energies

Article
Maintenance Optimization of Offshore Wind
Turbines Based on an Opportunistic
Maintenance Strategy
Lubing Xie *, Xiaoming Rui, Shuai Li and Xin Hu
School of Energy Power and Mechanical Engineering, North China Electric Power University,
Beijing 102206, China
* Correspondence: [email protected]




Received: 19 May 2019; Accepted: 8 July 2019; Published: 10 July 2019


Abstract: Owing to the late development of offshore wind power in China, operational data and
maintenance experience are relatively scarce. Due to the harsh environmental conditions, a reliability
analysis based on limited sample fault data has been regarded as an effective way to investigate
maintenance optimization for offshore wind farms. The chief aim of the present work is to develop
an effective strategy to reduce the maintenance costs of offshore wind turbines in consideration
of their accessibility. The three-parameter Weibull distribution method was applied to failure rate
estimation based on limited data. Moreover, considering the impacts of weather conditions on the
marine maintenance activities, the Markov method and dynamic time window were used to depict
the weather conditions. The opportunistic maintenance strategy was introduced to cut down on
the maintenance costs through optimization of the preventive maintenance age and opportunistic
maintenance age. The simulation analysis we have performed showed that the maintenance costs of
the opportunistic maintenance strategy were 10% lower than those of the preventive maintenance
strategy, verifying the effectiveness of the proposed maintenance strategy.

Keywords: offshore wind turbine; reliability; accessibility; three-parameter Weibull distribution;

opportunistic maintenance

1. Introduction
In recent years, the wind power industry has made rapid progress globally. Due to abundant
offshore wind energy resources, the installed capacity and power generation of offshore wind power
are growing rapidly [1]. However, considering the large investment and high risk involved in offshore
wind power, and the influence of weather conditions, the requirements for offshore wind farms in
terms of operation and maintenance vessels, spare-parts management, and other influential factors are
more strict. The special operating environment of offshore wind turbines, in which the equipment
is affected by natural conditions, such as typhoons, tides, waves, can accelerate the failure of unit
components, and the increased failure rate of electrical and mechanical systems finally leads to lower
reliability. Thus, the reliability of wind turbines has become an important issue.
The Weibull distribution is an important statistical measure in reliability engineering, known to
have good adaptability and availability for various forms of failure rate simulation in relation to
mechanical and electrical products. In this paper, the construction of a Weibull equation is pursued.
Based on the two-parameter Weibull equation, a location parameter was introduced to create what is
known as a three-parameter Weibull equation. Many researchers are interested in three-parameter
Weibull equations. In [2], a three-parameter Weibull distribution was used in the asphalt-concrete
fatigue assessment of aging. In [3], optimization of distribution was studied for an inventory system

Energies 2019, 12, 2650; doi:10.3390/en12142650 www.mdpi.com/journal/energies

Energies 2019, 12, 2650 2 of 26

established by Weibull. In [4], a three-parameter Weibull distribution was proposed in a study on the
mechanical sector. In [5], the three-parameter Weibull distribution equation was applied to predict
wind power density. In [6], a mixed Weibull distribution parameter estimation method was discussed.
In [7], the Weibull distribution was studied by using the three best linear unbiased parameter estimation
based on previous unbiased estimation and quasi-optimal linearity. This literature review reveals that
the three-parameter Weibull equation has been used in many fields, including machinery, architecture,
and aerospace; however, so far few applications of the three-parameter Weibull distribution have been
used in fault prediction for offshore wind turbines, especially for those with limited fault data.
Moreover, access to offshore wind farms is greatly restricted by wind and wave conditions,
which hinders the operation and maintenance of the unit, thus resulting in an increase in outage time,
a decrease in the availability of units, and an increase in operation and maintenance costs. It is necessary
to consider the impacts of sea weather conditions related to unit operation and maintenance. The mpact
of weather conditions on wind turbine maintenance strategies has been investigated in previous
works. In China, in [8], a wind turbine reliability evaluation method was proposed. Considering the
special operating environment of offshore wind turbines, in [9], the impacts of weather, accessibility,
and maintenance time on the maintenance strategy of offshore wind turbines were analyzed. In [10],
an offshore wind turbine maintenance strategy with minimum cost and maximum reliability was
established, considering the influence of wind speed and maintenance waiting time. In [11,12], a normal
behavior model was proposed to further investigate wind turbine vibration and fatigue load by using
neural networks and a stochastic approach considering the influence of wind speed. In [13], a survey
of stochastic models for sea state and different offshore wind farms was made according to the different
maintenance schemes of offshore wind turbines, combined with the failure rate and maintenance time
for various components of the unit. In [14], a novel method for simulating wind and wave conditions
for offshore sites was revealed, and the result indicated that the persistence of weather windows for
significant wave height values can be captured by this approach. All these tests show promising results,
and the weather model has been deemed accurate enough for simulations of offshore wind parks.
Moreover, in [15], a Markov model based on a statistical analysis of wind speed and wave height was
established, and finally, the predictions were made for maintenance waiting time. Although there have
been some related studies on the effect of weather conditions on offshore wind turbine maintenance
strategies, wind speed and wave height have not yet been simultaneously considered to evaluate the
accessibility of offshore wind turbine maintenance.
Meanwhile, many researchers have theoretically investigated the opportunistic maintenance
strategy from a different perspective. In [16], a two-level maintenance threshold strategy for wind farms
was developed, considering opportunistic maintenance and imperfect maintenance based on reliability.
In [17], opportunistic condition-based maintenance for systems subjected to degradation and shocks was
proposed to determine an optimal maintenance policy for multi-bladed offshore wind turbines. In [18],
a new bi-objective opportunistic maintenance optimization model was proposed, and a three-phase
discrete event simulation was used to evaluate the performance measures considering the stochastic
behavior of wind and limited maintenance capacity. In [19], an opportunistic maintenance approach for
wind farms was developed to take advantage of the maintenance opportunities, considering imperfect
maintenance actions. In [20], the trade-off between wind farm configuration and the maintenance
strategy was investigated by a new bi-objective redundancy and maintenance optimization model.
The primary goal of this research is to carry out studies in an effort to reduce the operating costs of
offshore wind turbines. In this paper, the impacts of weather conditions on the maintenance activities
are considered, the Markov chain method and dynamic time window are introduced to represent
the weather conditions, and a maintenance waiting time is proposed for offshore wind turbines.
In addition, the opportunistic maintenance strategy was used to optimize the key components of
the maintenance of the offshore wind turbines. Furthermore, the minimum maintenance cost within
the maintenance duration was deemed the optimal objective, and the preventive maintenance time
and opportunistic maintenance time have been optimized for the main components of wind turbines.
Energies 2019, 12, 2650 3 of 26

In particular, the novelties can be reflected in the following aspects: (1)The three-parameter Weibull
method, instead of the two-parameter Weibull method, was applied to the reliability analysis of offshore
wind turbines with limited fault data; (2) the Markov chain method was adopted to evaluate the
prediction of maintenance waiting time, considering the wind speed and wave height simultaneously;
(3) the preventive opportunistic maintenance strategy has proven to be one of the methods available to
generate maintenance cost efficiency, instead of the preventive maintenance strategy.
The remainder of this paper is organized as follows. Section 2 introduces the two-parameter
Weibull model of onshore wind turbines by the least squares and maximum likelihood methods,
and the three-parameter Weibull model of offshore wind turbines by the correlation coefficient,
probability-weighted moment, and bilinear regression methods. In Section 3, we describe the
construction of a Weibull equation for offshore wind turbines based on scant sample fault data. Section 4
describes the Markov prediction method. In Section 5, we develop an opportunistic maintenance
strategy based on accessibility evaluation. Section 6 gives the opportunistic maintenance simulation
based on maintenance waiting time. In Section 7, we propose conclusions and recommendations from
our study and suggest areas for further research.

2. Reliability Analysis Model for Wind Turbines

For wind turbines, the commonly used reliability analysis model is the Weibull distribution.
This distribution is derived from the analysis of material strength by Weibull. It is a very important
distribution form in reliability engineering [21]. The probability density function of the Weibull model
obtained from the weakest link theory is
  β−1   β 
β t−γ t−γ
exp − α (t ≥ γ)

α α


f (t) =  (1)
(t < γ)


 0

Its distribution function is given by

R   β 
t t−γ
 0 f (t)dt = 1 − exp − α (t ≥ γ)



F(t) =  (2)
(t < γ)


 0

Its inefficiency function is given by

f (t)
λ(t) = (3)
1 − F(t)

In Equations (1) and (2), α is the scale parameter, β is the shape parameter, β > 0, and γ is the
position parameter, γ > 0. γ is the threshold value for failure, β is used to describe the dispersion of the
measured values, and α is related to the average measured values.
In the failure analysis of the product, β is associated with the failure mechanism of the product,
and the different values of β are accompanied by different fault mechanisms. When β < 1, the failure-rate
function is a decreasing function, showing the life distribution of products under the wear-out failure
period. When β = 1, the failure-rate function is constant, representing the products in the life
distribution of the random failure period. When β > 1, the failure-rate function is an increasing
function, showing the distribution of products in the period of life [22].
Energies 2019, 12, 2650 4 of 26

2.1. Weibull Model of Onshore Wind Turbines

For onshore wind turbines, because of their simple operating conditions and abundant operation
and fault data, in practical applications, the common assumption is that the device fails at time γ = 0,
and the corresponding expressions, Equations (1) and (2), respectively, can be simplified as follows:

β t β−1
    β 
f (t) = α α exp − αt (t ≥ 0) , (4)

  β 
F(t) = 1 − exp − αt (t ≥ 0) . (5)

At present, the two-parameter Weibull analysis method is roughly divided into several categories,
i.e., least-squares estimation, median rank regression, maximum-likelihood estimation, and method
of moments. For engineering purposes, maximum likelihood is usually considered first due to its
high accuracy, as compared to method of moment, which tends to be adopted only if the maximum
likelihood function will be difficult to construct. In this paper, we used maximum likelihood rather
than method of moment.
Practically, median-rank regression can perform well in the case of extreme values occurring
when the number of failures is small. Generally, based on the least squares method, the mean variable
will be replaced by the median-rank variable so as to alleviate the estimated deviation caused by the
extreme value. In this study, the maintenance optimization was based on a small sample, and the
extreme value was carved out; therefore, the least squares method, rather than median-rank regression,
was pursued, as follows.

