energies

Article
Impact Assessment of Dynamic Loading Induced by the
Provision of Frequency Containment Reserve on the Main
Bearing Lifetime of a Wind Turbine
Narender Singh 1,2, * , Dibakor Boruah 1 , Jeroen D. M. De Kooning 1,2 , Wim De Waele 1
and Lieven Vandevelde 1,2

1 Department of Electromechanical, Systems & Metal Engineering, Faculty of Engineering & Architecture,
Ghent University, Tech Lane Ghent Science Park—Campus A, Technologiepark-Zwijnaarde 131,
B-9052 Ghent, Belgium
2 FlandersMake@UGent—Corelab MIRO, Flanders Make, B-9052 Ghent, Belgium
* Correspondence: narender.singh@ugent.be

Abstract: The components of an operational wind turbine are continuously impacted by both static
and dynamic loads. Regular inspections and maintenance are required to keep these components
healthy. The main bearing of a wind turbine is one such component that experiences heavy loading
forces during operation. These forces depend on various parameters such as wind speed, operating
regime and control actions. When a wind turbine provides frequency containment reserve (FCR) to
support the grid frequency, the forces acting upon the main bearing are also expected to exhibit more
dynamic variations. These forces have a direct impact on the lifetime of the main bearing. With an
increasing trend of wind turbines participating in the frequency ancillary services market, an analysis
of these dynamic forces becomes necessary. To this end, this paper assesses the effect of FCR-based
control on the main bearing lifetime of the wind turbine. Firstly, a control algorithm is implemented
such that the output power of the wind turbine is regulated as a function of grid frequency and the
amount of FCR. Simulations are performed for a range of FCR to study the changing behaviour of
Citation: Singh, N.; Boruah, D.; De dynamical forces acting on the main bearing with respect to the amount of FCR provided. Then,
Kooning, J.D.M.; De Waele, W.; based on the outputs from these simulations and using 2 years of LiDAR wind data, the lifetime of
Vandevelde, L. Impact Assessment of
the main bearing of the wind turbine is calculated and compared for each of the cases. Finally, based
Dynamic Loading Induced by the
on the results obtained from this study, the impact of FCR provision on the main bearing lifetime is
Provision of Frequency Containment
quantified and recommendations are made, that could be taken into account in the operation strategy
Reserve on the Main Bearing Lifetime
of a wind farm.
of a Wind Turbine. Energies 2023, 16,
2851. https://doi.org/10.3390/
en16062851
Keywords: frequency containment reserve; structural loading; wind energy; wind turbine control;
wind turbine main bearing
Academic Editors: Galih Bangga and
Martin Otto Laver Hansen

Received: 21 February 2023

Revised: 16 March 2023 1. Introduction
Accepted: 16 March 2023 The wind power sector has shown a remarkable growth in the past decade with
Published: 19 March 2023 no foreseeable sign of diminishing. Figure 1 presents the globally installed wind energy
capacity from 2012 to 2021. In 2021, nearly 94 GW of new wind power was installed
globally, including 22.1 GW of offshore wind capacity. At the end of 2021 the total global
Copyright: © 2023 by the authors.
wind power capacity reached 824 GW. It is expected that by 2026 the global installed wind
Licensee MDPI, Basel, Switzerland. capacity will reach 1394 GW [1,2]. With the targets of the European Union (EU) to reduce
This article is an open access article greenhouse gas emissions, a trend of increasing renewable energy penetration in the power
distributed under the terms and systems has been observed [3]. In 2021, Europe added 17 GW of new wind power capacity.
conditions of the Creative Commons This adds up to a total of 236 GW of installed wind energy capacity in Europe. In 2021, 15%
Attribution (CC BY) license (https:// of the total electricity consumed by all EU-nations was generated by using wind energy [4].
creativecommons.org/licenses/by/
4.0/).

Energies 2023, 16, 2851. https://doi.org/10.3390/en16062851 https://www.mdpi.com/journal/energies

Energies 2023, 16, 2851 2 of 14

900

Global wind energy capacity (GW)

800
700
600
500
400
300
200
100
0
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
Year

Figure 1. Global wind energy installed capacity.

