Adapting Optimal Preventive Maintenance

Strategies for Floating Offshore Wind in Atlantic

Areas by Integrating O&M Modelling and FMECA
Mitra Kamidelivand Peter Deeney Fiona Devoy McAuliffe
Environmental Research Institute Environmental Research Institute Environmental Research Institute
University College Cork University College Cork University College Cork
Cork, Ireland Cork, Ireland Cork, Ireland
mitra.kamidelivand@ucc.ie peter.deeney@ucc.ie f.devoymcauliffe@ucc.ie

Ross O’Connell Isaac Mulcahy Jimmy Murphy

OCEANS 2023 - Limerick | 979-8-3503-3226-1/23/$31.00 ©2023 IEEE | DOI: 10.1109/OCEANSLimerick52467.2023.10244469

Environmental Research Institute Civil, Structural and Environmental Civil, Structural and Environmental
University College Cork Engineering Engineering
Cork, Ireland University College Cork University College Cork
ross.oconnell@ucc.ie Cork, Ireland Cork, Ireland
jimmy.murphy@ucc.ie

Abstract— Ireland is among the countries best suited for the
floating offshore wind (FLOW) boom. The Irish government's
goal is to produce 80% of its power from renewable sources
achieving 7 GW of offshore wind capacity by 2030. This will be
aided by the significant effort that has already been made in
developing FLOW projects in Atlantic and Celtic Sea areas.
However, there are issues that must be resolved before floating
turbines can be widely deployed particularly in areas like the
Atlantic. One of these challenges is the harsh environment,
which limits the accessibility of the vessels needed for Operation
and Maintenance (O&M), especially in the winter. This paper
has conducted an O&M cost criticality assessment of a 1 GW
scale floating offshore wind (FLOW) project in two case studies
off the West and South coasts of Ireland. The consequences of
failures, and therefore the criticality of each component, are
calculated in an integrated O&M-Failure Mode, Effects and
Criticality (FMECA) model. In the model the mean time
between failure, the type of vessel and their costs, technician and
spare part costs, weather window and farm availability are
determining factors of the criticality scores of components in
each location. The ability to analyze the impact of different
Planned Maintenance (PM) strategies on the failure rate of
components is a particular functionality offered in this O&M
model. A PM strategy that directly affects O&M expenses can
help the project by improving farm availability, maintaining a
low rate of unexpected failure and lowering O&M costs as a way
to balance costs and availability. Several PM intervals are
looked at for the two Projected FLOW locations.
The importance of minor failure mode of components like
the ‘Electric component, ‘Generator leak’, ‘Pitch hydraulic
minor’, ‘Floating structure’, and ‘Tower’ is emphasized in both
places, but especially on the West coast. For the FLOW system
at both locations, the component criticality ratings are ranked
relative to one another. The total lifecycle O&M cost, including
the cost of energy losses, of a 15MW FLOW turbine over 25-
year lifetime is found €55.8 m in the West and €29.4 m in the
South. At subsystem level, ‘Structural subsystem’ typically
accounts for 31% of the cost-criticality, followed by auxiliary
subsystem with 29%, pitch hydraulic with 18%, generator with
13%, and gearbox with 9%. Implementing a semi-annual PM
for turbines, a PM every two years for substructure, and a
thorough PM every 10 years may, in the best scenario, boost the
availability of the FLOW farm in the West from 80.0% to 95%
in the best case. This resulted from a drop in remedial
interventions from 5.5 to 1.2 per turbine per year in the West
and from 6.8 to 2.8 in the South. Due to the lack of experience
with operating FLOW turbines in the sea, the input data for
modeling the O&M strategies of a large-scale FLOW farm are
somewhat uncertain. These uncertainties can be reduced when
more exact facts become available. The study did show that
scenario studies can, at this point of development, disclose some
potential PM solutions that, if correctly implemented, might
result in a large reduction in O&M expenses because of the
reduction in unexpected incidents and the rise in farm
availability. This area of research may benefit from increased
component life expectancy after PM in the model, which lowers
unexpected model intervention. The model illustrates a method
to balance the costs and benefits of a PM strategy.
Keywords— floating offshore wind, operation and
maintenance, component criticality, preventive maintenance
strategies
I. INTRODUCTION
The Floating Offshore Wind (FLOW) industry has been in
development for over a decade. The total capacity of small
scale or demonstrator FLOW projects installed to date is
around 126MW [1]. Examples of successful pilot FLOW
farms already in the water are: Hywind (6×5 MW spar-type)
by Equinor in Scotland that achieved an average capacity
factor of 57.1% in 2021; Kincardine (5×9.5 MW semi-
submersible) in Aberdeen, Scotland; and WindFloat Atlantic
(3×8.4 semisubmersible) in Portugal [2] that generated 75
GWh electricity production in the first year of operation in
2014. Currently, the world’s largest FLOW farm, Hywind
Tampen (11×8 MW) is under construction by Equinor in
Norway which expects to meet 35% of the annual power
requirement of five offshore oil and gas fields in Norway.
The installation of FLOW projects is anticipated to
increase to 20.9 GW by 2035 and to 264 GW by 2050, or more
than 3,000 times the Hywind Tampen farm [1]. The urgent
need to satisfy the high FLOW capacity demand requires a
rapid shift from pre-commercial to full-scale arrays [3]. From
a marketing perspective, it is anticipated that the global
FLOW power market will increase from €0.64bn in 2022 to
€14.65bn by 2029, growing at a compound annual growth rate
of 56% during the projected period . The potential for greater

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO DE JANEIRO. Downloaded on April 19,2024 at 12:51:09 UTC from IEEE Xplore. Restrictions apply.

