Asset Management Strategies For Power Electronic Converters in Transmission Networks: Application To HVDC and FACTS Devices

Received February 21, 2018, accepted April 5, 2018, date of publication April 12, 2018, date of current version
May 2, 2018.
Digital Object Identifier 10.1109/ACCESS.2018.2826360
Asset Management Strategies for Power

Electronic Converters in Transmission Networks:
Application to Hvdc and FACTS Devices
TIBIN JOSEPH 1 , (Member, IEEE), CARLOS E. UGALDE-LOO1 , (Member, IEEE),
JUN LIANG1 , (Senior Member, IEEE), AND PAUL F. COVENTRY2 , (Senior Member, IEEE)
1 School of Engineering, Cardiff University, Wales CF24 3AA, U.K.
2 Asset Policy, National Grid, National Grid House, Warwick CV34 6DA, U.K.
Corresponding author: Tibin Joseph (josepht@cardiff.ac.uk)
This work was supported in part by the People Program (Marie Curie Actions) of the European Union’s Seventh Framework
Program under Grant FP7/2007-2013, in part by the MEDOW Project under REA Grant 317221, and in part by the National Grid through
the NIA Project under Grant NIA_NGET0060.
ABSTRACT The urgency for an increased capacity boost bounded by enhanced reliability and sustainability
through operating cost reduction has become the major objective of electric utilities worldwide. Power
electronics have contributed to this goal for decades by providing additional flexibility and controllability to
the power systems. Among power electronic based assets, high-voltage dc (HVdc) transmission systems
and flexible ac transmission systems (FACTS) controllers have played a substantial role on sustainable
grid infrastructure. Recent advancements in power semiconductor devices, in particular in voltage source
converter based technology, have facilitated the widespread application of HVdc systems and FACTS devices
in transmission networks. Converters with larger power ratings and higher number of switches have been
increasingly deployed for bulk power transfer and large scale renewable integration—increasing the need
of managing power converter assets optimally and in an efficient way. To this end, this paper reviews the
state-of-the-art of asset management strategies in the power industry and indicates the research challenges
associated with the management of high power converter assets. Emphasis is made on the following
aspects: condition monitoring, maintenance policies, and ageing and failure mechanisms. Within this context,
the use of a physics-of-failure based assessment for the life-cycle management of power converter assets is
introduced and discussed.
INDEX TERMS Asset management, FACTS devices, HVdc, power electronic assets, physics-of-failure.
I. INTRODUCTION traditional transmission assets such as power line conductors,

Significant developments in power semiconductor devices, insulators, breakers and transformers have been well stud-
together with the need for interconnectors and increased ied and are common practice for utilities [1], the increased
transmission capability, have led to a revolution in power incorporation of power electronics into the existing grids has
electronics assets over the past decades. This transition has become an escalated management challenge [2], [3].
caused a paradigm shift from traditional power transmission Power electronics device installations in the grid date back
substation assets to substations incorporating power electron- to the early years of the 20th century with the invention of
ics. Out of these devices, high-voltage dc (HVdc) convert- the mercury-arc rectifier, marking the beginning of the ‘‘clas-
ers, particularly voltage source converter (VSC) based HVdc sical era’’ of power electronics [3], [4]. Power electronics
and flexible ac transmission systems (FACTS) devices like has emerged as a complex and multidisciplinary discipline
thyristor-controlled series compensators (TCSCs) and static following several decades of evolution. It embraces all areas
synchronous compensators (STATCOMs), are frontrunners in of the power system, thereby ushering in a new kind of
the transmission grid. These devices consist of power elec- industrial revolution due to its significant role in sustain-
tronics equipment such as converter valves, cooling systems able energy generation, renewable energy systems integra-
and auxiliary components. Although strategies for managing tion and bulk power transfer [3], [5], [6]. In recent years,
2169-3536
2018 IEEE. Translations and content mining are permitted for academic research only.
21084 Personal use is also permitted, but republication/redistribution requires IEEE permission. VOLUME 6, 2018
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
T. Joseph et al.: Asset Management Strategies for Power Electronic Converters in Transmission Networks
an increased number of power electronic assets has been

deployed to enable additional power flow capacity through
existing transmission assets to circumvent overhead trans-
mission expansion limitations. The proliferated application of
power electronics to support renewable generation, to boost
power transfer without building new transmission infrastruc-
ture, and to facilitate a sustainable evolution towards a low-
carbon future has made of these assets prominent members of
the power system industry. Therefore, revisited management
approaches are required [6], [15].
FIGURE 1. FEU breakdown for thyristor HVdc systems (2011-2012).
Significant investment has been taking place around the
world for the integration of non fossil fuel resources utilizing
power electronic assets, especially HVdc and FACTS, which
places new stress on existing equipment and interconnecting
systems [3], [6]. For instance, the US nearly doubled their
spending on network upgrade and expansion from $28 bil-
lion in 2010 to $44 billion in 2013 [7]. It is foreseen that
over $300 billion will be invested over the next 20 years
for refurbishment and deployment of transmission and dis-
tribution assets [7]. Similar scenarios are faced elsewhere.
For example, the net investment as per the UK government
obligation to meet low carbon network policies is expected
to reach £30-50 billion by 2030 as compared to the current FIGURE 2. Installation cost of an HVdc station.
value of less than £13 billion [7], [8].
Owing to the fact that in the future a major part of this
investment could be made on power flow control devices in the substation operation and maintenance can be found
like HVdc and FACTS, it is a matter of interest to look into in [17]–[20].
their effective management [8], [9]. The inherent capability On the other end of the power transmission spectrum, open
of these devices to control and stabilize networks has been market systems and deregulation have encouraged power util-
another aspect that renders their potential application in trans- ities to find optimal strategies for managing existing and new
mission systems. In addition, they aid to optimally utilize assets over their lifespan. This has caused asset management
existing assets—thereby increasing capital and operational to become one of the most prominent sectors [21], [22]. Gen-
savings [6]–[9]. However, the need for an effective asset eral aspects of asset management practices are well known
management is high; for instance in the UK the cost of a in power systems. Throughout the last 10 to 15 years, asset
new HVdc link of 2 GW capacity is around £1 billion and management has been adopted as the reigning paradigm for
the loss of revenue due to failure can cost £1 million per the management of electricity networks, contributing to the
day as a result of the loss in electricity transmission [7], [8]. understanding of the value of assets [3], [9]. This follows
Furthermore, a loss of 2 GW transmission capacity can poten- stringent standards like PAS 55 and ISO 55000, which pro-
tially jeopardize the UK network as it signifies a considerable vide clear definitions and requirements for establishing and
loss of total capacity (80 GW) [8]. These scenarios stem up verifying an optimal management for all kinds of physical
the ever-increasing demand from utilities around the world to assets related to the power industry [9], [30]. To increase
formulate efficient and optimal asset management practices the understanding of the effect of emerging capital assets
for ageing HVdc and FACTS assets [10]–[15]. on existing power networks, a roadmap of new research
It is evident that power converters are subjected to a high directions considering the power electronic assets should be
level of examination, monitoring and standardized report- identified [4]. Through this paper we are trying to bridge this
ing. For instance, CIGRE has produced biannual reports gap by addressing some of the research questions.
on HVdc converter station reliability and energy utiliza- Focus has been recently placed on the monitoring, main-
tion [16]. The impact that different components have on the tenance, and ageing and failure mechanisms, as reported
forced energy utilization (FEU) is depicted in Figure 1. Con- in [23]–[26] and references therein. Qiao et al. [23] (and cor-
verter valves contribute three fourths of the forced outages, responding references) have presented different monitoring
which emphasizes the need for their effective management. and health management approaches to improve the reliabil-
The cost breakdown of a new HVdc installation compared ity and lifespan of wind energy conversion systems. Simi-
to other components is shown in Figure 2. As it can be larly, state-of-the-art technologies and best practices in asset
observed, converter assets incur almost one fourth of the management for smart grid networks have been presented
cost [15]. Additional publications and guidelines for HVdc in [24]. However, due to the complexity of power converter
converter reliability based on power electronic components structures and their degradation mechanisms, interpreting the
VOLUME 6, 2018 21085

monitoring (CM) [26]–[29]. CM systems have evolved over

decades, with an existing HVdc and FACTS asset base span-
ning over 50 to 60 years. Traditional approaches like keep-
ing maintenance records and visual inspections are probably
the best sources of condition assessment. Fortunately, power
electronic substations are not much different than standard
electrical substations. This means that there is a substantial
FIGURE 3. Business needs for power electronic asset management. amount of experience in place already. However, there are
some special design characteristics that should be accounted
for which are not common for ac substations [15]. In general,
information obtained through existing techniques and sub- present CM systems are mainly focussed on ac substations
sequently evaluating their condition is a challenging task [27]–[30]; furthermore, the monitoring function is single and
(especially for non-experts). Moreover, current asset man- completely independent, protocols are not compatible and the
agement systems are mainly focussed on ac substations interface is not unified [31]–[36].
and, furthermore, there is lack of information, methods and To address these shortcomings and better satisfy the
understanding of the impact of new age power electronic needs of monitoring levels in power electronic substations,
devices [27]–[32]. Therefore, there is a need to investigate the design and implementation of CM and decision sup-
new techniques and their applications for analysing data port systems in HVdc and FACTS converter stations are
and identifying information which, in turn, may improve the required [34]–[36]. It is relevant to highlight that a recent
design of power converters in terms of reliability. Among industry-based survey on advancements in CM practices
these methods, the assessment through physics-of-failure and reported that more than 50% of the participants believed
degradation mechanisms constitute promising options. that monitoring techniques and assessment tools should be
Initial efforts on the application of the aforementioned improved [37]. The report also specified that semiconductor
techniques for power electronic device monitoring and diag- devices are often the weakest link in converters and comprise
nosis were reported in [25] and [33]. Nevertheless, such 30% of the total failures. Thus, it is evident that CM could
efforts are scattered over different aspects, such as the inter- play a crucial role in optimizing the operation of power
pretation of measurement data for diagnosis, maintenance electronic based assets. Therefore, in this section the CM and
and ageing. There is still lack of a common framework within assessment approaches used in power converter based assets
which new methods can be effectively deployed and executed are categorized (see Figure 4).
to perform a unified asset management approach in transmis-
sion networks. A. TRADITIONAL APPROACH
In this context, research efforts in this paper focus on pre- The traditional CM approach applied to combined ac and
senting the state-of-the-art of asset management strategies in dc substation components can be divided into five main
transmission networks. An emphasis is made on the following categories: visual inspection; monitoring the hot spot tem-
aspects: condition monitoring, maintenance policies, and the perature; assessment of the vibration of wall and winding;
estimation of ageing and failure mechanisms. In addition, sensor analysis for examining dissolved gases and partial
existing applications of physics-of-failure based diagnosis discharges; and the voltage and current signature measure-
and prognosis in power electronic assets condition assessment. To obtain meaningful information from these moni-
ment have been summarized. To provide value, these have tored parameters, data should be analyzed to assess the state
been linked to the transmission network asset management of the components [32]–[35]. Each CM data category can
regime, with a particular focus on prospective future research be evaluated using condition assessment (CA) techniques,
roadmaps. Following this line, power electronic assets can as illustrated in Figure 4. Each CM and CA technique will
be mainly characterized by the following business needs be discussed in the following subsections.
(see Figure 3):
1) Business Need 1: Trends in condition monitoring and 1) ACOUSTIC MEASUREMENT/VIBRATION MONITORING
assessment. Acoustic measurement and vibration analysis constitute the
2) Business Need 2: Strategies for effective maintenance traditionally used approaches for CM of rotating equip-
policies. ment. They have been recently employed in the power con-
3) Business Need 3: Estimating failure and ageing verter monitoring discipline due to the advancements in
mechanisms. the technology used for CA [30]. Acoustic and vibration
sensors have found a place in the market, with the for-
II. CONDITION MONITORING mer being attached to components to listen to their state
The technique or process for monitoring the operating char- using sound level meters, while the latter being mounted
acteristics of a component or an equipment, in such a to register local motion [38], [39]. The use of these mon-
way as to predict the need for maintenance before seri- itoring and diagnosis devices is well established in the
ous deterioration or breakdown, is known as condition wind turbine converter industry to estimate the health of a
21086 VOLUME 6, 2018

