1 s2.0 S1364032121001805 Main

Renewable and Sustainable Energy Reviews 144 (2021) 110886
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews

journal homepage: http://www.elsevier.com/locate/rser
Offshore wind turbine operations and maintenance: A

state-of-the-art review
Zhengru Ren a, Amrit Shankar Verma b, c, Ye Li d, *, Julie J.E. Teuwen b, Zhiyu Jiang e
a
Department of Marine Technology, Norwegian University of Science and Technology, NO-7491, Trondheim, Norway
b
Faculty of Aerospace Engineering, Delft University of Technology, Delft, the Netherlands
c
SINTEF Ocean AS, Trondheim, Norway
d
School of Naval Architecture, Ocean and Civil Engineering, Shanghai Jiaotong University, 200240, Shanghai, China
e
Department of Engineering Sciences, University of Agder, NO-4879, Grimstad, Norway
A R T I C L E I N F O A B S T R A C T
Keywords: Operations and maintenance of offshore wind turbines (OWTs) play an important role in the development of
Offshore wind turbine offshore wind farms. Compared with operations, maintenance is a critical element in the levelized cost of energy,
Operation and maintenance given the practical constraints imposed by offshore operations and the relatively high costs. The effects of
Maintenance strategy
maintenance on the life cycle of an offshore wind farm are highly complex and uncertain. The selection of
Onsite maintenance
maintenance strategies influences the overall efficiency, profit margin, safety, and sustainability of offshore wind
Maintenance scheduling
Environmental issues farms. For an offshore wind project, after a maintenance strategy is selected, schedule planning will be
considered, which is an optimization problem. Onsite maintenance will involve complex marine operations
whose efficiency and safety depend on practical factors. Moreover, negative environmental impacts due to
offshore maintenance deserve attention. To address these issues, this paper reviews the state-of-the-art research
on OWT maintenance, covering strategy selection, schedule optimization, onsite operations, repair, assessment
criteria, recycling, and environmental concerns. Many methods are summarized and compared. Limitations in
the research and shortcomings in industrial development of OWT operations and maintenance are described.
Finally, promising areas are identified with regard to future studies of maintenance strategies.
With rapid growth in wind power demand over the past decade and
the depletion of land resources, OWTs have become the focus of wind
1. Introduction technology development. Compared with onshore wind turbines, OWTs
have many advantages, e.g., abundant wind resources, lower turbu
1.1. Background lence, substantial space for establishment, lower transmission and dis
tribution losses, less visual impact, and less noise pollution. Given these
Among different renewable energy sources, wind power shows great notable advantages that guarantee reliable energy production, there has
promise due to its relatively high technological readiness level, abun been a rapid increase in the demand of OWTs in the last two decades; see
dant availability, and relatively low environmental footprint. Energy Fig. 1. The first OWT was constructed in Sweden in 1990 [3]. Since then,
harvesting via conventional wind turbines is achieved by converting the OWT projects have proliferated across Sweden, Denmark, the
kinetic energy of the wind into mechanical power through blade rota Netherlands, and the UK [4]. Europe has always been at the front runner
tion, and then into electrical power through generators. Based on their of OWT development, and the evolution of offshore wind capacity in
locations, wind turbines can be categorized as onshore or offshore wind Europe can be clearly seen from Fig. 1(b). By the end of 2019, the UK
turbines (OWTs). Although definitions exist for nearshore wind turbines had the highest total installed capacity of 9945 MW (representing 45.0%
[1] or unconventional turbine technologies such as airborne wind en of the total installed capacity in Europe), followed by Germany with
ergy systems [2], we broadly consider any off-the-coast turbines to be 7445 MW installed capacity [5].
OWTs and focus on three-bladed horizontal-axis technologies in this Despite the steady growth of the OWTs in recent years, their
paper.
* Corresponding author.
E-mail addresses: [email protected] (Z. Ren), [email protected] (A.S. Verma), [email protected] (Y. Li), [email protected] (J.J.E. Teuwen), zhiyu.
[email protected] (Z. Jiang).
https://doi.org/10.1016/j.rser.2021.110886
Received 13 August 2020; Received in revised form 25 January 2021; Accepted 24 February 2021
Available online 30 March 2021
1364-0321/© 2021 Published by Elsevier Ltd.
Z. Ren et al. Renewable and Sustainable Energy Reviews 144 (2021) 110886
three times higher than those of onshore wind farms [20]. High main
List of abbreviations tenance costs are a vital factor that restricts the development of offshore
wind farms. Though the performance of a wind farm degrades over time,
CapEx Capital expenditure reasonable and efficient maintenance strategies and procedures can
CMS Condition monitoring system reduce the downtime caused by aging equipment [21,22]. Hereafter, we
CTV Crew transfer vessel focus on OWT maintenance.
DecEx Decommissioning expenditure
DP Dynamic positioning 1.3. Challenges to OWT maintenance activities
GHG Greenhouse gas
LCOE Levelized cost of energy Maintenance activities are considered one of the most critical tasks
O&M Operations and maintenance for OWTs, and the challenges associated with them are due to many
OWT Offshore wind turbine reasons. First, the distance from an offshore wind farm to a port or shore
OpEx Operational expenditure reduces the accessibility and increases the downtime. The ownership or
RAMS Reliability, availability, maintainability, and safety hiring of a maintenance fleet and an increased number of technicians is
SCADA Supervisory control data acquisition costly. In addition, the complexity of OWTs is high due to the intro
duction of bottom-fixed and floating foundations. Moreover, weather
conditions, especially significant wave heights and wind speeds, limit
the accessibility of OWTs for service vessels and personnel transfer from
development lags far behind that of onshore turbines likely due to the the vessel to the OWT. Offshore access systems with motion-
high cost of power production from OWTs. The levelized cost of energy compensated gangways have been widely applied together with ser
(LCOE), which represents the average life-cycle price of the electricity vice operation vessels in the past decade, although such devices are still
generated from a given power source per megawatt-hour, is employed to heavy and costly [23]. If a maintenance task must be postponed due to
compare different power sources. As of 2018, the LCOE for offshore weather issues, a longer waiting period and greater loss of power gen
wind power is higher than that of other competitive energy resources, eration during downtimewill likely occur. Even without considering the
such as coal, hydro, and nuclear power [9]. Fig. 2 compares the LCOE of effects of weather, OWT maintenance costs are higher than that of
offshore wind power to that of onshore wind power. This figure shows equivalent tasks onshore due to the specialized equipment required.
that the cost of energy produced from onshore wind is still much lower Furthermore, a severe offshore working environment, higher wind
than that of offshore wind, though the deviation is becoming smaller. speed, wave-induced motions, and structural vibrations result in higher
Several strategies have been considered to reduce the LCOE related to failure rates of OWT components. Additionally, the growing size of
offshore wind power, for example, installing turbines in deep waters OWTs in recent decades, which aim to improve power generation effi
farther from shore, as well as installing wind turbines with increased ciency, requires larger and more specific devices for offshore mainte
power capacity and rotor sizes. Although the trend to install larger wind nance and repairs.
turbines provides a number of benefits, these would be counterbalanced Given that it is expected that 50% of electricity demand will be
by higher failure rates, thereby contributing to higher repair and fulfilled by wind energy by 2050, significant amounts of maintenance
maintenance cost [10]. and repair activities will be required in future decades [24]. Accord
The advancement of offshore wind farms is hindered by the harsher ingly, it is equally important to explore the effect of OWT maintenance
conditions to which offshore installations are exposed [13,14], difficult on environmental impact. Hence, the overall aim of a suitable repair and
and expensive maintenance [15], and the inherently unpredictable na maintenance strategy must balance maximizing profitability and mini
ture of wind. Minimizing the total lifetime expenditure of offshore wind mizing environmental impacts, thereby contributing to the sustainable
power is crucial to enhance their competitiveness. As shown schemati development of offshore wind energy over the long run. Based on the
cally in Fig. 3, the total costs that contribute to the total lifetime above discussion, it is clear that OWT maintenance is challenging, and
expenditure of a wind turbine can be divided into three components, i.e., proper maintenance will ensure a decrease in downtime while reducing
capital expenditure (CapEx), operational expenditure (OpEx), and losses in energy output.
decommissioning expenditure (DecEx). CapEx can be further divided The broad topic of OWT O&M can be separated into several unre
into the cost associated with wind turbine components and the cost of lated research questions, such as overall cost management and logistics
associated power production components; OpEx can be subdivided into planning, onsite operations and mechanical designs for specific opera
operating and maintenance costs. tions, and forward-looking evaluation of potential effects. Although
each subproblem has been studied by the researchers and engineers
1.2. Importance of maintenance from corresponding disciplines, an amalgamation of these technologies
is still in its infancy. Therefore, the target of this review is to provide a
Operating and maintenance (O&M) costs accounts for a large portion comprehensive framework for interested researchers and engineers with
of the LCOE of an offshore wind farm, constituting 23% of their total different backgrounds to gain a broad picture of OWT O&M.
investment cost, compared to only 5% for onshore wind turbines [18,
19]. Hence, reducing O&M costs is an effective way to control the LCOE. 1.4. Scope of this review
Compared to operating costs, maintenance costs are more important
in controlling the LCOE. In the composition of O&M costs, equipment This paper reviews the state-of-the-art research of OWT operations
costs are highest, followed by revenue losses. These two costs are and maintenance, including strategies, planning, onsite operations, and
explicitly associated with maintenance costs, with the former repre assessment criteria. Promising areas are identified concerning the future
senting the direct cost of maintenance, and the latter being associated development of maintenance strategies. Furthermore, the negative im
with the cost resulting from a lack of maintenance [18]. pacts of offshore maintenance on greenhouse gas emissions marine
Maintenance has a strong influence on downtime duration over the wildlife, and waste recycling are discussed. This review presents a
lifetime of an offshore wind farm and consequently contributes comprehensive overview of the literature on OWT maintenance (see
considerably to the LCOE. Maintenance activities of any engineering Fig. 4) and provide a basis for the development of maintenance strate
structures involve regular inspections and repairs to correct any failure gies in the future for offshore wind power facilities. Research gaps are
or replace faulty components. OWT maintenance costs vary with foun also identified through gathering and comparing many scientific pub
dation type and location. In general, the maintenance costs are two to lications, technical reports, and open databases.
2
The review is structured as follows. In Section 2, several maintenance presented and conclusions are drawn regarding these maintenance-
strategies are introduced and discussed, including their development, related issues.
benefits, shortages, and challenges, and critical factors that affect
maintenance costs are analyzed. Based on the selected maintenance 2. Maintenance strategies
strategy, optimal maintenance routing and scheduling are discussed in
Section 3. Several aspects of the associated optimization problem are An effective and reliable maintenance strategy is an indispensable
discussed including their developments and limitations. Onsite main part of OWTs’ daily operations. Since technicians have to visit the wind
tenance activities are summarized in Section 4. Three onsite operations farm from a port, it is impossible to achieve around-the-clock operations
are introduced, i.e., transferring, docking operation, and lifting opera without any interruptions of onsite maintenance. To prevent a failure
tion. Maintenance and repair of two vulnerable components, namely, from occurring, a maintenance team should visit the wind farm
the blade and the gearbox, are highlighted. Numerical analyses are frequently. However, unnecessarily frequent visits, on the one hand, are
conducted to evaluate the operational safety and critical environmental inefficient and expensive due to the high amount of maintenance vessels
conditions. In Section 5, the environmental impacts of OWT O&M are and personnel required. On the other hand, a lower visit frequency may
discussed, such as greenhouse gas emissions, negative impacts on ma result in a higher failure rate and, consequently, longer downtime.
rine wildlife, and waste recycling. In Section 6, further discussion is Therefore, maintenance frequency is a trade-off among risks, vessel
Fig. 1. (a) Global wind capacity (data sourced from Ref. [6]) and (b) newly installed capability in European countries and the percentage of OWT between 2009 and
2019 (data sourced from Refs. [7,8]).
3
Fig. 2. Comparison of LCOE for onshore and OWTs between 2009 and 2019 (data sourced from Refs. [11,12]).
long-term operation and aging, and others are caused by short-term

overload and sudden breakdown [26]. Since the rotor and drivetrain
rotate, and the structures are exposed to waves, the failure rates are
frequently caused by wear and fatigue during operation, and some
failures are considered to happen randomly without explicit trends and
predictions. The major failures of these components are listed as follows:
• Rotor and blade: deterioration, adjustment error, rotor imbalance,

blades and hub corrosion, crack, and serious aeroelastic deflections
[27–29];
• Shaft: shaft imbalance, shaft misalignment, shaft damage, and
broken shaft [30];
• Gearbox: wearing, fatigue, pitting, gear tooth damage, braking in
teeth, eccentricity of toothed wheels, displacement, oil leakage,
insufficient lubrication, high oil temperature, and poor lubrication
[31];
Fig. 3. Cost breakdown of a floating wind turbine (sourced from Refs. [16,17]). • Generator: overspeed, overheating, wearing, excessive vibration,
rotor asymmetries, bar break, electrical, problems, insulation dam
capacities, human resources, and so forth. A successful maintenance age, slip rigs, winding damage, and abnormal noises [32];
strategy aims to maximize economic benefit, extend components life • Bearings: overheating, spalling, wear, defect of bearing shells, and
spans, reduce the number of emergency repairs, decrease overtime labor bearing damage [33];
costs, and relieve the working stress of unpredictable equipment • Nacelle: fire and yaw error [34];
failures. • Tower: fatigue, vibration, foundation weakness, and crack formation
Maintenance strategies are typically categorized as corrective [35–37];
(reactive) maintenance, proactive maintenance, and opportunistic
maintenance according to when maintenance is conducted [25]. These
classifications are shown in Figs. 5 and 6. The meanings of the color
changes between the different lines are:
• From green to red: the wind turbine stops due to a failure;

• From red to green: the wind turbine is repaired and can continue to
work;
• From blue to orange: a maintenance vessel is used to execute tasks;
• From orange to blue: a maintenance vessel is back at the port and
waits for new tasks.
Details of these classifications are explained and discussed in sections