With the increasing share of wind energy, an active participation of the wind farms
in the ancillary services market is expected, and even obliged by the grid code in some
countries [5]. Some ancillary services such as frequency containment reserve (FCR), fast fre-
quency response (FFR), synchronous inertial response (SIR), enhanced frequency response
(EFR) and fast post-fault active power recovery (FPFAPR), provide opportunities for wind
farms to participate in the ancillary services market [6].
Along with the increasing share of wind energy in the energy mix and an increased
participation of wind farms in ancillary service markets, there is a growing concern about
the structural health and maintenance of wind turbine components. As a wind turbine or
wind farm ages, the overall performance is reduced. The age-based performance of a wind
farm was investigated in [7]. A survey of failures of Swedish wind farms is presented in [8].
Reliability analysis and probabilistic methods for wind farm performance are presented
in [9,10]. Methods to quantify the availability of wind farms were proposed in [11]. An
extensive data study presented in [12] found that wind turbines lose around 1.6% of their
nominal output per year due to aging. It has also been seen that some wind turbine
components fail before their expected lifetime [13]. To fill this gap, several studies have
been conducted that explore new methods of wind turbine monitoring and maintenance.
A deep learning approach was explored to predict the remaining useful life of rotating
components in [14]. A cost-effective condition monitoring for wind turbines was proposed
in [15]. Different methods of wind turbine maintenance management were suggested
in [16,17]. The approach for a remote condition monitoring system was explored in [18].
Wind turbine failure detection and condition-based maintenance strategies are presented
in [19] and [20], respectively. In terms of the overall optimization of a wind farm, a study
exploring wind farm layout optimization is presented in [21].
These studies are focused on overall wind farms or single turbines as a whole. How-
ever, specific studies based on important individual wind turbine components are rare.
A crucial component of the wind turbine is its main bearing. A downtime due to a main
bearing failure can cause a significant loss of revenue in addition to high replacement costs.
There are several factors that affect the health of a bearing. Foremost, the manufacturing
quality of the bearing is critical to a long service life. Logistics of storage, application
and mounting also play an important role. Furthermore, it is crucial for the bearing to
be well lubricated. The lubrication specifications and lubrication change intervals are
specific to the application and help ensure a smooth and healthy bearing operation [22].
According to SKF, the leading causes of bearing damage are lubrication (36%), contami-
nation (23%), application (18%), interface (8%), handling (8%), electrical (5%) and fatigue
(2%). A few studies have been conducted regarding the health of the wind turbine’s main
bearing. In [23], a novel prognostic approach to predict the remaining useful lifetime of
bearings in a wind turbine’s gearbox was proposed. The study relied on artificial neural
networks to predict the short-term tendencies. The data-driven approach was specific
to the bearings in the wind turbine gearbox. Another data-driven approach to analyse
Energies 2023, 16, 2851 3 of 14

bearing faults was discussed in [24], where data mining was applied to identify bearing
faults in a wind turbine. The research relied on a neural network algorithm using the
data collected from several wind turbines. The results from this study were the prediction
of over-temperature events ahead of the fault occurrence. A study about the prediction
and diagnosis of wind turbine faults is presented in [25]. The temperature parameters of
SCADA data, such as bearing temperature, were used for fault detection. A long-term
fault prediction framework was discussed in [26]. This study also utilised SCADA data
and secondary measurements from 108 wind turbines with main-bearing failures. Historic
wind turbine data for run-to-failure and non-run-to-failure wind turbines was utilized
to predict the remaining power generation before failure in [27]. All these studies were
data-based and focused on main-bearing-related issues such as the prediction of time-to-
failure, over-temperature event prediction and remaining power generation before failure.
However, there are no studies that use detailed wind turbine models to study the lifetime
of the main bearing of the wind turbine. The impact of the provision of ancillary services
on the health of the wind turbine’s main bearing remains unknown. The main bearings
of a wind turbine are subjected to dynamically varying forces during its operation. The
forces are significantly more dynamic when providing ancillary services such as FCR. This
originates from the control algorithms implemented to provide frequency control, which
change the radial and axial forces acting upon the main bearing, hence affecting the overall
health of the main bearing. A study of these forces and their impact on the lifetime of the
main bearing of the wind turbine is important in times when wind turbines are expected to
provide ancillary services. A study of the lifetime of the main bearing is also particularly
relevant as it is a comparatively expensive component with a high replacement cost and
a significant downtime for replacement. To this end, this study analyses and quantifies
the effect of providing ancillary services on the lifetime and health of the wind turbine
main bearing.
The article is organized as follows: Section 2 describes the models and data used for
the study. These include the wind turbine, torque control, pitch control, main bearing, wind
field design and data processing. In Section 3, the methodology used to calculate the forces
acting on the wind turbine main bearing and the lifetime are explained. Section 4 presents
and discusses the results from the simulated cases. The conclusions from the study are
presented in Section 5.

2. Models and Data

The models used in this study include the wind turbine rotor, the permanent magnet
synchronous generator, the control system and the main bearing. Additionally, wind data
are used as an input for the simulations. These models and data are detailed in this section.