FIGURE 4. CM and assessment methods for power electronic assets.
device [39]. In a dc converter station this analysis has been Sensor or SCADA-based monitoring is reliable and one of the
mainly used to estimate the noises and vibrations generated most cost-effective CM methods since it relies on the interpre-
from filter capacitors, valve cooling systems, valve reactors, tation of data only. In a dc sub-station it collects information
converter transformers and switching devices like circuit from key components and converter assemblies using the
breakers [38]–[42]. Another important indicator for the noise sensors fitted to valve-based units, heat-sinks, protection and
resulting from dc magnetization are the currents arising due control systems and switches. The status of the valves can be
to the asymmetry in the firing of thyristor valves; this in turn interpreted from operational data and from the measurements
depends on the accuracy of the firing control system [38]. of signals, such as temperature, pressure, and current. This
Additional estimation methods for diagnosis include the use information reflects the real-time state of valve units and,
of root-mean-square value of the signals, frequency response by proper assessment, a relationship between different sig-
and amplitude, energy, skewness, crest factor, counts and nals can be observed—in turn providing information on the
events [39]–[44]. health of the converter [27], [46]. Many wind turbine methods
can be effectively used for dc substation CM following an
2) SIGNATURE ANALYSIS adequate implementation. Given that it was not designed for
Signature analysis (SA) is one of the easiest yet comprehen- CM purposes, the main shortfall of SCADA-based analysis is
sive CM method employed to predict electrical and mechan- that it cannot be applied systematically as the collected data
ical faults. The signals that can be detected may be either may not be comprehensive enough to represent the whole
electrical (voltage, current, power or flux) or mechanical. Var- dc substation. The variation of SCADA data subjected to
ious types of faults can be detected using SA methods, which operational conditions poses the difficulty of detecting an
are mainly categorized as: 1) model reference, or 2) feature incipient fault from raw data without an appropriate data
extraction methods [28], [38]. The model reference method analysis tool [48]. Furthermore, the data capture rate of 5
spots faults by comparing the results of measurements with to 10 min is too slow for most fault diagnoses of fast-acting
likelihoods of mathematical-based or artificial intelligence- power electronic assets as compared to other equipments such
based models. Conversely, the feature extraction method uses as circuit breakers or transformers [49].
frequency and time-domain signal processing techniques to
capture signatures representing normal and faulty perfor- 4) TEMPERATURE MEASUREMENT/THERMAL ANALYSIS
mances [28]. The application of SA in estimating the health Temperature measurement (TM) and thermal analysis (TA)
of thyristor valves has been reported in [43] and [44]. SA are among the most exhaustive ways for assessing the con-
can identify misbehavior of valves, such as inconsistency in dition of power electronic devices. TM aids in identifying
the conduction and firing sequences due to disturbances in potential failures related to temperature changes in the equip-
ac or dc systems [43]. Fuzzy logic seems to be suitable for the ment due to faults, overloading and deformations. Monitoring
implementation of the natural language definitions of these is performed using sensors, such as optical pyrometers, resis-
misbehaviors [44]. Estimation of the frequency energy lev- tant thermometers and thermocouples [50]–[53]. TA using
els using frequency spectra characteristics in dc switchyards infrared (IR) thermography has become one of the most
and valve halls has been applied to analyze the electric and reliable and prominent prognosis methods for asset health
magnetic field strength of HVdc converter valve units [45]. monitoring of dc systems [50]. This method covers a wide
range of components from HVdc converter stations, including
3) SENSOR/SCADA DATA ANALYSIS thyristor valves, diodes, snubber circuits, pumps and expan-
Supervisory control and data acquisition (SCADA) sys- sion tanks [51]. Temperature-based CM relies on the fact that
tems are collectively used in most transmission networks. all working components emit heat; thus when a component
VOLUME 6, 2018 21087

in the system malfunctions, its temperature increases beyond taken place within the electronics industry and has focussed
normal values [52]. IR temperature transmitters and high- on optimizing the reliability of complex circuits [62]. Con-
resolution IR cameras constitute the sensors applied in TA, tinuous improvement has reduced failure rates of IGBT mod-
with results typically interpreted visually. For example, an IR ules in traction applications from 1000 failures-in-time (FIT,
thermal image showing the temperature distribution of the the number of failures in one billion hours) in 1995 to 20 FIT
surface between thyristors and the radiator in an HVdc con- in 2000, and very few in recent years [63].
verter has been evaluated in [51] to estimate the temperature Despite the extensive efforts to improve reliability, failures
distribution among the upper and lower thyristors valves. in power electronics have been observed continuously [63].
The measurement of the thyristor valve temperature is kept Surveys indicate that IGBTs have been the most frequently
as the base line for estimating the overload capability and used devices (42%) among power semiconductors, followed
rating of redundant thyristors for a 12-pulse converter scheme by metal-oxide semiconductor field-effect transistors (MOS-
subjected to the worst case thermal and voltage stresses [53]. FETs) (27%), thyristors (14%) and PiN diodes (10%) [37].
Most of the work in recent years has concentrated on IGBTs
5) VISUAL INSPECTION for power electronic converters [63]. In contrast with tra-
One of the most challenging monitoring techniques is the ditional CM approaches, PoF based CM focusses on the
visual inspection (VI) method, which is based on human lifetime estimation of new generation power semiconductor
sensory capabilities. VI may be used as a complement to other devices such as IGBTs, IGCTs, MOSFETs, SiCs, among oth-
CM approaches. It includes touch (temperature and vibration ers. A threefold process is employed to monitor the strength
checking), deformation and aspects (DA), and detection of of power electronic components; this involves diagnosis,
sounds emitted by a functioning system. VI is generally used prognosis and condition analysis. Figure 5 illustrates a typical
to monitor thyristor valve cooling system electrodes, water life estimation flowchart for voltage-controlled converters in
level sensors, corrosion and cracking [54]. In several cases, wind turbine applications. The interdependency of different
this method can be very effective in identifying problems physical models and their relevance in PoF based CM are
that were not previously determined by other techniques. evident from the diagram.
Such cases includes loose parts, connections, terminals, and
components; visibly worn or broken parts; excessive tempera-
tures; and oil leakages, corrosion, and leaks on the valve water
piping [51]. Since VI is a labor intensive and highly subjective
task, the results will depend on the experience and judgement
of the inspector [55], [56].
Nowadays, developments in VI technologies have reached
the pace of the wind industry. For instance, remote VI tech-
nologies have been implemented to inspect gearboxes,
wind turbine blades and other critical components [57].
Recent developments include the use of gyroscopes to
inspect interior areas that may not be accessible otherwise—
aiming to reveal hairline cracks, corrosion, pitting and other FIGURE 5. PoF based life estimation flowchart for voltage-controlled
defects [58]. Flying remote devices are the latest addition to converters [66].
the VI family, which appear to be effective in reaching out
remote and offshore installations [59].
Some of the discussed monitoring methods are still in their 1) DIAGNOSIS
infancy or development stages, although a few are mature Diagnosis is used to estimate the root cause failure in a power
techniques. The transition from conventional CM approaches electronic asset upon occurrence of a fault [63]. Based on the
is inevitable due to the reliability-based design and physics- understanding of failure mechanisms, reliability models for
based failure mechanisms adopted for modern power con- these devices can be built to aid the failure analysis and life-
verters. Physics-of-failure based monitoring and analysis is time prediction without performing costly tests. In a diagnosis
discussed in the next subsection. process, monitoring the contributors for degradation is a vital
route to understand the physics behind the damage caused by
B. PHYSICS-OF-FAILURE BASED CM it [64]. Device parameters are major contributors for compo-
Ensuring a high availability and reliability is the top pri- nent degradation. These include the on-state voltage or resis-
ority for transmission network operators. To achieve this tance (VCEsat /RON ), the internal thermal resistance (Rth ),
through power electronic assets, a physics-of-failure (PoF) and the threshold gate voltage (VGEth ) [63]–[66]. Although
based assessment during the design stages has been widely these parameters are difficult to measure in practice, they
accepted. The combination of degradation calculation with may cause changes in the operational characteristics of the
root cause failure analyses and probabilistic methods forms devices or converters at system level. Substantial research has
the basis of PoF [60], [61]. The PoF approach has mostly been conducted in this area [65], [66]. However, most of the
21088 VOLUME 6, 2018

work has looked at a device level instead of system level— to degradation and thereby failure since they are dependent on
which needs considerable attention since health indicators the application and power module technology. For instance,
may vary depending on the application (e.g. HVdc, FACTS) the bond wire degradation in an IGBT or diode is often
and mission profile (e.g. offshore, onshore). For instance, determined by an increase of forward voltage Vfwd . However,
different types of degradation may influence the forward different proposals for a critical level of Vfwd have been
voltage drop in an IGBT, but a single degradation mechanism reported [75].
may influence several damage indicators. Therefore, the use
of a single parameter to monitor degradation is insufficient to 4) DISCUSSION
develop a complete CM system [65]. PoF based CM represents a new addition to existing monitor-
ing approaches. It has recently received attention in ’mission
2) PROGNOSIS critical’ applications such as offshore wind farms and modu-
The process of estimating the remaining useful life (RUL) by lar multilevel converter (MMC) based HVdc stations. Appli-
predicting the health conditions at some point in the future is cation of the PoF approach in wind turbine condition based
known as prognosis. This method has a significant effect in monitoring and maintenance has been addressed in [87].
the decision making process as it provides flexibility to deter- Following the threefold process as discussed in previous
mine the RUL while minimizing the risk of failure [33], [68]. sections (i.e. diagnosis, prognosis and condition analysis),
Prognostics of IGBTs received attention primarily from a continuous online estimation of the damage accumulation
mission-critical systems such as avionics, high-speed rail, can be performed using standard converter parameters. Pre-
and offshore wind turbines. In such systems, no failure is cise assessments may be then carried out in real time consid-
desirable until the next maintenance period [33]. The usual ering the probability of failure for specific failure modes and
prognoses used in power electronics assets are data-driven components. In turn, these can reduce the investment cost of
and model-based approaches [33]. The model-based prognos- additional measurement equipment.
tic approach relies on the physical models that describe the Reliability based design for emerging semiconductor
relationship between lifespan and design parameters, mate- devices in wind turbines and photovoltaics has resulted in the
rial properties, and loading conditions. The damage-based, advent of a PoF approach in power systems and power elec-
strain-based, and fatigue life models are widely employed tronic applications [90], [91]. Moreover, new technologies
prognosis approaches [70]. Model-based CM is particularly and devices which can effectively monitor, diagnose and even
attractive for devices working under variable operating con- automatically detect and rectify faults have been installed
ditions and varying environments, such as HVdc converters, and/or are under research [29]. There is an urgent need
onshore and offshore converter stations and renewable energy for a standard and unified approach for the CM of devices
systems. Conversely, the data-driven approach exploits com- used in the transmission assets, irrespective of them being
ponent historical data and health condition. These attributes ac or dc [91]. Multi-vendor devices should follow strict guide-
are extracted from the device information without relying lines or type registrations as required by the utilities to have
on failure modes. The particle filtering method and power- an effective integrated monitoring and prognosis framework.
cycling testing are examples of data-driven approaches for To address this, Section V-A outlines a future perspective
prognosis [71]. on component level asset management which is required for
the effective monitoring and diagnosis of converter assets,
3) CONDITION ANALYSIS together with the challenges and opportunities it creates.
Condition analysis requires the measurement of one or more
damage indicators during the power converter operation. III. MAINTENANCE METHODOLOGIES
However, simple knowledge is not sufficient to form the In offshore wind farms, HVdc stations and FACTS con-
overall CM strategy [33], [72]. To lay down the main realistic trollers, maintenance of power converters is usually done by
degradation mechanisms, accelerated ageing tests that apply manufacturers for an initial period of 5 years. After this, either
mission profile-based stresses to the devices are used [72]. the operator or third parties perform the job or maintenance
Several types of degradation can influence a single damage contracts with the original equipment manufacturer (OEM)
indicator; for instance, a variation in the forward voltage of are extended [14], [15], [77].
an IGBT can be affected by the temperature, the gate oxide One of the main driving forces behind the transition from
integrity, the quality of electrical connections (including bond mercury arc to thyristor valves and later onto IGBTs for
wire lift-off), and the metallization reconstruction [73]. It is power conversion in the high voltage power electronic indus-
very difficult to measure these indicators in a functional try was the burden on maintenance requirements [7]–[9].
system; for example, it is necessary to estimate the chip However, many of the existing and new power electronic
temperature to measure the thermal resistance. In a functional assets are equipped with CM systems along with SCADA
system, such measurements would require an interruption for operation. This provision provides a platform for the
of the converter operation, which would lead to a complex development of intelligent and device dependent maintenance
monitoring system [74], [75]. The most complex task in con- policies. The use of a thyristor monitoring system to aid the
dition assessment is the determination of critical contributors HVdc maintenance policy has been reported in [76] and [77].
VOLUME 6, 2018 21089