2.2–2.4.
2.1. OWT failure modes

Fig. 5. Classification of maintenance strategies.
Failures can be categorized into two sources; i.e., some are caused by
Fig. 4. Development of maintenance strategies for an offshore wind farm.
4
Fig. 6. Diagrams of maintenance strategies (The green, red, and yellow colors denote the normally operated OWT, the stopped OWT due to failures, and the stopped
wind turbine due to maintenance, respectively. The blue and orange colors stand for the waiting maintenance vessel and the vessel performing tasks, respectively.).
• Electrical system: short circuit, component fault, bad connection, offshore wind farms due to a high failure rate and relatively low sys
contamination, and arcs [38]; tem reliability [41]. Unexpected failures may cost more than expected
• Mooring system: mooring line breakage and fatigue [39]. downtime. In addition, the marine environment reduces accessibility
and decreases reliability; for example, a failure may be noticed by the
There are many critical components in OWTs, and their failure rates maintenance team after a long downtime (see Fig. 6(a)).
vary. The failure rates depend on the location of the wind farm, foun
dation type, and drivetrain type. The failure rate increases and reli
2.3. Proactive maintenance strategy
ability decreases with the application of less mature techniques, i.e.,
larger scale and more complex drivetrain. The total failure rates for
Proposed in the early 1970s, proactive maintenance is a more
direct-drive and indirect-drive systems are nearly identical, but the
advanced approach [42] where scheduled inspection and replacement is
failure rates for different components vary. Compared with a direct-
carried out before failure to prevent minor faults from developing into a
drive wind turbine, the failure rates of the gearbox, inverters and elec
major failure. Major failures (only 25% of all failures) contribute to 95%
tronics, and generator in an indirect-drive wind turbine are higher,
of downtime [43]. Proactive maintenance is a relatively mature tech
lower, and lower, respectively [40].
nique, and proactive maintenance strategies mainly comprise preven
tive and condition-based maintenance strategies.
2.2. Corrective maintenance strategy
2.3.1. Preventive maintenance strategy
Corrective maintenance, or reactive maintenance, is a failure-based A preventive strategy usually refers to scheduled maintenance that
maintenance strategy in which maintenance is carried out only when takes place at (i) a predetermined period, or (ii) a given level of power
a failure has already occurred; see Fig. 6(a). The corrective maintenance generation.
strategy can effective achieve high availability while avoiding unnec
essary maintenance visits and inspection. It is thus suitable for a system (i) The selection of a planned intervention depends on the reliability
with negligible downtime loss. However, the corrective maintenance of each component and the overall cost. If a failure happens be
strategy turns out to be impractical and undesirable for large-scale tween two intervention intervals, the wind turbine will remain
5
out of operation until the next planned visit, as shown in Fig. 6(b). This type of maintenance is also guided by the status of the components.
Thus, it is possible to carry out repairs and regular maintenance Maintenance repairs occur when a failure occurs, as shown in Fig. 6(c).
in the meantime, which achieves efficient use of resources. The aim is to prevent major failures from happening [52]. Maintenance
Because the maintenance cost of different components varies repairs are used in the prospective health condition maintenance, and
markedly, increasing reliability and mitigating expensive main allow the planning and selection of the most effective repair methods
tenance tasks will help minimize maintenance cost. The number based on the wind turbine’s condition, faults, the costs of maintenance,
of planned intervention intervals in a year is calculated by resource depletion, and production efficiency. Asensio et al. [53] eval
considering capacity factors, weather-related accessibility, and uated the economic viability of a predictive maintenance strategy from
levelized production cost of each site [41]. the perspective of the life-cycle cost of CMS. The model takes account of
(ii) A preventive maintenance strategy that considers power gener the investment costs and O&M of the CMS and the cost reduction due to
ation considers the effect of power generation rate on the degree CMS implementation. Walgern et al. [54] compared a condition-based
of deterioration on the turbine and consequently on the mainte maintenance strategy with corrective and preventative maintenance
nance strategy [44]. strategies and found it to achieve the best performance of many
methods. Combining CMS with weekly scheduled maintenance was
The goal of preventive maintenance strategy is to optimize the pro shown to be the most cost-effective approach. Condition-based main
duction plan and the economic maintenance plan. Compared with tenance strategies minimize maintenance costs and increase OWT reli
corrective maintenance, the advantages of this strategy are (1) elimi ability, while the monitoring devices require extra costs. Many condition
nation of unplanned maintenance, (2) availability of a sufficient main monitoring techniques applied to monitor and inspect the components
tenance weather window, (3) minimization of the effect of in a wind turbine are listed in Table 1, and include vibration, acoustic
unpredictable weather, (4) reasonable use of service vessels, (5) avoid emission, ultrasonic measurement, and thermography techniques [55].
ance of excessive spare stock, (6) combined maintenance and repairs, (7) Sensors play a significant role in CMS. Many types of sensor systems
optimization of maintenance tasks, and (8) contribution to an effective have been introduced to analyze OWT system performance, and their
asset maintenance plan [45]. prices have gradually decreased in recent decades. Sensor measurements
Preventive maintenance tasks can be planned based on the age provide technicians with a clear and comprehensive image of the OWTs’
groups of different components. An optimum selection of extreme age real-time conditions. The topics of structural health monitoring, feature
thresholds and a number of age groups allows maintenance costs to be extraction, and fault detection have been intensively reviewed in early
minimized by reducing the setup and labor costs of repeated visits. This studies, e.g., Refs. [30,55–57]. Surveys of specific wind turbine com
approach is preferred for large offshore wind farms that require repet ponents are proposed, e.g., bearing [58], generators [59], gearbox [60],
itive maintenance. This age-based method is also used by Santos et al. energy conversion systems [61], and drivetrain [62] have also been
[46] with imperfect repairs and is compared with a corrective mainte proposed.
nance strategy and a classic preventive maintenance strategy with fixed According to measured data, frequency-frequency analysis is widely
time intervals. In that study, the preventive maintenance strategy used in fault detection and isolation, e.g., Fourier transformation and
considered that the age reduction ratio contributes to cost reduction wavelet transformation. The costs and levels of and deployment of these
with regard to the use of large vessels (55.51%) and replacements techniques are presented in Fig. 7. Visual inspection cannot achieve on-
(60.28%). Although the cost of the supply vessels and crew increased by line monitoring since it is impossible for a technician to remain at an
166.4%, the overall benefits are significant, yielding a total cost OWT. It is noted that the level of deployment declines with the cost.
reduction of 24.2%. Hence, attention should be placed not only on newly developed tech
Some efforts have been made to improve the preventive maintenance nologies but also on the budget reduction of existing solutions.
strategy. Dui et al. [47] proposed a cost-based measure to identify the The advanced data collection techniques provided by supervisory
maintenance priority of a component based on the joint effect of control data acquisition (SCADA) and CMS are significant due to their
component reliability and maintenance cost on system reliability. Nejad roles in the supervising operational conditions, thereby increasing reli
et al. [48] applied a the reliability-based maintenance strategy to ability and optimizing maintenance plans [65]. In addition, the
gearbox components that have a higher probability of fatigue failure and involvement of condition monitoring can improve planning and avoid
a lower level of reliability. The authors proposed a “vulnerability map” over-maintenance or under-maintenance. For example, the remaining
to reduce downtime and increase the efficiency of finding faulty com useful life could be predictive based on condition monitoring data [66].
ponents during routine inspection and maintenance. Many factors affect the performance of a condition-based mainte
Preventative maintenance strategies can frequently be described as nance strategy, such as the CMS detection rate and the false alarm rate.
an optimal maintenance scheduling problem, which should aim to May and Mcmillan [67] investigated the effects of these two factors and
reduce the maintenance cost and increase OWT availability without pointed out that an increase in the number of false alarms resulting from
threats to the system, ship crews, or the environment [49]. One effective a decrease in the reliability of the CMS will lead to a reduction in the
method is to optimize the selection of the preventive maintenance in availability of the wind farm. One way to improve the fault detection
terval, which will be reviewed on more detail in Section 3. success rate is to add more CMSs to the system. May et al. [68] per
formed an economic analysis of improvements in the use of CMS. Among
2.3.2. Maintenance strategy using sensors various approaches in which CMSs are added to the drivetrain, gearbox,
and generator, or the tower or the blades, only the additional blade
2.3.2.1. Condition-based maintenance strategy. OWTs are prone to CMSs improve the cost-effectiveness of the maintenance strategy [69].
deterioration due to fatigue, corrosion, erosion, and wear. Combined There is a 95% improvement compared with the use of a CMS on the
with a risk-based life-cycle approach based on the per-posterior drivetrain alone, taking both fault detection and the extra expense of the
Bayesian decision theory, condition-based maintenance, which is also additional CMS into consideration. In Ref. [70], the geographical clus
referred to as predictive maintenance, can be used to observe the degree tering of OWTs, such as the layout of the wind farm, is considered in
of the deterioration and thus increase the reliability of predictions [50, order to optimize a condition-based maintenance strategy. Dividing
51]. wind turbines into different clusters based on the optimum offshore
Condition-based maintenance is a strategy that combines relevant wind farm layout leads to further improvements in the convenience of
information measured by a condition monitoring system (CMS) and the maintenance.
results of an online or offline health diagnosis or fault analysis system. The level of automation and intelligence is thus improved. Data-
driven approaches, e.g., machine learning, has become popular in
6
Table 1
The monitoring and analysis methods to different components.
Nacele Tower Blade Bearings Shaft Gearbox Generator
Vibration analysis ✓ ✓ ✓ ✓ ✓ ✓
Torsional vibration ✓ ✓
Acoustics Emission ✓ ✓ ✓ ✓ ✓
Oil analysis ✓ ✓ ✓
Strain measurement ✓ ✓
Optical fiber monitoring ✓
Electrical effects ✓ ✓
Temperature ✓ ✓ ✓ ✓
Ultrasonic testing techniques ✓ ✓
Thermography ✓ ✓ ✓ ✓ ✓ ✓
Visual inspection ✓ ✓ ✓ ✓ ✓
Radiographic inspection. ✓ ✓
Generator power output ✓
Fig. 7. Costs and deployment levels of different wind turbine condition monitoring techniques (sourced from Refs. [56,63,64]).
recent years, and it has been applied to optimal maintenance scheduling maintenance should be performed. By combining measured data and
[71]. Supervised learning is the most widely used approach. A black-box virtual models, failures can be predicted before they occur. This method
neural network model is trained to fit the labeled data, and the network can be applied to both OWTs and service vessels. Although several
can be applied to conduct various analyses, classification monitoring, digital-twin platforms have been proposed [73–77], systematic and
and prediction [71]. This approach is especially suitable for scenarios convincing research is still lacking.
that are difficult to model due to high complexity and uncertainty.
However, there are a few shortcomings of learning approaches. First, the 2.3.2.3. Limitations. The use of sensors in practical applications is
method highly relies on the quality and quantity of measured data. A challenging due to their growing number. Although, more sensors
lack of necessary measurements degrades the neural network. Addi markedly improve the measurement accurately and redundancy, they
tionally, it is hard to prove stability. The network architecture influences also markedly increase system cost and complexity, and introduce new
the computational speed and robustness. If the failure scenarios are not problems, such as sensor failures and misreporting. Studies of effective
included in the trained data, failure can hardly be detected. There is no and robust approaches to fuse sensor signals and handle fault conflicts
guarantee that the key parameters tuned in the design period work well are remain to be performed. Wang et al. [78] introduced a monitoring
in practical applications since neural networks are not good at system for use in a condition-based maintenance strategy with a SCADA
extrapolation. database to collect and analyze monitoring information. The former
provides low-resolution monitoring to supervise the operation of the
2.3.2.2. Predictive maintenance strategy. Compared with condition- wind turbine, collects data, and alarms; the latter is employed to di
based maintenance, another similar but more advanced proactive agnose and predict subassembly faults through high-resolution moni
maintenance strategy is predictive maintenance. According to sensor toring [65]. However, it is challenging to distinguish whether a fault is
measurements, parametric analyses are conducted to determine when real or fake using SCADA analysis; thus, accuracy and robustness should
maintenance should be performed before a failure occurs; see Fig. 6(d). be improved by employing more advanced fault detection algorithms
The main idea is to minimize the downtime and maximize the reliability; and artificial intelligence.
i.e., the maintenance events are conducted when they are indeed The extensive monitoring of turbine conditions and supervision of
necessary. Though the associated equipment cost is higher, the benefits mechanical performance generates large quantities of data, in addition
of this strategy include reduced maintenance frequency and time, to the O&M information recorded during the turbine lifespan. The
downtime, and cost of spare parts and supplies. problems of collecting, filtering, analyzing, and storing these large
Digital-twin platforms are the latest popular research topic and are amounts of information have received much attention.
used to predict the remaining useful life of OWT components [72]. One shortcoming of existing data-collection schemes is the lack of
Practical physics and virtual models are paired to predict when detail that they record: merely recording failed components is far from
7
satisfactory. Reliability, availability, maintainability, and safety (RAMS) Zhang et al. [85] calculated a scheduled time for preventive mainte
databases have been proposed to provide more detailed information, nance based on reliability requirements and determined the opportu
such as the causes of failure, corresponding maintenance tasks, and the nistic maintenance interval by optimizing the total maintenance cost.
effects on future failure behaviors. This database serves as a basis for This method reduced downtime and overall maintenance costs
condition-based maintenance by determining the periods of preventive compared to the classic preventive maintenance strategy.
maintenance and contributing to maintenance planning, scheduling The opportunistic preventive maintenance strategy replaces failed
optimization, life-cycle cost minimization, and profit analysis [79]. Data components and takes the opportunity to replace or maintain operating
stored in a RAMS database will also serve as essential input to determine components preventively when onsite [86]. Group maintenance plan
and design a function-behavior-state model and functional redundancy ning is determined by the optimal maintenance plan for each individual
designer [79]. components [87] and maintenance cost [88].
Regarding the incompleteness of current operational data collection OWTs often suffer from the internal system deterioration and
and the loss of valuable data resulting from the rescaling of traditional external damages due to the harsh offshore environment. Shafiee et al.
databases, methods designed for big data are used to manage detailed [66] proposed an opportunistic condition-based maintenance strategy
operational data collection and reuse [80]. All data can be stored in a for multiple-blade OWTs subjected to deterioration and environmental
data chain. The streaming data processing tools employed allow the use shocks and verified that the strategy can reduce maintenance setup
of more sophisticated wind-farm-level alarms and warnings. The scal costs, greenhouse gas emissions, and O&M costs. Data collected by a
ability of these methods allows all historical data to be considered with SCADA system was also used to verify the proposed algorithm.
no need for data archiving. Hence, these methods can manage growing Both opportunistic preventive maintenance [85,86,89–91] and
wind farms predictably due to the comparatively simple and opportunistic condition-based maintenance [66,91,92] are described in
cost-effective features of the extended distributed big data systems the literature. Based on monitoring systems, condition-based mainte
compared with the traditional databases. nance has recently been extended to become opportunistic. Maintenance
Cyber-security is another critical issue in practical applications, e.g., should be conducted when the designed maintenance index reaches a
remote sensing. The digital network and rapid development of remote given threshold. If this threshold varies, the strategy is called dynamic
communication have significantly enhanced OWT O&M convenience opportunistic maintenance [91]. Maintenance costs can be dynamic.
and efficiency. However, cyber-security in the wind industry is rela Zhang et al. [91] reported that the dynamic opportunistic maintenance
tively unexplored, and issues of concern likely include information strategy yields 11% and 18% decreases in life cycle O&M costs
disclosure and cyber attacks. The system puts the reliability of the grid at compared with a static opportunistic maintenance strategy and a strat
a major risk. Systematic improvement and design are needed. egy that does not consider opportunistic maintenance, respectively.
Grouping periodic maintenance planning and reactive maintenance is
2.3.3. Summary studied in Zhu et al. [93].
A summary of different maintenance strategies is tabulated in
Table 2. 3. Optimization models for maintenance planning
Ensuring system reliability and minimizing the maintenance LCOE

2.4. Opportunistic maintenance strategy
represents a complex management problem with a number of un
certainties when considering a long-term perspective. There are many
The first opportunistic maintenance strategy was proposed in the
time-varying, unpredictable, or partly unpredictable factors, including
1960s [81], and the concept has since been extended and developed
the environment and climate, management, aging, supply chain, elec
since then. However, its definition is still not consensually defined [82].
tricity price fluctuations, technology advancements, risk analysis, in
The notion of opportunistic maintenance is often referred to as a
terest rates, political tendencies, and the global market. Therefore, most
grouping of diverse planned preventive maintenance tasks or the com
maintenance policies and decision-making algorithms tend to model and
bination of preventive and corrective maintenance actions. Different
maximize short-term benefits, i.e., ensuring that the maintenance fleet
types of maintenance tasks are typically scheduled within the same
and OWTs work efficiently.
period, or even during the same visit [83]. For example, additional
The optimum scheduling of maintenance tasks and fleet routing must
unplanned service actions that should be undertaken in the future are
consider costs, weather, maintenance intervals, personnel, downtime,
carried out together with a planned service at its corresponding down
repair time, and fleet size. (i) Maintenance scheduling refers to the
time when a failure occurs or when the reliability of a component rea
detailed arrangement of maintenance tasks for a set of target OWT
ches its predetermined preventive maintenance threshold. The
during recommended periods while considering environmental condi
maintenance team can take the opportunity to maintain other healthy
tions, resource availability, and the loss of revenue due to turbine fail
components whose maintenance thresholds have not yet been reached.
ure. (ii) Route planning refers to the choice of an optimum route for each
By taking advantage of wind forecasts and corrective maintenance of
service vessel to perform maintenance tasks for a group of target OWTs
low power generation periods or of unexpected failures to perform
within a specified weather window. The service objective expands from
preventive maintenance, the opportunistic preventive maintenance
one O&M base and one wind farm to multiple O&M bases and multiple
strategy leads to a 43% reduction in preventive maintenance cost [84].
Table 2
Comparison among different maintenance strategies.
Corrective maintenance Preventive maintenance Condition-based maintenance Predictive maintenance
Trigger Failure Planned date Real-time measurement Real-time measurement

Initial cost Low Medium High High
Operating cost High Medium Medium Low
Number of failures High Low Medium Low
Unnecessary visits High Medium Low Low to medium
Unplanned maintenance Low Low High Medium
Maintenance regarding failures After Before or after Shortly after Before
Downtime High Medium Medium Low
Level of automation Low Low to medium Medium to high High
8
wind farms while considering the number of available technicians and minimization problem.
spare parts, and the capacities of the service vessels. Once the schedule The quantitative model is based on a series of assumptions and
has been determined, service vessels are selected; routes are planned for simplifications and must be simplified with an educated guess. Signifi
each vessel to access the corresponding wind turbines, and personnel are cant amounts of models have been proposed to describe this process
assigned. Optimal route planning is achieved by balancing energy effi [97], and most are deterministic linear models that can be solved with
ciency and time consumption. Sea currents and winds are the primary commercial solvers. While some considerations introduce nonlinearities
environmental parameters that affect this problem. The environment is to a part or parts of the cost functions and constraints. Quantitative
typically assumed to be steady [94] or spatiotemporally variant Niu coefficients are determined based on project experiences and historical
et al. [95]. Some other problems can include, e.g., the optimal vessel data. However, other aspects are difficult to evaluate if they are
fleet composition [96]. time-varying, uncertain, or nonlinear. Table 3 lists the similarities and
Maintenance strategies can be solved as optimization problems. In differences among some state-of-the-art decision support algorithms
this section, cost functions and constraints are discussed, and the that perform optimal maintenance scheduling with respect to the opti
development and limitations of associated methods are reviewed. mization problem/solutions.
3.1. Optimization problem 3.1.1. Cost function

This section begins by discussing single-objective optimizations.
The core of route planning and maintenance scheduling is a con Single-objective cost functions are time-based [98], cost-based [83,86,
strained optimization problem. A standard form is given by 87,99–105], reliability-based [106], or sometimes availability-based.
For an OWT, reliability and availability are related and similar, but
min f (x)
x not equivalent. The most widely used cost functions are cost minimi
s.t. gi (x) ≤ 0, i = 1, ⋯, m (1) zation and reliability maximization.
hj (x) = 0, j = 1, ⋯, n The most widely studied optimum assignment is determined in terms
of total maintenance costs, which are a sum of revenues and penalty
where m ≥ 0 and n ≥ 0. Two parts are elementary to the constrained
costs. Zhang [100] and Dai et al. [83] only consider the fundamental
optimization problem in eq. (1), i.e., the cost function(s) (f(x)) and
travel cost and downtime penalty cost. The travel cost relates to travel
constraints (gi (x) and hj (x)). According to the objective functions, the
distance and vessel capacity. The lost revenue in the downtime is
optimization problems can be categorized into single-objective and influenced by the types of maintained components, technician skills,
multi-objective optimizations. The optimum solution minimizes the cost electricity price, wind speed, and OWT size. Additionally, the cost of a
functions under a number of inequality (gi (x)) and equality constraints repair or replacement is considered in the cost function in some studies.
(hj (x)). In this way, a maximization problem can be transferred to a
Table 3
Comparison of optimal routing and scheduling approaches (Duo-ACO:duo-ant colony optimization, MIP:mixed integer programming, MO:Multi-objective optimiza
tion, GA:genetic algorithm).
[99] [100] [83] [101] [102] [103] [86] [106] [104] [87] [108] [105] [96] [112]
Cost Downtime cost ⨯ ✓ ✓ ✓ ✓ ✓ ⨯ ✓ ✓ ✓ ✓ – ✓ ⨯