2.1.1. Least-Squares Estimation

By the principle of extremum, the equation can be defined as follows:
 n
∂F P
= −2 xi ( yi − axi − b) = 0


∂a


i=0

(6)

 n
∂F
 P



 ∂b
= −2 ( yi − axi − b) = 0
i=0

The solution equation is shown as follows:

 n
P n
P n
P


 n xi yi − xi yi
i=0 i=0 i=0
a=


 !2

 n n
x2i −
P P
n xi (7)


 i=0 i=0
n n


1 a
 P P
 b= yi − xi


 n n
i=1 i=1

The two-logarithm transformation of Equation (5) is

!!
1
ln ln = β ln(t) − β ln(α) (8)
1 − F(t)

Comparison with the one-element linear-regression equation gives:




 x=  (t) 
 ln

 y = ln ln 1−F1(t)



 (9)
a=β





b = −β ln(α)



Parameters α and β can be obtained from Equation (7).

Energies 2019, 12, 2650 5 of 26

2.1.2. Maximum-Likelihood Estimation

Maximum-likelihood estimation is an effective method of parameter estimation. It has strong
applicability, especially in the case of incomplete life. The basic idea is to select the undetermined
parameters first so that the probability of samples appearing in the field of observation is greatest;
then the sample value is taken as the unknown parameter’s point estimated value [23].
The likelihood function is
n
β  ti β−1
"  β #
Y t
L(t1 , t2 , . . . , tn , α, β) = exp − i (10)
α α α
i=1

For the convenience of calculation, the logarithm of Equation (10) is taken as a natural logarithm:

n  β

n t  P ti

 ln L(t1 , t2 , . . . , tn , α, β) = n(ln β − ln α) + (β − 1) ln α −

 P i
α

 i = 1 i = 1 (11)
∂ ∂
 ln L ln L
= ∂β = 0


∂α


Then, the likelihood equation is given by


β Pn  β nβ
 ti
α − α =0

α


i=1

. (12)

 n t  P n  t β  t 
n
 P
β + ln αi − ln αi ln αi =0




i=1 i=1

Solving the equation gives the parameters α and β.

2.2. Weibull Model of Offshore Wind Turbines

The three-parameter Weibull distribution is used in the life tests of materials with a low stress
level, which have better characteristics of curve fitting as compared to the two-parameter Weibull
distribution. In the Weibull distribution function, the positional parameter is the threshold value
for the failure time of targeted devices and is used to estimate the earliest failure time. Changes in
positional parameters will affect the curve displacement of the probability density curve. Generally,
when the unit operation time exceeds the positional parameter, the unit starts to fail.
In contrast with onshore wind farms, the conditions of operation and maintenance of offshore
wind farms are extremely complicated due to tidewater, typhoons, and corrosion, which result in
an increasing probability of the unit components’ failure as well as increased maintenance costs.
In addition, due to the particularity of the environment, when a unit failure occurs at an offshore
wind farm, maintenance personnel may not be able to reach the site for several months due to harsh
weather conditions, which results in difficulty in obtaining component failure data and failure models.
Therefore, it is crucial to carry out reliability prediction as soon as possible.
As part of our research, we performed a comparative analysis of the two-parameter Weibull
distribution and the three-parameter Weibull distribution. The Weibull three-parameter equation
prevails in short-term failure prediction and can be used effectively before t = γ. With regard to
short-term failure rate prediction, an abrupt and abnormal rise in the failure rate has taken place
under the two-parameter Weibull distribution. On the contrary, the failure rate curve under the
three-parameter Weibull distribution displays a rational and smooth rising trend.
However, three parameters exist at the same time, so estimating the parameters of the
three-parameter Weibull distribution is difficult. The least-squares method and the maximum-likelihood
method are no longer applicable to the three-parameter Weibull distribution. MATLAB is a mathematics
software produced by MathWorks. It is a high-level technical computing language and interactive
environment for algorithm development, data visualization, data analysis, and numerical calculation,
Energies 2019, 12, 2650 6 of 26

mainly including MATLAB and Simulink. The paper applied MATLAB software for data calculation
and simulation.

2.2.1. Correlation Coefficient Method

The two-logarithm transformations of Equation (2) are obtained by
!!
1
ln ln = β ln(t − γ) − β ln(α). (13)
1 − F(t)

Equation (13) is changed into

a=β




b = −β ln(α)



(14)

X = ln(t − γ)




Y = β ln(t − γ) − β ln(α) = aX + b




The relationship between X and Y is a linear relationship, as shown in Equation (14). Equation (14)
shows that when the estimate of γ is correct, a linear relationship exists between X and Y, i.e.,
the correlation coefficient between X and Y is a maximum. The correlation coefficient r between X and
Y is obtained by
Cov(X, Y)
r= p (15)
Var[X]Var[Y]
According to the flowchart shown in Figure 1, column vectors containing (k − 1),
correlation coefficients are obtained. The step size p can be adjusted according to the actual situation
and is selected as 0.01 in the paper. The correlation coefficient r initially increases and then decreases.
The position of the corresponding maximum correlation coefficient is found as N, and the optimal
estimation value of the position parameter is recorded as

ti = ti − (N2 − N )p/2
(
(16)
γ̂ = (N2 − N )p/2

After obtaining the estimated value of the position parameter, we can obtain the estimated value
of the parameters
Energies byPEER
2019, 12, x FOR least-squares
REVIEW or maximum-likelihood estimation. 7 of 27

Figure 1. Correlation
Figure 1. Correlation coefficient
coefficient method.
method.

2.2.2. Probability-Weighted Moment Method

For small samples, probability-weighted moments are unbiased and have small errors; this is
suitable for the reliability assessment of offshore wind turbines with a small amount of sample data.
Therefore, the probability-weighted moments of the sample are used to estimate the distribution
parameters, and the accuracy is high.
The inverse probability distribution function for the random variables X of the Weibull
three-parameter distribution, which is shown in Equation (17), is obtained using Equations (1) and
Energies 2019, 12, 2650 7 of 26

2.2.2. Probability-Weighted Moment Method

For small samples, probability-weighted moments are unbiased and have small errors; this is
suitable for the reliability assessment of offshore wind turbines with a small amount of sample data.
Therefore, the probability-weighted moments of the sample are used to estimate the distribution
parameters, and the accuracy is high.
The inverse probability distribution function for the random variables X of the Weibull three-parameter
distribution, which is shown in Equation (17), is obtained using Equations (1) and (2).

x = γ + α[− ln(1 − F)]1/β (17)

From this equation, the following can be obtained:

γ αΓ(1 + 1/β)
M1,0,k = + (18)
1 + k (1 + k)(1+1/β)

Let M1,0,k = M(k) ; then

M(0) = γ + αΓ(1 + 1/β)




 M = γ + αΓ(1+1/β)



 (1) 2 2(1+1/β)
(19)
 M(3) = γ4 + αΓ((11++1/β
1/β)



4 )

The equation is solved, and the estimated values of parameters α, β, γ can be acquired.
 M(0) −γ

 α̂ =   
  (0) −2M(1) ) 
(

M

  
Γ
  
ln /ln(2)

 


   (M(1) −2M(3) ) 
 


ln(2)

 β̂ =


  (M(0) −2M(1) ) 
 
(20)
ln

 



 ( 2M(1) −4M(3) )  

4 M(3) M(0) −M2(1)


 γ̂ =



(4M(3) +M(0) −4M(1) )

2.2.3. Bilinear Regression Method

Least-squares estimation is a linear-regression analysis method for a linear equation. It can only
be used to solve two-parameter estimation problems. For the estimation of three parameters, two linear
equations must be combined.
Let α = t0 1/β ; then, Equation (2) can be transformed as follows:
!!
1
ln ln = β ln(t − γ) − ln(t0 ) (21)
1 − F(t)
!!1/β
1 t γ
ln = − (22)
1 − F(t) α α
Equations (21) and (22) are linearly independent; they are linear equations that can be expressed
as yk = ak xk + bk , k = 1, 2 and xi , yi can be depicted, respectively, as
   
y1 = ln ln 1−F1(t) a1 = β





x1 = ln(t − γ)

b1 = − ln(t0 )



(23)
   1/β

y2 = ln 1−F1(t) 1

a2 =

α



γ


x2 = t b2 = − α


Energies 2019, 12, 2650 8 of 26

According to the least-squares method, Equations (21) and (22) can be substituted into Equation (7)
separately, and the combination can be obtained:
    1/β
2
t2 −(t) ln 1−F1(t)


γ = t−  





 1/β   1/β
t ln 1−F1(t) −t· ln 1−F1(t)




!#1/β

 "
(24)
  
α = exp β · ln(t − γ) − ln ln 1−F(t) 1







      
ln(t−γ)·ln ln 1−F1(t) ln(t−γ)·ln ln 1−F1(t)



 β=

2 −


2 2 2
(ln(t−γ)) −(ln(t−γ)) (ln(t−γ)) −(ln(t−γ))

As long as the accuracy is given, the values of β and γ can be estimated iteratively, and Equation (24)
is substituted to obtain the estimated value of α.

3. Analysis of Results

3.1. Verification of Reliability Analysis Method for Wind Turbines

Taking the failure shutdown data of a 3 MW wind turbine at an offshore wind farm in Jiangsu
province in China as an example, the method of parameter estimation is verified in this work. The wind
farm has 19 wind turbine units installed, each operating at 3 MW, and first operated in June 2017.
Typical fault data were collected from the Supervisory Control and Data Acquisition system (SCADA)
and selected from July 2017 to July 2018, as shown in Table 1.