Therefore, CM leads to the next business need: developing B. PROACTIVE MAINTENANCE

of a proper maintenance strategy which includes corrective, Safety, reliability and cost effectiveness are the major con-
preventive, reliability and PoF centred maintenance. This is straints while developing a maintenance policy for power
shown schematically in Figure 6. The use of PoF as the system assets. Policies should be developed to reduce degra-
new indicator for developing maintenance policy for power dation in equipments and to ensure that the devices oper-
electronic assets is in its early stages compared to other ate smoothly throughout their operating lifespan [12], [13].
strategies. In line with these objectives, the proactive maintenance (PM)
strategy is a combination of preventive and predictive main-
tenance practices including monitoring, inspection, testing,
cleaning, overhauling, and replacement activities (at fixed
intervals) [18], [77]. PM not only contributes to an increased
availability and efficiency, but also supports maintenance
planning and budgeting, as well as contributing to the for-
mulation of effective asset management strategies [78], [79].
Typical PM activities include shutting down the equipments
at regular time intervals to perform time-based maintenance
(TBM), using CM systems to perform condition-based main-
FIGURE 6. Maintenance strategies. tenance (CBM), and using risk-based or reliability-centred
maintenance (RCM). The RCM approach combines the ben-
efits of TBM and CBM.
A. RUN-TO-FAILURE/REACTIVE MAINTENANCE
1) TIME-BASED MAINTENANCE OR PREVENTIVE
This is arguably the oldest and simplest maintenance strategy MAINTENANCE
for low-cost machinery currently used in the industry that
It is the most currently applied maintenance activity for power
does not affect plant reliability. In general, run-to-failure
converter devices in utilities and industries. By carrying out
implies refurbishment or replacement once the device has
regular maintenance and asset examination at constant inter-
failed to operate. Therefore, a component is allowed to func-
vals [12], TBM may prevent many failures. However, it may
tion and is replaced or repaired only upon failure. This strat-
also cause unnecessary outages, wasting labor, time, and
egy cannot be applied for major equipment as their failure
money if the maintenance interval is too small. In addition,
would cost much more than their maintenance [16]–[18]. For
unexpected incidents may still occur if the intervals between
a long time run-to-failure was the main maintenance activity
maintenance tasks are too large [32]. For a typical HVdc
carried out reduce the downtime of HVdc converters. In some
asset, maintenance is performed once every year or every two
cases, the failure consequences can be catastrophic and thus
years, aiming to keep the scheduled energy unavailability to
this type of maintenance has been reserved for non-critical
the minimum [12], [77]. Given that the operating conditions
faults. Additionally, it depends on the availability of spare
and environment boundaries where a VSC performs have
parts [18], [77]. The advantages and disadvantages of run-
been given great importance, CBM and RCM are preferred
to-failure are listed below.
over TBM due to their more preventive nature and cost versus
risk benefits [80]. The benefits and disadvantages of TBM are
Advantages listed below.
1) It is an economical way of maintenance requiring only
limited man-hours. Advantages
2) It reduces unnecessary shut-downs and interruptions, 1) The procedure is easy to plan and well understood by
with maintenance only performed when needed. maintenance engineers and technicians.
3) It is a widely understood strategy among maintenance 2) Fault inception and damage can be incurred if the
technicians. inspection rates are reduced.
3) The lifespan of assets can be increased due to regular
Disadvantages examinations and maintenance.
1) Failures in high power converters become costly to
repair and may need expensive spare parts. Disadvantages
2) Degradation leading to failures in power electronic 1) The procedure is expensive and requires additional
assets may sometimes be irreparable if not detected man-hours and unnecessary interruptions.
early. 2) In some cases, TBM is unable to detect faults—
3) Complete shut-down of the plant or station may some- especially when the inspection interval is large.
times happen due to certain failures, which in some 3) TBM may produce unnecessary shut-downs and addi-
cases exceeds the cost of regular inspection—thereby tional maintenance costs, increasing the FEU in HVdc
losing revenue. assets.
21090 VOLUME 6, 2018

2) CONDITION-BASED MAINTENANCE OR PREDICTIVE applied to the converter stations at Manitoba Hydro has
MAINTENANCE increased the availability of the dc transmission system by
In this strategy, maintenance is carried out in response to reducing the number of outages [77], [86]. This approach
the observed degradation in component condition. Surveil- even reduced the maintenance frequency from being annual
lance, real-time monitoring, diagnostic, and trending tools to once in four years for major equipment [86]. The advan-
are used to assess component ageing and thus plan main- tages and disadvantages of RCM as compared to other main-
tenance accordingly. Maintenance is only performed when tenance policies are summarized below.
needed, so the strategy is very responsive to equipment
conditions [55]. For instance, in an HVdc asset specific Advantages
maintenance work is required depending on the converter 1) The probability versus risk approach reduces the cost
topology [35], [77]. In traditional line-commuted converter of maintenance operation, resulting in less shut-downs
(LCC) based HVdc systems, TBM is generally preferred over for low-risk failures.
CBM due to its operating experience and low failure rates. 2) The inspection rates are reduced, thereby saving cost
However, CBM may be necessary in some cases for a prudent associated with them.
operation of the system following a scheduled maintenance, 3) As the programme focuses on the reliability of the
as described in [81]. Conversely, CBM is preferred over TBM device, the possibility of high-risk failures can be
for VSC-based assets. This is due to the smaller amount of reduced.
implemented projects, the environmental conditions and the
degradation phenomena of IGBTs (or other power electronic Disadvantages
switches) [82]. The major advantages and disadvantages of 1) The formulation of risk matrices and the prioritization
using CBM are listed below. of components based on them are complex, thus con-
suming significant time when developing maintenance
Advantages models.
1) Maintenance is performed only when it is necessary, 2) The procedure and processes related with RCM are not
thereby providing savings by minimizing unnecessary well developed, and are thus less understood by the
inspections and man-hours. maintenance personnel.
2) It reduces unnecessary shut-downs and risks associated 3) Maintenance models require a background of well-
with them. judged data and information.
3) A real-time monitoring of the assets condition delivers
C. PHYSICS-OF-FAILURE BASED MAINTENANCE
an increased utilization.
The physics behind failure of a device and the develop-
ment of strategies for maintenance based on these factors
Disadvantages
is known as PoF based maintenance (PoFBM). Identifying
1) CBM depends on the monitored data, but continuous the failure mechanism is the key for developing an effective
monitoring is not economical. PoFBM strategy. Compared to other maintenance practices,
2) The processing and extraction of information from the this is a one-person exercise where the different ways in
monitored data is not well understood in industry. which a component can physically fail are identified or fore-
3) It requires fast data communication and manipulation casted [87]. The solution reached through PoFBM can be
facilities for successful on-line monitoring and experi- directly applied to any other identical component regard-
enced personnel for the effective interpretation of data. less of the equipment where it is located at—implying a
universal failure addressing nature [88], [89]. PoFBM is an
3) RISK OR RELIABILITY-CENTRED MAINTENANCE ideal maintenance tactic for power electronic assets where
The key objectives of the RCM approach are to optimize the trade-off between reliability and criticality versus cost has
maintenance costs, increase system reliability and improve always been a concern for the optimal life-cycle operation and
asset utilization [77], [84]. Several utilities have implemented maintenance (O&M) [87]. Control of the O&M cost of power
RCM strategies. These are based on the significance and state electronic assets is an area of growing interest to utilities as
of a component. During the maintenance policy development the assets come to the end of manufacturer’s warranty. To this
stages, the main consideration is given to the current state of end, a maintenance model which considers the constraints
the component and the consequences of its failure [78], [83]. and system parameters applied to the physical model of the
Based on these constraints, a risk and criticality list is created device under investigation can be developed (see Figure 7).
and the highest priority in maintenance scheduling is given Using this model the damage indicators and critical oper-
to components whose state has depreciated considerably and ating regions of the assets can be identified. PoFBM has
which are at the top in a risk matrix [78]. Existing power con- been applied to the electronic industry [62], with potential
verter assets like HVdc converters and wind power installa- application in wind turbines being discussed in [61] and [90].
tions use RCM as their main maintenance strategy [78], [86]. In offshore applications, PoFBM can be combined with
As compared to traditional approaches, an RCM strategy TBM or CBM to reduce the down-time resulting from outages
VOLUME 6, 2018 21091

IV. ESTIMATING FAILURE AND LIFETIME MODES

Since their appearance in the transmission regime, manu-
facturers have claimed an infinite lifespan of semiconductor
devices as they have no moving parts and thus no wear and
tear. Even CIGRE reported that ‘‘semiconductor devices do
not wear out’’ [92]. However, refurbishment and replacement
strategies for HVdc and FACTS controllers have reached
a level in which utilities and manufactures alongside are
looking for options to extend their lifespan through upgrades,
FIGURE 7. PoF centred operation and maintenance model.
renovations and extension methods [10]–[15].
The three different concepts of lifetime mechanisms for
power system assets can be applied to power semiconductors
and to increase reliability [90], [91]. The major advantages
as well. These are illustrated in Figure 8 and are physical,
and disadvantages of this approach are outlined below.
technical and economic lifetime:
• Physical lifetime: The selection of assets based on the
Advantages application where they will be used defines the physical
1) The pre-operational and operational environments are lifetime of the equipment. Recent developments in PoF
taken into consideration for developing maintenance and reliability-based design can extend the physical life
models of the entire life-cycle. of power semiconductor assets.
2) PoFBM provides an early warning of potential failure, • Technical lifetime: Rapid technological developments
thereby minimizing unscheduled maintenance. have caused the conventional power electronic assets to
3) The use of PoFBM can extend maintenance cycles, be replaced by new large capacity devices. For exam-
thereby maintaining effectiveness through timely repair ple, mercury arc valves have been replaced by thyristor
actions with less unwanted costs related to inspections. valves and IGBTs in HVdc and FACTS applications.
4) It is a single-person process for identifying ways to • Economic lifetime: The economical valuation of an
transfer a common maintenance strategy across every equipment depreciates with time even when it is still
asset into operation; thus, it saves time and avoids physically usable. New assets should be installed to
repetition. enhance the value of existing assets; for example, HVdc
interconnectors and FACTS devices may be incorpo-
Disadvantages rated to enhance power transfer capabilities.
1) It is a very recent approach and thus less understood by
maintenance engineers and technicians. A. PHYSICAL AGEING MECHANISM
2) Uncertainties in loading, mission profile and environ- Thanks to developments over 10 decades, the range of power
mental conditions can be a hindrance for a proper main- semiconductor devices used in converters for HVdc and
tenance plan development. FACTS controllers is extensive [91], [92]. The low switching
3) In-depth knowledge of the equipment is inevitable for frequency and high power capability of thyristors makes them
the understanding of physics behind failure. suitable for HVdc applications, while IGBTs are suitable for
4) Testing for damage prediction and robust validation, motor drives where high switching frequency is favored. For
vital for the estimation of the physical life of a device, renewable energy converters where maximum power trans-
is resource-consuming. mission and a limitation of harmonics is desired, IGBTs have
As previously discussed, the main aim of asset manage- been preferred due to their high power capacity and high
ment is to maximize the benefits for any utility subjected to switching frequencies [91], [92].
specific constraints. This may not only be achieved through The most important design consideration for the lifetime
suitable CM techniques, but also through efficient mainte- estimation of semiconductor devices is reliability. It is often
nance plans which can maximize the use, reduce the out- represented by a bath tub curve, as shown in Figure 9. This
age time, and increase the lifetime of the assets. However, is an idealized plot representing the regions of operation
research directions in asset utilization at a process level need of a device over its lifespan [91], [95]. The curve consists
to be revisited. To this end, the roadmap in Section V-B is of three regions: an initial stage, a normal operating region
proposed aiming to achieve a maximum benefit from new and and a wear out or failure period. The primary stage takes
existing asset bases. place during the initial years of an installation and includes
Proper management practices to extend the lifespan are design faults, packing procedures and installation problems.
related to the way in which the state of assets are monitored, The normal operating region period includes random failures
maintained and valued. Based on these assessments the next bounded by environments, which are very low. The final
business need is to estimate the state of health, modes of stage considers the total failure resulting from wear out,
failures and ageing mechanisms. These aspects discussed in which causes an increase in failure rates following the normal
the following sections. period [33], [91], [95].
21092 VOLUME 6, 2018