979-8-3503-3226-1/23/$31.00 ©2023 IEEE

 turbine sizes and access to a significantly larger maritime
region with high-quality wind resources are driving factors in
the market. Fixed offshore wind, like onshore wind, has
limited space to develop alongside the needs of other users of
the sea, but FLOW significantly expands the available range.
In comparison to fixed offshore, FLOW allows access to
higher and more reliable wind speeds farther offshore. Ireland
is one of the most suitable countries for the FLOW boom. In
Ireland there has already been considerable work at
developing FLOW projects in Atlantic and Celtic Sea
locations which will contribute to the Irish government’s
target of generating 80% of its electricity from renewable
sources and achieving 7 GW offshore wind capacity by 2030.
However, there are challenges that need to be addressed
before the large-scale rollout of floating turbines,
specifically, in regions like the Atlantic. The primary
problem is the extreme environment, which places
restrictions on the accessibility of vessels required for
Operation and Maintenance (O&M), particularly during the
winter. These limitations also represent a major risk of
revenue loss from extended turbine outages (downtime) and
potential asset damage, making the annual O&M costs
uncertain. As an illustration, in the NREL reference (57×8)
MW submersible FLOW farm, the O&M can account for
27.5% of the Levelized Cost of Energy (LCoE) [4].
Increased reliability (i.e., fewer unexpected failures) and
lower maintenance costs, for example, by utilizing efficient
maintenance strategies, can both reduce O&M costs.
Numerous O&M tools have developed to examine the
availability, energy production and maintenance costs of
offshore wind energy. Reviews of the existing O&M tools
can be found, for example, in [5]–[7] . A probability density
function (PDF) of the time to failure derived from historical
data or a mean time between failures (MTBF) derived from
observed failures are typically used in failure modelling. The
model then generates random simulations to determine when
failures occur [8].
To determine if it is feasible to implement acceptable
maintenance procedures, it is also vital to be aware of the most
significant failure modes of the FLOW subsystem as well as
their associated failure causes [9]. Some recent studies have
used the Failure Mode, Effect and Criticality Analysis
(FMECA) technique [10], for example, to rank systems for
which condition monitoring would produce the greatest value
for offshore wind systems [9] or to prioritize FLOW
components based on both the risk of failure and the risk
priority number [11]. Reference [12] expanded the traditional
FMECA methodology to examine support structure failures in
FLOW turbines. The investigation led to certain
recommendations for maintenance measures intended to
guarantee the efficient and secure operation of support
structures. For qualitative and quantitative failure evaluations
of semi-submersible FLOW, [13] used the full tree analysis
(FTA) method. It has been demonstrated that in the marine
environment, salt spray, and strong winds have the biggest
effects on FLOW turbines.
The impact of lost energy revenue due to failure and
associated repair times could be significant for offshore wind
turbines situated far from the shore. For instance, for a 500
MW offshore wind farm in the North Sea, [14] estimated that
the annual output losses from downtime might be as much as
high as €12m. The impacts of the lost energy revenue on the
cost critically number of components of fixed turbines have
Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO DE JANEIRO. Downloaded on April 19,2024 at 12:51:09 UTC from IEEE Xplore. Restrictions apply.
been examined in an O&M-based cost criticality study [15],
[16]. These studies, however, have not clearly modelled the
utilization of various vessels, their operating limits and the
weather window waiting time, which are greatly influenced by
local weather conditions, failure types and distance to the port.
This study tries to contribute to this knowledge gap for FLOW
turbines.
The purpose of the study is to evaluate the criticality of
each FLOW subsystem for two case studies for FLOW
turbines on Ireland’s West and South coasts with an installed
capacity of 1GW using an O&M-based FMECA approach.
The ability to analyse the impact of different Planned
Maintenance (PM) strategies on the failure rate of components
is a particular functionality offered in this O&M model. This
is based on the idea that PM can lengthen the lifespan and
lower the failure rates of components. The final goal is to
evaluate both the O&M costs and availability in order to
identify possible optimal PM intervals for each case study,
leading to lower O&M costs and higher availability.
The rest of this paper is divided into the following
sections: Section II explains the study’s methodology, data
and assumptions. The findings are presented and discussed in
Section III, and the study is concluded in Section IV with ideas
for other research.
II. METHOD
The study uses an integrated offshore O&M model and
FMECA approach to investigate optimal PM strategy
intervals which can improve the wind farm availability and
reduce O&M costs (FIGURE I). The O&M model uses hourly
time-series weather data for locations in the West Coast and
South Coast of Ireland; these are applied to case studies of 1
GW scale for a semisubmersible FLOW project. The power
curve and dimensions of the 15 MW wind turbine are assumed
to be similar to the reference model of IEA Wind 15MW [17]
at the rated wind speed of 10.56 m/s, hub height of 150m and
rotor diameter of 240m [18]. Assuming a staggered farm
configuration and around 2 km distance between turbines
(~8D), the farm comprises an area of around 345 km2 for
spacing 66×15 MW FLOW turbines. The catenary system of
mooring lines comprising of steel wire and chain connects the
FLOW turbine to the sea bed held by three drag embedded
anchors [19]. Keep your text and graphic files separate until
after the text has been formatted and styled. Do not use hard
tabs, and limit use of hard returns to only one return at the end
of a paragraph. Do not add any kind of pagination anywhere
in the paper. Do not number text heads-the template will do
that for you.
The model uses the input data to calculate the mean annual
energy production, repair costs including the costs of vessels
(hiring and fuel), technicians and spare-part/material costs, as
well as components failure counts, downtime, income loss
from down energy and the availability of turbines over the 25
year project lifetime for a large number of simulations. In the
model, each component’s failure time follows a Weibull
distribution which is one of the standard methods of failure
distribution used in many systems including engineering
systems .The two parameters of the distribution, shape (k) and
scale (λ) are calculated using Eq. 1 and Eq. 2.
MTTF= λΓ (1+ (1 )/k) Eq. 1
CDF=1- e^(-(x/λ)k) Eq. 2
 where MTTF is mean time to fail (an inverse of a 2.9 m in the west and 9.6 m/s and 1.7 m in the south,
component’s failure rate), CDF is the cumulative distribution respectively.
function and x is time. TABLE I. SPECIFICATION OF SITES & FLOW FARMS