Cost of vessels + personnel ⨯ ⨯ ⨯ ✓ ✓ ✓ ⨯ ⨯ ✓ ✓ ✓ – ✓ ✓
Travel/transportation cost ⨯ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ – ✓ ✓
Fixed cost ⨯ ⨯ ⨯ ✓ ✓ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ – ✓ ✓
Incompleted maintenance ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ ✓ ⨯ ⨯ – ✓ ⨯
Equipment maintenance cost ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ✓ ⨯ ✓ ⨯ – ⨯ ✓
Adjustment cost ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ ⨯ – ⨯ ⨯
Startup + CRM cost ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ – ⨯ ⨯
Mother ship cost ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ ⨯ – ⨯ ⨯
Consideration Multiple vessels ⨯ ✓ ✓ ✓ ✓ ✓ ⨯ ✓ ✓ ⨯ ✓ ✓ ✓ ✓
Multiple ports ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ✓ ⨯
Multiple services/ ⨯ ⨯ ⨯ ✓ ⨯ ✓ ✓ ⨯ ✓ ✓ ⨯ ✓ ✓ ⨯
components
Multiple OWT farms ✓ ✓ ✓ ✓ ⨯ ⨯ ✓ ✓ ✓ ✓ ✓ ✓ ⨯ ⨯
Multiple technician types ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ⨯ ⨯
Weather condition ⨯ ⨯ ⨯ ✓ ✓ ✓ ⨯ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Parallel maintenance tasks ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ ⨯
Sensor update ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯
opportunistic ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ✓ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯
Constraints Visit every farm only once ⨯ ✓ ✓ ✓ ⨯ ⨯ ⨯ ✓ ✓ ⨯ ✓ ✓ ⨯ ⨯
Limited Leave and return ⨯ ✓ ✓ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ ⨯
harbor
Vessel capacity ⨯ ✓ ✓ ✓ ✓ ✓ ⨯ ✓ ✓ ⨯ ⨯ ✓ ✓ ✓
Maximum offshore/travel ✓ ✓ ✓ ✓ ✓ ✓ ⨯ ⨯ ✓ ⨯ ⨯ ✓ ⨯ ✓
time
Personnel onboard ✓ ✓ ✓ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ ✓ ✓ ✓ ✓
Total vessels in base ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ✓ ✓ ⨯ ✓ ✓ ✓ ⨯
Waiting period ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯
Greenhouse gas emission + ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ✓
wildlife
Seasonal constraints ✓ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯
Total no. of constraints 16 17 19 29 – 17 – 10 62 – 16 – 8 14
Loss function Cost ⨯ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Reliability ✓ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓ ⨯ ⨯ ✓ ⨯ ⨯ ⨯
Power generation ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ⨯ ✓
Solver – Duo- MIP MIP – MIP – – MIP – MO GA MIP MO
ACO
9
The costs of the service fleet and associated personnel can be categorized simulation model is proposed by Ref. [116] to combine a continuous
into fixed and variable costs. Fixed costs, which are independent of the system dynamics model, a discrete agent-based simulation (ABS), and a
vessel usage, can include the cost of lease contracts, onshore/offshore discrete-event simulation (DES) [117,118].
bases, and the maintenance team; variable costs depend on how much
the vessel is used. Adjustment cost is related to schedule uncertainties 3.2. Development
due to the weather changes or other unexpected situations. Unplanned
downtime and speed losses are considered in Krokoszinski [107]. The optimal-scheduling problem is modeled and solved using a more
Compensation cost for incomplete tasks is studied in some studies [85, realistic and flexible perspective. A decade ago, this problem was still
87,101]. Raknes et al. [104] proposed a mathematical model that can modeled using static and deterministic parameters, while dynamic and
model several work shifts and corresponding vessels, and accurately stochastic methods become increasingly popular. In addition, the
calculate revenue losses resulting from turbine failures. This model can complexity of the optimal scheduling problem grows in time due to the
also be employed to evaluate decisions regarding vessel size and mix, as increasing number of considered factors.
well as the consequences of these decisions. Startup cost and customer Possible realistic operations and issues were not considered in early
relationship management cost are only considered in Zhong et al. [108] studies. The research was extended to multiple port bases [83,105],
and Hajej et al. [109]. Unlike a hydroelectric or a thermal power plant, multiple types of maintenance services [86,87,89,101,103–105], and
the startup cost of OWT maintenance scheduling is not significant since a multiple types of technicians [101]. All these factors greatly increase the
wind turbine can startup in 60 s [110]. From a long-term perspective, optimization complexity, resulting in more complex cost functions and
the spare-parts inventory cost affects the OWT life cycle cost and is increasing numbers of variables and constraints; see Table 3. For
related to the ordering, purchasing, and holding costs [91]. example, Irawan et al. [119] overcame the limitations of previous
In other studies, reliability criteria are considered. There are several models, that are restricted to a single O&M base and a single wind farm,
critical components and services for each component in an operational by proposing a model and solution for multiple O&M bases and wind
wind turbine. The system reliability can be calculated as the average of farms at different locations, which is more representative of real OWT
components’ individual reliabilities [108] or using fuzzy system reli developments. Dai et al. [83] considered only small cases, in which only
ability [111]. four, six, or eight OWTs required maintenance. In another study of this
However, reliability (or availability) maximization and cost mini topic, Zhang [100] considered the priority of the maintenance tasks and
mization are inversely related, resulting in the abovementioned optima suitable environmental conditions, and proposed the application of a
being partial and circumscribed in only one aspect. To overcome these duo ant colony optimization method to the scheduling and routing of a
limitations, another type of single-objective cost function that minimizes maintenance fleet for offshore wind farms. This method performs well,
the cost/reliability (or cost/availability) ratio can be used. As an alter even with many OWTs. The most popular approach currently is to
native approach, multi-objective optimization has been used in recent decompose the routing problem into a master problem (allocating routes
studies, where cost functions are cost-reliability-based [106,108,111], to each vessel) and subproblems (producing new routes) [119].
cost-power-based [90,112], and cost-reliability-availability-based Spare-parts inventory management is sometimes considered [91,120],
[113]. because spare parts are not always available.
Cooperative maintenance and fleet sharing enhance overall main
3.1.2. Constraints tenance efficiency. Maintenance tasks in parallel are studied in Raknes
The cost functions are restricted by specific constraints to achieve et al. [104]. The fleet leaves maintenance personnel at a specific OWT
specific considerations and ensure the practical meanings of optima. The and continues onto other wind turbines/farms. The technicians are then
number of constraints grows with the number of considerations and picked up after finishing their maintenance tasks. Some studies consider
requirements. To satisfy a specific requirement, several constraint reliability with [99] or without costs [106,108]. A study uit het Broek
inequations/equations are needed. Table 3 summarizes the most widely et al. [121] showed that the vessel and harbor sharing policy greatly
used selection of requirements. For a specific vessel, its maximum reduces overall maintenance costs.
offshore travel and maintenance time is affected by its loading capacity, Time-varying parameters have been considered in recent studies,
and the total number of onboard technicians are known. The accessi such as time-varying power harvesting [109], time-varying maintenance
bility of each wind farm depends on the vessel capacity and environ cost [122], and the time-varying reliability threshold of maintenance
mental conditions. The frequency of leaving and entering harbors is also [91]. Since the failure rate of an OWT increases over its lifespan, these
constrained in the optimization in case of unnecessary high-frequency factors grow linearly or exponentially with time. The problem can be
fluctuations in the optima. The total number of serving vessels must solved by transforming continuous variables in a discretized set.
be predetermined. For overnight service, whether technicians will be More realistic environmental models have been developed, where
back to onshore or accommodated onboard was to be determined. significant wave heights and wind speeds are critical parameters. The
Seasonal constraints also exist because maintenance is not allowed weather forecast can be assumed to be perfect for occasional travel, but
during certain periods. Environmental effects, e.g., greenhouse gases its uncertainty surges with a longer time windows. A one-time route and
and seabirds, are taken into account in only a small part of works [108, regular routes may not be the same due to weather uncertainties. Sto
112]. chastic modeling and Monte Carlo simulations are thus widely adopted,
more variations appear in the windy environment. In addition to the
3.1.3. Solver mean wind speed, gust measurement and estimation are introduced to
After constructing the optimization problem, the next step is to solve minimize the total duration of scheduled tasks [98]. Wind direction and
the problem and find the optima. Most linearly constrained program wake effects are modeled to improve the local OWT maintenance order
ming problems are solved by commercial solvers, such as Xpress, using in a specific wind farm [112]. The benefits of weather measurement and
mixed-integer programming. However, nonlinear cost functions are prediction reduces uncertainty during modeling. Improving predictions
used some works [108]. Multi-objective optimization problems can be within a given weather window reduces uncertainty in maintenance
solved using many approaches, including duo-Ant colony optimization schedules. Considering wave height, autoregressive models and artificial
[100], a nondominated sorting genetic algorithm [108], a genetic al neural networks with different lookahead time steps are compared based
gorithm [105], and ε-constraint method [112]. Optimal results can be on a data mining approach [123]. Online monitoring is integrated into
verified by simulations. The offshore wind farm O&M process can be planning in Zhu et al. [93].
simulated by, for example, distributed simulation using a multi-agent One trend is from deterministic modeling to stochastic modeling; i.e.,
system [114] and business process simulations [115]. A hybrid the considered uncertainties of OWT maintenance scheduling enhance
10
in the state-of-the-art studies. These uncertainties come from several the models, and the model uncertainty can be amplified by the improper
aspects, e.g., OWT component failure, weather condition, technician selection of model coefficients.
skills, defective repair, and vessel conditions. In deterministic models, The flexibility of current approaches remains limited. The optimi
failures are assumed to happen periodically according to historical data. zation is performed according to some specific requirements; however,
Stochastic failures occur randomly based on a predefined probability these requirements may change over time, and these changes are typi
distribution function using the collected data, such as the Weibull dis cally unforeseen. The performance and robustness of a specific algo
tribution [101,124], Bernoulli distribution, and binomial distribution rithm in unconsidered scenarios are not guaranteed. Hence, a
[125]. It is possible to extend the deterministic algorithm by probabi mechanism to reasonably and intelligently switch or fuse among all
listic modeling, such as [101]. The stochastic optimization problem these algorithms is valuable.
could be solved by transforming the stochastic programming formula
tion into its deterministic equivalents [126,127]. The effects of gov 4. Onsite maintenance
ernment subsidy are studied in Nguyen and Chou [124]. Due to a limited
number of studies, the stochastic modeling of the uncertainties should After maintenance tasks are planned, three operations related to the
be studied and discussed in future research [91]. onsite maintenance make up a considerable proportion of maintenance
To sum up, research on optimal scheduling has been intensively cost, i.e., (1) the delivery of personnel and equipment to an offshore
developed to improve planning performance and complexity. The wind farm, (2) the docking operation to transfer onboard technicians
number and size of maintenance fleets and wind farms continue between the service vessel and the wind turbine, and (3) the lifting
increasing with time, and more complete considerations and subtle operation when large components such as blades and the generator need
factors have been investigated. Maintenance activities are becoming replacement or maintenance. Since blades and gearbox are the two most
more diverse and flexible, and stochastic and time-varying models are vulnerable components of an OWT, their maintenance needs to be
being used to describe the environment more accurately. specified further. Innovative remote self-maintenance has become
increasingly popular.
3.3. Limitations
4.1. Equipment and crews transfer
However, there are some shortages and limitations in existing algo
rithms, including a lack of online updates, vessel failures, lack of vessel It is essential to choose a suitable maintenance fleet that provides
interaction and cooperation, extreme weather conditions, a large num sufficient accessibility while minimizing the extra costs of power gen
ber of constraints, and limited flexibility. eration. As offshore wind farms become larger and farther away from
First, the project schedule is normally decided offline without real- shore, the demands imposed on service vessels will increase.
time updating. The coefficients and parameters in the models are pre Various modes of transport are employed for different maintenance
determined by statistical metrics and project experience. Inaccurate purposes, i.e., transport of crews, the shipment of large spare parts, and
parameters result in misplanning. However, it is impossible to correct implementation of lifting operations; the corresponding vessels are crew
the parameters and coefficients by online update. Due to advances in transfer vessels (CTVs), supply vessels, multipurpose vessels, and
remote-sensing and communication techniques, more knowledge and floating cranes. Wind speed and significant wave height are represen
information are available, e.g., the health of technicians, vessel failures, tative parameters that limit the accessibility of helicopters or service
weather windows, project delays, and emergency issues. Real-time vessels and therefore maintenance. The use of each vessel type is limited
adjustment and replanning have the potential to improve solution by environmental conditions [126]. CTVs are limited by environmental
robustness. conditions. Climbing up a turbine is not allowed when the wind speed is
Second, the failures of maintenance vessels and devices are not higher than 20 m/s. In addition, helicopters are employed for mainte
included in literature. An offshore supply vessel approaches an OWT nance, but their use is limited by wind speed (which usually must be
using a dynamic positioning (DP) system. Since the DP system has a under 20 m/s) and visibility [83]. In the absence of timely maintenance,
relatively high failure rate, delays and other issues caused by mainte the downtime of a wind farm will be prolonged, resulting in massive
nance vessels should be considered [128]. losses of power generation, especially given the increasing capacities of
More realistic cooperative planning among several vessels is not OWTs.
considered in most studies. Instead of going back to the ports repeatedly The sea states can be measured and detected by several measurement
or using more vessels, interactions among vessels can improve the effi instruments, such as onsite wave buoys, onboard wave radars, and sat
ciency of all maintenance tasks and use vessel capacities fully. Consid ellites. Although wave buoy and wave radar can provide real-time sea
ering the following example. A normal-size vessel (A) is adopted to state information, they are costly due to the extra costs of their mea
conduct a complex maintenance task, and its deck space (or crane ca surement instruments. Satellite signals also have an hour-level delay.
pacity) is not enough for some components (or operations). It is possible However, an interesting research topic is to estimate real-time direc
to use another larger vessel (B) to carry the other components (or tional wave spectrum based on vessel responses, which is called the
conduct the corresponding tasks). After unloading these components (or wave buoy analogy [129,130]. However, this estimation’s accuracy
accomplishing the tasks), vessel (B) leaves and continues to its next strongly depends on the calculated response amplitude operators
project at another wind farm. If so, then there is no need to assign two (RAOs). Conversely, the vessel model can be tuned by vessel motions
normal-size vessels to the project. However, all studies neglected such a and environmental data [131]. A decision support system based on a
possibility due to the correspondingly high variations and complexity. wave height forecaster is proposed in Catterson et al. [132]. Concerning
Hence, the cooperation among vessels can improve the overall mainte the environmental conditions of offshore wind farms, long-term average
nance efficiency. wind speed estimates based on a forecast dataset are studied in James
Moreover, some extreme weather conditions are disregarded in et al. [133]. The estimate accuracy increases as the dataset widens,
current research. For example, a failure caused by ice in a cold climate advanced physical models are used, and better data assimilation tech
requires special icebreakers to conduct maintenance. A powerful niques are employed. Probabilistic forecasting is used in Taylor and Jeon
typhoon not only threatens wind turbines but also influences mainte [134] to calculate the probability of wave heights falling within the
nance safety. safety limit and to determine whether to send a service vessel. The re
The number of constraints in the optimization problem reduces the sults show that the proposed probabilistic method is more cost-effective
solver’s robustness and computational speed. For example, there are 62 than a deterministic approach based on point forecasting.
constraints in Raknes et al. [104]. A longer duration is needed to tune The optimum selection of a CTV plays a central role in organization
11
of maintenance logistics. The main target is the maximization of overall structures, such as the crane boom and OWT blades. Multibody dy
economic benefits, and its capacity should provide sufficient support to namics is used to simulate the dynamic system interaction.
the maintenance tasks with minimal cost. The economic benefits grow Simulating environmental loads is computationally expensive
with the capacity of the CTV if it is below the optimum size. However, because many simulations are required to calculate the critical envi
the benefits of using CTVs that are too large become decrease due to ronmental conditions in sensitivity studies. Therefore, hydrodynamics
insufficient usage. Table 4 lists the factors that should be considered and aerodynamics loads are simplified and calculated by RAOs and the
when selecting CTVs. Van Bussel and Bierbooms [135] investigated cross-flow principle [140–142]. To improve simulation fidelity, many
three access systems (rubber boats, an offshore access system, and he theories have been developed to balance computational efficiency and
licopters) and showed that 90% availability could be achieved if rubber accuracy when solving the Navier-Stokes equations. Real-time hybrid
boats were not used alone. Environmental conditions, failures of turbine simulations are powerful when evaluating a complex system. These
components, and assessment of the vessel’s operation were also shown simulations separate the entire system into two parts, i.e., a numerical
to affect maintenance tasks [102,136]. component that can be accurately simulated numerically and an
There are several maintenance optimization models that have been experimental component that is difficult to model. Sensors and actuators
developed individually. Sperstad et al. [137] employed six strategic are used as the interface between these two parts. However, these
decision support tools with different modeling methodologies to deter methods have not been adopted to simulate OWT O&M activities.
mine the best maintenance vessel fleet and rank the sensitivity of the
vessel fleet to various input assumptions. Their results show that the 4.2.2. Docking operation
decision support tools generally agree on the best selection, partially on After approaching an OWT, a docking operation between a service
the overall ranking of each vessel fleet, and on the ranking of the vessel and an OWT is carried out. This operation uses a simple fender or
sensitivity to input assumptions. Among the input assumptions, that of an active motion-compensated access device. The aim of docking is to
limiting significant wave height is the most important, while the vessel transfer personnel and equipment in an efficient and safe way. A passive
speed assumption is appreciably less important, and the assumptions of gangway can also be also used to connect a jackup vessel and an OWT.
failure rates and vessel day rates are of intermediate importance. Since Since there is no lifting crane on an OWT, typical personnel transfer
various tools yield similar results, decision makers should ensure that methods used on oil and gas platforms, such as the Reflex Marine
input assumptions are representative of a specific wind farm and try to deliver, are not applicable in OWT maintenance.
reduce uncertainties in input data while ensuring the completion of A fender is the simplest type of docking device and is typically made
preventive maintenance. Series games are used to help O&M planners, of rubber or similar materials. The vessel’s propulsion system provides a
engineers, and researchers gain a better understanding of the effects of pushing force to keep the bow tightly attached to the tower, relying on
their decisions and to prevent revenue loss due to inadequate mainte friction to control the relative motion. The maintenance crew can then
nance [138,139]. Van Bussel and Zaaijer [15] pointed out that one of the get onto the wind turbine from a ladder [143]. Fenders are inexpensive
main causes of high maintenance costs is using a large external crane and easy to install on the service vessels. To improve boarding perfor
vessel. Two methods are proposed to solve this problem. One approach mance, automated control of air cushion pressure can be used to reduce
is to design OWTs that can rely completely on built-in facilities to vertical accelerations at bow [144]. The turbine is assumed to be vertical
transfer failed parts and their replacements. The other approach is to and rigid. The motions in only three degrees of freedom are taken into
adopt the offshore wind energy conversion system (Opti-OWECS) design consideration, i.e., surge, heave, and pitch. The vessel and turbine in
solution, which involves expenditure on special maintenance facilities teractions are modeled as a linear spring.
as an overall investment. In this approach, a self-propelled jack-up Active motion-compensation access devices, i.e., hydraulic gang
platform is modified to perform the required lifting actions and main ways, can be installed on service vessels regardless of their size,
tenance base. providing sufficient deck space and weight capacity.These devices can
cancel the relative motion of the vessel within the hydraulic system’s
4.2. Docking and lifting operation limits, resulting in a higher working limit than a fender [23]. However,
these devices are more expensive than fenders. Compared with heave
4.2.1. Numerical simulations compensators and DP systems, an active motion-compensation gangway
Instead of time-consuming and costly model-scale and full-scale ex must cancel all six of vessel degrees of freedom, including second-order
periments, numerical simulation is an efficient and budget-friendly wave motions [145]. The mechanical system is similar to an industrial
approach to evaluate marine operations. Using numerical simulations, robotic arm, but with a larger size and rated power. Due to
it is possible to conduct an integrated aerodynamics-hydrodynamic- wave-induced motions, the desired trajectory is calculated based on
structural analysis of a maintenance project and identify operational inverse kinematics and the relative motion between the vessel and wind
limitations. Critical environmental conditions can be evaluated based on turbine. Because gangway designs are normally over-actuated, the
static results from finite element analysis. desired joint rotation angles can be calculated by the pseudo inverse
In current commercial marine operation software, the vessel and the method and other optimization approaches. The relative motion can be
lumped-mass payloads are normally simplified to be rigid bodies in measured and estimated through an inertial measurement unit and
scenarios where structural flexibility is negligible. Furthermore, struc LIDAR system [146,147]. Feedback control can be achieved by many
tural stiffness contributes to the local vibration and deformation of long control methods, such as a typical linear PD controller with feedforward
[148] and a model predictive controller [149]. Several companies have
developed motion-compensation gangways that are available on the
Table 4
Factors related to CTV selection.
current market, e.g., Ampelmann, Barge Master, Kenz Figee Group,
Royal IHC, Van Aalst Group, SMST, Uptime, ZTechnologies, Osbit, and
Environmental Failure CTV specification Financial
Lift2Work [150,151].
conditions characteristics attributes
Because the docking operation is governed by the interactions of the
• Wave height • Number of • Size • Electricity vessel’s structure, swell, and the relative motion of the docking device
and period components • Capacities (fuel, cost
• Wind speed • Components accommodation, • Fuel cost
and the tower, simulating the docking operation and evaluating crew
• Distance to configuration deck) • Vessel & safety and the process of equipment transfer has become an important
port • Failure rates • Speed technician research topic. When the vessel is equipped with a fender at its bow to
• Repair time • Operability cost access the tower, Brändli et al. [152] presented a comprehensive
• Repair cost
framework to analyze the docking, in which a partitioned approach is
12
proposed to solve the coupled motion while managing the governing barges, and jack-up vessels [52]. The day rate for lifting equipment for
fluid-structure interaction. González et al. [153] combined numerical offshore use is at least 10 times higher than that of onshore crane lifting
simulations with experiments to investigate the landing maneuvers of a for similar lifting heights because the cranes needed for offshore con
catamaran vessel. The simulation results were able to quantify the risk of ditions must be sufficiently over-dimensioned in terms of lifting weight
a fender suddenly slipping during docking. König et al. [154] developed [15]. The trend of the day rate for crane vessels versus hoisting height
a software framework to implement a partitioned numerical solution shows that there is a sudden surge at around a height of 85 m [163].
strategy to optimize service vessel access to an OWT. Ren et al. [142] Therefore, it makes sense to install built-in lifting facilities to reduce
developed a MATLAB/Simulink toolbox to simulate complex marine height requirements on external lifting equipment when replacing and
operations for control purposes, where the crane module could be used maintaining large OWT components.
to simulate the gangway. The crashworthiness and damage between A relatively small built-in lifting device installed on an offshore wind
several types of ships and different types of OWT foundations was also tower from a floating vessel was proposed to reduce the maintenance
evaluated [155,156], wh ere crashworthiness is determined by the costs by avoiding the need for a specialized maintenance vessel to
mechanical properties of the foundation structures. replace the gearbox [164]. The crane would be attached to the tower by
Attention must be paid to the risk of collision between maintenance a clamping mechanism and fixed in position by friction. However, this
vessels and other commercial vessels that pass close by at high speed, approach provides only limited lifting capacity and has a limited scope
and to the risk of collision between a ship and an OWT. Severe damages of application. The use of a modified self-propelling jack-up platform is a
can be caused to OWT foundations and to vessels. For example, an oil cost-effective method for the comparatively large wind farms [165]. A
leak resulting from an oil tanker colliding with an OWT would cause crane mounted on one of the legs of the platform could draw itself up to
environmental pollution. the required working height, and the associated platform can serve as a
Finite element analysis shows that the collision force is affected by base for the maintenance crew and tasks as well as a stock store.
the impact velocity, rubber hardness, and rubber thickness [157,158]. Automated control theories were applied to enhance the efficiency of
The critical relative motions for structural collisions are found through OWT maintenance. For example, an automatic lifting scheme was
finite element analysis. Wu [143] suggested that the docking capabilities studied to reduce dynamic tension during lifting and lowering processes
of service vessels should be considered when evaluating operational [166]. Active tugger line control was also proposed in Ren et al. [167,
limits. A linear frequency-domain method is proposed to assess the 168].
docking performance of various vessels employing either a fender or an The risks related to lifting operations using offshore crane vessels
active motion-compensated access device. Sperstad et al. [159] used were studied using numerical simulations [169]. As discussed before,
such a method to derive multi-parameter wave criteria to analyze the installation’s weather window is an important constraint that is
accessing systems. A numerical nonlinear finite element analysis method imposed during onsite maintenance. Currently, these weather windows
was used to investigate collisions between a vessel and OWTs with are determined using experience-based operational limits; the typical
monopile or jacket fixed-bottom foundations. For collisions with a allowable weather limit used in the industry is a 1.5 m significant wave
monopile foundation, the critical factors were found to be collision en height for crane-assisted lifting operation with a mean wind speed (Uw )
ergy, the height of the vessel, and the impact area [160]. For collisions below 10 m/s [12,170]. A more scientific method is required to estimate
with a jacket foundation, vessel speed, and impact area are the dominant these limits based on numerical modeling rather than just based on in
factors. Presencia and Shafiee [161] investigated the collisions of dustrial experiences.
maintenance ships with OWTs from another perspective, comparing A response-based method to assess the operational limits of blade
collision risks in terms of corrective maintenance and preventive installation using an offshore crane vessel was proposed by Refs.
maintenance strategies. The probability of occurrence of a collision is [171–173]. The emphasis was placed on collision risk of the hoisted
related to corrective repair and replacement, and an analysis of the blade with surrounding structures, such as a hub or the turbine tower
damage magnitude found that collision risk is closely related to that could occur due to dynamic motion responses of the blade instal
corrective replacement activities as part of a corrective maintenance lation system [174]. A detailed list of factors and collision scenarios that
strategy. In contrast to Dai et al. [156], in which considered factors can occur during blade installation was also identified [174,175], and a
included various external aspects related to the collision, such as per blade root impact with the hub was deemed the most critical. For
sonal characteristics of the crew and administrative controls, Moulas instance, Fig. 9 presents different collision scenarios that could occur
et al. [162] examined internal factors that are closely related to the during the blade root mating phase [176] - a head-on impact that could
collision and that determines the magnitude of damage, such as collision occur due to misaligned wind-wave conditions; and a sideways impact
direction and angle, and type of ship.
A risk assessment is essential to assess the magnitude of the collision
risk and to determine the critical factors involved. A specific risk anal
ysis framework involves six main steps, i.e., initial analysis, hazard
identification, probability analysis, consequence analysis, risk descrip
tion and evaluation, and risk reduction [156]. The critical values of force
and energy are identified to describe the likely structural damage in each
case. Risk-influencing factors are analyzed using Bayesian networks.
Based on the energy equation, the critical vessel speeds at which
structural damage of the OWT could occur turned out to be very low,
indicating that risk-reduction measures are essential.
4.2.3. Lifting operation