FIGURE 8. Classification of ageing mechanism for power semiconductor assets.
60700-1 [99]. It states that most manufacturers have given

less importance to the ageing process of thyristor valves due
to their intrinsic failure rates. This is because the voltage
withstanding capabilities of the valves can be maintained
by replacements during accelerated testing and through field
failure experiments [97], [99]. However, this fact stands only
when there is no systematic degradation in the module [100].
The key parameters for thyristor failure estimation include
electrical characteristics such as the valve leakage current,
blocking and on-state voltages, reverse recovery charge and
the anode voltage. For converters enabled with IGBT devices,
FIGURE 9. Bath tub curve for failure rate estimate over lifetime.
the estimation of life models and ageing is based on the
knowledge of the stress and degradation within the device.
According to CIGRE, the two main types of semiconductor The physical methods used for analysing typical solder
devices used for HVdc and FACTS applications are thyristors joint failures can be categorized into damage, energy, and
and IGBTs [92]. The physical ageing mechanism of these stress-strain dependent methods [100]. Among these, energy-
power devices can be categorized as intrinsic or extrinsic based methods outperform others in the assimilation of test
(see Figure 8). These mechanisms are graded through qualifi- results [103]. In an energy-based model, a device is pre-
cation procedures, theoretical calculations, field failure expe- sumed to be unworthy for operation if the deformation energy
riences and PoF-based processes [87], [96], [97]. reaches a predefined critical value at a certain period of
time [103], [104].
1) INTRINSIC AGEING
b: THEORETICAL OR ANALYTICAL CALCULATIONS
Degradation and failure analysis have attracted a signifi-
cant interest with regards to failure modes in semiconduc- The use of mathematical and analytical methods for the lifes-
tor devices used for high voltage applications [37]. This is pan estimation of semiconductor devices has been reported in
explained by the increased deployment of these devices facil- [102]–[105]. The major factors that influence the life mod-
itated by undergoing a decrease in size [98]. In this context, els are temperature, pressure cycles, frequency and current.
understanding the intrinsic ageing mechanism of a device A common estimation model used in analytical calculations
during the design process is of great importance to guarantee which determines the amount of stress (and thereby lifespan)
a prolonged life-cycle. The intrinsic ageing phenomenon is of a press-pack IGBT is given by [104]:
influenced by the different tests and procedures carried out
σ

to qualify the product for final installation. These tests and 100 · = (A − B)[ln(t)]2 (1)
σ0
procedures discussed next.
where σ0 and σ are the initial and remaining stress after
a: PROCESS QUALIFICATION t hours. The other key parameters influencing the model
This is the procedure to test a device with defined conditions. ageing mechanism are the material and temperature constants
It is bounded by international standard organizations such as represented by A and B, respectively.
IEC and is industry sector specific. The application of quali- The use of accelerated testing for power semiconductor
fication procedures to test thyristor valves used in HVdc and devices under different conditions to emulate real situations
FACTS converters has been reported in the IEC publication has become a standard industrial practice. An example is
VOLUME 6, 2018 21093

the temperature estimation of thyristor PN junctions used in been analyzed in [109]. The two main factors affecting filed
HVdc converters to predict their lifespan [81], [92]. Ther- failure mechanisms are discussed next.
mal and power cycling tests for accelerated life expectancy
measurement for high power IGBT modules in wind power b: IMPACT OF HARMONICS
applications have been reported in [102]–[105]. These tests Harmonics resulting from the switching operation of valves
replicate real-time conditions to realistically reproduce cor- and their negative impact on power system assets have
rect measurement gradients, but do not include real-time been discussed in [110]. The greatest impact is faced by
measurements. The analytical equations governing the accel- the converter transformer, which could result in capacity
erated tests are given by the Coffin-Manson relation, which derating; increase in power losses, temperature rise, vibra-
is widely used for estimating the number of cycles to tion and noise, and operating cost; and decrease in effi-
failure [83], [105]. This is given by ciency, insulation strength and service life [111]. CIGRE has
reported the harmonic impact of different converter topolo-
c b

Nf = (2) gies in [112] and [113]. Compared to LCC HVdc, VSC based
1Tj MMC topologies have the inherent capability of generating
1Tja b nearly sinusoidal waveforms at the ac side, which results in

AF = (3) a reduction on harmonic filtering requirements [112], [113].
1Tjb
Furthermore, this reduces the impact on the converter trans-
where Nf is the number of cycles to failure, b and c are formers derating and ageing; for instance, vendors claim that
extrapolated constants from accelerated test results, and Tj is normal ac transformers could be employed for MMC based
the junction temperature. stations [112].
Another aspect is the harmonic transfer through the dc side,
2) EXTRINSIC AGEING which may generate sub-harmonic currents and, in turn, could
This is an ageing process which is amplified by failures affect the ac system assets such as turbine generators and
related to the application and handling of power electronic might result in shaft failure [114]–[116]. However, these neg-
devices used in HVdc and FACTS converters. The major ative interactions are not that severe for VSC based converters
contributors for extrinsic ageing are the environment where due to the decoupling of the dc side [112]. Tsai et al. [115]
the device is installed and the electric stress resulting from and Yacamini [116] identified torsional oscillations due to
different operating conditions [101]. As these change the the HVdc scheme. The variable-frequency ripple current
physical performance of the device, the ageing process is superimposed on dc currents would excite subsynchronous
amplified, as depicted in Figure 8 [98]. Another contributor vibrations in turbine shafts, and could reduce the life of the
for this ageing mechanism is the latent defects regime of shafts [116]. It should also be noted that there is an obvious
those assets that can make a considerable impact on the connection between the harmonics generated at the converter
operating expenditure (OPEX) [106]. The two main indica- station and the control of the converters, which will be dis-
tors that may reveal this ageing phenomenon are the field cussed in the next section.
failure and PoF—which are discussed in the forthcoming
paragraphs. c: IMPACT OF CONTROL
The control system acts as the brain of the HVdc con-
a: FIELD FAILURE MEASUREMENTS verters or FACTS controllers. Compared to the rest of the
Assimilation of failure experiences from field data and components in an HVdc converter, the control system rep-
analysis for estimation of reliability measurements is the resents only about 7% of the total cost of the converter
key process behind field failure measurement. Two com- station (see Figure 2) [15]. A number of components can
mon methods for understanding the ageing process in power be substituted before the replacement of the entire instal-
electronic devices are long-term surveillance and statistical lation is required. Many utilities have followed this route
comparison [81]. For example, a number of thyristor devices for asset management. As shown in Table 2, the failure rate
are selected and their characteristics measured and recorded and repair time of the control system of converter assets
once or twice every year in the long term surveillance method. is very small compared to other components. Furthermore,
Results are then processed to find a pattern of failure modes. significant planning is necessary during the design stage to
In [107], specific field measurements taken from a static VAr estimate the impact of the control system on the converter
compensator (SVC) indicated that thyristor failure resulted stations. Operating experience has shown to be a key aspect
from the degradation of associated auxiliary assets. On the behind the control system selection for thyristor-based HVdc
other hand, few statistics and information on VSCs are avail- and FACTS devices [10]–[15]. The case is similar for VSC
able. Recent CIGRE publications reported that the highest stations. As an example, the Trans Bay Cable MMC project
failure rate for any converter is found in the power electronics built in 2010 is undergoing an upgrade to provide black start
component [80], [108]. Additionally, the application of dif- capability through both control and valve modifications [10].
ferent thermal cyclic and loading conditions on wind power CIGRE has made a detailed analysis of different control
converters and their impact on the failure mechanisms has aspects in HVdc converter stations and their impact on the
21094 VOLUME 6, 2018

reliability of the power supply compared to high voltage ac

(HVac) transmission systems [112]. The robustness of the
converter control ensures reliable and high quality power
supply from the dc network to the ac system.
Active thermal control strategies represent an alter-
native which can extend the lifetime of the converter
assets [117]–[119]. The main objective is to minimize the
high failure rate attributed primarily to the thermal cycling
in power semiconductor devices [119].
FIGURE 10. MCF curve represented as an adaption of bathtub curve [91].
d: RELIABILITY/PHYSICS ORIENTED AGEING MECHANISM
Physics or reliability-based design is a new method
for estimating the ageing process of semiconductor existing transmission assets need to be redefined to incorpo-
devices [33], [98]. Until recently, engineers have used a rate a depreciation cost. Furthermore, the increased complex-
standardized stress-based methodology for qualification tests ity to construct new transmission lines and the difficulties
and life assessments of power electronic assets. However, to obtain rights of way have motivated a modification of
the development of these devices has been accompanied existing transmission assets. HVdc and FACTS controllers
by the widespread application of PoF-based knowledge for have changed the way electric power has been transmitted
reliability assessment [126], [127]. Moreover, the physics- and controlled [94]. Economic studies are conducted at the
based health assessment and prognosis has allowed opera- planning stages to estimate the most advantageous project
tors to evaluate individual power modules under real-time long-term scenario. In terms of HVdc assets, this is a period
operational conditions. This has enabled an estimation of of 40-50 years [11]. In general the economic life service
the remaining lifetime of the assets [126]–[128]. Another (ELS) of a transmission line asset is determined by the period
important aspect in determining the physical oriented mech- of time in which profit is made by replacing or abandoning
anism is the failure rate estimation. This is assumed as it. This may be due to [10]–[15]:
constant in many applications although in truth it varies over • The deterioration or degradation rate of an asset has
the operational time. An alternative technique to obtain the reached its ultimate stage, where abandoning it is
failure level and time is the mean cumulative function (MCF) cheaper than refurbishing it to extend its life-cycle.
curve [91]. An example is shown in Figure 10, which can be A typical example is the paradigm shift in HVdc
considered as an integration of the bath tub curve where the and FACTS assets from mercury arc to thyristor-based
different lines indicate a range of failure rates. converter units—and presently into transistor-based
A summary of the most common types of failure arising modules.
in power electronic devices is provided by Table 1. This • Technological development, displacing older assets to
shows the different contributors to the failure, ageing, and add greater value. Innovation and progress in the power
health degradation of power electronic assets. The classifi- semiconductor technology can be related to ELS depre-
cation provided in this work has been inspired from wind ciation.
turbine converter regimes and the associated equipment fail- • Asset displacement resulting from competitive markets
ures [124], [125]. This includes semiconductor, control, dc which focus on the substitution of devices and processes
link, cable and connection, filter, and cooling system failures. by new products, aiming to provide added value. A dis-
Respective failure modes, mechanisms, behavior and critical tinctive example of this process includes the European
contributors are also presented in Table 1. Supergrid developments, which have been facilitated
It should be emphasized that the mechanisms employed by the increased controllability from VSC-based HVdc
to estimate the physical ageing of power electronic assets assets.
have evolved over the years, with new failure indicators and
degradation mechanisms related to their specific application C. TECHNICAL LIFESPAN
and environmental condition being adopted. Most of the Transmission networks including HVdc and FACTS assets
modern power electronic assets are cross-linked with their which were installed in the 1970’s and 80’s require some kind
power system counterparts and thus require to be strategically of replacement or refurbishment [13]–[15]. The usual lifetime
managed for their effective utilization. In this line, the main of power converter assets is in the range of 40-50 years,
challenges and opportunities associated with such an strategic although during this time newer and better technologies may
level asset management are discussed in Section V-B. appear. Continuing operation with existing assets following
refurbishments and upgrades at regular intervals can be suf-
B. ECONOMIC LIFESPAN ficient. However, new assets may provide an increased func-
An asset starts depreciating its value from the time it is tionality and performance [13]. Therefore, a good knowledge
purchased until it reaches its retire value, which is the esti- of the technical lifetime of converter-based assets in trans-
mated value of the asset at the time of its disposal. The mission grids is essential to develop a proper replacement
VOLUME 6, 2018 21095