Table 1. Failure times.

Sequence 1 2 3 4 5 6 7 8 9 10 11 12 13
Failure time (h) 4560 4568 4660 4756 4879 4899 4904 4967 4988 4990 4995 4996 4998
Sequence 14 15 16 17 18 19 20 21 22 23 24 25 26
Failure time (h) 4999 5034 5096 5137 5138 5278 5289 5367 5467 5567 5678 5778 5879
Sequence 27 28 29 30 31 32 33 34 35 36 37 38 39
Failure time (h) 5889 5900 6067 6078 6178 6278 6378 6478 6578 6678 6789 6879 6888
Sequence 40 41 42 43 44 45 46 47 48 49 50 51 52
Failure time (h) 6978 6988 6990 7089 7115 7245 7345 7356 7456 7468 7568 7569 7709
Sequence 53 54 55 56
Failure time (h) 7809 7889 7900 7908

The method of Weibull parameter estimation proposed in this paper is used to estimate the
parameters and failure-rate function of the fault data, as shown in Table 2. The failure-rate curves of
two- and three-parameter Weibull distribution are shown in Figure 2.

Table 2. Parameters estimated.

Weibull Distribution Parameters

Method Failure-Rate Functions
α β γ
Least-squares estimation 4.9979 × 103 18.012 - 4.7388 × 10−66 t17.023
Maximum-likelihood
4.9941 × 103 17.963 - 7.2376 × 10−66 t16.976
estimation
Correlation coefficient method 4.298 × 102 1.487 4462 1.8379 × 10−4 (t − 4457)0.484
Probability-weighted moment 4.069 × 102 1.383 4472 3.2298 × 10−4 (t − 4486)0.393
Bilinear regression method 3.859 × 102 1.144 4513 1.4007 × 10−3 (t − 4543)0.124
 method
Probability-weighted
4.069 × 102 1.383 4472 3.2298 × 10−4 (t − 4486)0.393
moment
Bilinear regression
3.859 × 102 1.144 4513 1.4007 × 10−3 (t − 4543)0.124
method
Energies 2019, 12, 2650 9 of 26

(a)

(b)

Figure 2. Failure-rate
Figure curves
2. Failure-rate of Weibull
curves distributions.
of Weibull (a) Comparison
distributions. of failure
(a) Comparison raterate
of failure for three- and and
for three-
two-parameter Weibull distributions. (b) Partial enlarged drawing of maximum likelihood
two-parameter Weibull distributions. (b) Partial enlarged drawing of maximum likelihood estimation
and estimation
least-squares
andestimation.
least-squares estimation.

According to the failure-rate function estimates given in Table 2, the reliability of the wind turbine
According to the failure-rate function estimates given in Table 2, the reliability of the wind
is predicted from the last fault time, as shown in Table 3.
turbine is predicted from the last fault time, as shown in Table 3.

Table 3. Short period prediction on failure rates.

 Two-Parameter Three-Parameter
Probability-W
Failure Correlation
Least-Squares Maximum-Likelih eighted
Rate Coefficient
Estimation ood Estimation Moment
Method
Energies 2019, 12, 2650 Method 10 of 26
λ (5378) 0.0129 0.0131 0.0050 0.0047
λ (8760) 51.7726 51.1463 0.0105 0.0086
Table 3. Short period prediction on failure rates.

Comparative analyses are given as follows:

Two-Parameter Three-Parameter
Correlation
1) As shown
Failure Rate in Table 2, for the two-parameter
Least-Squares Weibull
Maximum-Likelihood distribution, Probability-Weighted
Coefficient
the least-squares, and
maximum-likelihood Estimation Estimation
methods are suitable to obtain similar Moment Method
Methodresults. For the three-parameter
Weibull
λ(5378) distribution,0.0129
the correlation 0.0131
coefficient method, bilinear regression
0.0050 method, and
0.0047
λ
probability-weighted moment method produce similar results.
( 8760 ) 51.7726 51.1463 0.0105 0.0086

2) As shown in Figure 2, given the two-parameter Weibull failure-rate curve from the beginning (t
Comparative analyses are given
= 0), and the three-parameter as follows:
Weibull distribution of the failure-rate curve from the beginning (t
1) =Asγ),shown
the Weibull three-parameter
in Table 2, for the equation
two-parameter prevailsWeibull
for short-term failure the
distribution, prediction of key
least-squares,
and maximum-likelihood
defects methods wind
occurring in newly operated are turbines
suitable and
to obtain
can be used similar results. avoid
to effectively For the
the
three-parameter Weibull distribution, the correlation coefficient method, bilinear regression
impact of short-term reliability prediction of atypical failure data before t = γ.
method, and probability-weighted moment method produce similar results.
2) As shown Analysis
3.2. Simulation in Figureof2,Maintenance
given the two-parameter Turbinesfailure-rate curve from the beginning (t =
Costs of Wind Weibull
0), and the three-parameter Weibull distribution of the failure-rate curve from the beginning (t =
As shown
γ), the in Figure
Weibull 3, according
three-parameter to theprevails
equation wind turbine preventive
for short-term maintenance
failure prediction ofstrategy, the
key defects
maintenance
occurring cost of component
in newly operated i iswind turbines and can be used to effectively avoid the impact of
( i ) t = γ.
( )+C ( ) + C
short-term reliability prediction of atypical ( i ) failure data before
Cic Fi Tp 1 − Fi Tp
( ) 
ip N
(i )
Ci Tp = Costs (of
3.2. Simulation Analysis of Maintenance . (25)
Tp ) Wind Turbines
i
∞
(i )
 tf ( t ) d t + Tp 
(i)
f (t ) d t
As shown in Figure 3, according to0 the wind turbineTppreventive maintenance strategy, the
maintenance cost of component i is
In this equation, Cic is the failure maintenance cost of component i ; Cip is the preventive
( i ) F T (i) + C 1 − F T (i) + C
  h  i
maintenance cost of component

(i)
i ; TC
p ic iis the p preventive
ip maintenance
i p Nservice life of component i ;
Ci Tp = R (i) . (25)
( )
Tp( i ) ∞
R
C N is the fixed maintenance cost; Ci T0 p t fis (t)the (
dt +expectedi )
Tp T (i)cost rate of long-term operation of
f (t)dt
p

component i (Chinese ¥/day).

Figure 3. Preventive
Figure 3. Preventive maintenance
maintenance strategy
strategy based
based on
on age.
age.

In thiskey
The equation, Cic is the
components of failure maintenance
the wind costsubject
turbine are of component i; Cip distribution;
to Weibull is the preventive
the maintenance
distribution
cost of component i; T (i) is the preventive maintenance service life of component i; C is the fixed
parameters are shown p
 in Table 4. The maintenance costs are shown in Table 5. N
maintenance cost; Ci Tp (i) is the expected cost rate of long-term operation of component i (Chinese ¥/day).
The key components of the wind turbine are subject to Weibull distribution; the distribution
parameters are shown in Table 4. The maintenance costs are shown in Table 5.
Energies 2019, 12, 2650 11 of 26
Energies 2019, 12, x FOR PEER REVIEW 12 of 27
Energies 2019, 12, x FOR PEER REVIEW 12 of 27
Table 4. Weibull distribution parameters for major components of wind turbines.
Table 4. Weibull distribution parameters for major components of wind turbines.
Table 4. Weibull distribution parameters for major components of wind turbines.
Component
Component
αα/Day
/day
β β
Component α /day β
Blade Blade 3000
3000 2 2
Blade Gearbox 3000
2400 3 2
Gearbox 2400 3
Gearbox Generator 2400
3300 3 3
Generator 3300 3
Generator 3300 3
Table 5.
Table Maintenancecosts
5. Maintenance costsfor
forthe
themajor
majorcomponents
componentsof of wind
wind turbines.
turbines.
Table 5. Maintenance costs for the major components of wind turbines.
Component
Component C/¥
Cic /¥ Cip /¥Cip /¥CN /¥ C /¥
Component ic
Cic /¥ Cip /¥ CNN /¥
Blade Blade 112,000
112,000 28,00028,000
Blade 112,000 28,000
Gearbox
Gearbox 152,000
152,000 38,000 38,00035,000 35,000
Gearbox
Generator 152,000
100,000 25,00038,000 35,000
Generator 100,000 25,000
Generator 100,000 25,000

Usingthe
Using thetwo-
two-and
andthree-parameter
three-parameterWeibull
Weibull reliability
reliability analysis
analysismethods,
methods,the
themaintenance
maintenancecosts
costs
Using the two- and three-parameter Weibull reliability analysis methods, the maintenance costs
areseen
are seen to
to vary
vary over
over time,
time, as
as illustrated
illustrated in
in Figures
Figures 4‒6.
4–6.
are seen to vary over time, as illustrated in Figures 4‒6.

(a) (b)
(a) (b)
Figure 4. The cost rate of blade maintenance based on two- and three-parameter Weibull: the
Figure 4.4.The
Thecost
cost rate
rate of blade
of blade maintenance
maintenance basedbased
on two-onand
two- and three-parameter
three-parameter Weibull: Weibull: the
the left-hand
left-hand side (a) shows the blade maintenance cost rate of the two-parameter Weibull, and the
left-hand side (a) shows the blade maintenance cost rate of the two-parameter Weibull, and
side (a) shows the blade maintenance cost rate of the two-parameter Weibull, and the right-hand side the
right-hand side (b) shows the three-parameter Weibull.
right-hand
(b) side
shows the (b) shows the three-parameter
three-parameter Weibull. Weibull.

