Renewable Energy 28 (2003) 2165–2178

www.elsevier.com/locate/renene

Efficiency and effectiveness of wind farms—keys

to cost optimized operation and maintenance
H.-J. Krokoszinski
ABB New Ventures GmbH, P.O. Box 10 03 51, 68128 Mannheim, Germany

Received 20 February 2003; accepted 7 March 2003

Abstract

Operation and maintenance (O&M) cost will be the key to the economic viability of large
offshore wind farms planned worldwide. In order to support investment decisions a systematic
mathematical approach to the O&M cost contributions is required prior to detailed engineering
or even construction of the wind farm.
Adopting the general terms of efficiency, productivity and effectiveness defined for pro-
duction processes we introduce the Wind Farm Process and its Total Overall Equipment Effec-
tiveness (TotalOEE) by considering wind farms performing a transformation of produced elec-
trical energy to delivered (sold) electrical energy. This transformation process consists of an
installation, i.e. properly selected wind energy converters and their arrangement to form a
wind farm, and of a process comprising operation and maintenance. Both are the subject of
optimization to maximize the annual energy output by minimizing the different kinds of losses.
In a systematic approach to the causes and nature of losses in wind farms the terms theoreti-
cal production time, available production time and valuable production time are redefined in
unit full load hours. Then, a calculation scheme is developed to quantify wind farm production
losses in terms of planned or unplanned downtimes and speed losses and to relate the associa-
ted reduction of revenues ⌬R to the theoretical maximum of annual wind park revenues
Rtheo(park). It leads to the simple equation ⌬R / Rtheo = TotalOEE – 1 ⬍ 0.
 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Wind farm; Operation; Maintenance; Service; Efficiency; Overall equipment effectiveness;
Productivity; Availability; Performance; Capacity factor; Failure rate; Downtime; Speed loss; Revenue
loss

E-mail address: [email protected] (H.-J. Krokoszinski).

0960-1481/03/$ - see front matter  2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0960-1481(03)00100-9
2166 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178

1. Introduction

Offshore wind farms are a rapidly growing source of renewable energy. After a
tremendous boom of the on-shore wind farm business experienced in the past few
years, especially in Germany, the first offshore wind farms have been installed close
to the shore and the first really big ones (0.5–1.5 GW) are being planned at larger
distances from the coast line. With the currently available size of wind turbines for
offshore applications (2.0–4.5 MW) the number of Wind Energy Converters (WEC)
per wind farm ranges up to several hundreds. Although constructed for the highest
reliability it can be anticipated that operation, service and maintenance of WECs
will be continuous tasks in order to maintain the energy output at the maximum
level. Since productivity and maintenance costs are very much influencing the econ-
omic viability of these high investment facilities, the generation of a comprehensive
O&M strategy is mandatory prior to engineering or even installation of wind farms.
Instead of the usual assessment of energy production from rated wind farm power
and an availability simply calculated as uptime divided by 8766 h we derive a math-
ematical model comprising correct relationships of parameters and variables
determining O&M cost and revenue losses. Since the model is not necessarily limited
to offshore installations we keep it in general terms.

2. The wind farm process

We define a Wind Farm Process comprising both an installation and a transform-

ation process on the basis of general definitions of inputs and outputs known from
the description of manufacturing processes (Fig. 1, [1,2]). We modified this concept
for the special case of wind farms by considering the wind farm transforming pro-
duced electrical energy to delivered electrical energy. The total annunal electrical

Fig. 1. General definitions used in production processes starting with raw material as reference input
and ending with products as actual output [1].
 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178 2167

energy at the terminals of N individual wind energy converters is taken for raw
material and reference input to the wind farm process. Hence, we consider the con-
version of mechanical (wind) energy into electrical energy being complete and optim-
ized (by selecting the right type of WECs) rather than being subject of our wind
farm process.
The total annual electrical energy of an individual WEC with rated power Prated
can be calculated from the annual wind speed distribution of a specific wind farm
site (Weibull curve h(vw)) and the selected wind turbine’s power curve P(vw):

Etot(i)(WEC) ⫽ 冕 a
P(vw(t))dt ⫽ 冕dE
a dt
dt ⫽ 冕
0
⬁
dEel(vw)
dvw
dvw (1)

⫽ 冘 k
h(vw(k))P(vw(k)) ⌬t

where ⌬t = time interval between wind speed data vw(k) and vw(k+1) and
Σkh(vw(k)) = 1

2.1. Reference electrical energy

The reference input is called reference electrical energy Eref(park) summing up

the total annual electrical energies Etot(i)(WEC) of all N WECs:

冘
N

Eref(park) ⫽ Etot(i)(WEC) ⫽ N ⬍Etot(WEC)⬎ (2)

i⫽1

with ⬍Etot(WEC)⬎ = 1 / N ΣNi= 1Etot(i)(WEC) and Etot(i)(WEC) according to (Eq. (1)).