TABLE 1. Failure modes and mechanisms of power electronic assets.
networks, their corresponding failure rates and repair time

are difficult to obtain in the open literature. Publications
alongside with the network operators and manufacturer data
have been used for this estimation and have been taken from
references [121]–[123]. An insight into the lifespan alongside
with failure rates of different dc substation components can
be also found in [15, Table 2]. To adequately address the
technical lifetime of power converter assets it is imperative
to discuss the developments in semiconductor technologies
over the years. This is carried out in the following sections.
FIGURE 11. Developments in the power semiconductor technology
(adapted from [135]). 1) MERCURY ARC VALVES-BASED CONVERTERS
With the availability of mercury arc valves in the early
1930’s, HVdc transmission reached certain parity with ac
transmission. Such a major development in semiconductor
strategy. Developments in IGBT devices during the recent technology paved the way for the construction of the
years constitute a typical example; for instance, the voltage first dc link. This was a 3 MW, 45 kV link between
rating of a single IGBT module has increased to 6500 V [131]. Germany and Switzerland [132]. However, the employed
Thus, the technical lifetime of a power semiconductor device high power valves were highly sensitive to vacuum condi-
is closely related to the developments in semiconductor tech- tions and the surface contamination resulting from the inher-
nology, which has experienced drastic progression over the ent concentration of blocking voltage into narrow regions
recent years (see Figure 11). of the device. These attributes eventually reduced their
Failure rates and unavailability can reduce the value of lifespan [132], [133]. By 1970’s, the use of mercury arc
assets and can become a burden for operators and asset own- valves in HVdc applications lost ground due to the unavail-
ers. In line with this, the failure rates of the different com- ability of spare parts and the concurrent progress in thyristor
ponents in a typical power electronic substations should be valves [133]. A significant hindrance in their application
evaluated carefully. The critical components required for the was the limitation to increase the blocking voltage, which
safe and reliable operation of the station and their respective in turn restricted the selection of converter ratings for high
failure rate in occurrence per year (o/y) and repair time in power applications. Additionally, the need for a rapid switch-
hours per year (h/o) are presented in Table 2. The reliability in and switch-off imposed operational limits to avoid the arch
data of LCC stations can be found in reference [120]. Since back phenomenon. This reduced the viability of these devices
VSCs represent a relatively new addition to transmission compared to thyristors [132]–[136]. Furthermore, the use
21096 VOLUME 6, 2018

TABLE 2. Major assets of typical power electronic substations: their lifetime and failure rates [15].
of mercury-based switches has an adverse environmental with modular MMC topologies suitable for high power and
impact; for instance, the amount of mercury vapors released high voltage applications due to their low power losses
to the atmosphere during O&M is of concern. Cautious mon- and minimal filter requirements [138], [139]. These HVdc
itoring has been employed in the valve halls to control mer- technologies are not only used for conventional high power
cury emissions. Each self-contained sealed module included transmission, but are also suitable for the construction of
2.64 qt (2.5 L) of mercury [132]. Nowadays mercury arc- dc grids for the integration of large-scale renewables—such
based HVdc assets have been replaced by thyristor valves as offshore wind farms [139], [140]. Until now, VSC-based
except for one link [135]. transmission has paralleled thyristor-based converter assets
wherever possible. With the demand for renewables to meet
2) THYRISTOR-BASED CONVERTERS climate change challenges and the increasing interconnec-
The key performance addition from thyristor-based convert- tions among countries, further development of transistor-
ers was the ability to bring together control and conversion based power converter topologies is imperative.
operations into one device, leading to substantial savings in Developments in high power semiconductors have shifted
space and weight [136], [137]. Even when thyristors cannot the nature of HVdc assets to a modular and cost effective
be turned off with a control pulse, their ability to achieve a low transmission. One of the key improvements in this line is the
forward voltage drop led to its widespread use in HVdc instal- transition from two-level to multilevel VSC topologies [138].
lations [133]–[139]. Once thyristors started to take over the As the traditional HVdc and FACTS assets were not orig-
HVdc regime from mercury valves, the availability of HVdc inally designed for modular expansions, changes intro-
projects increased significantly from 83 to 98% [120]–[125]. duced following installation and commissioning have been
Since then, thyristor devices have been substantially deployed restricted to control system updates and pole upgrades [10].
in transmission networks for HVdc and FACTS controllers. It could be argued that this approach has limited the utilization
Their simple design, added with low maintenance costs, pre- of the upgraded transmission assets [138]. However, it is
dictable performance and reduced area for installations are foreseen that future HVdc and FACTS device installations
the key factors behind this fast deployment [136], [137]. For will be modular and multilevel in nature and, hence, the new
example, the typical space utilization of mercury arc-based topologies will have an enhanced value [138], [138], [140].
converters is about 3.5 m2 /MW compared to 1 m2 /MW for Furthermore, by adopting the modular topologies and in turn
thyristor valves [76]. Another vital aspect is the possibility avoiding the use of additional components such as filters
to arrange thyristor valves in series, which gives the advan- and transformers, the burden in maintenance and monitoring
tage of a high performance while making the modules less could be reduced [112]. The industry is already moving in
prone to fractional failures. This in turn enables the design this direction as indicated by the first MMC project in the US
optimization of an HVdc installation by having full control and the recent INELFE project linking Italy and Spain [140].
over ratings [138].
V. FUTURE WORK: RESEARCH IN POWER ELECTRONIC
3) TRANSISTOR-BASED CONVERTERS ASSET MANAGEMENT REQUIRES A ROADMAP
Voltage-controlled transistor-based semiconductor devices While the aforementioned procedures provide a systematic
like IGBTs were developed during the late 1970’s. They way for the selection of power converter assets considering an
have found their place in the lower end of the power range optimized cost, size and lifetime, their high complexity and
for HVdc applications [133]. Their self-turn off capability, the need for further research may restrict their application.
together with an independent control of active and reac- This is highlighted in Figure 12. The review of industrial
tive power, leads to a smaller footprint when employed experience documented in this work, together with recent
in VSC stations [76], [136]. As of now, VSC-based high developments in power electronic assets, has allowed the
voltage transmission assets are in their fourth generation, identification of a range of needs and gaps faced by power
VOLUME 6, 2018 21097

FIGURE 12. Research roadmap in terms of power electronic asset management.
transmission utilities. The primary gaps have been grouped plan can be developed using updated tools. Development of
into three categories—in line with the present industry asset an effective framework for maintenance optimization, spares
management needs built on sustainability. and inventory management, and economic constraints can
be also integrated into the tools. To achieve this level of
A. CATEGORY 1: COMPONENT LEVEL ASSET control over maintenance management a balance between
MANAGEMENT preventive and reactive maintenance should be adopted. All
Existing methods to gather and categorize data from ac sys- these attributes can be accommodated into the platform and
tems are widely and efficiently applied by utilities. Dissemi- simulated over the life-cycle to assess the impact on asset
nation of this knowledge is considerable; however, a revised utilization. The major opportunities and obstacles for the
asset management strategy is inevitable at device and system development of an integrated tool or software for managing
level due to the extensive deployment of power converter- power electronic assets can be summarized below:
based assets. Initial efforts in this direction have been carried
out by many utilities. Practices include the utilization of CM Challenges
techniques to assess the state of components; rescheduling of 1) Tests that estimate the failure rates of components are
maintenance plans according to CM information; and a tran- resource consuming and differ from traditional con-
sition into RCM maintenance approaches. The major obsta- stant failure rate models.
cles and research opportunities in the management of power 2) Improved technologies for locating faults in inacces-
electronic assets at a component level can be summarized as sible areas (e.g. HVdc underground and submarine
follows: cables, temperature estimation in power semiconductor
modules).
Challenges 3) Coherence between conventional power system assets
1) Prevalent and increased application of power electron- and converter-based assets in the operation and control
ics in a large range of applications and different envi- regime—especially for power balancing.
ronmental exposures.
2) Outdated procedures to estimate the health and failure Opportunities
rates of power semiconductor equipment and a lack of 1) Minimizing latent defects by robust reliability-based
understanding of critical failure modes. fault tolerant design, which may reduce failure rates
3) Increased complexity in used components and in pack- and extend the lifetime.
aging. 2) Emerging modular topologies of power converters and
capacitors, coupled with developments in on-line CM
Opportunities methods.
1) Research in microelectronics and methodologies can be 3) Effective utilization of dynamic time constants of grids
effectively used as a foundation for ongoing and future with power electronic assets for balancing and increas-
work in the field of power electronics. ing grid availability.
2) Focus should be given to device reliability by under-
standing the operating regime through their lifetime. C. CATEGORY 3: STRATEGIC LEVEL ASSET MANAGEMENT
3) Effective utilization of mission profiles and on-line CM This category establishes gaps closely aligned with the devel-
data to understand the PoF. opments of an organization at a business level with relation
to training, development and implementation of asset policies
B. CATEGORY 2: PROCESS LEVEL ASSET MANAGEMENT and practices. These may be either technical or scientific
This type of asset management deals with gaps in tools for related materials and understanding. The main drivers are
the effective utilization and processing of assets. New assets cost, policies from regulators and investors, the knowledge
can benefit from this as an effective life-cycle management gap associated with new technologies, and the risk associated
21098 VOLUME 6, 2018