(a) (b)
(a) (b)
Figure
Figure 5. The cost
5. The costrate
rate
of of gearbox
gearbox maintenance
maintenance basedbased on and
on two- two- and three-parameter
three-parameter Weibull:Weibull: the
the left-hand
Figure 5. The cost rate of gearbox maintenance based on two- and three-parameter Weibull: the
left-hand side (a)
side (a) shows theshows
gearbox themaintenance
gearbox maintenance cost
cost rate of therate of the two-parameter
two-parameter Weibull,
Weibull, and and the
the right-hand
left-hand side (a) shows the gearbox maintenance cost rate of the two-parameter Weibull, and the
right-hand side the
side (b) shows (b) three-parameter
shows the three-parameter
Weibull. Weibull.
right-hand side (b) shows the three-parameter Weibull.
Energies 2019,12,
Energies2019, 12,2650
x FOR PEER REVIEW 12
13ofof26
27

(a) (b)
Figure6.6.The
Figure Thecost
costrate
rateof
ofgenerator
generatormaintenance
maintenancebased
basedonontwo-
two-and
andthree-parameter
three-parameterWeibull,
Weibull,thetheleft
left
figure(a)
figure (a)shows
shows
thethe generator
generator maintenance
maintenance costcost
rate rate oftwo-parameter
of the the two-parameter Weibull,
Weibull, and theand thefigure
right right
figure
(b) (b)the
shows shows the three-parameter
three-parameter Weibull.Weibull.

Using
Usingthe
thetwo- and and
two- three-parameter Weibull
three-parameter reliability
Weibull analysis methods,
reliability the optimalthe
analysis methods, preventive
optimal
maintenance time and maintenance
preventive maintenance cost of key components
time and maintenance of the wind turbine
cost of key components can be
of the wind determined;
turbine can be
these are shown
determined; in Table
these 6.
are shown in Table 6.

Table6.6.Optimal
Table Optimalmaintenance
maintenancetime
timeand
andmaintenance
maintenancecost
costofofunit
unitcomponents.
components.
Two-Parameter
Two-Parameter Three-Parameter
Three-Parameter
Component
Component (i )
Tp /day ( ) (i )
TpTp/Day (T )
( )
    (i)
T(i)
p
C
/Day
i T C i p Tp
(i)
i
(i) /day Ci Tp(i) Ci p

Blade 2770
Blade 51.7344 51.7344
2770 2490
2490 49.8622 49.8622
Gearbox 1664
Gearbox 68.4917 68.4917
1664 1505
1505 63.5657 63.5657
Generator 2470
Generator 38.2659 38.2659
2470 2284
2284 31.0392 31.0392

It can be seen from Table 6 that the maintenance cost of the unit can be reduced by 8% using the
It can be seen from Table 6 that the maintenance cost of the unit can be reduced by 8% using the
reliability analysis method of the three-parameter Weibull model.
reliability analysis method of the three-parameter Weibull model.

4.4.Markov
MarkovPrediction
PredictionMethod
Method
Thismethod
This methoduses usesaamathematical
mathematicalmodel
modelto toanalyze
analyzethe
theevolution
evolutionof
ofobjects
objects[24].
[24].In
Inthe
theMarkov
Markov
process, the time series is regarded as a stochastic process. By studying the initial probability
process, the time series is regarded as a stochastic process. By studying the initial probability of of
different states of things and the state transition matrix, the state change trend is determined,
different states of things and the state transition matrix, the state change trend is determined, and the and
the future
future state state of things
of things can becan be predicted.
predicted. This method
This method has
has been been used
widely widely used in communications,
in communications, biology,
biology, the social sciences, and other
the social sciences, and other fields [25]. fields [25].

4.1.Markov
4.1. MarkovTheory
Theory

4.1.1.
4.1.1.Markov
MarkovProcess
Process
process X{(Xt)(,t ), t ∈TT }satisfies

IfIfthe
therandom
randomprocess t∈ thethe
satisfies following condition
following [26]:
condition [26]:
(1) If
(1) If the
the state
state of
of the
the stochastic
stochastic process
process X { X(t()t,),tt∈∈TT } atattime
timet tisisknown,
known,and andthe thestate
stateatattime
timet t++11isis
only related
only related to
to the
the state
state at
at time
time tt and and independent
independentof ofthe
thestate
statebefore
beforetime timet,t,then
thenititisisconsidered
considered
that the
the stochastic
stochastic process
process X { X(t()t,),t t∈∈TT }hashasMarkov
Markovproperty.
property.

that
process {XX(t()t ), T} isis S,S, ifif for

(2)
(2) When
When the the state
state space
space ofof the
the stochastic
stochastic process , tt ∈
∈T any nn≥≥ 2,2, and
for any and any
any
t1 < t2 < . . . < tn ∈ T in the condition X(ti ) = xi , xi ∈ S, i = 1, 2, . . . , n − 1, the conditional
t1 < t2 < ... < tn ∈ T in the condition X (ti ) = xi , xi ∈ S , i = 1, 2,..., n − 1 , the conditional
probability distribution function of X(tn ) is equal to its probability distribution function
under the condition X (tn −1 ) = xn −1 , which is
Energies 2019, 12, 2650 13 of 26

probability distribution function of X(tn ) is equal to its probability distribution function under
the condition X(tn−1 ) = xn−1 , which is

P(X(tn ) ≤ xn X(t1 ) = x1 , X(t2 ) = x2 , · · · , X(tn−1 ) = xn−1 )

(26)
= P(X(tn ) ≤ xn X(tn−1 ) = xn−1 )

4.1.2. Markov Chain

The Markov chain is a discrete-time stochastic process with Markov property, and the model is
usually expressed as λ = (S, P, Q), where the meaning of each element is as follows [27]:

1) S represents a non-empty set of all possible states in a random process. State is the result of
a random process occurring at a certain moment. P is a one-step state transition probability
matrix. Conditional probability can be expressed as follows:

P X(t + 1) = j X(t) = i = Pij (27)

2) Q is the initial state probability distribution vector, Let Q = [q1 , q2 , . . . , qn ] denote the probability
that the stochastic process is in state xi at time t = 0 is qi , then
n
X
qi = 1 (28)
i=1

4.2. Data Processing Based on Markov Method

When using the Markov chain model for predictive analysis, the random variables need to be
processed as follows:

(1) State division of random variables;

(2) Calculation of state transition probability;
(3) Test of “Markov property”.

4.2.1. State Division of Random Variables

For the state division of random variables, the following methods are possibilities [28]:

(1) Sample mean value-mean variance partition method;

(2) Ordered clustering method;
(3) Fuzzy clustering method.

In this paper, only the first method is used and will be briefly described below.
Suppose a random observation set of random variable X is x1 , x2 , . . . , xn , the sample mean is x,
and the sample mean variance is s. If the absolute value |r| of the auto-correlation coefficients of this set
is ≤0.30, it can be approximated as an independent and identically distributed sequence. According to
the central limit theorem:
( 
P x − s ≤ x ≤ x + s = 2Φ(1.0) − 1 ≈ 0.68
 (29)
P x − 1.5s ≤ x ≤ x + 1.5s = 2Φ(1.5) − 1 ≈ 0.87

4.2.2. Calculation of State Transition Probability

Before using the Markov chain for prediction, it is necessary to estimate the transition probability
of the Markov chain from the historical observation sequence.
Suppose a set of observation sequences in a random variable X is x1 , x2 , . . . , xm , which contains
n states, namely the state space S = {1, 2, . . . , n}. Use fij to indicate the frequency of the observation
Energies 2019, 12, 2650 14 of 26

sequence from state i to state j through one step, i, j ∈ S. A matrix ( fij )i,j∈S composed of fij (i, j ∈ S) is
called a transfer frequency matrix. The value obtained by dividing the element of the i-th row and the
j-th column in the transfer frequency matrix by the sum of the elements of the i-th row is called the
maximum likelihood estimation of the transition probability, P̂ij , i, j ∈ S, as follows:

fij
P̂ij = n
(30)
P
fij
j=1

4.2.3. Test of “Markov Property”

It is necessary to test whether the random variable sequence has “Markov property” before
applying the Markov chain model to solve practical problems. Usually, a discrete sequence of the
Markov chain can be tested using the χ2 statistic.
The marginal probability is denoted as P· j , as follows:

n
P
fij
i=1
P·j = n P n
(31)
P
fij
i=1 j=1

Then, when n is sufficiently large, the statistics obeys the χ2 distribution with a degree of freedom
of (n − 1)2 , where Pij is the transition probability.

n X
n
X Pij
χ2 = 2 fij log (32)
P·j
i=1 j=1

Given the significance level α, we can get the value of χ2α (n − 1)2 from the lookup table, and finally
calculate the value of the available statistic χ2α . If χ2α > χ2α (n − 1)2 , it can be considered that the random
sequences x1 , x2 , . . . , xm have “Markov property” [29].

4.3. Markov Prediction Model

The operation and maintenance of offshore wind turbines are affected by weather conditions,
such as sea wind speed and wind direction, and sea conditions, such as wave height and cycle.
Among them, wind speed and wave height are the two most important factors affecting the accessibility
of wind turbines [30]. As a consequence, in this paper, the accessibility of offshore wind turbines is
studied mainly by considering the wind speed and wave height.
Offshore wind speed and wave height are characterized by time variation and randomness.
Commonly used research methods include Gaussian statistics, the Markov method, and the
auto-regressive moving average method. The Markov method was used to build the model
since the method can accurately describe the long-term and seasonal wind speed and wave
height distribution [31].
There are three main methods for Markov chain prediction [32]:

(1) Markov chain prediction based on absolute distribution;

(2) Superimposed Markov chain prediction;
(3) Weighted Markov chain prediction.