It represents the maximum energy that could theoretically be delivered into the
grid at the grid connection point if it wasn’t for the losses or outages. The energy
can be converted to reference production time Tref by dividing it by the rated wind
farm power (N Prated (WEC)):
Eref(park) ⬍Etot(WEC)⬎
Tref ⫽ ⫽ [full load hours] (3)
N Prated Prated

2.2. Theoretical electrical energy

The reference output is called theoretical electrical energy Etheo(park) (see Fig.
2). It is the maximum possible output of electrical energy that could be fed into the
grid per year, if the complete reference electrical energy Eref was transformed
(transported to grid connection) with exactly the specified reference productivity.
For wind farms we name it the Layout Factor LF. It combines the specified (=
simulated) wake losses due to the arrangement of WECs in the wind farm (i.e. the
park efficiency η(park)) and the calculated electrical losses (described by the electri-
cal efficiency η(el)) of the cables and devices all the way down the line from the
WEC-terminals to the grid connection point.
2168 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178

Fig. 2. The Wind Farm Process: production process definitions of Fig. 1 interpreted for wind farms.

LF ⫽ η(park) η(el) (4)

Etheo(park) ⫽ Eref LF (5)
The corresponding time span in units of full load hours is the theoretical pro-
duction time:
Etheo(park)
Ttheo ⫽ ⫽ Tref LF ⫽ Tref h(park) h(el) [full load hours] (6)
N Prated

2.3. Available electrical energy

The actual input is called available electrical energy Eavail and reflects the amount
of energy that is available per year to be fed into the grid after subtracting scheduled
(i.e. inevitable) downtime losses from the theoretical electrical energy of the wind
farm. These losses are called external losses (section 3.1) since they are determined
by external influences and planned before starting operation of the wind farm. For-
mally, they are taken into account by the Planning Factor PF:
Eavail(park) ⫽ Etheo(park) PF ⫽ Eref(park) LF PF (7)
The corresponding time span in units of full load hours is the available pro-
duction time:
Eavail(park)
Tavail ⫽ ⫽ Ttheo PF ⫽ Tref LF PF (8)
N Prated
The total efficiency η of the wind farm describes the losses that are already determ-
ined in the engineering and operation planning phases of the wind farm, i.e. the park
 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178 2169

efficiency and the electrical efficiency due to positioning and cabling of the WECs
(combined in the Layout Factor) and the planned downtimes:
Efficiency η ⫽ LF PF ⫽ η(park) η(el) PF (9)

2.4. Valuable electrical energy

The available electrical energy is not completely harvested either since technical
losses reduce it before being fed into the grid. The actual output is called valuable
electrical energy Evalu. The productivity p ⬍ 1 of the wind farm process describes
the energy losses associated with all technical problems encountered during operation
and maintenance processes. With respect to the reference energy Eref(park) the valu-
able (i.e. finally sold) electrical energy Evalu is the result of the efficiency of the wind
farm installation and the productivity of the O&M process:
Evalu ⫽ Eavail(park) p ⫽ Eref(park) η p (10)
In equivalence to the energy, technical problems reduce the number of full load
hours operated per year from the available production time Tavail down to the valuable
production time Tvalu:
Evalu(park)
Tvalu ⫽ ⫽ Tavail OEE ⫽ Ttheo PF OEE (11)
N Prated
⫽ Tref LF PF OEE
The factor between valuable production time and available production time is
called Overall Equipment Effectiveness (OEE). It is clear that OEE in the time scale
is equal to productivitiy p in the energy scheme. Hence, compared to the theoretical
production time (the theoretical maximum of deliverable electrical energy) the actu-
ally valuable production time (the finally sold electrical energy) is described by the
Total Overall Equipment Effectiveness (TotalOEE):
Tvalu ⫽ Ttheo TotalOEE (12)
The TotalOEE combines the Planning Factor PF and the OEE comprising all losses
on top of the inevitable losses (wake and electrical losses), i.e. unplanned downtimes
and other technical losses incurred during operation (see section 3.2):
TotalOEE ⫽ PF OEE ⬍ OEE since PF ⬍ 1 (13)
In Fig. 3 the relations between the time periods are depicted. It is clear that, once
the production facility (the wind farm) is built, the reference production time cannot
be changed any more since the site and the turbines were selected. Similarly, the
theoretical production time (resulting from LF) is also fixed, since park efficiency
and electrical losses are results of engineering work. Additionally, the external losses
are basically determined by planned downtimes (see Table 1 in section ). Hence, by
definition, they are pre-fixed in the engineering and operation planning phase (Fig.
4) and described by a Planning Factor PF which has to be carefully determined in
agreement of wind farm developer, wind farm operator, the owner and the utility.
2170 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178