Item FLOW South Ireland FLOW West Ireland

Approx. 51.5N, 7.9W 52.7N, 10.2W
coordinates (site
centre)
Port distance 35 (Cork – Ringaskiddy) 78 (Shannon – Foynes)
(km)
Grid connection 31 (Aghada 220 kV S/S) 44 (Moneypoint 400 kV
distance (km) S/S)
Depth (m) 77 – 93 87 – 103
Seabed Coarse Substrate, Sand, Coarse Substrate, Sand,
character (Folk- Muddy Sand Muddy Sand
7) Sandy Mud
Wind data Data point 42284 Data point 47215
(1996-2005) at (Copernicus Climate Change (Copernicus Climate
10m above sea C3S) Change C3S)
surface
Method Hindcasted data and forecast data based on future trends
Significant Data point 252054, ERA-5 Data point 254222,
wave height reanalysis from the European ECMWF ERA5 Reanalysis
(1996-2005) centre for Medium-Range dataset (ERA5-ECMWF
FIGURE I. A HIGH-LEVEL DESCRIPTION OF THE Weather Forcecasts
MODEL (ECMWF)
Method A mixture of modelled data and satellite observations
The two equations above may be solved numerically
assuming that there is a 10% chance of component’s failure
by half of its MTTF, so that CDF = 0.1 and x=MTTF/2, B. Reliability and maintenance data
yielding the shape factor of 2.7801, which is consistent with Knowledge of FLOW turbines’ reliability is still at an
other studies, e.g. [20]. The scale factor for each component is early stage, but some literature has provided failure data (such
calculated using the MTTF of the specific component. The as the failure ratios of various failure modes and failure
Weibull distribution and Monte Carlo method determine the mechanisms) based on reliability analyses like fault tree
outcomes for the proposed 1GW FLOW wind farm in terms analysis and failure mode and effect analysis, for instance
of mean number of failures and the other costs mentioned [11], [13], [21]. The categories of subsystems and components
above. are based on [21] which offers extensive data on failure modes
A specific functionality included in this O&M model is the and failure mechanisms for FLOW. The failure rates found in
ability to analyse the effect of various Planned Maintenance [21] have undergone several modifications. It is assumed that
(PM) strategies on the failure rates of components. This is the tower and floating offshore wind structures will be well-
based on the assumption that PM can improve components life engineered to endure the harsh offshore environment,
expectancy and reduces their failure rates. The outcome of the therefore catastrophic failure such as collapses are not
PM can be defined for each component by setting a value envisioned. The FLOW mooring lines are presumptively
between 0 (no effect) and 1 (perfect effect, meaning the based on the DNV-required standard, which states the failure
component is as good as a new one). This value will affect the rate should be between 1-2%. Also, the failure rate for
MTTF for the component after the PM is carried out. The PM anchors (a component of the support subsystem), and cables
tasks may be set at any frequency. The start month may be set (an electrical component of the auxiliary subsystem) are
to make best use of the weather (e.g. summer). A single PM added from [2] and [22].
visit to a turbine may affect one or more components. A The failure modes are divided into three categories: minor
similar analysis of costs is produced for the PM as for repair, major repair, and major replacement based on the ratio
corrective maintenance (CM) mentioned above. The costs of of failure mechanisms (failure causes) in [21], the data from
PM tasks including the vessel, technician and spare-part costs [23], as well as expert judgements. For example, the gearbox
along with PM time and the effect of the PM strategy on the failure caused by tooth wear and gear cracks is categorized as
reduction of unexpected failures for each scenario will be a “major replacement”, but a lubrication failure caused by
modelled. dirty oil is categorized as a “minor repair”. The numbers of
A. Location specification data technicians for minor repair, major repair and major
replacement are assumed 2, 5 and 7, respectively. The average
A summary of the study’s location data is provided in hourly cost for a technician is €200/hr, whereas the hourly cost
TABLE I. The Copernicus Climate Change database1 and for a diver is 2000€/hr.
ECCMWF ERA52 are used to obtain a 10-year time series of
wind and wave data with a resolution of one hour. The With certain modifications for the FLOW situation, for
locations' respective average wind speeds at the hub height of instance, estimated higher spare costs relevant to turbine
the turbine and significant wave height (Hs) are 10.9 m/s and investment costs for a 15 FLOW device, the information on
the costs of interim repairs and spare parts is primarily based

1 2
https://cds.climate.copernicus.eu/#!/home https://www.ecmwf.int/en/forecasts/dataset/ecmwf-
reanalysis-v5