2.1. Wind Turbine

The wind turbine model used for this paper is a 5 MW reference wind turbine for
offshore system development [28]. The wind turbine model uses broad design information
from published documents of turbine manufacturers with a high correspondence with the
REpower 5M machine. The model is developed in FAST v8, a high-fidelity wind turbine
simulator developed by the National Renewable Energy Laboratory (NREL). The simula-
tions are performed using the blade element momentum (BEM) option in the AeroDyn v15
module of FAST v8 [29]. Within AeroDyn there are four sub-models for rotor induction,
blade airfoil aerodynamics, tower influence on the fluid local to the blade nodes and tower
drag. The module calculates aerodynamic loads on the blades and the tower. The structural
and aerodynamic properties of the blades are based on the 62.6 m long glass fibre blade
used in the study presented in [30]. The structural model of the wind turbine is the Offshore
Code Comparison Collaboration (OC3) monopile [31]. The main properties of this wind
turbine are listed in Table 1.
The wind turbine of the defined type operates in three different regions. These regions,
as presented in Figure 2, are based on the free-flow wind speed measured on the wind
Energies 2023, 16, 2851 4 of 14

turbine rotor. It can be seen in Figure 2 that Region 1 corresponds to wind speeds below
3 m/s, i.e., below the cut-in speed of the wind turbine. There is no torque generated in
Region 1 and no power is extracted. In Region 2, the wind turbine operates in maximum
power point tracking (MPPT) mode. In Region 3, the wind speeds are higher than the
nominal wind speed. In this region, the torque, speed and power are limited using control
methods in order to avoid overloading of the drivetrain components. In case the wind
speed exceeds 25 m/s, the wind turbine is in the cut-out region and no power is produced.

Table 1. Wind turbine properties.

Property Specification
Power rating 5 MW
Nominal torque 4 MN-m
Rotor orientation & configuration Upwind, 3 blades
Rated rotor speed 12.1 rpm
Rotor and hub diameter 126 m and 3 m
Hub height 90 m
Cut-in, Rated and Cut-out wind speed 3 m/s, 11.4 m/s and 25 m/s

5
Power (MW)

4
1 2 3 Cut-out
3
region
2
1
0
3 5 7 9 11 13 15 17 19 21 23 25
Vs (m/s)
Figure 2. Power curve of a 5 MW wind turbine.

2.2. Main Bearing

The original wind turbine model developed by NREL does not consider the detailed
features of the main bearing. However, since the wind turbine is largely based on the
REpower 5M machine, the available literature and data about the bearing design and its
properties were used. The schematic of the main bearing positioning in the wind turbine is
shown in Figure 3. The bearing used in this machine is a spherical roller bearing, weighing
3320 kg [32].

Vertical force

Pitch

Hub
Shaft
Axial force
Lateral force
Main
bearing

PMSG

Figure 3. Main bearing position.

Since a detailed datasheet of the turbine’s main bearing is not publicly available, some
assumptions and estimates were made about the bearing’s properties. The main property
of the bearing that was required to calculate the lifetime is its basic dynamic load rating C
Energies 2023, 16, 2851 5 of 14

in Newtons, which represents the maximum load that a bearing can be subjected to for a
rating life of one million revolutions. In order to estimate the C rating of the bearing, data
from similar bearing types available through SKF were used in curve fitting. To double
check, simulations were performed to calculate the load on the main bearing of the wind
turbine to make sure that the load was always lower than the bearing rating.

2.3. Permanent Magnet Synchronous Generator

The generator used for this study was a permanent magnet synchronous generator
(PMSG). The generator was modelled in a rotating direct, quadrature (d,q) reference frame.
The generator parameters used in this model are listed in Table 2.

Table 2. Generator properties.

Property Specification
Rated power 5 MW
Rated speed 12.1 rpm
Nominal efficiency 93%
Pole pairs 117
Nominal voltage 1950 V
Nominal current 876 A
Stator resistance 98.5 mΩ
Quadrature stator inductance 5.86 mH

2.4. Control
The different control schemes used in the study are detailed in this section.

2.4.1. Torque Control

A crucial part of this study is an effective torque control of the PMSG. To serve this
purpose, field-oriented control (FOC) was used. In FOC, the quadrature current component
(îq ) was proportionally regulated to the torque setpoint. The direct current component
(îd ) was kept at zero to realize field orientation. The torque control scheme is presented in
Figure 4. The proportional and integral gains of the proportional integral (PI) controller
were set equal to 0.0001. At each time step, a current signal based on a comparison between
the reference power ( P̂) and actual power (P) was generated by the controller. The reference
power was based on the grid frequency and the amount of contracted ancillary reserve. The
outputs of the PI controllers were then converted to three-phase abc signals using the inverse
Park transformation. As a result, pulse width-modulated (PWM) signals corresponding
to the three phases were generated. The PI controller was set up to minimize the tracking
error by minimizing the overshoot and settling time. The performance of the developed
controller varied with changing wind speed and the rate of change of the reference power.
However, the controller was able to track the reference power with negligible error.

Figure 4. Field-oriented control.