Lifting operations are widely used to execute the replacement and
maintenance of large-scale OWT components, such as the generators,
gearboxes, and blades (Fig. 8). Compared with onshore lifting opera
tions, offshore lifting operations are difficult owing to the unpredictable
wind and wave conditions. Special, expensive, and sometimes scarce
equipment is often required to perform lifting operations.
Offshore service vessels include crane vessels, flat-bottom sheer leg Fig. 8. Lifting operation [courtesy by Mrs Eva Boeckling of DEME].
13
that could occur due to collinear wind-wave conditions. Global motion removal of material due to continuous exposure to rain, ice, insects, and
responses were used to calculate the impact velocities for the hoisted dust, is a highly complex problem that degrades turbine performance
wind turbine blade for different operational sea states, and damage as [188]. As a result, blades are required to be regularly inspected, cleaned,
sessments were performed to evaluate operational limits for blade and repair. A typical manual cleaning is conducted by fully stopping the
installation using jack-up crane vessels. A sensitivity study [177] also turbine in a low-wind-speed environment. In recent years, automatic
used a tuned mass damper in the hub to control the vibrations of the hub blade inspecting and cleaning robots have been developed, such as
during installation in the absence of aerodynamic damping. The tuned climbing robots that move along the tower [189], inchworm-type robots
mass damper was found to be efficient at inhibiting resonance-induced that move along the blade [190], and unmanned aerial vehicles [191].
vibrations in top of the tower, reducing impact velocities, while Parallel cleaning was introduced in Deb et al. [192], and an
expanding the operational limits and weather window of the task. Other artificial-rain cleaning device from BladeCleaning was equipped on a
novel lifting operation concepts, e.g., Refs. [141,178], have also been tower to spray water with detergent. While current technologies require
recently proposed. Nevertheless, the technology readiness levels of these rope access and manual repair of leading edges using solutions such as
concepts are low, and further research is required before they can be leading edge tapes from 3 M [193], a large emphasis is currently being
applied to onsite maintenance tasks. placed on drone-based applications [194] and robotic-assisted solutions
[195]. Given that the current repair and maintenance cost of wind tur
bine blades requires millions of euros every year, more research and
4.3. Maintenance of the most vulnerable components development are required on this topic.
The typical blade maintenance strategy currently is corrective
4.3.1. Blade maintenance. However, proactive maintenance becomes feasible with
Due to complex long-term working conditions, OWT blades tend to the fast development of various evolution algorithms and structural
experience many internal and external damages [179]. Damages include health monitoring techniques [84,196]. Blade structural health and the
the fatigue failure of materials, wear, corrosion, erosion, and cracks interval between inspections can be estimated by the life-cycle model
induced by system degradation or deterioration [180]. Environmental [196], an optimization model using knowledge-based force analysis
conditions can cause damage to blades, both internal and external, e.g., [197], a crack length model using the stochastic gamma process [66],
rain/hail/ice, lightning, wave slamming, and wind gusts. For instance, and predictive modeling using curve fitting [198].
rain causes erosion, which then decreases AEP and eventually leads to An optimal opportunistic condition-based maintenance method was
damage to the blades themselves [181]. Lightning could also cause investigated in Ref. [66], and found that major maintenance needs to be
splitting of the blade from the tip towards the inside. Blade failures make carried out when the crack length in any blade exceeds a given
up a high proportion of all wind turbine failures [182]. threshold, and preventive maintenance could be performed on the other
Based on a database of 1013 wind turbine blades, the percentage blades; otherwise, the scheduled preventive maintenance would be
breakdown of damage locations and types clearly shows that the ma carried out for the pre-determined operational age. Optimal values were
jority of the damage is located on the coating surface and adhesive simultaneously determined by the model to minimize the average
bonds, whereas the major blade structure damage modes are transverse long-term maintenance cost per blade per unit time. To be more
cracks, spalling, leading edge adhesive bond failure, delamination in consistent with the practical operation, a two-level maintenance
load-carrying laminate, sandwich/core debonding, and trailing edge threshold (the preventive maintenance threshold and corrective main
adhesive bond failure [183]. Minor external damages tends to lead to a tenance threshold) were proposed [85].
loss in AEP. Damage inspection and detection can be accomplished by Integrating in-situ structural health monitoring techniques based on
acoustic emission sensors [184], visual cameras [185], tomography acoustic emissions into a condition-based maintenance method have
[186], and vibration-based estimation using accelerometers [187]. been shown illustrated to be practicable and promising [199,200]. The
Among all damage types, leading edge erosion which involves the
Fig. 9. Blade root impact scenarios [175].
14
knowledge-based methods for load analysis can also be employed to strategy until full maintenance was possible. This method was based on
optimize proactive maintenance of OWT blades, which allows for a qualitative approach and consisted of a fault diagnosis system
monitoring blade performance in real-time, leading to advanced alarms composed of two modules, i.e., a functional redundancy designer and a
when needed [197]. This process contributes to scheduling maintenance model-based reasoner. When a fault occurs, the system analyzes the
effectively. A fracture-mechanics-based model for estimating the available information and reconfigures alternative components to
remaining life of a blade was used for risk-based maintenance to perform the function of the faulty component. Both modules take
improve the maintenance schedule for the blade lifetime [196]. Niche advantage of a function-behavior-state model that provides information
nametla et al. [198] used predictive analytics to optimize the opera on potential existing system redundancies. This method forms a foun
tional life cycle cost and improve the reliability of the wind turbine dation for self-maintained wind turbines and is able to optimize the
blades to reduce maintenance costs. Machine learning was also used to capabilities of OWT components, thereby enhancing system capabilities
extract blade features in Jiménez et al. [201]. against faults.
4.3.2. Gearbox 5. Environmental issues