with implementation. Approaches to be developed by the component, process, and organization level developments
asset owner include the coordination and linkage with differ- has been discussed—together with the identification of chal-
ent utilities, collaboration with the academia and the devel- lenges and opportunities that may arise.
opment of in-house tools to match and validate the models REFERENCES
from the asset provider where needed. The main challenges [1] IEC Market Strategy Board, ‘‘Strategic asset management of power
and opportunities associated with strategic asset development networks,’’ IEC, Geneva, Switzerland, White Paper, ISBN 978-2-8322-
are summarized below: 2810-4, 2015.
[2] ENTSO-E Research and Development Road Map 2013–2020, ENTSO-E,
Challenges Brussels, Belgium, 2013.
[3] R. Adapa, ‘‘High-wire act: HVdc technology: The state of the Art,’’ IEEE
1) Resources needed for an effective application of asset Power Energy Mag., vol. 10, no. 6, pp. 18–29, Nov./Dec. 2012.
management in line with the development of new com- [4] B. K. Bose., ‘‘Global energy scenario and impact of power electronics in
ponents and technologies. 21st century,’’ IEEE Trans. Ind. Electron., vol. 60, no. 7, pp. 2638–2651,
Jul. 2013.
2) Cost-benefit analysis linked with the application of new [5] D. Kumar, F. Zare, and A. Ghosh, ‘‘DC microgrid technology: System
asset management strategies and selection of the opti- architectures, AC grid interfaces, grounding schemes, power quality,
mal method—constrained by environmental changes. communication networks, applications, and standardizations aspects,’’
IEEE Access, vol. 5, pp. 12230–12256, 2017.
3) Approaches to cope with culture changes in support of [6] United States Department of Energy, Grid 2030—A National Vision for
asset management, new age assets, and future assets. Electricity’s Second 100 Years. Damascus, MD, USA: Penny Hill Press,
2015.
Opportunities [7] G. Strbac et al., ‘‘Delivering future-proof energy infrastructure,’’
A Report for the National Infrastructure Commission, Imperial
1) Training and communication with the academia to College London and Univ. Cambridge, London, U.K., EPRG, Working
nurture individuals who can cope with the recent devel- Paper, 2016, accessed: Apr. 14, 2018. [Online]. Available: https://
opments and new technologies in physics and engineer- www.gov.uk/government/uploads/system/uploads/attachment_data/file/
507256/Future-proof_energy_infrastructure_Imp_Cam_Feb_2016.pdf
ing. [8] UK Department of Energy and Climate Report, Delivering UK Energy
2) Development of a framework for asset manage- Investment, London, U.K., 2016.
ment including vision and cornerstones, with training [9] U. Bryan, ‘‘Leveraging PAS 55 within a national grid,’’ in Proc.
IET Seminar PAS 55 Measuring Asset Manage., London, U.K., 2007,
focussed in cost-benefit analysis, life-cycle cost analy- pp. 45–51.
sis and social cost. [10] W. Litzenberger, K. Mitsch, and M. Bhuiyan, ‘‘When it’s time to upgrade:
3) Developing data frameworks within utilities to support HVdc and FACTS renovation in the western power system,’’ IEEE Power
Energy Mag., vol. 14, no. 2, pp. 32–41, Mar. 2016.
enterprise-wide asset management with linkage to dif- [11] H. Bilodeau et al., ‘‘Making old new again: HVdc and FACTS in the
ferent components. Northeastern United States and Canada,’’ IEEE Power Energy Mag.,
vol. 14, no. 2, pp. 42–56, Mar./Apr. 2016.
VI. CONCLUSION [12] N. M. Kirby et al., ‘‘Extending their lifetimes: Keeping HVdc and FACTS
The large-scale deployment of power converter assets into installations in service longer,’’ IEEE Power Energy Mag., vol. 14, no. 2,
pp. 57–65, Mar. 2016.
transmission networks has reached a level of parity with ac [13] H. Bjorklund, A. Thelenius, U. Elgqvist, J. P. Kjaergaard, and C. Reddy,
system assets. Their management and maintenance has thus ‘‘Keeping the power flowing: A commitment throughout the HVdc and
become an extremely difficult job for asset managers and FACTS life cycles,’’ IEEE Power Energy Mag., vol. 14, no. 2, pp. 66–71,
Mar./Apr. 2016.
engineers. To tackle these difficulties, this paper has char- [14] B. Gemmell and M. Korytowski, ‘‘Refurbishments in Australasia:
acterized power electronic assets into three business trends. Upgrades of HVdc in New Zealand and FACTS in Australia,’’ IEEE
In particular, the different aspects that need to be consid- Power Energy Mag., vol. 14, no. 2, pp. 72–79, Mar./Apr. 2016.
[15] R. Johnson and G. Wolf, ‘‘Refurbish rather than replace: Resuscitating
ered for the development of a sustainable asset manage- aging HVdc and FACTS projects,’’ IEEE Power Energy Mag., vol. 14,
ment framework for HVdc and FACTS devices have been no. 2, pp. 22–31, Mar./Apr. 2016.
illustrated. It has been concluded that an imminent change [16] I. Vancers, D. J. Christofersen, A. Leirbukt, and M. G. Bennett, ‘‘A
survey of the reliability of HVDC systems throughout the world during
in the monitoring and assessment methods (ranging from 2011–2012,’’ in Proc. 44th CIGRE Session Paris, Paris, France, 2014,
traditional, reliability, or PoF-based approaches) is needed pp. 1–9.
due to the inherent design of power electronic components. [17] CIGRE Study Committee B4, ‘‘Protocol for reporting the operational
performance of HVDC transmission systems,’’ CIGRE, Paris, France,
Moreover, the incorporation of dc substations into ac grids Tech. Rep. WG B4.04, 2014, pp. 14–97.
will pose the challenge of maintaining them in a reliable and [18] J. J. Cochrane, M. P. Emerson, J. A. Donahue, and G. Wolf, ‘‘A sur-
optimal way, thus making of PoF a potential new philosophy vey of HVDC operating and maintenance practices and their impact on
reliability and performance,’’ IEEE Trans. Power Del., vol. 11, no. 1,
for maintenance practice.
pp. 514–518, Jan. 1996.
The failure and degradation processes of power semicon- [19] IEEE Guide for the Evaluation of the Reliability of Hvdc Converter
ductor devices have a significant impact on their ageing Stations, IEEE Standards 1240-2000, Feb. 2001.
and thus lifespan. Therefore, recent trends in power con- [20] I. Vancers and F. J. Hormozi, ‘‘A summary of North American HVDC
converter station reliability specifications,’’ IEEE Trans. Power Del.,
verter design, operation and management, bound by PoF and vol. 8, no. 3, pp. 1114–1122, Jul. 1993.
reliability-based approaches, can provide value for money [21] J. Schneider et al., ‘‘Asset management techniques,’’ Int. J. Elect. Power
in terms of business needs for asset owners. To provide Energy Syst., vol. 8, no. 9, pp. 643–654, 2006.
[22] S. Bartlett et al., ‘‘Asset Management in a de-regulated environment,’’
some guidance into future work, a research roadmap for an in Proc. CIGRE Joint Task Force Austral. Work. Groups, Paris, France,
effective management of power converter assets in line with 2002, pp. 23–30.
VOLUME 6, 2018 21099

[23] Q. Wei, Z. Pinjia, and C. Mo-Yuen, ‘‘Condition monitoring, diagnosis, [48] W. Yang, R. Court, and J. Jiang, ‘‘Wind turbine condition monitoring
prognosis, and health management for wind energy conversion systems,’’ by the approach of SCADA data analysis,’’ Renew. Energy., vol. 53,
IEEE Trans. Ind. Electron., vol. 62, no. 10, pp. 6533–6535, Oct. 2015. pp. 365–376, May 2013.
[24] S. Bahramirad et al., ‘‘Guest editorial: Special section on asset manage- [49] B. Chen, D. Zappalá, C. J. Crabtree, and P. J. Tavner, ‘‘Survey of
ment in smart grid,’’ IEEE Trans. Smart Grid, vol. 6, no. 2, pp. 953–954, commercially available SCADA data analysis tools for wind turbine
Mar. 2015. health monitoring,’’ School Eng. Comput. Sci., Durham Univ., Durham,
[25] F. Blaabjerg and M. M. Pecht, ‘‘Special issue on robust design and U.K., Tech. Rep., 2010, accessed: Apr. 14, 2018. [Online]. Available:
reliability of power electronics,’’ IEEE Trans. Power Electron., vol. 30, http://dro.dur.ac.uk/12563/1/12563.pdf
no. 5, pp. 2373–2374, May 2015. [50] G. Jianhong, W. Haifeng, A. Chengliu, and G. Guobiao, ‘‘The application
[26] Asset Management of Transmission Systems and Associated CIGRE of evaporative cooling technology in HDVC converter station,’’ in Proc.
Activities, CIGRE Techn. Brochure, Paris, France, 2006, p. 309. Int. Conf. Electr. Mach. Syst., 2013, pp. 2267–2270.
[27] Obtaining Value From On-Line Substation Condition Monitoring, docu- [51] H. P. Lips, ‘‘Water cooling of HVDC thyristor valves,’’ IEEE Trans.
ment WG B 3.12, CIGRE Brochure 14-97, CIGRE, Paris, France, 2011. Power Deliv., vol. 9, no. 4, pp. 1830–1837, Oct. 1994.
[28] Y. Han and Y. H. Song, ‘‘Condition monitoring techniques for electrical [52] C. J. Hellier, Handbook of Non-Destructive Evaluation. New York, NY,
equipment—A literature survey,’’ IEEE Trans. Power Del., vol. 18, no. 1, USA: McGraw-Hill, 2003.
pp. 4–13, Jan. 2003. [53] M. H. Jodeyri and A. Dzus, ‘‘Thyristor valve for the 12-pulse converter for
[29] D. Kwon, M. R. Hodkiewicz, J. Fan, T. Shibutani, and M. G. Pecht, the Champa-Kurukshetra HVDC transmission scheme,’’ in Proc. IEEE
‘‘IoT-based prognostics and systems health management for industrial ISGT, Nov. 2013, pp. 1–6.
applications,’’ IEEE Access, vol. 4, pp. 3659–3670, 2016. [54] High-Voltage Direct Current (HVDC) Installations System Tests, Stan-
[30] A. Azmi, J. Jasni, N. Azis, and M. Z. A. A. Kadir, ‘‘Evolution of dard Indian Standards 16076, Bureau of Indian Standards, India, 2013.
transformer health index in the form of mathematical equation,’’ Renew. [55] F. Besnard and L. Bertling, ‘‘An approach for condition-based mainte-
Sustain. Energy Rev., vol. 76, pp. 687–700, Sep. 2017. nance optimization applied to wind turbine blades,’’ IEEE Trans. Sustain.
[31] J. L. Velasquez-Contreras, M. A. Sanz-Bobi, and S. G. Arellano, Energy, vol. 1, no. 2, pp. 77–83, Jul. 2010.
‘‘General asset management model in the context of an electric utility: [56] H. Q. Niu, A. Cavallini, and G. C. Montanari, ‘‘Identification of partial
Application to power transformers,’’ Electr. Power Syst. Res., vol. 81, discharge phenomena in HVDC apparatus,’’ in Proc. IEEE Int. Symp.
no. 11, pp. 2015–2037, 2011. Electr. Insul., Jun. 2008, pp. 373–376.
[32] A. E. B. Abu-Elanien and M. M. A. Salama, ‘‘Asset management [57] K. Smarsly et al., ‘‘Advanced structural health monitoring based on multi-
techniques for transformers,’’ Electr. Power Syst. Res., vol. 80, no. 1, agent technology,’’ in Computation for Humanity: Information Technol-
pp. 456–464, 2010. ogy to Advance Society. New York, NY, USA: Taylor & Francis, 2012.
[33] H. Oh, B. Han, P. McCluskey, C. Han, and B. D. Youn, ‘‘Physics-of- [58] M. McGugan, G. C. Larsen, B. F. Sørensen, K. K. Borum, and
failure, condition monitoring, and prognostics of insulated gate bipolar J. Engelhardt, ‘‘Fundamentals for remote condition monitoring of off-
transistor modules: A review,’’ IEEE Trans. Power Electron., vol. 30, shore wind turbines,’’ Danmarks Tekniske Univ., Riso Nat. Lab., Lyngby,
no. 5, pp. 2413–2426, May 2015. Denmark, Tech. Rep. 1639, 2008.
[34] E. U. Ogbodo, D. Dorrell, and A. M. Abu-Mahfouz, ‘‘Cognitive radio [59] D. Effren, Automated Turbine Inspection. Charlotte, NC, USA:
based sensor network in smart grid: Architectures, applications and com- AutoCopter, 2011.
munication technologies,’’ IEEE Access, vol. 5, pp. 19084–19098, 2017. [60] H. Lu, C. Bailey, and C. Yin, ‘‘Design for reliability of power electronics
[35] N. M. Kirby et al., ‘‘Optimizing the availability of HVDC systems modules,’’ Microelectron. Rel., vol. 4, nos. 9–11, pp. 1250–1255, 2009.
via effective asset management,’’ in Proc. CIGRE Canada Conf., 2015, [61] E. E. Kostandyan and J. D. Sørensen, ‘‘Physics of failure as a basis for
pp. 1–8. solder elements reliability assessment in wind turbines,’’ Rel. Eng. Syst.
[36] P. L. Lewin et al., ‘‘The state of the art of condition monitoring: Where Safety, vol. 108, pp. 100–107, Dec. 2012.
do we go from here?’’ in Proc. Int. Elec. Ins. Conf., 2013, pp. 58–66. [62] S. Salemi, ‘‘Physics-of-failure based handbook of microelectronic sys-
[37] S. Yang, A. Bryant, P. Mawby, D. Xiang, L. Ran, and P. Tavner, tems,’’ in The Reliability Information Analysis Center. Utica, NY, USA:
‘‘An industry-based survey of reliability in power electronic converters,’’ RIAC Press, 2008.
IEEE Trans. Ind. Appl., vol. 47, no. 3, pp. 1441–1451, May/Jun. 2011. [63] S. Yang, D. Xiang, A. Bryant, P. Mawby, L. Ran, and P. Tavner,
[38] HVDC Stations Audible Noise, document WG SC 14.26, CIGRE Tech. ‘‘Condition monitoring for device reliability in power electronic con-
Brochure 202, CIGRE, 2002. verters: A review,’’ IEEE Trans. Power Electron., vol. 25, no. 11,
[39] H. D. M. de Azevedo, A. M. Araújo, and N. Bouchonneau, ‘‘A review pp. 2734–2752, Nov. 2010.
of wind turbine bearing condition monitoring: State of the art and chal- [64] M. Ciappa, ‘‘Lifetime prediction on the base of mission profiles,’’ Micro-
lenges,’’ Renew. Sustain. Energy Rev., vol. 56, pp. 368–379, Apr. 2016. electron. Rel., vol. 45, nos. 9–11, pp. 1293–1298, 2005.
[40] H. Smede, C.-G. Johansson, O. Winroth, and H.-P. Schutt, ‘‘Design of [65] Y. Avenas, L. Dupont, N. Baker, H. Zara, and F. Barruel, ‘‘Condition
HVDC converter stations with respect to audible noise requirements,’’ monitoring: A decade of proposed techniques,’’ IEEE Ind. Electron. Mag.,
IEEE Trans. Power Del., vol. 10, no. 2, pp. 747–758, Apr. 1995. vol. 9, no. 4, pp. 22–36, Dec. 2015.
[41] J. Shengchang, W. Peng, Z. Qiaogen, and L. Yanming, ‘‘Study on [66] H. Huang and P. A. Mawby, ‘‘A lifetime estimation technique for volt-
the noise-level calculation method for capacitor stacks in HVDC con- age source inverters,’’ IEEE Trans. Power Electron., vol. 28, no. 8,
verter station,’’ IEEE Trans. Power Del., vol. 25, no. 3, pp. 1866–1873, pp. 4113–4119, Aug. 2013.
Jul. 2010. [67] P. Cova and F. Fantini, ‘‘On the effect of power cycling stress on IGBT
[42] J. Li, S. Ji, L. Zhu, P. Wu, and T. Cao, ‘‘Vibration characteristics of filter modules,’’ Microelectron. Rel., vol. 38, nos. 6–8, pp. 1347–1352, 1998.
capacitors used in HVDC converter stations,’’ IEEE Trans. Power Del., [68] G. Haddad, P. A. Sandborn, and M. G. Pecht, ‘‘An options approach for
vol. 31, no. 5, pp. 2045–2053, Oct. 2016. decision support of systems with prognostic capabilities,’’ IEEE Trans.
[43] K. L. Thakre, A. G. Kothari, and H. M. Suryawanshi, ‘‘Identification of Rel., vol. 61, no. 4, pp. 872–883, Dec. 2012.
converter faults using sequence detectors as signature analyzers,’’ Electr. [69] N. M. Vichare and M. G. Pecht, ‘‘Prognostics and health management
Power Compon. Syst., vol. 30, no. 1, pp. 31–44, 2002. of electronics,’’ IEEE Trans. Compon. Packag. Technol., vol. 29, no. 1,
[44] K. L. Thakre, A. G. Kothari, H. M. Suryawanshi, and S. S. Tanavade, pp. 222–229, Mar. 2006.
‘‘Quick identification of converter valve misbehavior using fuzzy logic,’’ [70] S. Ramminger, N. Seliger, and G. Wachutka, ‘‘Reliability model for Al
Electr. Power Compon. Syst., vol. 34, no. 4, pp. 445–460, 2006. wire bonds subjected to heel crack failures,’’ Microele. Rel., vol. 40, no. 8,
[45] R. V. DeVore, R. Caldecott, P. D. McWane, and S. A. Sebo, ‘‘Radio pp. 1521–1525, 2000.
frequency spectral characteristics of an HVDC converter station,’’ Electr. [71] N. Patil, D. Das, and M. Pecht, ‘‘A prognostic approach for non-punch
Power Syst. Res., vol. 9, no. 3, pp. 201–212, 1985. through and field stop IGBTs,’’ Microelectron. Rel., vol. 52, no. 3,
[46] S. Fang, W. Luo, H. Wang, and H. Hu, ‘‘Operational performance of the pp. 482–488, 2012.
valve cooling system in Guangzhou converter station,’’ in Proc. 11th IET [72] W. Zhou, X. Zhong, and K. Sheng, ‘‘High temperature stability and
Int. Conf. AC DC Power Trans., 2015, pp. 5–11. the performance degradation of SiC MOSFETs,’’ IEEE Trans. Power
[47] M. Schlechtingen, I. F. Santos, and S. Achiche, ‘‘Wind turbine condition Electron., vol. 29, no. 5, pp. 2329–2337, May 2014.
monitoring based on SCADA data using normal behavior models. Part 1: [73] H. Berg and E. Wolfgang, ‘‘Advanced IGBT modules for railway traction
System description,’’ Appl. Soft Comput., vol. 13, no. 1, pp. 259–270, applications: Reliability testing,’’ Microelectron. Rel., vol. 38, no. 6,
2013. pp. 1319–1323, 1998.
21100 VOLUME 6, 2018