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO DE JANEIRO. Downloaded on April 19,2024 at 12:51:09 UTC from IEEE Xplore. Restrictions apply.
 on [23]. A 15MW FLOW turbine will cost an estimated €14 C. Vessel Data
million after 2030 [24]. According to Catapult and BVG The model takes into account seven different types of
association ; the investment cost of a fixed 10 MW turbine is vessels for maintenance tasks, whether corrective or planned:
£1000 per MW (in 2021). If a 15 MW FLOW turbine
investment costs €15 million, the mean annual material and • Crew Transfer Vessel (CTV) which severs as primary
spare-part cost for all minor repairs and major repairs in maintenance vessel for majority of the models’ CM
TABLE I would be 0.2% and 2.6% of the FLOW turbine and PM duties.
price, respectively. It is assumed that minor repair, major
repair, and major replacement will each require spare parts • Helicopter for certain of PM activities requiring light
that cost less than €10,000, up to €90,000, and more than equipment and few technicians.
€90,000. • Service operations vessel (SOV) only for some major
The replacement costs for mooring, cables and anchors in maintenance with offshore mean time to repair ≥22
TABLE I are only approximations. An estimation of the cost hours (TABLE I).
of the anchor, inter-array cables, and export cables has been • Cabling Laying Vessel (CLV) for major repair of
made based on [19]. Estimated from [2] is the cost of the spare cables.
parts needed to replace mooring lines (a steel and chain
combination). Using the information and presumptions from • Anchor Handling Tug Support (AHTS) vessel for
the aforementioned literature, the reliability and maintenance major replacement of mooring lines and anchors.
data of the FLOW components is compiled in TABLE II. The
• Tugboats for onshore PM retrofitting and major
FLOW turbine’s annual failure rate in the study is estimated
replacement tasks including replacing gearbox and
to be 7.7, with more than 85% of those failures being minor.
blades. As an alternative, it is presumable that a
TABLE II. COMPONENTS RELIABILITY & REPAIR DATA climbing crane and a small barge will perform
offshore PM retrofitting in place of tugboats.
Subsystem/ ID Failure Mode MTTR Material
Component # (FM) Failure Failure causes (FC)
/year
(hr) € • Diving Support Vessel (DVS) along with CTV for the
1 Tower 0.96 Strong wave/wind 7 8,000 inspection or doing minor maintenance of mooring
2 Floating 0.11 22 60,000 lines, cables and anchors (e.g. maintaining joints,
platform Storm, lightning,
major welding defects snagging, etc).
3 Floating 0.86 7 7,000
Substructure/ The specification of the vessels assumed in the study are
platform Strong wave/wind
Support
structure
4 Mooring 0.09 12 420,000 summed up in TABLE II using data from the literature [2] [7]
lines failure Mooring lines broken
major a
(e.g., wear) and [25]–[28], with few alternations and reasonable
5 Mooring 0.06 Friction Chain wear, 10 7,500 estimations, such as increasing the cost of CTV with
lines failure Other
6 Anchors 0.16 12 110,000 gangways, adding mobilization times and costs where
failure appropriate.
majora Material failure
7 Hydraulic 0.40 26 40,000 TABLE III. VESSEL DATA SPECIFICATIONS
system
failure major Over Pressure
8 Hydraulic 0.84 4 4,000
Vessel Speed Hs Wind Mob. Mob. Average Average
system (number) knots m m/s hour cost € rate Fuel l/h
a
Pitch failure Oil failure, Other €/day (Transit/
9 Alarm 4 3,000 Operation)
facility Lighting protection, CTV (10) 22 1.75 20 24 30,000 6,000 380/230
failure 0.26 Limit Switch
10 Wrong blade 85 b
600,000 SOV (1) 12 2.5 25 72 150,000 30,000 900
angle majora 0.02 Wrong blade angle CLV (1) 3 2 25 144 875,000 175,000 780/1560
11 Bearing 0.28 Corrosion, Wear, 6 4,500 AHTS (1) 10 1.5 25 144 480,000 96,000 1046
failure Other
Helicopter 50 4 15 24 - 7,000 500
12 Gear failures 0.08 Vibration, Wear, 72 b 350,000
Gearbox
majora Crack, Pitting (1)
13 Lubrication 0.04 Abnormal filter, Dirty 5 1,000 DSV (1) 17 1.5 15 - - 4,000 50
failure oil Tugboat 5 2 15 144 350,000 70,000 800/1100
14 Leak 0.93 Leak 5 4,000 (A pair)b
15 Measurement 0.18 6 2,500 a
The number of CTVs, was determined by a preliminary analysis
facilities Measurement
Generator b
Similar specification if climbing Crane and small barge are used for PM
failures facilities failure
16 Bearing 0.10 60 90,000
failure major Asymmetry
17 Electric 1.06 6 5,500
components D. Planned maintenance data
failure Short circuit, dirt, etc.
18 Blade system 0.49 6 2,500 In terms of O&M, a PM strategy is required to ensure the
failure Hub damage, Other best performance of a turbine, including the extension of its
19 Yaw 5 5,000
Auxiliary subsystem useful life. However, PM tasks come with expenses and
failure 0.49 Abnormal vibration downtime. The study's assumptions for the PM costs data in
20 Speed train 0.09 Break failure and 28 25,000
failure major other TABLE IV are based on expert opinions. For the ideal PM
21 Cable 0.05 20 2,100,000 (perfect PM), it is envisioned that the component would
majorc/a Harsh condition
22 Cable 0.13 Minor damage 10 9,500
improve to the level of a brand-new component following PM.
a.
Major replacement In the imperfect PM, it is presumable that the component's
c.
Onshore repair, 12 hours for the offshore preparation work, if the device is towed back to the port
b.
The South Cost has a 32% and a 19% lower annual failure rate for cables and its spare-part cost
health will significantly improve after PM, but it won't be as
excellent as a brand-new component. A factor of 0.8 is
presumed for the study's imperfect PM, meaning that the

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO DE JANEIRO. Downloaded on April 19,2024 at 12:51:09 UTC from IEEE Xplore. Restrictions apply.
 component’s expected MTTF is 80% of that of a new results in a downtime that is substantially longer (on average
component. 279%) than the South coast.