Energies 2023, 16, 2851 6 of 14

The simulations were performed with four different cases. In the first case no FCR is
provided, i.e., the wind turbine operates in the MPPT mode. In the second case, the wind
turbine is de-loaded by 20%, i.e., the turbine operates at 80% of the maximum power curve.
In the third and fourth cases 20 and 40% FCR is provided, i.e., 20 and 40% of the available
power is reserved for FCR, respectively. For the 20% FCR case, the wind turbine control
design is as shown in Figure 5. In Figures 5 and 6, the blue and red curves represent the
upper and lower bound of the output power. The black curve represents the base power.
The shaded area demonstrates the area of operation, i.e., the reference power at any given
point lies within this area based on the grid frequency. The reference power for these cases
was calculated by using Algorithm 1. The algorithm was designed such that a symmetrical
200 mHz frequency support is provided. A frequency response dead-band of 10 mHz
centred at a nominal frequency (50 Hz) was present to reduce excessive control activities
and turbine mechanical wear for normal power system frequency variations. Here, PUB and
PLB are the upper and lower bound powers, respectively, as presented in Figures 5 and 6.
The reference power Pref was calculated as the sum of base power Pbase and a time varying
term Pfreq (t). Pfreq (t) was calculated using α which is proportional to the percentage of
FCR provided, p FCR . k is the proportionality constant with the value 0.25. α is defined
as follows:
α = k · p FCR (1)
Pbase is defined according to the black curve in Figures 5 and 6 for the 20 and 40% FCR
cases, respectively. Furthermore, Pfreq (t) is based on the varying grid frequency f and the
contracted FCR bid PFCR , i.e., 20 or 40%.

P upper bound
5
P lower bound
Area of operation
4
P base
Power (MW)

0
3 5 7 9 11 13 15 17 19 21 23 25
Vs (m/s)

Figure 5. Control design for 20% FCR.

P upper bound
5
P lower bound
Area of operation
4 P base
Power (MW)

0
3 5 7 9 11 13 15 17 19 21 23 25
Vs (m/s)

Figure 6. Control design for 40% FCR.

Energies 2023, 16, 2851 7 of 14

Algorithm 1 Reference power decision.

Require: Pref ≤ PUB
if f ≤ 49.8 Hz then
Pref (t) = PUB
else if 49.80 Hz < f < 49.99 Hz then
Pref (t) = Pbase + Pfreq . Pfreq =α(50-f)
else if 49.99 Hz < f < 50.01 Hz then
Pref (t) = Pbase . 10 mHz deadband
else if 50.01 Hz < f < 50.20 Hz then
Pref (t) = Pbase + Pfreq . Pfreq =α(50-f)
else if f ≥ 50.20 Hz then
Pref (t) = PLB
end if . all powers given in MW

2.4.2. Pitch Control

The pitch control system developed in Simulink was connected to the wind turbine
model in FAST. On every loop of the simulation, the pitch control system sends a pitch
command to the wind turbine for each of the three blades. The block diagram of the pitch
control system is presented in Figure 7. The pitch control is a PI controller with proportional
and integral gains equal to 206.3 and 25, respectively. The reference rotor speed Ωref is
a set parameter to which the actual rotor speed Ω is compared at each time step by the
PI controller. The gain scheduling is implemented by a dimensionless gain correction
factor G as defined below. Here, θ is the pitch angle and θd is a tuning parameter set at
0.055 radians [33].

1.6
G=   (2)
θ
1+ 2θd

+
-

Figure 7. Pitch control of the wind turbine blades.

2.5. Wind Field Design and Data Processing

The wind data used for the simulations in this research work are from the publicly
available data repository of meteorological data provided by Ørsted. The data are available
for several offshore meteorological stations [34]. The LiDAR data are 10 min statistics for
wind measurements from LiDARs installed at offshore wind farms. A total of 2 years
of data sampled at 10 min were used. In order to fit the data to the operating range of
the wind turbine, the data were processed to fit the requirements. Firstly, only the data
within the range of wind velocities between 3 and 25 m/s were retained. The wind data
were then rounded off to the nearest integer. These data can be visualised in graphical
form as presented in Figure 8. Figure 8 shows the probability of occurrence of different
levels of wind speed within the operating range of the wind turbine. The simulations were
performed for each of these wind levels, and their probability of occurrences was used to
deduce the final results.
Energies 2023, 16, 2851 8 of 14

10

Probability of occurences (%)

8

0
0 5 10 15 20
Wind speed (m/s)
Figure 8. Wind speed percentage occurrences.

3. Methodology
Different outputs from these simulations were analysed and used for the calculation
of bearing loads and lifetime. These parameters are presented in the following subsections.