[74] A. Rashed et al., ‘‘On-line [TJ , Vce ] monitoring of IGBTs stressed by [101] C. I. Ossai, ‘‘Optimal renewable energy generation—Approaches for
fast power cycling tests,’’ in Proc. 15th Eur. Conf. Power Electron. Appl. managing ageing assets mechanisms,’’ Renew. Sustain. Energy Rev.,
(EPE), 2013, pp. 1–9. vol. 72, pp. 269–280, May 2017.
[75] L. Dupont, G. Coquery, K. Kriegel, and A. Melkonyan, ‘‘Accelerated [102] W. Nelson, Accelerated Testing Statistical Models, Test Plans
active ageing test on SiC JFETs power module with silver joining tech- and Data Analysis, vol. 8. Hoboken, NJ, USA: Wiley, 2004,
nology for high temperature application,’’ Microelectron. Rel., vol. 49, p. 135.
no. 11, pp. 1375–1380, 2009. [103] M. Ciappa et al., ‘‘Lifetime modelling of thermos mechanics related fail-
[76] A. Ekstrom and L. Eklund, ‘‘HVDC thyristor valve development,’’ IEEE ure mechanisms in high power IGBT modules for traction applications,’’
Trans. Power Electron., vol. PE-2, no. 3, pp. 177–185, Jul. 1987. ISPSD, Cambridge, U.K., Tech. Rep., 2003, pp. 295–298.
[77] Guidelines for Life Extension of Existing HVDC Systems, CIGRE Work- [104] C. Busca et al., ‘‘An overview of the reliability prediction related aspects
ing Group B 4.54 Brochure, Paris, France, 2016. of high power IGBTs in wind power applications,’’ Microelectron. Rel.,
[78] M. R. Allison, K. G. Lewis, and M. L. Winfield, ‘‘An integrated approach vol. 51, nos. 9–11, pp. 1903–1907, 2011.
to reliability assessment, maintenance and life cycle costs in the National [105] L. R. GopiReddy, L. M. Tolbert, B. Ozpineci, and J. O. P. Pinto,
Grid Company,’’ in Proc. IET 2nd Int. Conf. Rel. Trans. Distrib. Equip., ‘‘Rainflow algorithm-based lifetime estimation of power semiconduc-
1995, pp. 180–185. tors in utility applications,’’ IEEE Trans. Ind. Appl., vol. 51, no. 4,
[79] A. Arab, A. Khodaei, Z. Han, and S. K. Khator, ‘‘Proactive recovery of pp. 3368–3375, Jul. 2015.
electric power assets for resiliency enhancement,’’ IEEE Access, vol. 3, [106] IEEE Recommended Practice for Test Procedures for High-Voltage Direct
pp. 99–109, 2015. Current Thyristor Valves, IEEE Standards 857-1996, (Revised 857-1990),
[80] S. Dodds, B. Railing, K. Akman, B. Jacobson, T. Worzyk, and B. Nilsson, IEEE, Piscataway, NJ, USA, 1996.
‘‘HVDC VSC (HVDC light) transmission—Operating experiences,’’ in [107] A. Janke et al., ‘‘SVC operation & reliability experiences,’’ in Proc. IEEE
Proc. CIGRE Paris Conf., 2010, pp. 1–9. Power Energy Soc. General Meeting, Minneapolis, MN, USA, Jul. 2010,
[81] M. Cepek and C. P. Krishnayya, ‘‘Thyristor aging,’’ in Proc. IEEE Int. pp. 1–8.
Conf. Power Syst. Technol., vol. 1. Aug. 1998, pp. 12–18. [108] J. Bjorn, K. Patrik, A. Gunnar, H. Lennart, and J. Tomas, ‘‘VSC-HVDC
[82] Components Testing of VSC System for HVDC Applications, document transmission with cascaded two-level converters,’’ in Proc. CIGRE Conf.,
WG B4.48, CIGRE, Paris, Paris, 2011. 2010, p. 69.
[83] M. E. Beehler, ‘‘Reliability centered maintenance for transmission sys- [109] K. Ma, M. Liserre, F. Blaabjerg, and T. Kerekes, ‘‘Thermal loading
tems,’’ IEEE Trans. Power Del., vol. 12, no. 2, pp. 1023–1028, Apr. 1997. and lifetime estimation for power device considering mission profiles
[84] R. A. Valiquette, ‘‘HVDC life extension for Nelson River HVDC sys- in wind power converter,’’ IEEE Trans. Power Electron., vol. 30, no. 2,
tem,’’ in Proc. IEEE Power Eng. Soc. Winter Meeting, vol. 2. Jan. 2002, pp. 590–602, Feb. 2015.
pp. 1059–1062. [110] V. E. Wagner et al., ‘‘Effects of harmonics on equipment,’’ IEEE Trans.
[85] F. Besnard, K. Fischer, and L. Bertling, ‘‘Reliability-centred asset Power Del., vol. 8, no. 2, pp. 672–680, Apr. 1993.
maintenance—A step towards enhanced reliability, availability, and prof- [111] C. Liang et al., ‘‘Harmonic elimination using parallel delta-connected
itability of wind power plants,’’ in Proc. IEEE ISGT Europe, Oct. 2010, filtering windings for converter transformers in HVDC systems,’’ IEEE
pp. 10–16. Trans. Power Del., vol. 32, no. 2, pp. 933–941, Apr. 2017.
[86] N. S. Dhaliwal et al., ‘‘Application of reliability centred maintenance [112] B. R. Andersen et al., ‘‘Voltage source converter (VSC) HVDC for power
(RCM) To HVDC converter station,’’ CIGRE Session, Paris, France, transmission—Economic aspects and comparison with other AC and DC
Tech. Rep. B4-107, 2008, pp. 1–5. technologies,’’ CIGRE Workgroup B4: HVDC Power Electron., Paris,
[87] M. Pecht and J. Gu, ‘‘Physics-of-failure-based prognostics for electronic France, Tech. Rep. 269, B4.37, 2012.
products,’’ Trans. Inst. Meas. Control, vol. 31, no. 4, pp. 309–322, 2009. [113] E. Kontos, G. Tsolaridis, R. Teodorescu, and P. Bauer, ‘‘High order
[88] (2016). Life Time Reliability Solutions: Report on Reliability voltage and current harmonic mitigation using the modular multilevel
Centred Maintenance. Accessed: Jan. 14, 2016. [Online]. Available: converter STATCOM,’’ IEEE Access, vol. 5, pp. 16684–16692, 2017.
http://www.lifetime-reliability.com/cms/consulting/lrs-reliability- [114] L. Hu and R. Yacamini, ‘‘Harmonic transfer through converters and
centered-maintenance-rcm6/ HVDC links,’’ IEEE Trans. Power Electron., vol. 7, no. 3, pp. 514–525,
[89] DfR Solutions. (2016). Physics of Failure for Maintenance Man- Jul. 1992.
agement. Accessed: Jan. 23, 2016. [Online]. Available: http://www. [115] J.-I. Tsai, C.-H. Lin, and T.-P. Tsao, ‘‘Assessment of long-term life
dfrsolutions.com/services/design-review/physics-of-failure/9 expenditure for steam turbine shafts due to noncharacteristic subharmonic
[90] C. S. Gray and S. J. Watson, ‘‘Physics of failure approach to wind turbine currents in asynchronous links,’’ IEEE Trans. Power Syst., vol. 19, no. 1,
condition based maintenance,’’ Wind Energy, vol. 13, no. 5, pp. 395–405, pp. 507–516, Feb. 2004.
2010. [116] R. Yacamini, ‘‘How HVDC schemes can excite torsional oscillations
[91] H. Wang et al., ‘‘Transitioning to physics-of-failure as a reliability driver in turbo-alternator shafts,’’ IEE Proc. C-Generat., Transmiss. Distrib.,
in power electronics,’’ IEEE J. Emerg. Sel. Topics Power Electron., vol. 2, vol. 133, no. 6, pp. 301–307, Sep. 1986.
no. 1, pp. 97–114, Mar. 2014. [117] I. Vernica, K. Ma, and F. Blaabjerg, ‘‘Advanced derating strategy for
[92] Semiconductor Power Devices for use in HVDC and FACTS Controllers, extended lifetime of power electronics in wind power applications,’’
document CIGRE Technical Brochure 112 Working Group 14.17, DC in Proc. IEEE 7th Int. Symp. Power Electron. Distrib. Generat. Syst.
Links and Power Electronic Equipment, 1997. (PEDG), Vancouver, BC, Canada, Jun. 2016, pp. 1–8.
[93] W. Li, E. Vaahedi, and P. Choudhury, ‘‘Power system equipment aging,’’ [118] A. Markus, G. Buticchi, and M. Liserre, ‘‘Study of reliability-efficiency
IEEE Power Energy Mag., vol. 4, no. 3, pp. 52–58, May/Jun. 2006. tradeoff of active thermal control for power electronic systems,’’ Micro-
[94] R. Raineri, ‘‘Asset life and pricing the use of electricity transmission electron. Rel., vol. 58, pp. 119–125, Mar. 2016.
infrastructure in Chile,’’ Energy Policy., vol. 38, no. 1, pp. 30–41, 2010. [119] M. Andresen, M. Liserre, and G. Buticchi, ‘‘Review of active thermal and
[95] A. S. Farag et al., ‘‘Failure analysis of composite dielectric of power lifetime control techniques for power electronic modules,’’ in Proc. 16th
capacitors used in distribution systems,’’ Electr. Power Syst. Res., vol. 44, Eur. Conf. Power Electron. Appl. (EPE), 2014, pp. 1–10.
no. 2, pp. 117–126, 1998. [120] K. Xie, B. Hu, and C. Singh, ‘‘Reliability evaluation of double 12-pulse
[96] D. Chamund, ‘‘IGBT module reliability application notes,’’ Dynex Semi- ultra HVDC transmission systems,’’ IEEE Trans. Power Del., vol. 31,
conductor, Lincoln, U.K., Tech Rep. AN 5945, 2010. no. 1, pp. 210–218, Feb. 2016.
[97] H. Huang, ‘‘Lifetime prediction for power converters,’’ Ph.D. dissertation, [121] S. Zadkhast, M. Fotuhi-Firuzabad, F. Aminifar, R. Billinton, S. O. Faried,
School Eng., Univ. Warwick, Coventry, U.K., 2012. and A. A. Edris, ‘‘Reliability evaluation of an HVDC transmission system
[98] N. Pascoe, Reliability Technology: Principles and Practice of Failure tapped by a VSC station,’’ IEEE Trans. Power Del., vol. 25, no. 3,
Prevention in Electronic Systems. Hoboken, NJ, USA: Wiley, 2011. pp. 1962–1970, Jul. 2010.
[99] Thyristor Valves for High Voltage Direct Current (HVDC) Power [122] F. Shouqiang et al., ‘‘The status and prospects of VSC-HVDC reliability
Transmission—Part 1: Electrical Testing, document IEC Publication research,’’ in Proc. IEEE China Int. Conf. Electron. Distrib. (CICED),
60700-1/Electrical Testing, 1997. Aug. 2016, pp. 1–10.
[100] M. Sampei, T. Yamada, S. Tanabe, H. Takeda, and S. Kobayashi, [123] Ø. Rui, C. Öhlén, J. Solvik, J. Thon, K. Karijord, and T. Gjengedal,
‘‘Secular change in characteristics of thyristors used in HVDC ‘‘Design operation and availability analysis of a multi-terminal HVDC
valve,’’ IEEE Trans. Power Del., vol. 12, no. 3, pp. 1159–1167, grid—A case study of a possible offshore grid in the norwegian sea,’’ in
Jul. 1997. Proc. IEEE PowerTech, Jun. 2011, pp. 1–7.
VOLUME 6, 2018 21101