TABLE IV. ASSUMPTIONS OF PM COST DATA TABLE VI. MEAN O&M COSTS PER COMPONENT PER TURBINE
OVER 25-YEAR FOR THE SOUTH BASE
Interval Semi-annual Annual Every-10-
CTV/Helicopter CTV year Comp. Down Down Incident O&M Total Cost
Tugboat/ ID # Time Energy Count Cost Incl.
Climbing Hour MWh € Energy
Crane €
Technician 5 5 10 1 1,013 8,192 22.07 1,141,524 1,960,709
Repair hour 10 20 80 2 214 1,544 2.04 669,991 824,419
Spare part 15,000 30,000 300,000 3 902 7,302 19.65 1,091,111 1,821,292
€/PM/Turbinea 4 321 2,312 1.65 1,726,681 1,957,841
% of Turbine price 0.1 0.2 2.0 5 114 947 0.98 89,991 184,665
a.
For the PM, no major replacement of substructure component is considered; CTV
and DSV are used, and the PM cost is approximated to be 17,000 for every-2-year
6 631 4,519 3.28 2,384,541 2,836,458
PM, with 4 technicians working on the task for 10 hours. 7 1,033 7,425 8.84 2,417,490 3,159,971
8 1,027 8,259 19.27 892,486 1,718,351
9 298 2,401 5.56 252,098 492,229
III. RESULTS AND DISCUSSION 10 14 98 0.04 77,170 86,931
11 358 2,847 6.11 325,583 610,323
A. O&M-based criticality score 12 444 3,145 1.45 2,195,629 2,510,136
The O&M outputs for each component are calculated over 13 26 211 0.47 21,008 42,121
14 1,187 9,568 21.25 1,001,207 1,957,969
the project life time using 100 simulations. The average results 15 224 1,800 3.81 176,965 356,945
for the 25-year lifetime of each turbine for the West case study 16 325 2,351 1.81 777,388 1,012,459
and the South case study are displayed in TABLE IV and 17 1,429 11,475 24.34 1,237,579 2,385,049
TABLE V, respectively. 18 644 5,179 10.97 509,293 1,027,170
19 614 4,958 10.99 526,852 1,022,657
20 189 1,354 1.56 431,579 566,967
TABLE V. MEAN O&M COSTS PER COMPONENT PER TURBINE 21 149 1,056 0.77 2,197,535 2,303,161
OVER 25-YEAR FOR THE WEST BASE 22 296 2,438 2.61 295,389 539,144
All 11,452 89,379 170 2,0439,090 29, 376,966
Comp. Down Down Incident O&M Total Cost
ID #a Time Energy Count Cost Incl.
Hour MWhb € Energy The availability of the FLOW farm in the west is only
€ 80%, while it is 95% for the same farm in the south (derived
1 4,254 37,970 18.0 967,308 4,764,278 from the downtime to total time ratio). Shorter port distance
2 371 3,186 1.6 555,378 873,938 and lower site accessibility are two important reasons that
3 3,781 33,765 16.0 922,358 4,298,824 account for the West's lower availability compared to the
4 1,281 10,785 1.3 1,562,146 2,640,676 South. Port distance for the example study on the West coast
5 597 5,199 0.7 69,810 589,726
is 78 km, compared to 35 km for the South coast. For example,
6 2,617 22,087 2.7 2,174,671 4,383,370
7 1,809 15,562 7.3 2,078,009 3,634,199
if a CTV traveling at 41 km/hr is required to complete 100 CM
8 4,287 38,229 15.7 760,671 4,583,612 jobs, the equivalent downtime (excluding weather window
9 1,204 10,698 4.6 215,889 1,285,669 waiting time) would be up to 4% in the West, which is twice
10 17 142 0.02 39,864 54,103 as much as in the South. Only 26% of the West coast is
11 1,511 13,412 5.0 275,949 1,617,179 reportedly accessible year-round for a 6-hour window gap and
12 950 8,114 1.1 1,875,779 2,687,184 a significant wave height of 2 meters [29].
13 88 766 0.3 14,947 91,576
14 5,027 44,676 17.4 852,804 5,320,417 Without considering the cost of energy lost due to
15 938 8,314 3.1 149,550 980,943 downtime, in both case studies, the first five critical
16 656 5,692 1.4 637,249 1,206,414 components to fix that account for more than 50% of the
17 6,219 55,346 19.8 1,046,974 6,581,578 expenses are the "Anchor major", "Pitch hydraulic major",
18 2,765 24,575 9.0 434,753 2,892,294
"Cable major", "Gearbox major", and "mooring major".
19 2,564 22,708 9.0 448,792 2,719,576
20 322 2,764 1.2 358,853 635,260 When the costs of lost energy are taken into account, the
21 406 3,526 0.6 1,904,391 2,256,996 importance of minor repairs of components like the "Electric
22 1,686 14,891 2.1 243,288 1,732,415
component", "Generator leak", "Pitch hydraulic minor",
Allc 43,352 382,408 138 17,589,433 55,830,227
a.
Component ID is explained in TABLE II "Floating structure" and "Tower" is underlined in both
c.
b.
Electricity rate is assumed €100/MWh locations but specially in the West coast. The relative ranking
Total expenses may be varied because of rounding
. of component criticality ratings for the FLOW system at both
The following are the key findings from the O&M-based locations is shown in FIGURE II using a normalized O&M
criticality results in TABLE IV and TABLE V: cost criticality score. The illustration amplifies the
significance of minor failures in both contexts, but more so in
The mean yearly incidents (i.e. corrective repairs), are 364 the West coast. In the West and the South, respectively, 67%
in the West and 448 in the South. The reason for the West and 48% of the total O&M expenses could be attributed to all
coast’s lower incident rate is that there is a longer wait time minor failure modes. Their contribution is roughly 36% in
for repairs, mostly because of location accessibility, which both sites even without accounting for energy loss expenses.