3.1. Dynamic Equivalent Force

There are different forces acting on the blade root of the wind turbine and hence the
main bearing. The axial, vertical and lateral forces are shown in Figure 3. These forces
were used to calculate the dynamic equivalent force. The axial force was measured at the
root of each blade. The three individual blade forces were then averaged, as shown in (3).
Here, Fb1 , Fb2 and Fb3 represent the axial forces, Fa (in Newtons) at the three blades of the
wind turbine.
The radial force, on the other hand, was given as a result of the lateral force Fl and the
vertical force Fv acting on the blade root, as shown in Figure 3. The equation is given in
Equation (4).

Fb1 + Fb2 + Fb3

Fa = (3)
3
q
Fr = Fl2 + Fv2 (4)

The dynamic equivalent force Pd in Newtons is defined in Equation (5), where bx and
by are the dimensionless empirical factors for the specific spherical roller bearing. The SKF
catalogue for rolling bearings [35] suggests that the mean force, Fm (in Newtons) should be
calculated considering their minimum and maximum values. Therefore, the mean values
of radial and axial forces were calculated using Equation (6), where Fmin and Fmax represent
the minimum and maximum values of the forces within the steady period of the entire
simulation time.
Pd = bx Fr + by Fa (5)

Fmin + 2Fmax
Fm = (6)
3

3.2. Lifetime
The linear fatigue damage accumulation method used in this study was based on [36].
The method incorporates the weight and dynamic equivalent force of each load level.
Hence, providing a good estimate for the bearing lifetime. In this method, firstly Li is
defined as the number of revolutions spent at a given wind speed by the main bearing. Li
Energies 2023, 16, 2851 9 of 14

is calculated as in Equation (7). Here, ti is the number of hours of load level i. Ω is the
speed in rpm for load level i.

Li = ti Ωi 60 (7)
Next, the basic rating life L10i is defined as the life required for 10 % of bearing samples
to fail, for an identical group of bearings at a given load level i. L10i was calculated as given
in Equation (8). Here, p is a bearing specific value, which is equal to 10/3 for roller bearings
as per SKF.
 p
C
L10i = 106 (8)
Pd
Thereafter, f i is a fraction of the level i, defined to assign weight to individual load
levels (1 to m). f i was calculated as in Equation (9).

Li
fi = (9)
∑im=1 Li
The combined lifetime (in years) including all load levels was determined by Equa-
tion (10). Here, a1 , a2 and a3 represent the life modification factors. a1 was set equal to 0.21
corresponding to a 99% reliability of the main bearing surviving the estimated lifetime. The
factors a2 and a3 correspond to the material of the bearing and lubrication conditions and
these were set equal to 1.

1 a1 a2 a3
LM = (10)
106
 
fi
∑im=1 L10i

4. Results and Discussion

For this study, four different cases were designed. For each case, 23 simulations were
performed. These 23 simulations were designed such that for each integer wind speed
within the operating range (3–25 m/s) of the wind turbine, a simulation was conducted. In
this manner, the entire operating range of the wind turbine was included in the study. The
main outputs from these simulations, namely axial, radial and dynamic equivalent forces
were then used to calculate the lifetime of the main bearing of the wind turbine. These
simulations considered the forces once the wind turbine had reached a steady operation.
These four cases differed from each other in the type of control applied. Case 1 is
where the wind turbine operates to provide maximum available power. Therefore, no
margin in the active power is retained to provide an FCR service. The wind turbine operates
on either MPPT or power-limiting control, based on the wind speed. In Case 2, the control
is such that the reference power is de-loaded by 20%. Apart from de-loading, the operation
of the wind turbine remains similar to Case 1, and no FCR is provided. Figure 9 presents
the reference power for curves for Case 1 and Case 2. The blue curve presents Case 1 and
the red presents Case 2. It can be seen here that for each individual wind speed load level,
a different reference power is set for the two cases.
Energies 2023, 16, 2851 10 of 14

Power (MW)
3

1
No FCR
20% Deloaded
0
5 10 15 20 25
Time (s)
Figure 9. Power reference for the no FCR and 20% de-loaded cases.

Case 3 and Case 4 were comparative cases that were simulated to observe the effect of
providing FCR on the main bearing of the wind turbine. In these cases the wind turbine
control systems based on Algorithm 1 were used. Figure 10 consists of three different
curves. The red curve corresponding to the right y-axis represents the grid frequency. The
first blue curve on the top, corresponding to the left y-axis represents Case 3 where the wind
turbine provides 20% FCR. The second blue curve at the bottom represents Case 4 where
the wind turbine provides 40% FCR. As can be observed from Figure 10, the reference
power changes to accommodate the real-time changes in FCR requirements based on the
grid frequency.
4.5 50.04

50.02
4
Frequency (Hz)

50
Power (MW)

3.5 49.98

49.96

3
49.94
FCR 20 FCR 40 Frequency
49.92
0 20 40 60 80 100
Time (s)

Figure 10. Power reference for the 20 and 40% FCR cases.

