measurement,vol.58,no.10,pp.3796–3805,October 2009.
[18] A.M.R.Amaral and A.J.M.Cardoso,“A Simple Offline Technique for Evaluating the Condition of
Aluminum-Electrolytic-Capacitors,” IEEE Transactions on Industrial Electronics,vol.56,no.8,pp.3230–
3237,August 2009.
[19] M.L.Gasperi,“Life Prediction Modeling of Bus Capacitors in ACVariable-Frequency Drives,” IEEE Trans-
actions on Industry Applications,vol.41,no.6,pp.1430–1435,November/December 2005.
[20] K.-W.Lee,M.Kim,J.Yoon,S.B.Lee,and J.-Y.Yoo,“Condition Monitoring of DC-Link Electrolytic
Capacitors in Adjustable-Speed Drives,” IEEE Transactions on Industry Applications,vol.44,no.5,pp.
1606–1613,September/October 2008.
[21] A.G.Abo-Khalil and D.-C.Lee,“DC-Link Capacitance Estimation in AC/DC/AC PWMConverters Using
Voltage Injection,” IEEE Transactions on Industry Applications,vol.44,no.5,pp.1631–1637,Septem-
ber/October 2008.
[22] B.Backlund,M.Rahimo,S.Klaka,and J.Siefken,“Topologies,Voltage Ratings and State of the Art High
Power Semiconductor Devices for MediumVoltage Wind Energy Conversion,” IEEE Power Electronics and
Machines in Wind Applications PEMWA,June 2009.
[23] M.Liserre,R.C´ardenas,M.Molinas,and J.Rodr´ıguez,“Overview of Multi-MWWind Turbines and Wind
Parks,” IEEE Transactions on Industrial Electronics,vol.58,no.4,pp.1081–1095,April 2011.
[24] B.Saha,J.R.Celaya,and K.F.Goebel,“Towards Prognostics for Electronics Components,” IEEEAerospace
Conference,pp.1–7,March 2009.
[25] I.F.Kovacevic,U.Drofenik,and J.W.Kolar,“New Physical Model for Lifetime Estimation of Power
Modules,” International Power Electronics Conference IPEC,pp.2106–2114,June 2010.
[26] S.Kouro,M.Malinowski,K.Gopakumar,J.Pou,L.G.Franquelo,B.Wu,J.Rodriguez,M.A.Perez,and
J.I.Leon,“Recent Advances and Industrial Applications of Multilevel Converters,” IEEE Transactions on
Industrial Electronics,vol.57,no.8,pp.2553–2580,August 2010.
[27] A.L.Julian and G.Oriti,“A Comparison of Redundant Inverter Topologies to Improve Voltage Source
Inverter Reliability,” IEEE Transactions on Industry Applications,vol.43,no.5,pp.1371–1378,Septem-
ber/October 2007.
[28] J.Birk and B.Andresen,“Parallel-Connected Converters for Optimizing Efficiency,Reliability and Grid
Harmonics in a Wind Turbine,” European Conference on Power Electronics and Applications,September
[29] P.Lezana,J.Pou,T.A.Meynard,J.Rodriguez,S.Ceballos,and F.Richardeau,“Survey on Fault Operation
on Multilevel Inverters,” IEEE Transactions on Industrial Electronics,vol.57,no.7,pp.2207–2218,July
[30] S.Ceballos,J.Pou,E.Robles,I.Gabiola,J.Zaragoza,V.J.L,and D.Boroyevich,“Three-Level Converter
Topologies with Switch Breakdown Fault-Tolerance Capability,” IEEE Transactions on Industrial Electron-
ics,vol.55,no.3,pp.982–995,March 2008.