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO DE JANEIRO. Downloaded on April 19,2024 at 12:51:09 UTC from IEEE Xplore. Restrictions apply.
 This underlines once more how crucial it is to lower incidence
of unanticipated minor failures.
On average, the West and South coasts could see annual
output losses of €101,975 per MW and €23,834 per MW,
respectively. In a 500 MW fixed offshore wind farm situated
50 km offshore in the North Sea, the average annual
production loss is stated to be €24,000/MW [14]. While the
lost production cost due to repair activities in the South case
study is comparable with [14], the production loss for the West
coast is noticeably greater, underscoring the fact that these
outputs are very site dependent.
At the subsystem level (see their assemblies in TABLE
I),’Structural and substructural subsystem’, on average,
accounts for 31% of the cost-criticality score, followed by
‘Auxiliary subsystem’ by 29%, ‘Pitch hydraulic’ by 18%,
‘Generator’ by 13% and ‘Gearbox’ by 9%.
The O&M-based criticality results of this study and those
in the literature that has explicitly used the O&M modelling to
designate the most expensive components as critical, for
example [15][16], show some similarities and some
discrepancies. Some of the variations have to do with whether
the study includes metocean data from particular sites and
vessel types. Some of it relates to the components that are
included by the study’s scope. For the FLOW turbines, for
instance, the distribution of the criticality score changes when
mooring, anchoring, and floating substructure are added.
Factors such as anchor and mooring complexities and the
turbine motion directly affect the O&M of FLOW [8].
FIGURE II. COMPARATIVE O&M-BASED-CRITICALITY SCORES FOR
COMPONENTS FAILURE MODES
In [15], the gearbox followed by generator and pitch
hydraulic system all received the highest cost severity scores.
In [16], the ‘Structure’ (including nacelle and foundation)
followed by ‘Gearbox’ and ‘Rotor-blade’ (blade and hub) are
identified as the costly critical components for the fixed
offshore wind of 2-3 MW with capacity factor of 0.30 3 .
However, neither of the weather window waiting time at the
sites, nor the vessel characteristics have been specified in their
analyses. In their studies, the critical expensive components
are those with the highest repair failure rate, time to repair,
3
The maximum capacity factor for the 15 MW FLOW turbine employed in
this study is 0.60, which represents the proportion of theoretical energy
generated over a certain time period to the maximum power output.
Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO DE JANEIRO. Downloaded on April 19,2024 at 12:51:09 UTC from IEEE Xplore. Restrictions apply.
spare-part costs, and technician expenses. Despite this, our
findings in FIGURE II show that the expensive components
to repair such as “Gearbox”, “Pitch hydraulic”, and “Cable”
are still gaining a high criticality score in the South case study,
which has a milder environment than the West. However,
minor failures of “Pitch hydraulic” and structural component
such as “Tower” and “Floating structure” can greatly
influence the O&M costs.
Making a trade-off between boosting farm availability,
maintaining a low rate of unexpected failure, and expenses is
essential. Evidently, the solution for these lies in increasing
component reliability. For example, upgrading to more
expensive and higher-quality components, adding
redundancies and implementing condition-based monitoring
and Supervisory control and data acquisition (SCADA) all
result in an increase in Capital Expenditure (CAPEX). The
adaption of appropriate PM strategies, which can also offer a
solution to balance costs and availability, has a direct bearing
on O&M expenses. The scenario analysis discussed in the next
part are focused on this solution.
B. PM scenario analysis
Studies have shown that offshore wind structural
components should be inspected at least once a year, while
submerged components such as cables, mooring, and other
structural systems below the water should be reviewed at least
every five years [29]. Inspection, alterations, cleaning,
painting, lubrication, repairs, and replacements are all
considered to be part of the preventive maintenance in the
study (see the PM cost data in section D).
According to TABLE VI, case studies have been
examined in various PM scenarios for West and Coast sites
based on the hypotheses of the PM impact on the components'
time between failure as well as based on the results in the
previous section. Perfect PM scenarios (S1a-S2a) in TABLE
VI mean that the component is as good as brand new after PM
while in partial (imperfect) PM scenarios (S1b-S2b)
components are like new but have 80% of their original mean
time to failure as a result of PM. Analysis is also done on the
no-tow-to-shore PM scenarios (S1a', S2a', S1b', and S2b')
where climbing cranes and supporting vessels will do the
same PM jobs. For each PM scenarios, availability and O&M
expenses of CM and PM are analysed. In comparison to the
Base case studies, the outcomes of the PM scenario are
displayed FIGURE III and FIGURE IV.
The paper recommends the following PM intervals for the
case studies: a semi-annual PM for turbines, a PM every two
years for substructures including mooring, anchoring, and
cable, and a PM offshore for retrofitting every ten years for
the West case study. Even in the summer, the CTVs in the
West are unable to complete all of the PM tasks, necessitating
the use of a helicopter. The PM intervals for the South are the
same as those for the West, with the exception that a yearly
PM of turbines is sufficient. The important findings derived
from the outcomes of the PM scenarios and the outcomes
shown in FIGURE III and FIGURE IV are discussed below.
Depending on the PM intervals specified in TABLE VI,
under perfect PM scenarios, the O&M costs for the West and
South projects might be reduced significantly by 37-64% and
 27-57%, respectively. This substantial decrease in O&M insurance, and maintenance of onshore and offshore
expenses is the result of the failure rates falling by 53-63% in substations, can range from 33 to 41% of the overall operation
West and 40-58% in the South, or by an average reduction of and maintenance costs for a 1 GW offshore wind project,
5.5 to 1.2 corrective interventions per turbine per year in the according to Catapult BVG's cost estimations. All these
West and 6.8 to 2.8 in the South. expenditures must be taken into account when modelling
TABLE VII. DESCRIPTION OF PM SCENARIOS
OPEX in a future study. The OPEX, CAPEX, and LCoE of a
proposed 1-GW FLOW farm on the West coast of Ireland are
PM scenarios a
further discussed [29].
PM Intervals and vessles S1a S1a’ S2a S2a’ In the best-case scenario (S1b, S2b in perfect PM), the
Intervals and vessels S1b S1b’ S2b S2b’
Annual (CTV) × ×
availability of the South case study can reach 97.2% and the
Semi-Annual (CTV & × × availability of the West farm can rise by up to 94.5%. These
Helicopter) availability levels appear to be strong for the FLOW turbines
Every 2-yr substructure × × × × even though they would drop to 92.1% in the West and 96.2%
(CTV & DSV) in the South for the imperfect PM (S1b', S2b'). According to
Every 10-yr Retrofit × × the analysis by [30] the availability for the FLOW case studies
(Tugboat)
of the 30 MW Hywind and 47.5 MW Kincardine in Scotland
Every 10-yr Retrofit × ×
(Climbing Crane) was 92.8% and 93.9%, respectively. However, there is no
a
Scenarios ending at (b) are for imperfect PM, which means the component has optimization in their research. The 15MW FLOW turbines
80% of the anticipated lifetime of a brand-new component.
b
In Scenarios with prime signs (') at the end, all PM duties will be completed offshore.
used in the study have a potential annual output of 67.1 GWh
and 78.6 GWh of power in the South and West, respectively.
A 1% increase in the availability of a 1-GW farm could
generate gross revenues of 59.2 thousand euros in the West
FIGURE III. PM SCENARIOS COMPARED TO BASE CASE STUDY and 50.6 thousand euros in the South, assuming an electricity
IN WEST COAST
rate of 100€/MWh. Once more, the emphasis here is on
increasing the FLOW's availability.
IV. CONCLUSION AND FUTURE STUDY

The primary motivations for this study were: the fact that
Ireland has plenty of deep-water sites suitable for FLOW
development along its South and Wwest coasts; that access to
FLOW turbines for O&M jobs is restricted by the abrasive
environment of the Atlantic region, particularly in the West;
FIGURE IV. PM SCENARIOS COMPARED TO BASE CASE STUDY and that and research was needed to offer a feasible approach
IN SOUTH COAST for lowering unanticipated failure occurrences and O&M
expenses, while raising energy output.
An O&M-based criticality model was developed. The
FLOW critical parts that pose the largest O&M expenses and
revenue losses were discovered. The model examined various
PM strategies and gave comparative O&M costs for both
corrective and planned maintenance actions for perfect and
imperfect PM scenarios. It pinpointed the optimal PM
scenarios that can lower O&M expenses and boost farm
availability.