Figure 11 presents nine different graphs to present the dynamic axial, radial and
equivalent forces acting on the main bearing. The load levels presented in Figure 11
correspond to the wind speeds of 4, 12 and 25 m/s. These levels were chosen such that the
forces in low-, medium- and high-wind speed ranges can be presented. However, it should
be noted that the simulations were performed for each load level from 3 to 25 m/s.
In Figure 11, each graph consists of four curves, representing the four cases, i.e., no
FCR, 20% de-loaded, 20% FCR, and 40% FCR. Figure 11a–c present the axial forces for the
three load levels. It can be seen that the first two curves, corresponding to no FCR and 20%
de-loaded, even though oscillating (Vs = 12 and 25 m/s, respectively), are relatively steady
compared to the last two curves, where FCR is provided. Considering the varying grid
frequency, the power output is changed through torque and pitch control, resulting in a
varying axial force.
Figure 11d–f present the radial forces for the three load levels. The radial forces for
each of these cases is relatively lower compared to the axial forces. For the low-wind speed
of 4 m/s, in Figure 11d, the radial forces are close for these cases. Providing FCR exhibits a
clear effect on the radial forces, as seen in Figure 11e.
Energies 2023, 16, 2851 11 of 14

Figure 11g–i present the dynamic equivalent force. The dynamic equivalent force was
calculated as explained in Section 3.1. The combined effect of both axial and radial forces
can be observed here. The impact of these two forces on the calculated Pd is dependent
on the empirical factors bx and by . The lifetime of the main bearing of the wind turbine
is directly related to its basic dynamic load rating C. The value of C varies based on the
strength of the bearing. In order to study the loading and lifetime depending on different
design choices, the range of 0.75C to 1.25C was used in the simulations, where C was as
defined in Section 2.2.
The results of bearing lifetime are presented in Table 3. The results show that the
highest lifetime was observed in the case with no FCR for all C values. It can be seen that
the difference in lifetime compared to the other three cases is higher for higher C values
and lower for lower C values. It can also be seen that the values for the 20% de-loaded
and 20% FCR cases are close. The least lifetime was observed for the 40% FCR case. The
results here firstly point to the effect of a higher C rating on the bearing lifetime. A higher C
value consistently gave a higher lifetime estimate for the main bearing. Secondly, a lifetime
reduction for the main bearing is associated with providing FCR. Figure 12 presents a graph
of the change in lifetime with increasing α, where α represents the amount of provided FCR.
The different curves in Figure 12 represent the lifetime for different C ratings corresponding
to the basic dynamic load rating values. It is seen that the main bearing lifetime reduces
with increasing α. However, for the higher C rating cases, the lifetime is higher than the
average lifetime of the studied wind turbine, i.e., 20 years. An informed decision can be
made based on this research and the economics of wind turbine operation, such that if the
cost of using a high C-rated main bearing can be compensated for by the revenue generated
from the provision of FCR, then it is indeed an advantage to provide FCR with a wind
turbine. On the other hand, if the C rating of the main bearing is low, it is not advisable to
provide FCR due to its detrimental impact on lifetime.

Figure 11. Dynamic forces acting on the wind turbine’s main bearing.
Energies 2023, 16, 2851 12 of 14

Table 3. Bearing life, in years.

C No FCR 20% deloaded 20% FCR 40% FCR

0.75 13.88 8.88 9.02 8.1
0.8 17.21 11.01 11.19 10.02
0.85 21.06 13.48 13.7 12.26
0.9 25.49 16.31 16.57 14.84
0.95 30.52 19.53 19.84 17.77
1 36.21 23.17 23.54 21.08
1.05 42.6 27.27 27.7 24.81
1.1 49.75 31.84 32.35 28.97
1.15 57.69 36.92 37.51 33.6
1.2 66.49 42.55 43.23 38.71
1.25 76.18 48.75 49.53 44.36

80

70

60
Bearing life (years)

50 0.75C
0.80C
40 0.85C
0.90C
0.95C
30
C
1.05C
20
1.10C
1.15C
10
1.20C
1.25C
0
0 2.5 5 7.5 10 12.5

Figure 12. Bearing life vs. α for the different basic dynamic load rating values.