A gearbox increases the rotational speed input to the generator and is
the most vulnerable and expensive component of the wind turbine There is no denying that offshore wind farms contribute greatly to
drivetrain due to its high work intensity and complex operation [202]. reducing reliance on fossil fuels, and wind power is more environmen
The gearbox is one of the OWT components with the highest failure rates tally friendly than traditional energy resources. However, it is never
[203], and its down-tower replacement requires the use of heavy lifting theless associated with some environmental concerns, such as noise
cranes and vessels, which are expensive. Fatigue damage is a major pollution, visual appearance, and consequences for nearby wildlife.
concern; thus, it is important to optimize its maintenance strategy. Although their impact is minor at present, because wind energy is likely
Increasing the gearbox reliability is particular important, as noted in the to become the main green energy source in the future, this may not al
Gearbox Reliability Collaborative project, which was found in 2004 ways be the case, and there could be serious consequences [215].
[204]. Kang [205] determined a reasonable interval to replace gear Further investigations need to be carried out, and an optimum strategy
boxes to minimize the life-cycle cost of OWT gearboxes and discussed should be developed for offshore wind farms so that wind energy can
the relationship between transition rates and failure probabilities. become an even more environmentally friendly and sustainable energy
Deng et al. [206] proposed a model of the optimal maintenance in resource during its operation life. Noise and visual aesthetics account for
terval for a gearbox to maximize its profit per unit time which con about 39% of the total damage (excluding effects on global warming) for
tributes to the maintenance interval schedule for the preventive onshore wind farms but amount to less than 1% for offshore wind farms
maintenance method. Li et al. [202] adapted a nonhomogeneous [216]. Therefore, only greenhouse gas (GHG) emissions, impacts on
continuous-time Markov process to manage the gearbox as a multistate wildlife, and waste recycling related to OWT O&M are discussed here.
degrading system due to its performance degradation to analyze gearbox
reliability and develop an optimal maintenance policy. Condition 5.1. Greenhouse gas emission
monitoring systems and models have been developed for gearboxes such
as the Gaussian process gearbox temperature model [207] and nonlinear GHG emissions are a critical environmental issue. According to Wang
state estimation technique model [208]. The monitoring model, which is and Sun [217], the lifetime emission intensity of current wind farms
based on echo state network modeling and the dynamic threshold from design to end-of-life is 5.0–8.2 g CO2/kWh electricity. Regarding
scheme, uses SCADA vibration data. The gearbox was also verified to O&M, GHG emissions will result from the burning of diesel by the ser
improve unsatisfactory detection accuracy and the adaptability of vice vessels’ engines and from the cleaning, repair, and replacement of
traditional static monitoring methods [209]. A drivetrain vulnerability OWT components. The required materials and equipment are trans
map can be calculated in numerical simulations according to the accu ported from shore to the assembly base and are then delivered to the
mulative damage hypothesis [210]. Igba et al. [211] proposed using wind farm mainly by barges, tugboats, and deck barges. The amounts of
historical failure data for a specific module or subassembly to select an CO2 emitted from coal-, oil-, and gas-fired power plants are 154, 117,
optimum preventive maintenance interval based on minimum mainte and 96 times that of wind power, respectively, with an average emission
nance cost and maximum availability to achieve the required reliability. of wind power of 6.3 g CO2/kWh. Significant reductions in GHG emis
This method was shown to be valid by applying it to the gearbox of a sions have thus been achieved. However, with the rapid expansion of
wind turbine. offshore wind farms, attention must be paid to the issue of GHG emis
sions to maintain sustainable development.
4.4. Remote O&M and self-maintenance of OWTs Life-cycle assessment was been widely adapted to quantify the
relation of energy and environmental impacts within the whole life span
Instead of manned inspection and maintenance, remote O&M of of products and services [218,219]. A process-based life-cycle inventory
OWTs is a promising solution to mitigate the issue of restricted acces model has been used to analyze life-cycle environmental emissions
sibility caused by the harsh weather conditions and to reduce the [218].
number of maintenance tasks [212] and the costs incurred by manned Adopting more efficient maintenance arrangements can effectively
maintenance. A remotely controlled robot prototype was perform in reduce GHG emissions that are produced by grid connection and
spections and the easiest maintenance tasks inside a wind turbine [213]. maintenance activities. A large reduction in the GHG produced during
This device was tested and found to be reasonable and effective after transport can be achieved by using alternative shorter transport routes.
comparison with manned inspections. A case study showed that CO2 emissions associated with the transport of
Making full use of system redundancies to decrease downtime is an OWTs and their components could be reduced by 33% with reasonable
effective way to reduce the costs of maintenance and energy generation. shorter transport routes; however, the operation only accounts for a very
This concept improves the continuous operation capability of OWTs. small portion of the emissions [217,220]. The use of steel and the
Therefore, significant economic value can be gained by the development replacement of OWTs makes up a larger proportion (3% planned, 47%
of fault-tolerant control; maintaining the operation of an OWT, even at a unplanned) of the GHG emissions during operation compared to vessel
lower energy output, when faults occur in some components can be transportation [221]. Of this amount, approximately 33% of the GHG
beneficial. An innovative maintenance system proposed in Echavarria emissions results from the use of specialized vessels in the replacement
et al. [214] could reconfigure the system or a subsystem to maintain of large components, while CTVs and helicopters account for only a
OWT operations even at a reduced capacity, and determined a repair minor part. A large percentage (46%) comes from the production and
15
decommissioning of lubricants and spare parts. accurately. The core problem of OWT maintenance is to ensure opera
During maintenance, failure rates are directly related to GHG emis tional safety, enhance economic profits, lower the LCOE, and minimize
sions, because they determine the need for transportation and conse negative effects. Significant amounts of theoretical innovations and
quently affect fuel consumption. Arvesen and Hertwich [222] noted technical advancements have improved every aspect of OWT mainte
certain obstacles to future life-cycle assessments of wind power gener nance in the recent decades.
ation, including the lack of knowledge of toxic materials emitted, As the scale of offshore wind farms expands rapidly, a corrective
inadequate considerations of the details of the offshore wind farm maintenance strategy is no longer suitable and is gradually being
operation, and insufficient experience of replacement of components. replaced by proactive maintenance strategies. These strategies primarily
Greater attention must be paid to these factors to optimize the life-cycle involve preventive maintenance based on a predetermined period,
environmental assessments and maintenance scheduling of offshore together with condition-based maintenance based on the use of a con
wind farm O&M. dition monitoring system to supervise the health. Preventive mainte
nance strategies can be optimized by (1) optimizing the selection of the
5.2. Effects on marine wildlife predetermined interval according to the failure probabilities of various
components; (2) taking the opportunity to carry out preventive main
The effects of offshore wind farm operations on marine wildlife, e.g., tenance by replacing or maintaining faulty parts in the meantime; (3)
fish, marine mammals, and seabirds, cannot be neglected. Sensitive dividing components into different age groups and applying the corre
creatures like cod and herring can detect piling noise at great distances sponding preventive maintenance tasks; and (4) employing queuing
(perhaps up to 80 km from the sound source), and dab and salmon are theory to determine the maintenance waiting time and carry out pre
also sensitive to pile-driving pulses [223]. Their behaviors can thus be ventative maintenance according to the chosen maintenance priority.
influenced by the presence of OWTs. Although the noise generated by Condition-based maintenance strategies can be improved by (1)
normal turbine operation cannot be heard at water depths below 20 m combining them with a risk-based life-cycle approach to monitor the
[224], it has been found that this noise does have the potential to in degree of deterioration and thereby increase the reliability of prediction;
fluence the physiology and behavior of harbor porpoises and seals at and (2) carrying out condition-based maintenance according to the alert
considerable range. Leaked oil and other waste during component threshold of a given type of deterioration. Opportunistic maintenance
replacement operations and from lubrication during maintenance are strategies combine these maintenance strategies. However, it is very
harmful to the wildlife [225]. The effects on birds resulting from OWT difficult to decide on the best maintenance strategy since the selection
O&M include flight-route changes due to the visual stimulus provided by always yields corresponding optimal scheduling problems. To evaluate
the turbines, physical habitat changes, and growing mortality rate the safety and economy of scheduling, several assessment methods have
resulting from collisions with the rotating blades or other superstruc been proposed, including economic assessment and risk assessment.
tures [226]. Furthermore, transportation by boat or helicopter associ To carry out maintenance efficiently, maintenance tasks must be
ated with maintenance may displace the activity space of birds. scheduled based on simple proper route planning and more complicated
Therefore, more environmentally friendly designs should be investi scenarios. Route planning for OWT maintenance has been achieved with
gated in future research. one or multiple O&M bases by considering available crews and spares, as
well as the capacity of the mode of transport. The aim of optimum route
5.3. Waste inventory recycling selection is simpler, i.e., highest efficiency and minimum transport cost,
as well as reduced GHG emissions. Optimal scheduling should consider
Waste inventory happens in every stage in the life cycle of an OWT, i. several more topics, including minimizing downtime, maximizing rev
e., transportation, installation, O&M, and disassembly and decom enue, improving system reliability, and realizing cooperation among
missioning. In this review, we focus on the waste that is produced during maintenance teams. The first step is to quantify the problem, and nu
maintenance. merical deterministic and probabilistic models are used to describe the
Among all OWT components, blade waste recycling and reuse are the process. The scheduling problem can be converted into an optimization
most important topics [227–229]. The blades are made from composite problem with a number of cost functions and constraints. According to
materials, which are energy-intensive to manufacture and environ the cost function, the problem can be further categorized into single-
mentally problematic. Therefore, the disposal and recycling of broken objective or multi-objective optimization problems. Cost, reliability, or
blades represent valuable research topics. Blade waste is predicted to their combinations makes up the cost functions. Maintenance strategies
significantly increase in upcoming decades; there is a clear linear trend have been improved to cope with the limited weather window due to
between blade mass and power rating [227]. harsh offshore environmental conditions and thereby achieve high
Blade recycling is achieved by mechanical, thermal, chemical ap availability and reduce revenue loss caused by downtime. The
proaches, e.g., decomposing the waste into other recyclable materials or complexity and reality of optimal scheduling algorithms has increased
raw materials for secondary use [229]. gradually. However, existing models still have their limitations. In
addition, more practical, uncertain, and complex operations are still not
6. Discussion and conclusion considered.
Onsite maintenance is the next step after the scheduling, and is
Maintenance of an offshore wind project is a broad topic. The cost of markedly different from that of an onshore turbine. First, unpredictable
maintenance makes up a larger part of the total energy generation cost weather conditions limit the transport of crews and equipment and
compared with onshore wind power. In this review, we present the state- impose more stringent requirements on the modes of transport. More
of-the-art development of OWT maintenance with regard to strategy over, an extra docking operation is required, and docking devices have
selection, schedule planning, onsite operations, and environmental been reviewed in this paper, including active motion-compensated ac
threats. Analyzing the maintenance of OWTs and optimizing the pro cess devices and simple fenders. The risk of collision between service
cedures involved contribute to describing the status quo of offshore vessels and the turbine should be accessed while evaluating the impor
wind power. The major challenges of OWT maintenance include long tance of various critical factors related to collisions. The requirements
distance from shore, weather uncertainty (including wind and wave for lifting operations are stricter for OWTs due to irregular wave heights.
conditions), a lack of information from remote monitoring, unpredicted Specialized and expensive lifting equipment are often required, whose
failures, aging, and subjective factors (such as technicians’ skills). daily rates are considerably higher than for onshore lifting to similar
Research in OWT maintenance involves a higher level of uncertainty and heights. A built-in lifting device has been proposed for installation on
complexity to make calculations and analyses resemble reality more OWT towers to reduce the height through which external cranes need to
16
lift large components. climate and weather. Based on historical data and weather forecasts,
Numerical simulations are powerful tools to evaluate and predict the this uncertainty could be minimized. However, a wind-farm planner
performance of onsite operations during planning. The critical impact should estimate the equivalent cost that the operator might
velocity was found through finite element analysis, and the environ encounter before the farm is designed, which can aid grid penetra
mental limitations are calculated based on time-domain multibody dy tion in terms of cost competitiveness.
namics simulations and FEM results. Both modeling approaches were • Additional offshore technologies (e.g., service platform and un
based on a series of simplifications. FEM modeling only considered the manned system) might be developed to optimize future O&M pro
instant impact, which is characterized by the impact area, impact speed, cesses. Although the automatic systems have significantly improved
and impact direction. To ensure the computational speed, multibody O&M efficiency, there is much more work to be done to enhance the
dynamics models are commonly considered under the rigid-body as levels of automation and intelligence in future research and appli
sumptions. The flexible structures, such as the towers, blades, and wire cations. Before achieving fully intelligent operations, human opera
ropes are simulated by one or a number of connected lumped-mass tors must supervise and make crucial decisions. Hence, human-
nodes. Wave-induced loads are also simplified to be a group of trans machine interaction and remote operations are meaningful. There
fer functions, namely, response amplitude operators. Higher-fidelity fore, unmanned or partly unmanned O&M exhibits along a signifi
simulations could be achieved by including computational fluid dy cant potential to reduce human resource cost, resulting in a lower
namics and aerodynamics [230–232]. LCOE.
The maintenance of OWTs can be optimized from two perspectives. • An O&M friendly wind turbine design should be proposed to reduce
One approach is to improve onsite maintenance by increasing the ability O&M costs and the LCOE, although a wind turbine’s capital cost
to predict the weather windows, which is fundamental for arranging might increase. For example, hydraulic transmission reduces the
onsite maintenance. The other approach is to replace onsite mainte height of the drivetrain resulting in more efficient maintenance. The
nance by remote-controlled maintenance through robots installed inside section of the tower that is Near water surface could also be rede
the tower to carry out simple maintenance tasks or to take advantage of signed for tug accessibility.