In the case of unrestricted visual conditions, the operation and maintenance of offshore wind farms
only need to consider the effects of wind speed and wave height. On the other hand, according to the
Energies 2019, 12, 2650 15 of 26

monitoring of coastal sea conditions by the National Ocean Forecasting Station, the relationship between
the average wind speed at sea and the effective wave height can be expressed as in Equation (33) [33]:

Hs = 0.038V 2.0 + 0.1950 (33)

where Hs is the effective wave height, and V represents the wind speed. Based on this, the paper
simplifies the weather conditions at sea, and the wind speed prediction value can be obtained according
to Equation (33).
Due to the fact that the wind speed and wave height at sea have obvious periodicity and seasonality
and the state transition probability satisfies the “time homogeneity” characteristic, this paper uses the
Markov chain prediction method based on absolute distribution to model the weather of offshore wind
farms. The detailed method and steps for modeling are as follows:

(1) State division of historical observation data of wind speed and wave height. In this paper, ∆Hs =
0.1 m, ∆V = 1 m/s or ∆Hs = 0.4 m, ∆V = 1 m/s were used as the standard interval grouping
method to classify the state of the sea wind and waves.
(2) According to the state group established by step (1), a frequency transfer matrix ( fij )i,j∈S
and a Markov chain state transition probability matrix (P̂ij )i,j∈S having a step size of one can
be obtained.
(3) Test of “Markov property”.
(4) If the state at time t is known at which the wind speed and the wave height is i, the initial
distribution can be considered to be as follows:

P(0) = (0, · · · , 0, 1, 0, · · · , 0) (34)

The absolute distribution of wind speed and wave height at t + 1 is

P(1) = P(0)P = (pi1 , pi2 , · · · , pin ) (35)

The predicted state j at time t + 1 satisfies Equation (36):

n o
pij = max pij , j ∈ S S = {1, 2, · · · , n} (36)

Then, the wind speed and wave height at time t + k can be obtained.
(5) Duration prediction of accessibility and inaccessible state. According to the current ship
performance and sea level in China, referring to the domestic and foreign related literature [34],
and combined with the weather and ocean conditions in the sea area where the offshore wind
farm is located, the operation and maintenance of offshore wind farms are generally operated at
wind speed V ≤ 10 m/s and wave height Hs ≤ 2 m. According to Equation (32), when the wave
height is less than 2 m, the wind speed can meet the ship’s sea conditions. Therefore, the paper
only needs to consider the wave height. Suppose H is a set of accessibility states that contains
multiple wind speeds and wave height states; according to Equation (36), the one-step state
transition probability matrix P
e for H and its complement S − H (denoted as H)
n o
e can be obtained.
π= e
The steady-state distribution vector is e πH , e
πH πH , e
e , where e πHe represent the probability of
state set H and H,
e respectively.
" #
e= e pHH e pHH
P e
(37)
pHH
ee pH
e eH
e
Energies 2019, 12, 2650 16 of 26

Assuming that τ(τ = 1, 2, . . . , l) is the duration of the state set H, the probability that the state set
H lasts for l durations is
n o
p{τ = l} = p X1 ∈ H, X2 ∈ H, · · · , Xl ∈ H X0 ∈ H, e Xl + 1 ∈ H
e
l−1 (38)
=eπ2ep21 (e
p11 ) e p12

where X1 , X2 , Xs , Xl represent the states of the first, second, s-th, and l-th periods, respectively.
Then, the average duration E(τ) can be obtained according to Equation (39):
∞
X
E(τ) = p{τ = i}i (39)
i=1

p{τ = l} of the state H

Similarly, the probability e e lasting for l periods of time and the average
duration E(τ) are
e
l−1
p{τ = l} = e
e π1 e
p12 (e
p22 ) e p21 (40)
∞
X
e(τ) =
E p{τ = i}i
e (41)
i=1

(6) Maintenance waiting time prediction analysis. The operation and maintenance of offshore wind
turbines are affected by climate and tides, resulting in increased waiting time for maintenance and
limited actual maintenance time. Assuming that personnel, vessels, spare parts, etc., are adequately
prepared, the waiting time for maintenance depends only on the weather conditions.

Assuming that the unit components fail at time t and the total time required for failure repair and
transportation is tn , the maintenance waiting time can be estimated using the dynamic time window.
The specific steps are as follows:

1) Determine the season w at time t, and then calculate the cumulative probability distribution based
on the probability (πw ) of each state in the season. A random number between (0, 1) is generated,
and the cumulative probability distribution interval in which the random number is located is
determined; thereby, a state Xt,w of the wind speed and the wave height at the same moment can
be obtained.
2) If Xt,w ∈ H, then initialize the duration tlo = ∆t of the reachable state and the maintenance waiting
time twa = 0; if not, let tlo = 0, twa = ∆t, where ∆t is the Markov model step size.
3) According to the one-step state transition matrix Pw , the probability vector of the state Xt,w
converted to the next moment state is PX,w , and Xt,w = PX,w . The next time state Xt+1,w is obtained
according to the method in step (1).
4) If Xt+1,w ∈ H, update the reachable state duration tlo = tlo + ∆t, t = t + 1; otherwise, tlo = 0,
twa = tlo + ∆t, t = t + 1.
5) Repeat steps (3) and (4) until tlo ≥ tn , where twa is the waiting time for this maintenance activity.

5. An Opportunistic Maintenance Strategy Based on Accessibility Evaluation

A wind turbine is a complex multi-component system. Once a component fails, the functions of
other components will also be affected, resulting in overall shutdown and the need for maintenance of
the unit. The paper intends to study a maintenance strategy for offshore wind turbines in combination
with equipment fault repair and preventive maintenance.

5.1. Opportunistic Maintenance Strategy

The opportunistic maintenance strategy refers to the fact that, when a component in the system
is repaired or replaced due to a malfunction or other reason, it also provides an opportunity for
Energies 2019, 12, x FOR PEER REVIEW 18 of 27

Energies 2019, 12, 2650 17 of 26

The opportunistic maintenance strategy refers to the fact that, when a component in the system
is repaired or replaced due to a malfunction or other reason, it also provides an opportunity for
preventivemaintenance
preventive maintenance or replacement
or replacement of components,
of other other components, considering
considering the
the fact that thefact that the
maintenance
maintenance
of of at
multiple parts multiple
the sameparts
timeatcan
theeffectively
same time can
save effectively
fixed save costs
maintenance fixed and
maintenance costs and
increase availability.
increase availability.
The principle of the opportunistic maintenance strategy for failure repair combined with preventive
The principle
maintenance of the opportunistic
of components maintenance
is shown in Figure 7. strategy for failure repair combined with
preventive maintenance of components is shown in Figure 7.

R( ) ( t )
i

T ( )
i

To (
i)
Tp (
i)

Corrective
Corrective maintenance or Preventive
maintenance Opportunity maintenance
maintenance

Figure 7. Opportunistic
Figure 7. Opportunistic maintenance
maintenance strategy.
strategy.

5.2. Opportunistic Maintenance Model

5.2. Opportunistic Maintenance Model
5.2.1. Opportunistic Maintenance Probability Density Function
5.2.1. Opportunistic Maintenance Probability Density Function
According to the opportunistic maintenance strategy in Figure 7, the expected replacement rate
of theAccording
component to consists
the opportunistic maintenance
of three parts, complete strategy in Figure
replacement, 7, the expected
incomplete repair,replacement
and minimum rate
of the component consists of three parts, complete replacement, incomplete
maintenance. This paper only considers the opportunistic maintenance strategy for a complete repair, and minimum
maintenance.According
replacement. This paper onlyupdate
to the considers thetheory,
process opportunistic
when themaintenance
component isstrategy
repairedforfor aa period
completeof
replacement. According to the update process theory, when the component is
time, the maintenance rate is close to constant, which is the reciprocal of the expected replacement repaired for a period
of time,
cycle of the
the maintenance
component. rate is close to
According to the
constant, which is maintenance
opportunistic the reciprocal strategy
of the expected
shown replacement
in Figure 7,
cycle of the component. According to the opportunistic maintenance strategy
the replacement rate λi of one component consists of three parts, as shown in Equation (42): shown in Figure 7, the
replacement rate λi of one component consists of three parts, as shown in Equation (42):
λi = 1/Ei (T ) = λci + λoi + λpi (42)
λi = 1 Ei (T ) = λci + λoi + λ pi (42)
where λi = Pi ( f ) × λi ; λoi = Pi (o) × λi ; λpi = Pi (p) × λi are the failure replacement rate, opportunistic
where λi = rate,
replacement ) × λpreventive
Pi ( f and i ; λoi = replacement
Pi (o) × λi ; rate
λ pi of= one ) × λi are respectively.
Pi ( pcomponent, the failurePireplacement
is the replacementrate,
probability
opportunistic Ei (T ) is the rate,
andreplacement expected replacement rate of one component, respectively. Pi
life of the component.
and preventive
Since the opportunistic replacement of component i is caused by the failure replacement or
is the replacement probability and Ei (T ) is the expected life of the component.
preventive replacement of other components, the opportunistic replacement probability density
Since the opportunistic replacement of component
function of the component i can be approximated as an exponential i is caused by the failure
distribution replacement
function, or
as follows:
preventive replacement of other components,
  the h opportunistici replacement probability density
t, To (i) = δi exp
function of the component i can begiapproximated as an (t − To (i) ) distribution function, as follows:
−δiexponential (43)

( δi =
)
gi t , To ( i ) = X
δNi exp
λci + λpi
(i )
  −δ i (t − To )  (43)
(44)
N
δ i =  ( λci + λ pi )
j,i
(44)
(i)
where gi (t, To ) represents the opportunistic replacement
j ≠i probability density function of the component
i, δi is the exponential
(i ) distribution function, and N is the number of key components of the unit.
where gi (t , To ) represents the opportunistic replacement probability density function of the
component i , δ i is the exponential distribution function, and N is the number of key components
of the unit.
Energies 2019, 12, 2650 18 of 26

5.2.2. Key Component Replacement Probability of the Wind Turbine

According to the opportunistic maintenance strategy shown in Figure 7, the probability of each
replacement is as follows:
Z To (i) Z Tp (i)  
Pi (c) = fi (t)dt + fi (t) × Gi t, To (i) dt, (45)
0 To ( i )

Z Tp (i)  
Pi (o) = Fi (t) × gi t, To (i) dt, (46)
To ( i )
   
Pi (p) = Fi Tp (i) × Gi Tp (i) , To (i) , (47)
Z t
Fi (t) = 1 − fi (u)du, (48)
0
  Z t  
(i)
Gi t, To = 1− gi u, To (i) du, (49)
Toi

where fi (t) is the failure probability density function of component i, which obeys the
Weibull distribution.