Fig. 3. Time based definitions describing the Wind Farm Process.

Fig. 4 clearly visualizes where the wind farm developer is responsible (lower left-
hand triangle) and where the wind farm operator is active (upper right-hand triangle):
Layout Factor and Planning Factor are fixed during wind farm development whereas
(Total)OEE is the result of an effective O&M strategy. However, it is obvious that
the planned downtimes (as part of PF) are on the boarderline between the Engineer-
ing Phase and the Operation and Maintenance Phase which implies the possibility
for the wind farm operator to reduce planned downtimes (compared to the plan) by
a clever O&M strategy including optimized spare part stock and skilled crews and,
thus, to increase the available production time and with it the valuable production
time (revenues).

3. Classification of losses

According to the theory of production processes [1] both external and technical
losses can be classified in terms of causes: downtime losses, speed losses and quality
losses (Fig. 5).

3.1. External losses

The external losses are subdivided according to Table 1 and interpreted in wind
farm specific terms. It can be inferred from that list that these losses are partly
planned (e.g. scheduled maintenance) and partly imposed by external factors like
weather conditions that deviate from the specified (statistical) average (too low or
too high wind speeds) or diseases or strikes. Thus, not all external losses are planned
although being described by the Planning Factor PF.
External losses in terms of annual average planned downtimes per WEC ⬍tdp⬎
are the result of the O&M planning process (Fig. 4) which has to fix number np and
average length of planned downtimes like scheduled maintenance [3] (auto control
stops, regular maintenance + major overhauls, see Table 1, row (a)) or external speed
losses. With np = number of planned actions per year, nsp = number of shifts per
action, 12h = shift length, msp = number of shifts per day, we define the annual
average planned downtime per WEC:
 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178 2171

Table 1
Subdivision of external losses developed on the basis of general theory [1].

Types of Losses Causes of External Losses Causes of External Losses (for Influence by
(General theory) wind farms) O&M?

(a) External downtime Planned downtimes for –Auto control stops, i.e. remote Yes
losses preventive maintenance controlled stop, system test and
(Scheduled maintenance) auto-restart
–Regular maintenance
–Major overhauls including
upgrades, modifications of hard-
/software
(b) External speed losses (i) Lack of demand –Limitation of transmission grid No
(ii) Lack of raw material –Less wind than expected No
according to statistics
–Lower farm efficiency by No
deviations from specified wind
rose
(ii) Too much input (wind) –Blade overload protection (Yes)
–Cable overload protection (Yes)
(iv) Lack of personnel Epidemic diseases, strikes No
(v) Environmental deals –Reduction of production to No
preserve nature
(c) External quality losses Missing of rated quality Derating of power delivery due No
(e.g. during start-up) to bad power quality

Fig. 4. Division of the Wind Farm Process concept in Engineering and Operation Planning Phase and
the Operation and Maintenance phase.

⬍tdp⬎ ⫽ np [ nsp 12h ⫹ (2 (14)

⫺ msp) integer(nsp) ] 12h [h / a](nsp non⫺integer)
Here, nightshifts can be considered by msp = 2 (no nightshifts: msp = 1, nighttime
2172 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178

Fig. 5. Classification of losses (both technical and external) in terms of downtime losses, speed losses
and quality losses based on [1].

= integer(nsp) 12h, where integer(x) := nsp - 1 for all x in interval (nsp - 1, nsp] ),
which influences both downtimes and O&M-cost [3]. In addition, we have to con-
sider a need for planned maintenance of the grid connection equipment since there
is a number of components / subsystems (MV-switchgear, MV/HV-transformers,
power converters, HV-switch-gear, cables) which need regular maintenance or even
service & repair in case of statistical failures (see section 3.2.1).
If we neglect in a first approach the external speed and quality losses, then the
Planning Factor can be calculated with (Eq. (8)) and (Eq. (14)):
⬍ tdp ⬎
PF ⫽ Tavail / Ttheo ⫽ 1 ⫺ (15)
8766h
Example: np = 1 / year, np = 5.5 shifts, msp = 2 =⬎PF = 1 – 66 h / 8766 h =
0.992. If we don’t plan for nightshifts we get with msp = 1: PF = 1 – 126 h /
8766 h = 0.986.