5. Conclusions
In this paper, the bearing lifetime of a wind turbine primary shaft was studied. The
importance of this study lies in the fact that there is an increasing participation of wind
turbines in the ancillary services market. This participation is expected to grow in the
coming years as a result of the higher penetration of wind power in the energy grid. In such
a scenario, it is crucial to conduct research on the impact of providing ancillary services on
wind turbine components. The scope of this study focussed on a specific component of the
wind turbine, its main bearing. The choice of this component was based on the relative cost
and downtime associated with the maintenance and replacement of this component. The
study explored several different cases and methods of FCR control. A common observation
from this study was the clear impact of providing FCR on the lifetime of a wind turbine’s
main bearing. The study concludes that the impact of FCR provision on the lifetime of the
main bearing is subjective to the amount of FCR provided and its C rating. Considering
the base case with no FCR provision, a lifetime reduction of 36% was observed for the case
when the wind turbine was operated with 20% de-loaded. A lifetime reduction of 35 and
41.8% was observed for the cases of 20 and 40% FCR, respectively. This trend holds for the
entire range of C ratings simulated in this study. It was also observed that to achieve the
same bearing lifetime as the case with no FCR provision, an oversizing of approximately 15
and 20% in the C rating was required for the 20 and 40% FCR cases, respectively. Based on
this study a higher FCR provision is suggested for a wind turbine consisting of bearings
with a higher basic dynamic load rating.
Energies 2023, 16, 2851 13 of 14

Author Contributions: Conceptualization, N.S., J.D.M.D.K. and L.V.; methodology, N.S. and D.B.;
software, N.S.; validation, N.S.; formal analysis, N.S.; writing—original draft preparation, N.S.;
writing—review and editing, N.S., J.D.M.D.K. and L.V.; visualization, N.S., J.D.M.D.K. and L.V.;
supervision, J.D.M.D.K., W.D.W. and L.V.; funding acquisition, L.V. and J.D.M.D.K. All authors have
read and agreed to the published version of the manuscript.
Funding: This research was conducted in the frame of the BEOWIND project, funded by the Energy
Transition Fund of the Belgian Federal Government.
Data Availability Statement: The data used in the current study are available from the corresponding
author upon reasonable request.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Global Wind Energy Council. Global Wind Report. 2022. Available online: https://gwec.net/global-wind-report-2022/ (accessed
on 1 September 2022).
2. IRENA. Renewable Energy Statistics 2022; The International Renewable Energy Agency: Abu Dhabi, United Arab Emirates, 2022.
Available online: https://www.irena.org/ (accessed on 1 September 2022).
3. Energy Roadmap 2050 Energy. 2012. Available online: https://ec.europa.eu/energy/ (accessed on 1 September 2022).
4. Wind Energy in Europe. Available online: https://proceedings.windeurope.org/ (accessed on 1 September 2022).
5. Technical Requirements for the Connection of Generating Stations to the Hydro-Québec Transmission System. 2019. Available
online: http://www.hydroquebec.com/ (accessed on 1 September 2022).
6. European Association for Storage of Energy. Ancillary Services. 2021. Available online: https://ease-storage.eu/wp-content/
uploads/2021/08/Ancillary-Services.pdf (accessed on 6 May 2022).
7. Hughes, G. The Performance of Wind Farms in the United Kingdom and Denmark; Renewable Energy Foundation: London, UK, 2012.
Available online: http://tinyurl.com/cn5qnqg (accessed on 1 September 2022).
8. Ribrant, J.; Bertling, L. Survey of Failures in Wind Power Systems With Focus on Swedish Wind Power Plants During 1997–2005.
IEEE Trans. Energy Convers. 2007, 22, 167–173. [CrossRef]
9. Tavner, P.; Xiang, J.; Spinato, F. Reliability analysis for wind turbines. Wind Energy 2007, 10, 1–18. [CrossRef]
10. Arwade, S.; Lackner, M.; Grigoriu, M. Probabilistic Models for Wind Turbine and Wind Farm Performance. J. Sol. Energy Eng.
2011, 133, 041006. [CrossRef]
11. Conroy, N.; Deane, J.; Gallachóir, B.Ó. Wind turbine availability: Should it be time or energy based?—A case study in Ireland.
Renew. Energy 2011, 36, 2967–2971. [CrossRef]
12. Staffell, I.; Green, R. How does wind farm performance decline with age? Renew. Energy 2014, 66, 775–786. [CrossRef]
13. Managing the wind: Reducing kilowatt-hour costs with condition monitoring. Refocus 2005, 6, 48–51. [CrossRef]
14. Deutsch, J.; He, D. Using Deep Learning-Based Approach to Predict Remaining Useful Life of Rotating Components. IEEE Trans.
Syst. Man Cybern. Syst. 2018, 48, 11–20. [CrossRef]
15. Yang, W.; Tavner, P.J.; Crabtree, C.J.; Wilkinson, M. Cost-Effective Condition Monitoring for Wind Turbines. IEEE Trans. Ind.
Electron. 2010, 57, 263–271. [CrossRef]
16. Pedregal, D.; García, F.; Roberts, C. An algorithmic approach for maintenance management based on advanced state space
systems and harmonic regressions. Ann. Oper. Res. 2008, 166, 109–124. [CrossRef]
17. Raouf, A. Maintenance Excellence: Optimizing Equipment Lifecycle Decision. J. Qual. Maint. Eng. 2004, 10, 75. [CrossRef]
18. Marquez, F.P.G. An Approach to Remote Condition Monitoring Systems Management. In Proceedings of the 2006 IET International
Conference on Railway Condition Monitoring, Birmingham, UK, 29–30 November 2006; pp. 156–160.
19. García, F.P.; Pedregal, D.J.; Roberts, C. Time series methods applied to failure prediction and detection. Reliab. Eng. Syst. Saf. 2010,
95, 698–703. . [CrossRef]
20. Byon, E.; Ding, Y. Season-Dependent Condition-Based Maintenance for a Wind Turbine Using a Partially Observed Markov
Decision Process. IEEE Trans. Power Syst. 2010, 25, 1823–1834. [CrossRef]
21. Yeghikian, M.; Ahmadi, A.; Dashti, R.; Esmaeilion, F.; Mahmoudan, A.; Hoseinzadeh, S.; Garcia, D.A. Wind Farm Layout
Optimization with Different Hub Heights in Manjil Wind Farm Using Particle Swarm Optimization. Appl. Sci. 2021, 11, 9746.
[CrossRef]
22. Bearing Damage and Failure Analysis. Report. Available online: https://www.skf.com (accessed on 3 March 2023).
23. Teng, W.; Zhang, X.; Liu, Y.; Kusiak, A.; Ma, Z. Prognosis of the Remaining Useful Life of Bearings in a Wind Turbine Gearbox.
Energies 2017, 10, 32. [CrossRef]
24. Kusiak, A.; Verma, A. Analyzing bearing faults in wind turbines: A data-mining approach. Renew. Energy 2012, 48, 110–116.
[CrossRef]
25. Kusiak, A.; Li, W. The prediction and diagnosis of wind turbine faults. Renew. Energy 2011, 36, 16–23. [CrossRef]
26. Herp, J.; Pedersen, N.; Nadimi, E. A Novel Probabilistic Long-Term Fault Prediction Framework Beyond SCADA Data—With
Applications in Main Bearing Failure. J. Phys. Conf. Ser. 2019, 1222, 012043. [CrossRef]
Energies 2023, 16, 2851 14 of 14