redundancies in the system to maintain the operation of the wind tur • Numerical simulations are widely used in onshore planning and
bine, even at a reduced capacity and thereby reduce maintenance fre onboard decision making. Simplified logistic models have been used
quency. Both perspectives require developments in data-collection to verify the proposed optimal scheduling approaches. Both FEM
capabilities. modeling and multibody dynamics modeling have investigated the
The important environmental issues arising from OWT maintenance criteria of the maintenance operations, such as docking, lifting, and
include GHG emissions and effects on wildlife. Improving route planning mating operations. Compared with rigid-body dynamics, high-
for transportation and employing reusable materials have been pro fidelity simulations could be achieved with more accurate
posed to reduce GHG emissions. With regard to effects on wildlife, no modeling approaches, such as real-time computational fluid
suitable approach has been proposed. The recycling and reuse of OWT dynamics.
components are also of concern. The environmental issues related to the • Due to reductions of sensor prices, a more complete and precise
maintenance of OWTs, therefore, cannot be neglected. image of OWT operational conditions could be built into mainte
The research on OWT maintenance has coevolved and accumulated nance planning and execution periods. System behaviors could be
with the technical advances and theoretical innovations in all relevant measured and predicted more accurately. Measurement availability,
realms. The LCOE is gradually reduced by these technologies and their reliability, and accuracy improve with the development of data sci
applications, which intensifies the market competition of offshore wind ence, sensor fusion, and remote communication. A digital-twin
energy. platform was adopted to predict future performance and possible
failure, combining the numerical models and various sensor data. In
• Newly developed supply vessels and onboard equipment can addition, large quantities of gathered data promote the development
improve the reliability and operational efficiency of maintenance of both onshore and onboard decision support systems. State-of-the-
tasks. Straightforward methods include improving wave and wind art algorithms are of significant interest to analyze and utilize the
resistance, which could result in better accessibility and onsite op collected data, such as via big data and machine learning ap
erations in more strict offshore environmental conditions, yielding proaches. The prediction of short-term weather and long-term
longer workable weather window for a specific operation and longer climate conditions is useful in the operational and maintenance
available maintenance time year-round for long-term planning. planning stages.
Hence, the maintenance costs, as well as the LCOE, are reduced due • Automatic control theories improve the operational efficiency of
to the smaller number of required vessels and technicians. OWT maintenance. Currently, there are many studies of the auton
• However, the involvement of specialized equipment increases the omous systems applied during OWT maintenance, such as dynamic
capital intensity of OWT maintenance. The scale and price of the positioning systems, climbing robots, heave compensators, and
devices increase with their sizes and OWT weights; for example, actively controlled tugger lines. Automatic systems exhibit remark
larger supply vessels, higher cranes, and more powerful tugboats able potential for unmanned maintenance in the future. The system
have been developed and deployed. Owing to these challenges, the redundancy can be improved by using fault-tolerant control.
financial safety of the wind power industry is at risk and making
companies less resistant to global economic fluctuations. Hence, Overall, this review provides a systematic knowledge set for wind
accomplishing maintenance tasks with the cooperation of small-scale farm operators and researchers, and provides guidance and suggestions
and commonly used equipment is a valuable issue. to policy decision-makers and technology developers. Additionally,
• Given that the scales of offshore wind farms are growing, and some information will be useful for related sister technologies such as
equipment is becoming more effective, there is a major trade-off tidal current energy farms and wave energy farms, which are being
between renting and buying O&M services. Maintenance planners readied for commercialization but for which only a handful literature
should evaluate many factors when deciding the percentage of self- are available [233,234].
operated maintenance, including budget and liquidity, outsourcing
agreement, the occurrence of emergencies, scheduling flexibility, Declaration of competing interest
technique and management levels, and strategy selection.
• The quantification of environmental impacts might be itemized into The authors declare that they have no known competing financial
LCOE and introduce extra costs. All O&M activities are influenced by interests or personal relationships that could have appeared to influence
17
the work reported in this paper. [29] Wen Binrong, Li Zhanwei, Jiang Zhihao, Tian Xinliang, Dong Xingjian,
Peng Zhike. Blade loading performance of a floating wind turbine in wave basin
model tests. Ocean Eng 2020;199:107061.
Acknowledgment [30] Lu Bin, Li Yaoyu, Wu Xin, Yang Zhongzhou. A review of recent advances in wind
turbine condition monitoring and fault diagnosis. In: 2009 IEEE power electronics
This work was supported by the Research Council of Norway (RCN) and machines in wind applications. IEEE; 2009. p. 1–7.
[31] Bhardwaj U, Teixeira AP, Guedes Soares C. Reliability prediction of an offshore
through the Centre for Research-based Innovation on Marine Operations wind turbine gearbox. Renew Energy 2019;141:693–706.
(RCN-project 237929), National Natural Science Foundation of China [32] Shipurkar Udai, Ma Ke, Henk Polinder, Blaabjerg Frede, Ferreira Jan A. A review
(Nos. 51761135012 and Nos. 11872248), Ministry of Science and of failure mechanisms in wind turbine generator systems. In: 2015 17th European
conference on power electronics and applications (EPE’15 ECCE-Europe). IEEE;
Technology of China (Nos. 2017YFE0132000), and National Key 2015. p. 1–10.
Research and Development Program of China (No. 2019YFE0102500 [33] Singh Gopal, Sundaram Kalpathy, Matuonto Marco. A solution to reduce
and 2019YFB1504402). The authors also express thanks to Prof. Amir overheating and increase wind turbine systems availability. Wind Engineering;
2020, 0309524X20910992.
Nejad of NTNU for discussions of the initial draft. [34] Sun Wei, Lin Wei-Cheng, You Fei, Shu Chi-Min, Qin Sheng-Hui. Prevention of
green energy loss: estimation of fire hazard potential in wind turbines. Renew
References Energy 2019;140:62–9.
[35] Price Seth J, Figueira Rita B. Corrosion protection systems and fatigue corrosion
in offshore wind structures: current status and future perspectives. Coatings 2017;
[1] Henrik Klinge Jacobsen, Hevia-Koch Pablo, Wolter Christoph. Nearshore and
7(2):25.
offshore wind development: costs and competitive advantage exemplified by
[36] Kang Jichuan, Sun Liping, Sun Hai, Wu Chunlin. Risk assessment of floating
nearshore wind in Denmark. Energy Sustain Dev 2019;50:91–100.
offshore wind turbine based on correlation-fmea. Ocean Eng 2017;129:382–8.
[2] Ahrens Uwe, Diehl Moritz, Roland Schmehl. Airborne wind energy. Springer
[37] Dong Xiaofeng, Lian Jijian, Wang Haijun, Yu Tongshun, Zhao Yue. Structural
Science & Business Media; 2013.
vibration monitoring and operational modal analysis of offshore wind turbine
[3] Sun Xiaojing, Huang Diangui, Wu Guoqing. The current state of offshore wind
structure. Ocean Eng 2018;150:280–97.
energy technology development. Energy 2012;41(1):298–312.
[38] Lu Bin, Sharma Santosh K. A literature review of igbt fault diagnostic and
[4] Zhang Da, Zhang Xiliang, He Jiankun, Chai Qimin. Offshore wind energy
protection methods for power inverters. IEEE Trans Ind Appl 2009;45(5):1770–7.
development in China: current status and future perspective. Renew Sustain
[39] Wu Haoyu, Zhao Yongsheng, He Yanping, Shao Yanlin, Mao Wengang,
Energy Rev 2011;15(9):4673–84.
Han Zhaolong, Huang Chao, Gu Xiaoli, Jiang Zhiyu. Transient response of a tlp-
[5] F Selot D Fraile, Brindley G, Walsh C. Offshore wind in europe–key trends and
type floating offshore wind turbine under tendon failure conditions. Ocean Eng
statistics 2018. Wind Europe; 2019. Technical report.
2021;220:108486.
[6] Dudley Bob. Bp statistical review of world energy. London, UK: BP Statistical
[40] Tavner PJ, van Bussel GJW, Spinato F. Machine and converter reliabilities in wind
Review; 2018. accessed Aug, 6:2018.
turbines. IET Conf Proc 2006;(3):127–30. January.
[7] Wind Europe. Wind energy in europe in 2018—trends and statistics. Brussels,
[41] Karyotakis A, Bucknall R. Planned intervention as a maintenance and repair
Belgium: Wind Europe; 2019.
strategy for offshore wind turbines. J Mar Eng Technol 2010;9(1):27–35.
[8] Soares-Ramos Emanuel PP, de Oliveira-Assis Lais, Sarrias-Mena Raúl, Fernández-
[42] Jiang Peng. Maintenance of wind turbine. Electr Equip 2011;28(6):68–71.
Ramírez Luis M. Current status and future trends of offshore wind power in
[43] Faulstich Stefan, Hahn Berthold, Peter J Tavner. Wind turbine downtime and its
europe. Energy 2020;202:117787.
importance for offshore deployment. Wind Energy 2011;14(3):327–37.
[9] Duan Fei. Wind energy cost analysis coe for offshore wind and lcoe financial
[44] Zied Hajej, Nidhal Rezg. An optimal preventive maintenance plan according to
modeling. Master’s thesis. Helsinki Metropolia University of Applied Sciences;
power generation for a wind turbine. In: International conference on industrial
2017.
engineering and operations management; 2016.
[10] Hofmann Matthias, Iver Bakken Sperstad. Will 10 mw wind turbines bring down
[45] Alexander Karyotakis. On the optimisation of operation and maintenance
the operation and maintenance cost of offshore wind farms? Energy Procedia
strategies for offshore wind farms. PhD thesis. University College London; 2011.
2014;53:231–8.
[46] Santos FP, Teixeira AP, Guedes Soares C. An age-based preventive maintenance
[11] Lerch Markus, De-Prada-Gil Mikel, Molins Climent, Benveniste Gabriela.
for offshore wind turbines. In: Safety and reliability: methodology and
Sensitivity analysis on the levelized cost of energy for floating offshore wind
applications. CRC Press; 2014. p. 1183–92.
farms. Sustain Energy Technol Assess 2018;30:77–90.
[47] Dui Hongyan, Si Shubin, Yam Richard CM. A cost-based integrated importance
[12] Verma Amrit Shankar. Modelling, analysis and response-based operability
measure of system components for preventive maintenance. Reliab Eng Syst Saf
assessment of offshore wind turbine blade installation with emphasis on impact
2017;168:98–104.
damages. PhD thesis. NTNU; 2019.
[48] Amir Rasekhi Nejad, Gao Zhen, Moan Torgeir. Fatigue reliability-based
[13] Wu Xiaoni, Hu Yu, Ye Li, Yang Jian, Duan Lei, Wang Tongguang, Adcock Thomas,
inspection and maintenance planning of gearbox components in wind turbine
Jiang Zhiyu, Gao Zhen, Lin Zhiliang, Borthwick Alistair, Liao Shijun. Foundations
drivetrains. Energy Procedia 2014;53:248–57.
of offshore wind turbines: a review. Renew Sustain Energy Rev 2019;104:379–93.
[49] Itamar Esdras Martínez García, Alejandro Sánchez Sánchez, Barbati Stefano.
[14] Jiang Zhiyu. Installation of offshore wind turbines: a technical review. Renew
Reliability and preventive maintenance. In: MARE-WINT: new materials and
Sustain Energy Rev 2021;139:110576.
reliability in offshore wind turbine technology. Springer International Publishing;
[15] Van Bussel GJW, Zaaijer MB. Reliability, availability and maintenance aspects of
2016. p. 235–72.
large-scale offshore wind farms, a concepts study. In: Proceedings of MAREC,
[50] Jannie Jessen Nielsen, John Dalsgaard Sørensen. On risk-based operation and
2001; 2001.
maintenance of offshore wind turbine components. Reliab Eng Syst Saf 2011;96
[16] Castro-Santos Laura, Diaz-Casas Vicente. Life-cycle cost analysis of floating
(1):218–29.
offshore wind farms. Renew Energy 2014;66:41–8.
[51] Ramírez and John Dalsgaard Sørensen José G Rangel. Maintenance planning of
[17] Maienza C, Avossa AM, Ricciardelli F, Coiro D, Troise G, Christos Thomas
offshore wind turbine using condition monitoring information. In: The
Georgakis. A life cycle cost model for floating offshore wind farms. Appl Energy
international conference on ocean, offshore and arctic engineering (OMAE).
2020;266:114716.
American Society of Mechanical Engineers; 2009. OMAE09–79159.
[18] Dinwoodie Iain, Mcmillan David. Operation and maintenance of offshore wind
[52] Van Bussel GJW, Henderson AR, Morgan CA, Smith B, Barthelmie R, Argyriadis K,
farms. Eng Technol Ref 2014;1(1).
Arena A, Niklasson G, Peltola E. State of the art and technology trends for
[19] Zhao Xin Gang, Ren Ling Zhi. Focus on the development of offshore wind power
offshore wind energy: operation and maintenance issues. In: Offshore Wind
in China: has the golden period come? Renew Energy 2015;81:644–57.
Energy EWEA special topic conference; 2001.
[20] Tyler J Stehly, Beiter Philipp C. Cost of wind energy review. Technical report.
[53] Asensio E Segura, Pérez JM Pinar, Márquez FP García. Economic viability study
Golden, CO (United States): National Renewable Energy Lab.(NREL); 2018. 2020.
for offshore wind turbines maintenance management. In: Proceedings of the ninth
[21] Ryan Wiser, Bolinger Mark. Wind technologies market report. Technical report.
international conference on management science and engineering management.
Golden, CO (United States): National Renewable Energy Lab.(NREL); 2010. 2011.
Springer; 2015. p. 235–44.
[22] Staffell Iain, Green Richard. How does wind farm performance decline with age?
[54] Walgern Julia, Peters Lennart, Madlener Reinhard. Economic evaluation of
Renew Energy 2014;66:775–86.
maintenance strategies for offshore wind turbines based on condition monitoring
[23] Guanche Raúl, Martini Michele, Jurado Alfonso, Iñigo J, Losada. Walk-to-work
systems. In: FCN working papers 8/2017, E.ON energy research center, future
accessibility assessment for floating offshore wind turbines. Ocean Eng 2016;116:
energy consumer needs and behavior (FCN); 2017.
216–25.
[55] Tchakoua Pierre, Wamkeue René, Ouhrouche Mohand, Slaoui-Hasnaoui Fouad, ,
[24] European Wind Energy Association. EU Energy policy to 2050. EWEA; 2011.
Tommy Andy Tameghe, Gabriel Ekemb. Wind turbine condition monitoring:
[25] Mahmood Shafiee. Maintenance logistics organization for offshore wind energy:
state-of-the-art review, new trends, and future challenges. Energies 2014;7(4):
current progress and future perspectives. Renew Energy 2015;77(2):182–93.
2595–630.
[26] Martinez Luengo Maria, Kolios Athanasios. Failure mode identification and end of
[56] Yang Wenxian, Peter J Tavner, Crabtree Christopher J, Feng Y, Qiu Y. Wind
life scenarios of offshore wind turbines: a review. Energies 2015;8(8):8339–54.
turbine condition monitoring: technical and commercial challenges. Wind Energy
[27] Jenny Niebsch, Ramlau Ronny, Nguyen Thien T. Mass and aerodynamic
2014;17(5):673–93.
imbalance estimates of wind turbines. Energies 2010;3(4):696–710.
[57] Liu WY, Tang BP, Han JG, Lu XN, Hu NN, He ZZ. The structure healthy condition
[28] Dong Wenbin, Moan Torgeir, Gao Zhen. Fatigue reliability analysis of the jacket
monitoring and fault diagnosis methods in wind turbines: a review. Renew
support structure for offshore wind turbine considering the effect of corrosion and
Sustain Energy Rev 2015;44:466–72.
inspection. Reliab Eng Syst Saf 2012;106:11–27.
18
[58] Dias Machado de Azevedo Henrique, , Alex Maurício Araújo, [85] Zhang Chen, Gao Wei, Guo Sheng, Li Youliang, Yang Tao. Opportunistic
Bouchonneau Nadège. A review of wind turbine bearing condition monitoring: maintenance for wind turbines considering imperfect, reliability-based
state of the art and challenges. Renew Sustain Energy Rev 2016;56:368–79. maintenance. Renew Energy 2017;103:606–12.
[59] Daneshi-Far Zahra, Gerard-Andre Capolino, Henao Humberto. Review of failures [86] Sarker Bhaba R, Tasnim Ibn Faiz. Minimizing maintenance cost for offshore wind
and condition monitoring in wind turbine generators. In: The XIX international turbines following multi-level opportunistic preventive strategy. Renew Energy
conference on electrical machines-ICEM 2010. IEEE; 2010. p. 1–6. 2016;85:104–13.
[60] Salameh Jack P, Cauet Sebastien, Etien Erik, Sakout Anas, Rambault Laurent. [87] Thi Anh Tuyet Nguyen, Chou Shuo-Yan. Maintenance strategy selection for
Gearbox condition monitoring in wind turbines: a review. Mech Syst Signal improving cost-effectiveness of offshore wind systems. Energy Convers Manag
Process 2018;111:251–64. 2018;157:86–95.
[61] Amirat Yassine, Benbouzid Mohamed EH, Bensaker Bachir, Wamkeue René. [88] Xie Lubing, Rui Xiaoming, Li Shuai, Hu Xin. An opportunistic maintenance
Condition monitoring and fault diagnosis in wind energy conversion systems: a strategy for offshore wind turbine based on accessibility evaluation. Wind
review. In: 2007 IEEE international electric machines & drives conference, vol. 2. Engineering; 2019. page 0309524X19852351.
IEEE; 2007. p. 1434–9. [89] Ding Fangfang, Tian Zhigang. Opportunistic maintenance for wind farms
[62] Maheswari R Uma, Umamaheswari R. Trends in non-stationary signal processing considering multi-level imperfect maintenance thresholds. Renew Energy 2012;