This analysis models a pessimistic impact of PMs on the
component's MTBF based on the assumption that 85% of
component failures are minor (TABLE II). Comparing the
effects of the perfect and imperfect PMs on the cost of O&M
and availability in FIGURE III and FIGURE IV, reveals that
the impact on the decrease in O&M cost and increase in
availability in the imperfect PM (with a 0.8 factor) becomes
smaller by, on average, 18% and 2% in both case studies. This
can provide insight into the areas where the favorable trade-
off is lost. In [4], it is anticipated a 53% reduction in OPEX
from advanced O&M strategies, improved vessel
accessibility, and remote maintenance strategies. The model
developed in this study can be used to investigate more
maintenance strategies in the future. It should be emphasized
that the variable O&M costs measured in the study only take
into account the expenses of servicing and maintaining the
structural and substructural parts of the turbine. The remaining
OPEX costs, which include the operation, administration,
The simulation results show that minor failures of
components can account for 67% and 48% of the total O&M
expenses, in the West and South coasts respectively. Even
without accounting for energy loss expenses, they still
represent around 36% of the costs in both sites. Under perfect
PM scenarios (i.e. components become as good as new ones
after PM), substantial reduction of the O&M costs of 37-54%
in West and 27-57% in South can be achieved. Availability of
the West coast case study can increase from 80% in the base
case to 95% in the best scenario. All these advantages are due
to the reduction of 5.5 to 1.2 corrective interventions per
turbine per year in the West and 6.8 to 2.8 in the South.
It is obvious that location has a large role in the O&M
metrics like costs and downtime. The yearly energy losses for
the South and West coasts might be in the range of €23,834
and €101,975 per MW, respectively. Due to prolonged
exposure to severe weather conditions, FLOW turbines
operating farther from the shore in rougher seas may face