5.2.3. Opportunistic Maintenance Optimization Model

During one replacement cycle of a wind turbine component, the maintenance cost of component i
consists of replacement costs, downtime losses, and rental vessel costs. The expected repair cost of
component i during a replacement cycle can be expressed as follows:

Ei (C) = [Chci + (twa + tn )Ce ]Pii (c) + Coi Pi (o)

(50)
+ Cpi + (twa + tn )Ce Pi (p) + CN

where Cci , Cpi , Coi represent the costs of failure maintenance, preventive maintenance, and opportunistic
maintenance, respectively, of component i; Ce is the loss per unit downtime, and CN is the rental
vessel cost.
During one replacement cycle of a wind turbine component, the expected life of component i can
be expressed as follows:
R To ( i )    
Ei ( T ) = t × f ( t ) dt + T (i) × F T (i) × G T (i) , T (i)
0 i p i p i p o
R Tp (i) h    i (51)
+ (i) t fi (t) × Gi t, To (i) + Fi (t) × gt t, To (i) dt
To

According to the above analysis, by optimizing the opportunistic maintenance age and preventive
maintenance service age of each key component of the unit to minimize the average cost rate,
the following optimization model is established:

N
Ei (C)
  P
minZ To , Tp = Ei (T )
i=1
(52)
s.t. 0 < To < Tp
tlo ≥ tn

where Z(To , Tp ) is the average cost rate of the unit.

Energies 2019, 12, 2650 19 of 26

Energies 2019, 12, x FOR PEER REVIEW 20 of 27

Energies 2019, 12, x FOR PEER REVIEW 20 of 27
6. Simulation Analysis
6.1. Accessibility Evaluation Simulation Analysis
6.1. Accessibility Evaluation
6.1. Accessibility Evaluation Simulation
Simulation Analysis
Analysis
Due to a lack of weather and ocean data for offshore wind farms, this paper takes the historical
Due to
observation
Due to aadata
lackfrom
lack of weather
of weather and ocean
an observation
and ocean data for
station
data forthe
in offshore
Yangtze
offshore wind
wind farms,
River this paper
estuary
farms, this paper
(30 takes
°N,takes
128 °E)thein
the historical
the East
historical
observation
China Sea as data
an from
example an observation
to model the station
sea in the
weather. Yangtze River estuary
observation data from an observation station in the Yangtze River estuary (30 N, 128 E) in the East (30◦ °N, 128 ◦ °E) in the East
China
China Seadata
The
Sea as an
as an example
onexample
wind speed to model
to model
and wavethe sea
the sea weather.
height in the article were recorded every half hour from 0:00
weather.
The
on 1 The data
January on wind
2014. The speed
17,306 and
datawave height
points
data on wind speed and wave height in the in
recorded theinarticle
article
2014 were recorded
were recorded
used to make everya half
every half houranalysis
forecast
hour from 0:00
from 0:00
of
on
wind
on 1 January
speed and
1 January 2014.
2014. The
wave 17,306
Theheight
17,306in data
2015.
data points recorded in 2014 were used to make a forecast
points recorded in 2014 were used to make a forecast analysis of analysis of
wind speed and wave height
wind speed and wave height in 2015. in 2015.
6.1.1. Wind Speed and Wave Height State Prediction
6.1.1. Wind Speed and Wave
6.1.1. Wave Height
Height State
StatePrediction
Prediction
Using the Markov method, the wind speed and wave height can be divided into n states, and
Using
Using
the state the
the
set Markov
isMarkov method,
method,
S = {1,2,⋯,n}. thewind
In the
this windspeed
paper, H sand
speed =and wave
wave
0.1 0.2height
m, height 0.3can
m, can m,be
be 0.4divided
divided into
intoused
m are nasstates,
n states, andand
the
grouping
state
the setto
state
criteria S =is{1,2,·
isset
predict ·{1,2,⋯,n}.
S =the · ,n}.
100 In this
data paper,
Inpoints ofHwind
this paper, Hspeed
s = 0.1 m, 0.2
s = 0.1andm, 0.3
0.2m,m,
m,wave 0.40.3
heightmm,areJanuary
in used
0.4 m as
aregrouping
used
2015. The criteria
asresults
groupingto
are
predict
criteria
shown in the 100
to Figures data
predict the points
8‒11. of wind speed and wave height in January 2015. The results
100 data points of wind speed and wave height in January 2015. The results are are shown in
Figures
shown in 8–11.
Figures 8‒11.

Figure 8. The comparison of predicted and observed values of wave height, wave height prediction
Figure
Figure 8. The
8.
curve with ∆Hcomparison
The comparison
s = 0.1 m. of
of predicted
predicted and
and observed
observed values
values of
of wave
wave height,
height, wave
wave height
height prediction
prediction
curve with ∆H = 0.1 m.
curve with ∆Hs = 0.1 m.

Figure 9. The
Figure 9. The comparison
comparison ofof predicted
predicted and
and observed
observed values
values of
of wave
wave height,
height, wave
wave height
height prediction
prediction
curve
curve with
Figure ∆H
9. The
with = 0.2
0.2 m.
∆Hcomparison
ss = m. of predicted and observed values of wave height, wave height prediction
curve with ∆Hs = 0.2 m.
 Energies 2019, 12, x FOR PEER REVIEW 21 of 27
Energies 2019,12,
Energies2019, 12,2650
x FOR PEER REVIEW 20
21 of
of 26
27

Figure 10. The comparison of predicted and observed values of wave height, wave height prediction
Figure
curve10.
Figure The
with
10. ∆H
The comparison
s = 0.3 m.
comparison of
of predicted
predicted and
and observed
observed values
values of
of wave
wave height,
height, wave
wave height
height prediction
prediction
curve with ∆H
curve with ∆Hss == 0.3
0.3 m.
m.

Figure The
11.11.
Figure comparison
The comparisonof of
predicted and
predicted observed
and values
observed of of
values wave height,
wave wave
height, height
wave prediction
height prediction
curve
Figurewith
curve with
11. ∆H∆H
The s = 0.4
= m.
0.4 m.
comparison
s of predicted and observed values of wave height, wave height prediction
curve with ∆Hs = 0.4 m.
The
Thedifference
difference between
between the predicted
the predicteddata and
data andthe observed
the observed data and
data andthe correlation
the correlationcoefficient
coefficient
areare
used as criteria
useddifference
The of the
as criteriabetweenquality of
of the quality the predicted data.
of the predicted
the predicted The
data.
data and the correlation coefficients
The correlation
observed of
data andcoefficientsthe observed
of the
the correlation and
observed
coefficient
predicted
areand
used data
predicted
as of data
average
criteria of of wind
theaverage speed
quality wind and
of the average
speed andwave
predicted height
average
data. Thewaveareheight
calculated
correlation separately,
are calculated
coefficients as shown
separately,
of the observedinas
Table
and 7.
shown in Table
predicted 7. of average wind speed and average wave height are calculated separately, as
data
shown in Table 7.
Table 7. Comparison of observation and prediction data.
Table 7. Comparison of observation and prediction data.
Table 7. Comparison
∆Hs (m) V(m/s) of observation
Hs (m) and
d(m)predictionr data.
ΔH s (m) V(m/s) H s (m) d(m) r
ΔH s (m)
0
0 V(m/s) 9.9300
0.1
9.9300
9.2756
H4.5618
(m)
4.5618
s
3.9360
0
16.0764
1.0000
d(m)0
0.8742
r
1.0000
00.1 9.27569.7762
0.29.9300 3.9360 17.7252 16.0764
4.5618
4.1960 00.8235 0.8742
1.0000
0.10.2 9.77628.1154
0.39.2756 4.1960 18.6377 16.0764
3.9360
3.6840 17.7252
0.8712 0.8235
0.8742
0.20.3 8.11548.7624
0.49.7762 3.6840 19.3732 17.7252
4.1960
3.5240 18.6377
0.8525 0.8712
0.8235
0.30.4 8.7624
8.1154 3.5240
3.6840 19.3732
18.6377 0.8525
0.8712
0.4 8.7624 3.5240 19.3732 0.8525
In Table 7, the data of the wind speed and wave height when ΔH s = 0 are observed data; r is
In Table 7, the
the correlation data of the
coefficients ofwind speed and
the observed andwave whend Δ
height data;
predicted isH s =absolute
the 0 are observed data; r is
error of prediction
data
the and observation
correlation dataofofthe
coefficients theobserved
wave height,
and expressed
predicted as in Equation
data; (52):
d is the absolute error of prediction
data and observation data of the wave height, expressed as in Equation (52):
Energies 2019, 12, 2650 21 of 26

In Table 7, the data of the wind speed and wave height when ∆Hs = 0 are observed data; r is the
correlation
Energies 2019,coefficients the observed and predicted data; d is the absolute error of prediction22data
ofREVIEW
12, x FOR PEER of 27
Energies 2019, 12, x FOR PEER REVIEW
and observation data of the wave height, expressed as in Equation (52): 22 of 27