techniques applied to vibration analysis of wind turbine drive train–a 45:175–82.
contemporary survey. Mech Syst Signal Process 2017;85:296–311. [90] Abdollahzadeh Hadi, Atashgar Karim, Abbasi Morteza. Multi-objective
[63] Yang Wenxian, Peter J Tavner, Crabtree Christopher J, Wilkinson Michael. Cost- opportunistic maintenance optimization of a wind farm considering limited
effective condition monitoring for wind turbines. IEEE Trans Ind Electron 2009; number of maintenance groups. Renew Energy 2016;88:247–61.
57(1):263–71. [91] Zhang Chen, Gao Wei, Yang Tao, Guo Sheng. Opportunistic maintenance strategy
[64] Lian Jijian, Cai Ou, Dong Xiaofeng, Jiang Qi, Zhao Yue. Health monitoring and for wind turbines considering weather conditions and spare parts inventory
safety evaluation of the offshore wind turbine structure: a review and discussion management. Renew Energy 2019;133:703–11.
of future development. Sustainability 2019;11(2):494. [92] Zhou P, Yin PT. An opportunistic condition-based maintenance strategy for
[65] Tavner Peter. Offshore wind turbines: reliability, availability and maintenance. offshore wind farm based on predictive analytics. Renew Sustain Energy Rev
Energy engineering. Institution of Engineering and Technology; 2012. 2019;109:1–9.
[66] Mahmood Shafiee, Finkelstein Maxim, Bérenguer Christophe. An opportunistic [93] Zhu Wenjin, Bruno Castanier, Bettayeb Belgacem. A dynamic programming-based
condition-based maintenance policy for offshore wind turbine blades subjected to maintenance model of offshore wind turbine considering logistic delay and
degradation and environmental shocks. Reliab Eng Syst Saf 2015;142:463–71. weather condition. Reliab Eng Syst Saf 2019;190:106512.
[67] Allan May and David Mcmillan. In: Condition based maintenance for offshore [94] Niu Hanlin, Al Savvaris, Tsourdos Antonios, Ji Ze. Voronoi-visibility roadmap-
wind turbines: the effects of false alarms from condition monitoring systems. based path planning algorithm for unmanned surface vehicles. J Navig 2019;72
Esrel; 2013. p. 218–29. (4):850–74.
[68] May Allan, McMillan David, Thöns Sebastian. Economic analysis of condition [95] Niu Hanlin, Ji Ze, Al Savvaris, Tsourdos Antonios. Energy efficient path planning
monitoring systems for offshore wind turbine sub-systems. Renew Power Gen Iet for unmanned surface vehicle in spatially-temporally variant environment. Ocean
2015;9(8):900–7. Eng 2020;196:106766.
[69] Wen Binrong, Tian Xinliang, Jiang Zhihao, Li Zhanwei, Dong Xingjian, [96] Gutierrez-Alcoba A, Hendrix EMT, Ortega G, Halvorsen-Weare EE, Haugland D.
Peng Zhike. Monitoring blade loads for a floating wind turbine in wave basin On offshore wind farm maintenance scheduling for decision support on vessel
model tests using fiber bragg grating sensors: a feasibility study. Mar Struct 2020; fleet composition. Eur J Oper Res 2019;279(1):124–31.
71:102729. [97] Helene Seyr, Muskulus Michael. Decision support models for operations and
[70] Song Sanling, Li Qing, Frank Felder, Wang Honggang, Coit David. Integrated maintenance for offshore wind farms: a review. Appl Sci 2019;9(2):278.
optimization of offshore wind farm layout design and turbine opportunistic [98] Nurseda Y, Yürüşen, Rowley Paul N, Watson Simon J, Melero Julio. Automated
condition-based maintenance. Comput Ind Eng 2018. wind turbine maintenance scheduling. Reliability Engineering & System Safety;
[71] Colone Lorenzo, Dimitrov Nikolay, Straub Daniel. Predictive repair scheduling of 2020. p. 106965.
wind turbine drive-train components based on machine learning. Wind Energy [99] Perez-Canto Salvador, Rubio-Romero Juan Carlos. A model for the preventive
2019;22(9):1230–42. maintenance scheduling of power plants including wind farms. Reliab Eng Syst
[72] Sivalingam Krishnamoorthi, Marco Sepulveda, Spring Mark, Davies Peter. Saf 2013;119:67–75.
A review and methodology development for remaining useful life prediction of [100] Zhang Zhenyou. Scheduling and routing optimization of maintenance fleet for
offshore fixed and floating wind turbine power converter with digital twin offshore wind farms using duo-aco. Adv Mater Res 2014;1039:294–301.
technology perspective. In: 2018 2nd international conference on green energy [101] Gundegjerde Christian, Halvorsen Ina B, Halvorsen-Weare Elin E, Lars Magnus
and applications (ICGEA). IEEE; 2018. p. 197–204. Hvattum, Lars Magne Nonås. A stochastic fleet size and mix model for
[73] Garcia Mari Cruz, Sanz-Bobi Miguel A, Del Pico Javier. Simap: intelligent system maintenance operations at offshore wind farms. Transport Res C Emerg Technol
for predictive maintenance: application to the health condition monitoring of a 2015;52:74–92.
windturbine gearbox. Comput Ind 2006;57(6):552–68. [102] Dalgic Yalcin, Lazakis Iraklis, Dinwoodie Iain, McMillan David, Revie Matthew.
[74] Canizo Mikel, Onieva Enrique, Conde Angel, Santiago Charramendieta, Advanced logistics planning for offshore wind farm operation and maintenance
Trujillo Salvador. Real-time predictive maintenance for wind turbines using big activities. Ocean Eng 2015;101:211–26.
data frameworks. In: 2017 IEEE international conference on prognostics and [103] Li Xiaodong, Ouelhadj Djamila, Song Xiang, Jones Dylan, Wall Graham,
health management (ICPHM). IEEE; 2017. p. 70–7. Howell Kerry E, Paul Igwe, Martin Simon, Song Dongping, Pertin Emmanuel.
[75] Geselschap Catina, Meskers Geert, Radboud Van Dijk, Van Winsen Ivan. Digital A decision support system for strategic maintenance planning in offshore wind
twin-engineering with the human factor in the loop. In: Offshore technology farms. Renew Energy 2016;99:784–99.
conference. Offshore Technology Conference; 2019. [104] Nora Tangen Raknes, Ødeskaug Katrine, Stålhane Magnus, Lars Magnus Hvattum.
[76] Meisam Jahanshahi Zeitouni, Ahmad Parvaresh, Abrazeh Saber, Mohseni Saeid- Scheduling of maintenance tasks and routing of a joint vessel fleet for multiple
Reza, Gheisarnejad Meysam, Khooban Mohammad-Hassan. Digital twins-assisted offshore wind farms. J Mar Sci Eng 2017;5(1):11.
design of next-generation advanced controllers for power systems and electronics: [105] Stock-Williams Clym, Krishna Swamy Siddharth. Automated daily maintenance
wind turbine as a case study. Inventions 2020;5(2):19. planning for offshore wind farms. Renew Energy 2019;133:1393–403.
[77] Jiang Zhiyu, Bjørnholm Marius, Guo Jiamin, Dong Wenbin, Ren Zhengru, [106] Yildirim Murat, Gebraeel Nagi Z, Xu Andy Sun. Integrated predictive analytics
Verma Amrit Shankar. Damage identification of a jacket support structure for and optimization for opportunistic maintenance and operations in wind farms.
offshore wind turbines. In: 2020 15th IEEE conference on industrial electronics IEEE Trans Power Syst 2017;32(6):4319–28.
and applications (ICIEA). IEEE; 2020. p. 995–1000. [107] Krokoszinski Hans-Joachim. Efficiency and effectiveness of wind farms-keys to
[78] Wang Ke Sheng, Sharma Vishal S, Zhang Zhen You. Scada data based condition cost optimized operation and maintenance. Renewable Energy; 2003.
monitoring of wind turbines. Adv Manuf 2014;2(1):61–9. [108] Zhong Shuya, Pantelous Athanasios A, Beer Michael, Zhou Jian. Constrained non-
[79] Hameed Z, Vatn J, Heggset J. Challenges in the reliability and maintainability linear multi-objective optimisation of preventive maintenance scheduling for
data collection for offshore wind turbines. Renew Energy 2011;36(8):2154–65. offshore wind farms. Mech Syst Signal Process 2018;104:347–69.
[80] Zsolt János Viharos, Csaba István Sidló, Benczúr András, Csempesz János, [109] Hajej Zied, Nidhal Rezg, Anis Chelbi, Bouzoubaa Maryem. An optimal integrated
Krisztián Balázs Kis, Petrás István, Garzó András. “big data” initiative as an it production and maintenance strategy for a multi-wind turbines system. Int J Prod
solution for improved operation and maintenance of wind turbines. In: European Res 2019:1–24.
wind energy association (EWEA) conference; 2013. p. 184–8. [110] Chi-Cong Nguyen, Thi-Hong-Hieu Le, Phat-Tai Tran. A numerical study of
[81] Radner Roy, Jorgenson Dale W. Opportunistic replacement of a single part in the thickness effect of the symmetric naca 4-digit airfoils on self starting capability of
presence of several monitored parts. Manag Sci 1963;10(1):70–84. a 1kw h-type vertical axis wind turbine. Int J Mech Eng Appl 2015;3(3):7.
[82] Thomas Édouard, Levrat Éric, Iung BenoÎt. Overview on opportunistic [111] Zhong Shuya, Pantelous Athanasios A, Goh Mark, Zhou Jian. A reliability-and-
maintenance. IFAC Proc Vol 2008;41(3):245–50. cost-based fuzzy approach to optimize preventive maintenance scheduling for
[83] Dai Lijuan, Stålhane Magnus, Ingrid B Utne. Routing and scheduling of offshore wind farms. Mech Syst Signal Process 2019;124:643–63.
maintenance fleet for offshore wind farms. Wind Eng 2015;39(1):15–30. [112] Ge Xiaolin, Chen Quan, Fu Yang, Chung CY, Yang Mi. Optimization of
[84] Besnard Frangois, Patrikssont Michael, Ann Brith Strombergt, maintenance scheduling for offshore wind turbines considering the wake effect of
Adam Wojciechowskit, Bertling Lina. An optimization framework for arbitrary wind direction. Elec Power Syst Res 2020;184:106298.
opportunistic maintenance of offshore wind power system. In: PowerTech. IEEE [113] Rinaldi Giovanni, Pillai Ajit C, Thies Philipp R, Johanning Lars. Multi-objective
Bucharest; 2009. p. 1–7. 2009. optimization of the operation and maintenance assets of an offshore wind farm
using genetic algorithms. Wind Engineering; 2019, 0309524X19849826.
19
[114] Sahnoun M’hammed, Baudry David, Mustafee Navonil, Louis Anne, Andi [141] Jiang Zhiyu, Yttervik Rune, Gao Zhen, Peter Christian Sandvik. Design,
Smart Philip, Godsiff Phil, Mazari Belahcene. Modelling and simulation of modelling, and analysis of a large floating dock for spar floating wind turbine
operation and maintenance strategy for offshore wind farms based on multi-agent installation. Mar Struct 2020;72:102781.
system. J Intell Manuf 2019;30(8):2981–97. [142] Ren Zhengru, Jiang Zhiyu, Roger Skjetne, Gao Zhen. Development and
[115] Joschko Philip, Andi H Widok, Appel Susanne, Greiner Saskia, Albers Henning, application of a simulator for offshore wind turbine blades installation. Ocean Eng
Page Bernd. Modeling and simulation of offshore wind farm o&m processes. 2018;166:380–95.
Environ Impact Assess Rev 2015;52:31–9. [143] Wu Mingkang. Numerical analysis of docking operation between service vessels
[116] Mustafee Navonil, Sahnoun M’Hammed, Smart Andi, Godsiff Phil. An application and offshore wind turbines. Ocean Eng 2014;91:379–88.
of distributed simulation for hybrid modeling of offshore wind farms. In: [144] Auestad Øyvind F, Gravdahl Jan T, Perez Tristan, Asgeir J Sørensen, Trygve H
Proceedings of the 3rd ACM SIGSIM conference on principles of advanced discrete Espeland. Boarding control system for improved accessibility to offshore wind
simulation; 2015. p. 171–2. turbines: full-scale testing. Contr Eng Pract 2015;45:207–18.
[117] Borucki Jakub, Pawel Pawlewski, Chowanski Wojciech. Mixing abs and des [145] Ren Zhengru, Roger Skjetne, Verma Amrit Shankar, Jiang Zhiyu, Gao Zhen, Karl
approach to modeling of a delivery process in the automotive industry. In: Henning Halse. Active heave compensation of floating wind turbine installation
International conference on practical applications of agents and multi-agent using a catamaran construction vessel. Mar Struct 2021;75:102868.
systems. Springer; 2014. p. 133–43. [146] Ren Zhengru, Roger Skjetne, Jiang Zhiyu, Gao Zhen, Verma Amrit Shankar.
[118] McAuliffe F Devoy, Desmond C, Chester R, Flannery B, Judge F, Lynch K, Integrated GNSS/IMU hub motion estimator for offshore wind turbine blade
Murphy J. A tool to simulate decommissioning offshore wind farms. In: Journal of installation. Mech Syst Signal Process 2019;123:222–43.
physics: conference series, vol. 1356. IOP Publishing; 2019, 012021. [147] Merriaux Pierre, Boutteau Rémi, Pascal Vasseur, Savatier Xavier. IMU/LIDAR
[119] Irawan Chandra Ade, Ouelhadj Djamila, Jones Dylan, Stålhane Magnus, Iver based positioning of a gangway for maintenance operations on wind farms. In:
Bakken Sperstad. Optimisation of maintenance routing and scheduling for 2014 IEEE/RSJ international conference on intelligent robots and systems. IEEE;
offshore wind farms. Eur J Oper Res 2016;256(1):76–89. 2014. p. 4213–9.
[120] Jin Tongdan, Tian Zhigang, Huerta Miguel, Piechota Jett. Coordinating [148] Liang Lihua, Le Zhiwen, Zhang Songtao, Li Jianfeng. Modeling and controller
maintenance with spares logistics to minimize levelized cost of wind energy. In: design of an active motion compensated gangway based on inverse dynamics in
2012 international conference on quality, reliability, risk, maintenance, and joint space. Ocean Eng 2020;197:106864.
safety engineering. IEEE; 2012. p. 1022–7. [149] Ranneh Mohamad Amr. Automated docking of an offshore gangway: a predictive
[121] Michiel AJ uit het Broek, Veldman Jasper, Fazi Stefano, Roy Greijdanus. control approach. Master’s thesis. Delft University of Technology; 2019.
Evaluating resource sharing for offshore wind farm maintenance: the case of jack- [150] David Julio Cerda Salzmann. Ampelmann: Development of the access System for
up vessels. Renew Sustain Energy Rev 2019;109:619–32. offshore wind turbines. PhD thesis. Delft University of Technology; 2010.
[122] Ir Thijs Nicolaas Schouten, Dekker Rommert, Ayse Sena Eruguz. Optimal [151] Hu B, Stumpf P, Deijl W vd. Offshore wind access 2019. Technical report. Petten,
maintenance policies for wind turbines under time-varying costs. Master’s thesis. The Netherlands: TNO; 2019.
Rotterdam, the Netherlands: Erasmus Universiteit; 2019. [152] Brändli Silvan, Höft Eyke, Abdel-Maksoud Moustafa, Alexander Düster.
[123] Dinwoodie I, Catterson VM, Mcmillan D. Wave height forecasting to improve off- Simulation of the interaction of service ships with offshore wind turbine plants.
shore access and maintenance scheduling. J Epidemiol Community 2013;67(8): In: Proceedings of the conference on maritime energy; 2013. p. 511–8. Hamburg,
1–5. Germany.
[124] Thi Anh Tuyet Nguyen, Chou Shuo-Yan. Improved maintenance optimization of [153] Ferreira González Daniel, Lemmerhirt Matthias, Abdel-Maksoud Moustafa,
offshore wind systems considering effects of government subsidies, lost König Marcel, Alexander Düster. Numerical and experimental investigation
production and discounted cost model. Energy 2019;187:115909. regarding the landing manoeuvre of a catamaran vessel at an offshore wind
[125] Stålhane Magnus, Halvorsen-Weare Elin E, Lars Magne Nonås, Pantuso Giovanni. turbine in waves. In: ASME 2015 34th international conference on ocean, offshore
Optimizing vessel fleet size and mix to support maintenance operations at and arctic engineering. American Society of Mechanical Engineers Digital
offshore wind farms. Eur J Oper Res 2019;276(2):495–509. Collection; 2015.
[126] Halvorsen-Weare Elin E, Gundegjerde Christian, Halvorsen Ina B, Lars Magnus [154] König Marcel, Ferreira González Daniel, Abdel-Maksoud Moustafa,
Hvattum, Lars Magne Nonås. Vessel fleet analysis for maintenance operations at Alexander Düster. Numerical investigation of the landing manoeuvre of a crew
offshore wind farms. Energy Procedia 2013;35:167–76. transfer vessel to an offshore wind turbine. Ships Offshore Struct 2017;12(sup1):
[127] Irawan Chandra Ade, Eskandarpour Majid, Ouelhadj Djamila, Jones Dylan. S115–33.
Simulation-based optimisation for stochastic maintenance routing in an offshore [155] Biehl Florian, Lehmann Eike. Collisions of ships with offshore wind turbines:
wind farm. Eur J Oper Res 2019. calculation and risk evaluation. In: International conference on offshore
[128] Roger Skjetne, Ren Zhengru. A survey on modeling and control of thruster- mechanics and arctic engineering; 2006. p. 281–304.
assisted position mooring systems. Mar Struct 2020;74:102830. [156] Dai Lijuan, Ehlers Soren, Utne, Bouwer Ingrid, Rausand, Marvin. Risk of collision
[129] Cheng Xu, Li Guoyuan, Andrè Listou Ellefsen, Chen Shengyong, Hildre Hans between service vessels and offshore wind turbines. Reliab Eng Syst Saf 2013;109
Petter, Zhang Houxiang. A novel densely connected convolutional neural network (109):18–31.
for sea state estimation using ship motion data. IEEE Trans Instrum Meas 2020. [157] Lee Kangsu. Effects on the various rubber fenders of a tripod offshore wind
[130] Ren Zhengru, Han Xu, Verma Amrit Shankar, Alexander Dirdal Johann, turbine substructure collision strength due to boat. Ocean Eng 2013;72:188–94.
Roger Skjetne. Sea state estimation based on vessel motion responses: improved [158] Liu Chunguang, Hao Ertong, Zhang Shibo. Optimization and application of a
smoothness and robustness using bézier surface and l1 optimization. Mar Struct crashworthy device for the monopile offshore wind turbine against ship impact.
2021;76. 102904. Appl Ocean Res 2015;51:129–37.
[131] Han Xu, Bernt Johan Leira, Sævik Svein. Vessel hydrodynamic model tuning by [159] Iver Bakken Sperstad, Halvorsen-Weare Elin E, Hofmann Matthias, Lars Magne
discrete bayesian updating using simulated onboard sensor data. Ocean Eng 2021; Nonås, Stålhane Magnus, Wu Mingkang. A comparison of single- and multi-
220:108407. parameter wave criteria for accessing wind turbines in strategic maintenance and
[132] M Catterson V, Mcmillan D, Dinwoodie I, Revie M, Dowell J, Quigley J, Wilson K. logistics models for offshore wind farms. Energy Procedia 2014;53(5939):221–30.
An economic impact metric for evaluating wave height forecasters for offshore [160] Song Ming, Jiang Zhiyu, Yuan Wei. Numerical and analytical analysis of a