27. Wiese, B.; Pedersen, N.; Nadimi, E.; Herp, J. Estimating the Remaining Power Generation of Wind Turbines—An Exploratory
Study for Main Bearing Failures. Energies 2020, 13, 3406. [CrossRef]
28. Jonkman, J.; Butterfield, S.; Musial, W.; Scott, G. Definition of a 5-MW Reference Wind Turbine for Offshore System Development.
2009. Available online: https://www.nrel.gov/docs/fy09osti/38060.pdf (accessed on 1 September 2022).
29. Jonkman, J.M.; Hayman, G.J.; Jonkman, B.J.; Damiani, R.R.; Murray, R.E. AeroDyn v15 User’s Guide and Theory Manual, NREL.
Available online: https://www.nrel.gov/wind/nwtc/assets/pdfs/aerodyn-manual.pdf (accessed on 1 September 2022).
30. Lindenburg, C. Aeroelastic Modelling of the LMH64-5 Blade; DOWEC Dutch Offshore Wind Energy Converter 1997–2003 Public
Reports; Energy Research Center of The Netherlands: Petten, The Netherlands, 2002.
31. Jonkman, J.; Butterfield, S.; Passon, P.; Larsen, T.; Camp, T.; Nichols, J.; Azcona, J.; Martinez, A. Offshore Code Comparison
Collaboration within IEA Wind Annex XXIII: Phase II Results Regarding Monopile Foundation Modeling. In Proceedings of the
2007 European Offshore Wind Conference & Exhibition, Berlin, Germany, 4–6 December 2007.
32. The Great Power. 2005. Available online: https://evolution.skf.com/the-great-power-3/ (accessed on 1 September 2022).
33. Samani, A.; De Kooning, J.; Kayedpour, N.; Singh, N.; Vandevelde, L. The Impact of Pitch-To-Stall and Pitch-To-Feather Control
on the Structural Loads and the Pitch Mechanism of a Wind Turbine. Energies 2020, 13, 4503. [CrossRef]
34. Offshore Wind Data. Orsted.com. 2022. Available online: https://orsted.com/en/our-business/offshore-wind/wind-data
(accessed on 1 September 2022).
35. SKF. Rolling Bearings. 2018. Available online: https://www.skf.com/group/products/rolling-bearings (accessed on 1 Septem-
ber 2022).
36. Zaretsky, E. A. Palmgren Revisited: A Basis For Bearing Life Prediction; National Aeronautics and Space Administration: Washington,
DC, USA, 1997.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.