[124] U. Shipurkar, K. Ma, H. Polinder, F. Blaabjerg, and J. A. Ferreira, ‘‘A CARLOS E. UGALDE-LOO (M’02) was born in
review of failure mechanisms in wind turbine generator systems,’’ in Proc. Mexico City. He received the B.Sc. degree in
17th Eur. Conf. Power Electron. Appl. (EPE), 2015, pp. 1–10. electronics and communications engineering from
[125] K. Fischer et al., ‘‘Field-experience based root-cause analysis of power- the Instituto Tecnologico y de Estudios Superi-
converter failure in wind turbines,’’ IEEE Trans. Power Electron., vol. 30, ores de Monterrey, Mexico City, Mexico, in 2002,
no. 5, pp. 2481–2492, May 2015. the M.Sc. degree in electrical engineering from
[126] L. Yang, P. A. Agyakwa, and C. M. Johnson, ‘‘Physics-of-failure lifetime
the Instituto Politcnico Nacional, Mexico City,
prediction models for wire bond interconnects in power electronic mod-
in 2005, and the Ph.D. degree in electronics and
ules,’’ IEEE Trans. Device Mater. Rel., vol. 13, no. 1, pp. 9–17, Mar. 2013.
[127] M. J. Cushing, D. E. Mortin, T. J. Stadterman, and A. Malhotra, ‘‘Compar- electrical engineering from the University of Glas-
ison of electronics-reliability assessment approaches,’’ IEEE Trans. Rel., gow, Scotland, U.K., in 2009.
vol. 42, no. 4, pp. 542–546, Dec. 1993. In 2010, he joined the School of Engineering, Cardiff University, Wales,
[128] M. Pecht and A. Dasgupta, ‘‘Physics-of-failure: An approach to reliable U.K., where he is currently a Senior Lecturer in electrical power systems.
product development,’’ J. IES, vol. 38, no. 5, pp. 30–34, 1995. His academic expertise includes power system stability and control, grid
[129] M. White and J. B. Bernstein, ‘‘Microelectronics reliability: Physics- integration and control of renewables, HVdc transmission, modeling of
of-failure based modeling and lifetime evaluation,’’ JPL, Pasadena, CA, dynamic systems, and multivariable control.
USA, NASA Tech. Rep. NASA WBS: 939904.01.11.10, 2008.
[130] M. S. Bowman, ‘‘applied economic analysis for technologists, engineers
and managers,’’ Environ. Sci., vol. 38, no. 5, pp. 30–34, 1999.
[131] P. Ladoux, N. Serbia, and E. I. Carroll, ‘‘On the potential of IGCTs
in HVDC,’’ IEEE J. Emerg. Sel. Topics Power Electron., vol. 3, no. 3,
pp. 780–793, Sep. 2015.
[132] D. Tiku, ‘‘DC power transmission: Mercury-arc to thyristor HVdc valves
[history],’’ IEEE Power Energy Mag., vol. 12, no. 2, pp. 76–96, Mar. 2014.
[133] N. G. Hingorani, ‘‘High-voltage DC transmission: A power electronics
workhorse,’’ IEEE Spectrum, vol. 33, no. 4, pp. 63–72, Apr. 1996.
[134] J. Arrillaga, High Voltage Direct Current Transmission. London, U.K.:
Inst. Electr. Eng., 1998.
[135] W. Long and S. Nilsson, ‘‘HVDC transmission: Yesterday and today,’’ JUN LIANG (M’02–SM’12) received the B.Sc.
IEEE Power Energy Mag., vol. 5, no. 2, pp. 22–31, Mar. 2007. degree from the Huazhong University of Sci-
[136] M. P. Bahrman and B. K. Johnson, ‘‘The ABCs of HVDC transmis- ence and Technology, Wuhan, China, in 1992,
sion technologies,’’ IEEE Power Energy Mag., vol. 5, no. 2, pp. 32–44,
and the M.Sc. and Ph.D. degrees from the China
Mar. 2007.
[137] B. Andersen and C. Barker, ‘‘A new era in HVDC?’’ IEE Rev., vol. 46,
Electric Power Research Institute, Beijing, China,
no. 2, pp. 33–39, Mar. 2000. in 1995 and 1998, respectively.
[138] A. Nami, L. Jiaqi, F. Dijkhuizen, and G. D. Demetriades, ‘‘Modu- From 1998 to 2001, he was a Senior Engineer
lar multilevel converters for HVDC applications: Review on converter with the China Electric Power Research Institute.
cells and functionalities,’’ IEEE Trans. Power Electron., vol. 30, no. 1, From 2001 to 2005, he was a Research Associate
pp. 18–36, Jan. 2015. at Imperial College, London, U.K. From 2005 to
[139] J. Liang, O. Gomis-Bellmunt, J. Ekanayake, N. Jenkins, and W. An, 2007, he was a Senior Lecturer at the University of Glamorgan, Wales,
‘‘A multi-terminal HVDC transmission system for offshore wind farms U.K. He is currently a Professor with the School of Engineering, Cardiff
with induction generators,’’ Int. J. Electr. Power Energy Syst., vol. 43, University, Wales. His research interests include FACTS devices/HVdc,
no. 1, pp. 54–62, 2012. power system stability and control, power electronics, and renewable power
[140] T. K. Vrana and S. Energi, ‘‘Review of HVDC component ratings: XLPE generation.
cables and VSC converters,’’ in Proc. IEEE Int. Energy Conf. (ENERGY-
CON), Apr. 2016, pp. 1–6.
TIBIN JOSEPH (M’16) received the B.Tech. and

M.Tech. degrees from Mahatma Gandhi Univer-
sity, Kerala, India, in 2008 and 2011, respectively,
all in electrical engineering. He is currently pursu-
ing the Ph.D. degree in electrical engineering with
Cardiff University, Wales, U.K.
From 2012 to 2013, he was a Lecturer with the
Saintgits College of Engineering, Kerala, India. He PAUL F. COVENTRY (M’07–SM’17) received the B.Sc. and Ph.D. degrees
was also a Marie Curie Early Stage Researcher in electrical engineering from the University of Southampton, Southampton,
with Cardiff University from 2014 to 2017. He has U.K., in 1979 and 1984, respectively. He joined the CEGB in 1988 and
been a Visiting Researcher with CEPRI, Beijing, China, and National Grid, was with the Asset Management Business Unit of National Grid, Warwick,
Warwick, U.K. Since 2016, he has been a Research Associate with Cardiff U.K. His work during that time focused on transmission substation plant and
University, on the ANGLE-DC project. His research interests include DC systems. He has managed a number of Research and Development projects,
transmission and distribution systems, power system stability and control, and is currently the Technical Leader in HVdc technologies with National
subsynchronous oscillations, power electronic converters control, and renew- Grid. He is a Chartered Engineer in the U.K.
able energy integration.
21102 VOLUME 6, 2018

Asset Management Strategies For Power Electronic Converters in Transmission Networks: Application To HVDC and FACTS Devices

Uploaded by

Copyright:

Available Formats

Asset Management Strategies For Power Electronic Converters in Transmission Networks: Application To HVDC and FACTS Devices

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Asset Management Strategies For Power Electronic Converters in Transmission Networks: Application To HVDC and FACTS Devices

Uploaded by

Copyright:

Available Formats

Received February 21, 2018, accepted April 5, 2018, date of publication April 12, 2018, date of current version

Asset Management Strategies for Power

I. INTRODUCTION traditional transmission assets such as power line conductors,

an increased number of power electronic assets has been

VOLUME 6, 2018 21085

monitoring (CM) [26]–[29]. CM systems have evolved over

21086 VOLUME 6, 2018

FIGURE 4. CM and assessment methods for power electronic assets.

VOLUME 6, 2018 21087

21088 VOLUME 6, 2018

VOLUME 6, 2018 21089

Therefore, CM leads to the next business need: developing B. PROACTIVE MAINTENANCE

21090 VOLUME 6, 2018

VOLUME 6, 2018 21091

IV. ESTIMATING FAILURE AND LIFETIME MODES

21092 VOLUME 6, 2018

FIGURE 8. Classification of ageing mechanism for power semiconductor assets.

60700-1 [99]. It states that most manufacturers have given

VOLUME 6, 2018 21093

21094 VOLUME 6, 2018

reliability of the power supply compared to high voltage ac

VOLUME 6, 2018 21095

TABLE 1. Failure modes and mechanisms of power electronic assets.

networks, their corresponding failure rates and repair time

21096 VOLUME 6, 2018

VOLUME 6, 2018 21097

FIGURE 12. Research roadmap in terms of power electronic asset management.

21098 VOLUME 6, 2018

VOLUME 6, 2018 21099

21100 VOLUME 6, 2018

VOLUME 6, 2018 21101

TIBIN JOSEPH (M’16) received the B.Tech. and

21102 VOLUME 6, 2018

You might also like