wind maintenance access. Wind Energy 2016;19(2):199–212. monopile-supported offshore wind turbine under ship impacts. Renewable Energy
[133] James Eric P, Benjamin Stanley G, Marquis Melinda. Offshore wind speed 2020.
estimates from a high-resolution rapidly updating numerical weather prediction [161] Carla E. Presencia and Mahmood Shafiee. Risk analysis of maintenance ship
model forecast dataset. Wind Energy 2017;21(4). collisions with offshore wind turbines. Int J Sustain Energy 2017:1–21.
[134] Taylor James W, Jeon Jooyoung. Probabilistic forecasting of wave height for [162] Moulas D, Shafiee M, Mehmanparast A. Damage analysis of ship collisions with
offshore wind turbine maintenance. Eur J Oper Res 2018;267. offshore wind turbine foundations. Ocean Eng 2017;143:149–62.
[135] Van Bussel GJW, Bierbooms WAAM. Analysis of different means of transport in [163] van Bussel GJW, Zaaijer MB, van den Broek W. Toward selection of concepts for
the operation and maintenance strategy for the reference dowec offshore wind offshore support structures for large scale wind turbines. In: Proceedings of
farm. Proc OW EMES, Naples 2003:1–12. MAREC 200: marine renewable energy conference; 2001.
[136] Dalgic Yalcin, Lazakis Iraklis, Osman Turan. Investigation of optimum crew [164] van Dun Luuk. Maintenance offshore wind: feasibility study into alternative
transfer vessel fleet for offshore wind farm maintenance operations. Wind Eng lifting systems for offshore wind turbine maintenance operations. Master’s thesis.
2015;39(1):31–52. Delft University of Technology; 2018.
[137] Iver Bakken Sperstad, Stålhane Magnus, Dinwoodie Iain, Endrerud Ole Erik V, [165] Van Bussel GJW, Schöntag Chr. Operation and maintenance aspects of large
Martin Rebecca, Warner Ethan, Iver Bakken Sperstad, Stålhane Magnus, offshore windfarms. In: EWEC conference. EWEC; 1997. p. 272–5.
Dinwoodie Iain, Endrerud Ole Erik V. Testing the robustness of optimal access [166] Ren Zhengru, Roger Skjetne, Gao Zhen. A crane overload protection controller for
vessel fleet selection for operation and maintenance of offshore wind farms. blade lifting operation based on model predictive control. Energies 2019;12(1):
Ocean Eng 2017;145:334–43. 50.
[138] Mustafee Navonil, Anna Wienke, Smart Andi, Godsiff Phil. Learning maintenance, [167] Ren Zhengru, Jiang Zhiyu, Gao Zhen, Roger Skjetne. Active tugger line force
repair and operations (mro) concepts in offshore wind industry through game- control for single blade installation. Wind Energy 2018;21(12):1344–58.
based learning. In: 2015 winter simulation conference (WSC). IEEE; 2015. [168] Ren Zhengru, Roger Skjetne, Jiang Zhiyu, Gao Zhen. Active single-blade
p. 1068–79. installation using tugger line tension control and optimal control allocation. Int J
[139] Dornhelm Esther, Helene Seyr, Muskulus Michael. Vindby—a serious offshore Offshore Polar Eng 2020;30(2):220–7.
wind farm design game. Energies 2019;12(8):1499. [169] Gao Zhen, Verma Amrit, Zhao Yuna, Jiang Zhiyu, Ren Zhengru. A summary of the
[140] Zhao Yuna, Cheng Zhengshun, Peter Christian Sandvik, Gao Zhen, Moan Torgeir. recent work at ntnu on marine operations related to installation of offshore wind
An integrated dynamic analysis method for simulating installation of single turbines. In: ASME 2018 37th international conference on ocean, offshore and
blades for wind turbines. Ocean Eng 2018;152:72–88.
20
arctic engineering. American Society of Mechanical Engineers Digital Collection; [196] Florian Mihai, Dalsgaard Sãrensen John. Wind turbine blade life-time assessment
2018. model for preventive planning of operation and maintenance. J Mar Sci Eng 2015;
[170] Verma Amrit Shankar, Ulrich Haselbach Philipp, Nils Petter Vedvik, Gao Zhen. 3(3):1027–40.
A global-local damage assessment methodology for impact damage on offshore [197] Nguyen Trinh Hoang, Prinz Andreas, Noll Josef. Proactive maintenance of
wind turbine blades during lifting operations. In: ASME 2018 37th international offshore wind turbine blades using knowledge-based force analysis. In: Third
conference on ocean, offshore and arctic engineering. American Society of international conference on innovative computing technology; 2013.
Mechanical Engineers Digital Collection; 2018. [198] Amith Nag Nichenametla, Nandipati Srikanth, Abhay Laxmanrao Waghmare.
[171] Verma Amrit Shankar, Jiang Zhiyu, Ren Zhengru, Gao Zhen, Nils Petter Vedvik. Optimizing life cycle cost of wind turbine blades using predictive analytics in
Response-based assessment of operational limits for mating blades on monopile- effective maintenance planning. In: 2017 annual reliability and maintainability
type offshore wind turbines. Energies 2019;12(10):1867. symposium (RAMS). IEEE; 2017. p. 1–6.
[172] Verma Amrit Shankar, Gao Zhen, Jiang Zhiyu, Ren Zhengru, Nils Petter Vedvik. [199] Bouzid, Mabrok Omar. In-situ health monitoring for wind turbine blade using
Structural safety assessment of marine operations from a long-term perspective: a acoustic wireless sensor networks at low sampling rates. University of Newcastle
case study of offshore wind turbine blade installation. In: ASME 2019 38th Upon Tyne; 2013.
international conference on ocean, offshore and arctic engineering. American [200] Chen Chia, Ciang, Lee Jung-Ryul, Hyung-Joon Bang. Structural health monitoring
Society of Mechanical Engineers Digital Collection; 2019. for a wind turbine system: a review of damage detection methods. Meas Sci
[173] Verma Amrit Shankar, Zhao Yuna, Gao Zhen, Nils Petter Vedvik. Explicit Technol 2008;19(12):122001.
structural response-based methodology for assessment of operational limits for [201] Alfredo Arcos Jiménez, Zhang Long, Carlos Quiterio Gómez Muñoz, Fausto Pedro
single blade installation for offshore wind turbines. In: Proceedings of the fourth García Márquez. Maintenance management based on machine learning and
international conference in ocean engineering (ICOE2018). Springer; 2019. nonlinear features in wind turbines. Renew Energy 2020;146:316–28.
p. 737–50. [202] Li Mingxin, Kang Jichuan, Sun Liping, Wang Mian. Development of optimal
[174] Verma Amrit Shankar, , Nils Petter Vedvik, Gao Zhen. A comprehensive maintenance policies for offshore wind turbine gearboxes based on the non-
numerical investigation of the impact behaviour of an offshore wind turbine blade homogeneous continuous-time markov process. J Mar Sci Appl 2019;18(1):93–8.
due to impact loads during installation. Ocean Eng 2019;172:127–45. [203] Kabir MJ, Amanullah M T Oo, Rabbani Mahbub. A brief review on offshore wind
[175] Verma Amrit Shankar, Jiang Zhiyu, , Nils Petter Vedvik, Gao Zhen, Ren Zhengru. turbine fault detection and recent development in condition monitoring based
Impact assessment of a wind turbine blade root during an offshore mating maintenance system. In: Power engineering conference; 2015. p. 1–7.
process. Eng Struct 2019;180:205–22. [204] Link H, Lacava W, Van Dam J, Mcniff B, Sheng S, Wallen R, Mcdade M, Lambert S,
[176] Verma Amrit Shankar, , Nils Petter Vedvik, Gao Zhen. Numerical assessment of Butterfield S, Oyague F. In: Gearbox reliability collaborative project report:
wind turbine blade damage due to contact/impact with tower during installation. findings from phase 1 and phase 2 testing, nrel technical report: nrel/tp-5000-
IOP Conf Ser Mater Sci Eng 2017;276(1):012025. 51885. Office of Scientific & Technical Information Technical Reports; 2011.
[177] Verma Amrit Shankar, Jiang Zhiyu, Gao Zhen, Nils Petter Vedvik. Effects of a [205] Kang Jichuan. In: Preventative maintenance optimization for offshore wind
passive tuned mass damper on blade root impacts during the offshore mating turbine gearbox. Twenty-Seventh; 2017.
process. Mar Struct 2020;72:102778. [206] Deng Meng Na, Yun Hai Yu, Chen Liang, Zhao Hong Shan. Optimal maintenance
[178] Guachamin-Acero Wilson, Jiang Zhiyu, Lin Li. Numerical study of a concept for interval for wind turbine gearbox. In: Applied mechanics and materials, vol. 130.
major repair and replacement of offshore wind turbine blades. Wind Energy; Trans Tech Publ; 2012. p. 112–8.
2020. [207] Wang Xueru, Jin Zhou, Guo Peng. Wind turbine gearbox forecast using Gaussian
[179] Chou Jui-Sheng, Chiu Chien-Kuo, Huang I-Kui, Chi Kai-Ning. Failure analysis of process model. In: Proceedings of the 26th China control and decision conference;
wind turbine blade under critical wind loads. Eng Fail Anal 2013;27:99–118. 2014.
[180] Verma Amrit Shankar, Jiang Zhiyu, Ren Zhengru, Hu Weifei, Julie JE Teuwen. [208] Wang Yue, Infield David. Supervisory control and data acquisition data-based
Effects of onshore and offshore environmental parameters on the leading edge non-linear state estimation technique for wind turbine gearbox condition
erosion of wind turbine blades: a comparative study. J Offshore Mech Arctic Eng monitoring. Renew Power Gen Iet 2013;7(4):350–8.
2021;143(4). [209] Wu Xin, Wang Hong, Jiang Guoqian, Xie Ping, Li Xiaoli. Monitoring wind turbine
[181] Verma Amrit S, Castro Saullo GP, Jiang Zhiyu, Hu Weifei, Julie JE Teuwen. gearbox with echo state network modeling and dynamic threshold using scada
Leading edge erosion of wind turbine blades: effects of blade surface curvature on vibration data. Energies 2019;12(6):982.
rain droplet impingement kinematics. In: Journal of physics: conference series, [210] Wang Shuaishuai, Amir R Nejad, Moan Torgeir. On design, modelling, and
vol. 1618. IOP Publishing; 2020, 052003. analysis of a 10-mw medium-speed drivetrain for offshore wind turbines. Wind
[182] Chou Jui-Sheng, Tu Wan-Ting. Failure analysis and risk management of a Energy; 2020. p. 1099–117.
collapsed large wind turbine tower. Eng Fail Anal 2011;18(1):295–313. [211] Joel Igba, Alemzadeh Kazem, Henningsen Keld, Durugbo Christopher. Effect of
[183] Macdonald Hamish. The infuence of hail on wind turbine blade leading edge preventive maintenance intervals on reliability and maintenance costs of wind
erosion and damage. PhD thesis. University of Strathclyde; 2017. turbine gearboxes. Wind Energy 2015;18(11):2013–24.
[184] Jüngert Anne. Damage detection in wind turbine blades using two different [212] Jeremías Moragues Pons. Practical experiments on the efficiency of the remote
acoustic techniques. In: The NDT database & journal (NDT); 2008. presence, remote inspection on an offshore wind turbine. Institutt for Teknisk
[185] Sarrafi Aral, Zhu Mao, Niezrecki Christopher, Poozesh Peyman. Vibration-based Kybernetikk; 2012.
damage detection in wind turbine blades using phase-based motion estimation [213] Netland Øyvind. Remote inspection of offshore wind turbines: a study of the
and motion magnification. J Sound Vib 2018;421:300–18. benefits, usability and feasibility. Technical Cybernetics; 2014.
[186] Waldron Kenneth J, P Sattar Tariq, Rodriguez Hernando Leon, Bridge Bryan. [214] Echavarria Erika, Tomiyama Tetsuo, Van Bussel Gerard JW. Fault diagnosis
Climbing ring robot for inspection of offshore wind turbines. Ind Robot: Int J approach based on a model-based reasoner and a functional designer for a wind
2009. turbine. an approach towards self-maintenance. J Phys Conf 2007;75(1):012078.
[187] Wang Yanfeng, Liang Ming, Xiang Jiawei. Damage detection method for wind IOP.
turbine blades based on dynamics analysis and mode shape difference curvature [215] Leung Dennis YC, Yuan Yang. Wind energy development and its environmental
information. Mech Syst Signal Process 2014;48(1–2):351–67. impact: a review. Renew Sustain Energy Rev 2012;16(1):1031–9.
[188] Verma Amrit Shankar, Castro Saullo GP, Jiang Zhiyu, Julie JE Teuwen. Numerical [216] Schleisner Lotte. Life cycle assessment of a wind farm and related externalities.
investigation of rain droplet impact on offshore wind turbine blades under Renew Energy 2000;20(3):279–88.
different rainfall conditions: a parametric study. Composite Structures; 2020. [217] Wang Yuxuan, Sun Tianye. Life cycle assessment of co2 emissions from wind
112096. power plants: methodology and case studies. Renew Energy 2012;43:30–6.
[189] Brahmbhatt Nisarg, Patel Mrunal, Deb Dipankar. Micro-controller driven wind [218] Yang Juhua, Chang Yuan, Zhang Lixiao, Yan Hao, Qin Yan, Wang Changbo. The
turbine blade cleaning peripheries. In: 2017 international conference on advances life-cycle energy and environmental emissions of a typical offshore wind farm in
in computing, communications and informatics (ICACCI). IEEE; 2017. p. 847–51. China. J Clean Prod 2018;180:316–24.
[190] Sun Lim, Park Chang-Woo, Hwang Jung-Hoon, Kim Dong-Yeop, Kim Tae-Keun. [219] Xue Bing, Ma Zhixiao, Geng Yong, Heck Peter, Ren Wanxia, Tobias Mario,
The inchworm type blade inspection robot system. In: 2012 9th international Maas Achim, Jiang Ping, Jose A, de Oliveira Puppim, Fujita Tsuyoshi. A life cycle
conference on ubiquitous robots and ambient intelligence (URAI). IEEE; 2012. co-benefits assessment of wind power in China. Renew Sustain Energy Rev 2015;
p. 604–7. 41:338–46. ISSN 1364-0321.
[191] Wang Long, Zhang Zijun. Automatic detection of wind turbine blade surface [220] Wang Shifeng, Wang Sicong, Liu Jinxiang. Life-cycle green-house gas emissions of
cracks based on uav-taken images. IEEE Trans Ind Electron 2017;64(9):7293–303. onshore and offshore wind turbines. J Clean Prod 2019;210:804–10.
[192] Deb Dipankar, Patel Mrunal, Singh Himmat. Automated cleaning of wind turbine [221] Reimers Britta, Özdirik Burcu, Kaltschmitt Martin. Greenhouse gas emissions
blades with no downtime. In: 2017 IEEE international conference on industrial from electricity generated by offshore wind farms. Renew Energy 2014;72(4):
technology (ICIT). IEEE; 2017. p. 394–9. 428–38.
[193] 3m wind blade protection coating w4600 technical data sheet and application [222] Anders Arvesen, Hertwich Edgar G. Assessing the life cycle environmental
guide. https://multimedia.3m. impacts of wind power: a review of present knowledge and research needs. Renew
com/mws/media/978868O/3m-wind-blade-coating-w4600-app-guide-and-tech Sustain Energy Rev 2012;16(8):5994–6006.
nical data.pdf; 2014. [223] Frank Thomsen, Lüdemann Karin, Kafemann Rudolf, Piper Werner. Effects of
[194] Drone application. https://terra-drone.eu/en/drone-inspection/drone-win offshore wind farm noise on marine mammals and fish. Technical report, Biola,
d-turbine-blade inspection/; 2019. Hamburg. Hamburg, Germany: Germany on behalf of COWRIE Ltd; 2006. p. 7.
[195] Robotic blade care systems-wtg maintenance using robotic technology to reduce [224] Henderson Andrew R, Morgan Colin, Smith Bernie, Sørensen Hans C,
downtime and increase aep. https://www.aerones.com; 2019. Barthelmie Rebecca J, Boesmans Bart. Offshore wind energy in europe - a review
of the state-of-the-art. Wind Energy 2010;6(1):35–52.
21
[225] Dai Kaoshan, Anthony Bergot, Liang Chao, Xiang Wei Ning, Huang Zhenhua. [230] Bai X, Avital EJ, Munjiza A, Williams JJR. Numerical simulation of a marine
Environmental issues associated with wind energy - a review. Renew Energy current turbine in free surface flow. Renew Energy 2014;63:715–23.
2015;75:911–21. [231] Ghasemian Masoud, Ashrafi Z Najafian, Ahmad Sedaghat. A review on
[226] Fox AD, Desholm Mark, Kahlert Johnny, Christensen Thomas Kjaer, Petersen Ib computational fluid dynamic simulation techniques for darrieus vertical axis
Krag. Information needs to support environmental impact assessment of the wind turbines. Energy Convers Manag 2017;149:87–100.
effects of european marine offshore wind farms on birds. Ibis 2006;148(s1): [232] Tang Ye, Shi Wei, Ning Dezhi, You Jikun, Michailides Constantine. Effects of
129–44. spilling and plunging type breaking waves acting on large monopile offshore wind
[227] Liu Pu, Claire Y Barlow. Wind turbine blade waste in 2050. Waste Manag 2017; turbines. Front Mar Sci 2020;7:427.
62:229–40. [233] Ye Li, Lence Babara J, Calisal Sander M. An integrated model for estimating
[228] Jonas Pagh Jensen, Skelton Kristen. Wind turbine blade recycling: experiences, energy cost of a tidal current turbine farm. Energy Convers Manag 2011;52(9):
challenges and possibilities in a circular economy. Renew Sustain Energy Rev 1677–87.
2018;97:165–76. [234] Guanche R, De Andrés A, Losada IJ, Vidal C. A global analysis of the operation
[229] Chen Junlei, Wang Jihui, Ni Aiqing. Recycling and reuse of composite materials and maintenance role on the placing of wave energy farms. Energy Convers
for wind turbine blades: an overview. J Reinforc Plast Compos 2019;38(12): Manag 2015;106:440–56.
567–77.
22

1 s2.0 S1364032121001805 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S1364032121001805 Main

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

1 s2.0 S1364032121001805 Main

Uploaded by

Copyright:

Available Formats

Renewable and Sustainable Energy Reviews 144 (2021) 110886

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

Offshore wind turbine operations and maintenance: A

long-term operation and aging, and others are caused by short-term

• Rotor and blade: deterioration, adjustment error, rotor imbalance,

• From green to red: the wind turbine stops due to a failure;

Details of these classifications are explained and discussed in sections

2.1. OWT failure modes

Fig. 4. Development of maintenance strategies for an offshore wind farm.

Ensuring system reliability and minimizing the maintenance LCOE

Trigger Failure Planned date Real-time measurement Real-time measurement

3.1. Optimization problem 3.1.1. Cost function

Cost Downtime cost ⨯ ✓ ✓ ✓ ✓ ✓ ⨯ ✓ ✓ ✓ ✓ – ✓ ⨯

4.2.3. Lifting operation

Fig. 9. Blade root impact scenarios [175].

4.3.2. Gearbox 5. Environmental issues

You might also like