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO DE JANEIRO. Downloaded on April 19,2024 at 12:51:09 UTC from IEEE Xplore. Restrictions apply.
 severe challenges in terms of accessibility and likelihood of
failure. To examine the likelihood of higher failure owing to
the weather at the project site, more research is required.
To discuss the various inputs to the model, additional
sensitivity analyses are necessary. For instance, it may take
longer to reach the turbine because of the FLOW turbines'
motion, lengthening the operation's offshore phase. According
to a basic sensitivity analysis carried out for this study, a 20%
increase in the time taken for offshore activities can decrease
the availability of FLOW farms by about 2.9% in the West,
but only by 0.4% in the South. Because there is minimal
experience with the operation of FLOW turbines in the sea,
there are some uncertainties in the input data for modelling the
O&M strategies of a large-scale FLOW farm. As more precise
data become available, these uncertainties will be reduced.
However, the study’s scenario analyses have demonstrated
that at this stage of development, there are some viable PM
solutions that, if properly implemented, could lead to a
significant decrease in O&M costs due to a decrease of CM
jobs and increase in farm availability. The reduction of
corrective failures as a result of PM tasks, as modelled by this
study, can offer assistance to researchers in O&M of marine
energy.
ACKNOWLEDGMENT
This research was funded by Science Foundation Ireland
(SFI) through MaREI, the SFI Research Centre Energy,
Climate, and Marine [under Grant No 12/RC/2302_ P2].
REFERENCES
[1] DNV, “Floating offshore wind: The next five years,” 2022.
[2] A. Garcia-Teruel, G. Rinaldi, P. R. Thies, L. Johanning, and H. Jeffrey,
“Life cycle assessment of floating offshore wind farms: An evaluation
of operation and maintenance,” Appl. Energy, vol. 307, no. November
2021, p. 118067, 2022, doi: 10.1016/j.apenergy.2021.118067.
[3] Timmington D. and Efthimiou L., “Mooring Systems for Floating
Offshore Wind: Integrity Management Concepts, Risks and
Mitigation,” World Forum Offshore Wind, 2022, [Online]. Available:
https://wfo-global.org/reports/.
[4] T. Stehly and P. Duffy, “2020 Cost of Wind Energy Review,” NREL
Rep., no. December, p. 68, 2021.
[5] A. Kolios and B. Feargal, “ROMEO porject. Deliverable report, D8 .
1 : Review of existing cost and O & M models , and development of a
high- fidelity cost / revenue model for impact assessment.,” no.
745625, 2020.
[6] H. Seyr and M. Muskulus, “Decision support models for operations and
maintenance for offshore wind farms: A review,” Appl. Sci., vol. 9, no.
2, 2019, doi: 10.3390/app9020278.
[7] J. McMorland, M. Collu, D. McMillan, and J. Carroll, “Operation and
maintenance for floating wind turbines: A review,” Renew. Sustain.
Energy Rev., vol. 163, no. May, p. 112499, 2022, doi:
10.1016/j.rser.2022.112499.
[8] J. McMorland et al., “A review of operations and maintenance
modelling with considerations for novel wind turbine concepts,”
Renew. Sustain. Energy Rev., vol. 165, no. May, p. 112581, 2022, doi:
10.1016/j.rser.2022.112581.
[9] M. N. Scheu, L. Tremps, U. Smolka, A. Kolios, and F. Brennan, “A
systematic Failure Mode Effects and Criticality Analysis for offshore
wind turbine systems towards integrated condition based maintenance
strategies,” Ocean Eng., vol. 176, no. October 2018, pp. 118–133,
2019, doi: 10.1016/j.oceaneng.2019.02.048.
[10] FMECA, “Reliability Analysis Centre, 1993. Failure Mode, Effects
and Criticality Analysis (FMECA). CRTA-FMECA.”
[11] M. Shafiee, “Failure analysis of spar buoy floating offshore wind
turbine systems,” Innov. Infrastruct. Solut., vol. 8, no. 1, pp. 1–19,
2023, doi: 10.1007/s41062-022-00982-x.
Authorized licensed use limited to: UNIVERSIDADE FEDERAL DO RIO DE JANEIRO. Downloaded on April 19,2024 at 12:51:09 UTC from IEEE Xplore. Restrictions apply.
[12] H. Li, H. Diaz, and C. Guedes Soares, “A developed failure mode and
effect analysis for floating offshore wind turbine support structures,”
Renew. Energy, vol. 164, pp. 133–145, 2021, doi:
10.1016/j.renene.2020.09.033.
[13] J. Kang, L. Sun, and C. Guedes Soares, “Fault Tree Analysis of floating
offshore wind turbines,” Renew. Energy, vol. 133, pp. 1455–1467,
2019, doi: 10.1016/j.renene.2018.08.097.
[14] G. Rinaldi, P. R. Thies, and L. Johanning, “Current status and future
trends in the operation and maintenance of offshore wind turbines: A
review,” Energies, vol. 14, no. 9, 2021, doi: 10.3390/en14092484.
[15] D. Cevasco, M. Collu, and Z. Lin, “O&M Cost-Based FMECA:
Identification and Ranking of the Most Critical Components for 2-4
MW Geared Offshore Wind Turbines,” J. Phys. Conf. Ser., vol. 1102,
no. 1, 2018, doi: 10.1088/1742-6596/1102/1/012039.
[16] N. Tazi, E. Châtelet, and Y. Bouzidi, “Using a hybrid cost-FMEA
analysis for wind turbine reliability analysis,” Energies, vol. 10, no. 3,
2017, doi: 10.3390/en10030276.
[17] E. Gaertner, J. Rinker, L. Sethuraman, B. Anderson, F. Zahle, and G.
Barter, “IEA Wind TCP Task 37: Definition of the IEA 15 MW
Offshore Reference Wind Turbine,” Tech. Rep. NREL/TP-5000-75698
March 2020, pp. 1–44, 2020, [Online]. Available:
https://www.nrel.gov/docs/fy20osti/75698.pdf.
[18] H. Y. Tong, T. Y. Lin, and S. W. Chau, “Normal Operating
Performance Study of 15 MW Floating Wind Turbine System Using
Semisubmersible Taida Floating Platform in Hsinchu Offshore Area,”
J. Mar. Sci. Eng., vol. 11, no. 2, 2023, doi: 10.3390/jmse11020457.
[19] A. Myhr, C. Bjerkseter, A. Ågotnes, and T. A. Nygaard, “Levelised
cost of energy for offshore floating wind turbines in a lifecycle
perspective,” Renew. Energy, vol. 66, pp. 714–728, 2014, doi:
10.1016/j.renene.2014.01.017.
[20] T. A. T. Nguyen and S. Y. Chou, “Improved maintenance optimization
of offshore wind systems considering effects of government subsidies,
lost production and discounted cost model,” Energy, vol. 187, p.
115909, 2019, doi: 10.1016/j.energy.2019.115909.
[21] H. Li and C. Guedes Soares, “Assessment of failure rates and reliability
of floating offshore wind turbines,” Reliab. Eng. Syst. Saf., vol. 228,
no. August, 2022, doi: 10.1016/j.ress.2022.108777.
[22] J. Warnock, D. McMillan, J. Pilgrim, and S. Shenton, “Failure rates of
offshore wind transmission systems,” Energies, vol. 12, no. 14, pp. 1–
12, 2019, doi: 10.3390/en12142682.
[23] J. Carroll, A. McDonald, and D. McMillan, “Failure rate, repair time
and unscheduled O&M cost analysis of offshore wind turbines,” Wind
Energy, vol. 19, no. 6, pp. 1107–1119, 2016, doi: 10.1002/we.1887.
[24] E. C. M. Ruijgrok, E. J. Van Druten, and B. H. Bulder, “Cost
Evaluation of North Sea Offshore Wind Post 2030,” no. February,
2019, [Online]. Available: www.witteveenbos.com.
[25] A. Dewan and M. Asgarpour, “Reference O&M concepts for near and
far offshore wind farms. ECN-E-16-055,” 2016.
[26] D. Ahn, S. C. Shin, S. Y. Kim, H. Kharoufi, and H. C. Kim,
“Comparative evaluation of different offshore wind turbine installation
vessels for Korean west–south wind farm,” Int. J. Nav. Archit. Ocean
Eng., vol. 9, no. 1, pp. 45–54, 2017, doi: 10.1016/j.ijnaoe.2016.07.004.
[27] Y. Dalgic, I. Lazakis, I. Dinwoodie, D. McMillan, M. Revie, and J.
Majumder, Cost benefit analysis of mothership concept and
investigation of optimum chartering strategy for offshore wind farms,
vol. 80, no. 0. Elsevier B.V., 2015.
[28] R. C. Ramachandran, C. Desmond, F. Judge, J.-J. Serraris, and J.
Murphy, “Floating offshore wind turbines: Installation, operation,
maintenance and decommissioning challenges and opportunities,”
Wind Energy Sci., vol. 7, no. 2, pp. 903–924, 2022, [Online].
Available: https://doi.org/10.5194/wes-2021-120.
[29] F. D. McAuliffe, M. Kamidelivand, F. Judge, J. Murphy, and R.
O’Connell, “Wild Atlantic Windfarms, are they cost-effective,” 2023.
[30] G. Rinaldi, A. Garcia-Teruel, H. Jeffrey, P. R. Thies, and L. Johanning,
“Incorporating stochastic operation and maintenance models into the
techno-economic analysis of floating offshore wind farms,” Appl.
Energy, vol. 301, no. June, p. 117420, 2021, doi:
10.1016/j.apenergy.2021.117420.