100

dd== 
d =X100 ( H ) −(H )
100
(53)
 s ) predictive value − ( Hs sobserved
((HHs )spredictive
predictive value− (H )s observed value
i =
i=
1 value )observedvalue
value
(53)
(53)
i =11
Taking the average error of the wave height prediction data as a standard to measure the
Taking
Taking the
advantages the
andaverage
average error
error of
disadvantages of the
of wave
the wave height
height
different prediction
prediction
state data
grouping as
as aa standard
datamethods, the to
standard to measure
measure
average the
the
error,
advantages
advantages and disadvantages
and of different
disadvantages of state grouping
different state methods,
grouping themethods,
average error,
the corresponding
average to
error,
corresponding to H s = 0.1 m, 0.2 m, 0.3 m, 0.4 m, is shown in Figure 12.
s =
Hcorresponding
0.1 m, 0.2 to H = 0.1 m, 0.2 m, 0.3 m, 0.4 m, is shown in Figure 12.
m, 0.3 m, 0.4 m, is shown in Figure 12.
s

Figure 12. The average error of the wave height prediction data with ∆Hs = 0.1 m, 0.2 m, 0.3 m, 0.4 m.
Figure 12.The
Figure12. Theaverage
averageerror
errorofofthe
thewave
waveheight
heightprediction
predictiondata with∆H
datawith s s==0.1
∆H 0.1m,
m,0.2
0.2m,
m,0.3
0.3m,
m,0.4
0.4m.
m.
From Figure 12, when ΔH s = 0.1 m, the average error of wave height prediction data and
From
From Figure
Figure 12, when ∆H
12, when ΔHs s== 0.1
0.1m,m,the
theaverage
averageerror
errorofofwave
waveheight
heightprediction
predictiondata
data and
and
observationdata
observation datareached
reachedthe theminimum,
minimum,which
whichmeans
meansthe thestate
stategrouping
groupingmethod
methodofof∆HΔHs s =
= 0.10.1 m
m is
observation
superior to the
is superior
data reached
to other
the
three three
the other
minimum,
groups. which means
As a consequence,
groups.
the state
this state
As a consequence,
grouping
grouping
this
method of
method ismethod
state grouping
Δ H = 0.1
used asisa grouping
sused as a
m
is superior
criterion
grouping for to the
the other
state
criterion inthe
for three
the groups.
following.
state As a consequence, this state grouping method is used as a
in the following.
grouping criterion for the state in the following.
6.1.2.
6.1.2.Duration
DurationPrediction
Predictionof ofWind
WindSpeed
SpeedandandWave
WaveHeight
Height
6.1.2.Markov
Duration Prediction of Wind Speed and Wave Height
Markovmodels
modelsareareestablished
establishedforforeach
eachofofthe
thefour
fourseasons.
seasons.TheTheoccurrence
occurrenceprobability
probabilityof ofthe
the
accessible
Markov H and are
statemodels the inaccessible
established state
for each 
H inofdifferent
the four seasons
seasons. is shown
The in Figure
occurrence
accessible state H and the inaccessible state H in different seasons is shown in Figure 13.
e 13.
probability of the
accessible state H and the inaccessible state H in different seasons is shown in Figure 13.

Figure 13.The
Figure13. Theprobability
probabilityof
ofeach
eachstate
stateoccurring
occurringin
indifferent
differentseasons.
seasons.
Figure 13. The probability of each state occurring in different seasons.
The
Theduration
durationof
ofeach
eachstate
stateunder
underdifferent
differentseasons
seasonsisisshown
shownin
inTable
Table8.8.One
Oneduration
durationisisthe
thestep
step
size of the
The Markov
duration model,
of each which
state is 0.5
under
size of the Markov model, which is 0.5 h.h.
different seasons is shown in Table 8. One duration is the step
size of the Markov model, which is 0.5 h.
Table 8. The duration of each state under each season.
Table 8. The duration of each state under each season.
Persistence
State
Spring Summer Persistence
Autumn Winter
State
H Spring
0.624 Summer
0.786 Autumn
0.570 Winter
0.455
H 0.624 0.786 0.570 0.455
Energies 2019, 12, 2650 22 of 26

Table 8. The duration of each state under each season.

Persistence
State
Spring Summer Autumn Winter
H REVIEW
Energies 2019, 12, x FOR PEER 0.624 0.786 0.570 0.455 23 of 27

H
e 0.376 0.214 0.430 0.545
H 0.376 0.214 0.430 0.545

Take the inaccessible state H

e in January 2015 as an example, the observed and predicted values
Take the inaccessible state H in January 2015 as an example, the observed and predicted
of the probability distribution and the cumulative distribution of the state duration are presented in
values of the probability distribution and the cumulative distribution of the state duration are
Figure 14, which shows the accuracy of the Markov chain model.
presented in Figure 14, which shows the accuracy of the Markov chain model.

e
Figure 14.
Figure The duration
14. The duration prediction
prediction of
of state H.
state H.

According
AccordingtotoFigure
Figure1313
and Table
and 8, the
Table 8, occurrence
the occurrenceprobability of accessibility
probability in summer
of accessibility is largeris
in summer
than
larger than in any other seasons, and the average duration is the longest, indicating that thespeed
in any other seasons, and the average duration is the longest, indicating that the wind wind
and wave
speed height
and waveinheight
summer in are more suitable
summer are moreforsuitable
sea. On the
for contrary,
sea. On the
the occurrence
contrary, theprobability of
occurrence
the inaccessible state in winter is the largest, indicating that the wind speed and wave height
probability of the inaccessible state in winter is the largest, indicating that the wind speed and wave are not
suitable, and
height are the
not bad weather
suitable, will
and the last
bad for a long
weather willtime.
last for a long time.
6.1.3. Maintenance Waiting Time Estimate
6.1.3. Maintenance Waiting Time Estimate
Taking the maintenance waiting time in January 2015 as an example, the relationship between the
Taking the maintenance waiting time in January 2015 as an example, the relationship between
maintenance waiting time and the state of the reachable state is shown in Figure 13. One duration is
the maintenance waiting time and the state of the reachable state is shown in Figure 13. One
set to be 0.5 h. It can be seen that the waiting time for maintenance increases with the duration of the
duration is set to be 0.5 h. It can be seen that the waiting time for maintenance increases with the
accessible state.
duration of the accessible state.
As seen in Figure 15, at the initial moment, there is a large difference between the predicted
As seen in Figure 15, at the initial moment, there is a large difference between the predicted
value and the observed value when ∆Hs = 0.1 m. As the duration increases, the observations
value andapproach
gradually the observed
the predicted value,Δand
value when = 0.1 only
H s finally, m. As the duration
a small increases,
gap is observed. When the ∆H
observations
s = 0.4 m,
gradually approach the predicted value, and finally, only a small gap is observed.
the predicted value is far from the observed value at the initial moment; as the duration increases, When
the
ΔHdecreases,
gap s = 0.4 m,butthe
in predicted
the end, it value
still is far from
maintains a the
large observed
gap. value at the initial moment; as the
duration increases, the gap decreases, but in the end, it still maintains a large gap.
duration of the accessible state.
As seen in Figure 15, at the initial moment, there is a large difference between the predicted
value and the observed value when ΔH s = 0.1 m. As the duration increases, the observations
gradually approach the predicted value, and finally, only a small gap is observed. When
ΔH s =2019,
Energies 0.412,m,
2650the predicted value is far from the observed value at the initial moment; 23
asofthe
26
duration increases, the gap decreases, but in the end, it still maintains a large gap.

Figure 15. Maintenance waiting time

Figure 15. time and
and duration
duration diagram.
diagram.

6.2. Simulation Analysis of the Opportunistic Maintenance Strategy

In this paper, a 3 MW wind turbine in an offshore wind farm in China is used as an example to
optimize the opportunistic maintenance strategy for the critical components of the unit, including the
gearbox, the main bearing, and the generator. According to the analysis of the statistical data of failure
record of the unit, the failure time of the component follows the Weibull distribution, and the specific
parameters are shown in Table 9 below, where 1, 2, 3, and 4 represent the blade, gearbox, main bearing,
and generator, respectively; α and β are the shape parameters and scale parameters of the components,
respectively; unit downtime loss Ce = 720 yuan/h; and the ship chartering fee CN = 150,000 yuan/day.

Table 9. Weibull distribution parameters and maintenance costs of critical components.

Component
αi /Day βi γi /Day C1i /Yuan C2i /Yuan
Number
1 2529 2.8 465 672,000 168,000
2 2228 2.5 136 912,000 228,000
3 4036 2.5 91 360,000 90,000
4 3154 2.2 36 600,000 150,000

When the traditional age-preventive maintenance strategy is adopted, the preventive maintenance
age of the critical components of the unit is optimized to minimize the average cost rate. The results
after optimization are shown in Table 10.

Table 10. Optimization results of age maintenance strategy.

Preventive
Minimum Cost Total Minimum Cost
Component Maintenance Age
Rate/¥ Day−2 Rate/¥ Day−1
Tp /Day
Blade 1816 184.8898
Gearbox 1423 300.5747
782.6744
Main bearing 2822 108.3843
Generator 2284 188.8256

When using a multi-component opportunistic maintenance strategy, the opportunistic maintenance

age, and preventive maintenance age of the components are optimized to minimize the average cost
rate. The results are shown in Table 11.
Energies 2019, 12, 2650 24 of 26

Table 11. Optimization results of the opportunistic maintenance strategy.

Opportunistic Preventive
Total Minimum Cost
Component Maintenance Age Maintenance Age
Rate/¥ Day−1
To /Day Tp /Day
Blade 721 2709
Gearbox 639 1812
702.5834
Main bearing 1001 3000
Generator 897 2430

From the comparison of the results in Tables 10 and 11, it can be seen that the maintenance cost of
the wind turbine using the opportunistic maintenance strategy is 10% lower than that of the preventive
maintenance strategy.
Although the proposed model gains competitiveness in terms of the unit maintenance cost,
its limitations are as follows:

(1) The research on the maintenance cost was aimed at the key components of a single wind turbine
rather than multiple wind turbines.
(2) Influential factors, such as the availability of vessels, the adequacy of the spare parts, and the
meteorological conditions, were linked to maintenance waiting time. In this paper, wind speed and
wave height were considered in the determination of maintenance waiting time, assuming vessels
and spare parts were available.

7. Conclusions
Before concluding, one additional point needs to be discussed. Because the operating environment
of offshore wind turbines is complex, wind is a crucial factor that should be considered in optimal
maintenance decisions. Moreover, the randomness of wind impacts on the power generated by wind
turbines, which contributes to unsatisfactory energy costs. In this paper, the influence that fluctuations
in wind speed had on the costs of offshore wind farming was considered in the determination of
maintenance waiting time. From the perspective of reducing the entire-life costs of an offshore wind
farm, we paid attention to how much maintenance costs will be reduced with the introduction of the
opportunistic maintenance strategy. Actually, in this sense, how the wind fluctuation will act on the
unit cost is not the research priority in this study. Indeed, the influence of wind fluctuation and grid
impact on the unit is worth intensively study in further research.
In this paper, we attempted to analyze the characteristics of operation data of offshore wind
turbines and investigated reliability analysis methods for offshore wind turbines based on limited
fault data. Considering the influence of weather factors, such as wind speed and wave height,
we studied maintenance waiting time prediction methods for offshore wind turbines. Combining failure
maintenance and preventive maintenance, we proposed an opportunity-based offshore wind turbine
maintenance strategy. The main study results are as follows:

(1) The construction of a Weibull equation for offshore wind turbines was based on a small amount of
sample fault data. Different to [8], based on the construction of a two-parameter Weibull equation,
a three-parameter Weibull equation was proposed. The results show that the maintenance costs
can be reduced by 8% with the adoption of a three-parameter Weibull model, and the fitting curve
and failure rate short-term prediction of the three-parameter Weibull distribution is superior to
the two-parameter Weibull distribution where there are limited fault data;
(2) A maintenance waiting time prediction method was introduced for offshore wind turbines.
The Markov chain method and dynamic time window were used to describe wind speed and
wave height, and a maintenance waiting time prediction model was established. Different to [9],
the impacts on maintenance waiting time arising from wind speed and wave height were
considered in this paper. The results show that the deviation of the predicted value of the wave
Energies 2019, 12, 2650 25 of 26

height obtained by grouping interval with ∆Hs = 0.1 m was the smallest, which was close to the
true value;
(3) Combining failure maintenance and preventive maintenance, an opportunistic maintenance
strategy was presented for offshore wind turbines. The minimum expected maintenance cost
was regarded as an objective function to optimize the opportunistic maintenance time and
preventive maintenance time. Compared with [10], the opportunistic maintenance strategy
reduces maintenance duration and decreases the maintenance waiting time and downtime,
thereby reducing maintenance costs. The results show that the maintenance cost was reduced
by 10% under the opportunistic maintenance strategy for offshore wind turbine maintenance,
which verified the effectiveness and superiority of the opportunistic maintenance strategy for
offshore wind turbine maintenance.

Author Contributions: Data curation, X.H.; Investigation, L.X.; Methodology, S.L.; Supervision, X.R.; Visualization,
S.L.; Writing—original draft, L.X.
Funding: This research received no specific grant from any funding agency in the public, commercial,
or not-for-profit sectors.
Acknowledgments: We would like to thank AIP Author Services for English language editing.
Conflicts of Interest: The authors declare no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.

References
1. Mahmood, S. Maintenance logistics organization for offshore wind energy: Current progress and future
perspectives. Renew. Energy 2015, 77, 182–193.
2. Li, X.; Liang, N.; Cheng, Z.J. Estimation of fatigue failure of asphalt concrete based on three-parameter
Weibull distribution. Chongqing Jiaotong Univ. (Nat. Sci.) 2017, 1, 178.
3. Sanni, S.S.; Chukwu, W.I.E. Inventory model with three-parameter Weibull deterioration, quadratic demand
rate and shortages. Am. J. Math. Manag. Sci. 2016, 35, 159. [CrossRef]
4. Deng, B.; Jiang, D.; Gong, J. Is a three-parameter Weibull function really necessary for the characterization of
the statistical variation of the strength of brittle ceramics? Eur. Ceram. Soc. 2017, 20, 123. [CrossRef]
5. Chaurasiya, P.K.; Ahmed, S.; Warudkar, V. Study of different parameters estimation methods of Weibull
distribution to determine wind power density using ground based Doppler SODAR instrument. Alex. Eng. J.
2017, 37, 140. [CrossRef]
6. Chai, H.; Shin, J.; Lim, T.; Trans, J. Failure rate calculation using the mixture Weibull distribution. Kim. Korean
Inst. Electr. Eng. 2017, 66, 500.
7. Xu, X.L. Statistical inference for incomplete data three-parameter Weibull distribution. Shanghai Teachers
Univ. (Nat. Sci.) 2002, 31, 22.
8. Fu, Y.; Xu, W.X.; Liu, L.J. An opportunistic maintenance strategy for offshore wind turbine based on
accessibility evaluation. electric power. Wind Eng. 2016, 49, 74–80.
9. Liu, L.J.; Fu, Y.; Ma, S.W. Preventive maintenance strategy for offshore wind turbine based on reliability and
maintenance priority. Proc. CSEE 2016, 36, 5732–5740.
10. Zheng, X.X.; Zhao, H.; Liu, L.J. A Combined maintenance strategy for offshore wind turbine considering
accessibility. Power Syst. Technol. 2014, 38, 3030–3036.
11. Pedro, G.L.; Luis, V.T.; Matthias, W. Normal behaviour models for wind turbine vibrations: Comparison of
neural networks and a stochastic approach. Energies 2017, 10, 1944–1957.
12. Pedro, G.L.; Iván, H.; Matthias, W. Fatigue load estimation through a simple stochastic model. Energies 2014,
7, 8279–8293.
13. Monbet, V.; Aillot, P.; Prevosto, M. Survey of stochastic models for wind and sea state time series.
Probab. Eng. Mech. 2007, 22, 113–126. [CrossRef]
14. Matha, D.; Scheu, M.; Muskulus, M. Validation of a Markov-based weather model for simulation of O&M for
offshore wind farms. In Proceedings of the 22th International Offshore and Polar Engineering Conference,
Rhodes, Greece, 17–22 June 2012; pp. 463–468.
Energies 2019, 12, 2650 26 of 26

15. Scheu, M.; Matha, D.; Hofmann, M.; Muskulus, M. Maintenance strategies for large offshore wind farms.
Eng. Proc. 2012, 24, 281–288. [CrossRef]
16. Zhang, C.; Gao, W.; Guo, S.; Li, Y.L.; Yang, T. Opportunistic maintenance for wind turbines considering
imperfect, reliability-based maintenance. Renew. Energy 2017, 103, 606–612. [CrossRef]
17. Mahmood, S.; Maxim, F.; Christophe, B. An opportunistic condition-based maintenance policy for offshore
wind turbine blades subjected to degradation and environmental shocks. Reliab. Eng. Syst. Saf. 2015, 142,
463–471.
18. Hadi, A.; Karim, A.; Morteza, A. Multi-objective opportunistic maintenance optimization of a wind farm
considering limited number of maintenance groups. Renew. Energy 2016, 88, 247–261.
19. Ding, F.F.; Tian, Z.G. Opportunistic maintenance for wind farms considering multi-level imperfect
maintenance thresholds. Renew. Energy 2012, 45, 175–182. [CrossRef]
20. Atashgar, K.; Abdollahzadeh, H. Reliability optimization of wind farms considering redundancy and
opportunistic maintenance strategy. Energy Convers. Manag. 2016, 112, 445–458. [CrossRef]
21. Hallinan, A.J. A review of the Weibull distribution. Qual. J. Technol. 1993, 25, 85–93. [CrossRef]
22. Zhu, W.Y. Design of Mechanical Reliability; Press of Shanghai Jiaotong University: Shanghai, China, 1992;
pp. 62–66.
23. Zhu, Y.H.; Tai, S.C.; Sun, Y.Y. Applied Mathematical Statistics; Press of Wuhan University of Hydraulic and
Electrical Engineering: Wuhan, China, 1999; pp. 68–72.
24. Lu, D.J. Stochastic Processes and Applications; Tsinghua Press: Beijing, China, 1986; pp. 100–115.
25. Liu, C.H. Stochastic Processes; Press of Huazhong University of Science and Technology: Wuhan, China, 2008;
pp. 42–45.
26. Liu, Y.F. Research on Reliability of Software System Based on Markov Chain Method; Changchun University of
Science and Technology: Changchun, China, 2005.
27. Sheldon, M.R. Stochastic Processes, 2nd ed.; John Wiley and Sons: Hoboken, NJ, USA, 1996.
28. Qian, M.P. Application on Stochastic Processes; Peking University Press: Beijing, China, 1998.
29. Luo, J.Y. Economic Statistical Analysis Methods and Forecasts; Tsinghua Press: Beijing, China, 1987; pp. 347–348.
30. Van Bussel, G.J.W.; Zaaijer, M.B. Reliability, availability and maintenance aspects of large-scale offshore wind
farms, a concepts study. In Proceedings of the MAREC, Newcastle, UK, 17–18 April 2001; pp. 5–9.
31. Hagen, B.; Simonsen, I.; Hofmann, M.; Muskulus, M. A multivariate Markov weather model for O&M
simulation of offshore wind parks. Energy Procedia 2013, 35, 137–147.
32. Lyding, P.; Faulstich, S.; Hahn, B. Reliability of the electrical parts of wind energy systems-a statistical
evaluation of practical experiences. In Proceedings of the EPE Wind Energy Chapter Symposium,
Staffordshire University, Stafford, UK, 15–16 April 2010.
33. Feng, W.B.; Peng, X.L.; Zhang, S.L. Analysis on wave condition of south ocean of China: No 14. In Proceedings
of the China Ocean Science and Technology Symposium, Hohhot, China, 16–17 May 2009.
34. Sinha, Y.; Steel, J.A. A progressive study into offshore wind farm maintenance optimization using risk based
failure analysis. Renew. Sust. Energ. Rev. 2015, 42, 735–742. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).