3.2. Technical losses

Each type of technical losses in Fig. 5 is quantified by a characteristic factor:

Availability A = gross operation time / available operation time =⬎(unpl.) down-

time losses
Performance P = net operation time / gross operation time =⬎speed losses
Quality Q = valuable production time / net operation time =⬎ quality losses
 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178 2173

From the definitions and Fig. 5 it is clear that:

OEE ⫽ A∗ P∗ Q (16)

3.2.1. Unplanned downtimes (Availability A)

From a statistical analysis in 1997 of several hundreds of onshore WECs of the
500 kW class in operation in Germany for more than one year, a failure mode and
MTBF distribution of the major WEC components was deduced [4]. The resulting
average failure frequency of 2.3 failures/year was considered unacceptable and
reduced by 25% to 1.79/a. This was defined base case and target for offshore WECs
with the same spread of MTBFs and failure frequencies (nf) as found onshore. Now-
adays the typical number of failures expected (or targeted) for individual offshore
wind energy converters is nf ⬍ 1 / year and is supposed to vary with lifetime (bath
tub curve) [5]. With τN = MTBF of N turbines within a wind farm [h], λN =
failure rate of N turbines within a wind farm [h - 1] and nf = frequency of occurrence
of a fault we get
λN ⫽ N λ ⫽ N / τ ⫽ N nf / 8766h ⫽ 1 / τ N (17)
Each time when a failure with switch-off occurs the total unplanned downtime
for a WEC depends on how fast the failure can be removed, i.e. what type of fault
occurred and how fast the service crew can get to the WEC to fix the fault: tdu
(WEC). It is the sum of time spans needed for identification of the fault, mobilisation
of the crew, wait for access and transportation of crew and spare parts [6]. Hence,
in the case of high numbers of WECs on a wind farm several failures will happen
within a relatively short time interval. In the case of offshore sites this will be
especially at times of non-accessibility which depend mainly on the transport vehicles
chosen to access the wind energy converter [6]. The net time for repair ⌬trep on site
is depending on the grade of destruction or complexity of fault which can be grouped
in fault classes, for example

Minor fault: alarm stoppage repaired by remote service

Small fault: alarm stoppage restored by short corrective maintenance action on site
Severe fault: system breakdown requiring considerable time, spare parts and work
to be repaired
Fatal fault: total damage of system, no possibility for repair, total system has to
be replaced

The total annual downtime due to unplanned stoppages is the sum of the individual
downtimes in these classes (k = 1,…,4) times the statistical frequency of occurrence

tdu(N) ⫽ 冘 k
λN(k) 8766htdu(k) ⫽ N 冘 k
nf(k) tdu(k) (18)

and per WEC

⬍tdu(WEC)⬎ ⫽ tdu(N) / N ⫽ 冘 k
nf(k) tdu(k) (19)
2174 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178

Besides the WECs we have to consider failure rates of the complex grid connection
equipment comprising offshore substation(s), subsea cables and onshore
substation(s). Let us define a combined failure rate λGC for the complete system:
λGC ⫽ 1 / τGC ⫽ nf(GC) / 8766h where τGC is the MTBF of a grid connection
(20)
If the system exists twice in a large wind farm then the total failure rate is, of
course, doubled. Similar to the WEC case we define an unplanned GC-downtime
due to statistical failures of grid connection equipment with need for switch-off and
immediate service and repair. The duration tdu(GC) = ⌬ttra(GC) + ⌬trep(GC) consists
of ⌬ttra(GC) for transportation of spare parts not available inside the wind farm and
⌬trep(GC), the time for actual repair work. Let nGC be the number of independent
grid connection branches (1 or 2) with equal share of wind farm power. In contrast
to WEC-failures, where each time only the production of one WEC is lost, each grid
connection failure causes the loss of electrical energy of the total or (with two
branches) twice as often half of the windpark (2 ∗ N/2 WECs). Then the GC-induced
downtime of each WEC is on the average:
⬍tdu(GC)⬎ ⫽ (N / nGC) nGC λGC 8766h tdu(GC) / N
(21)
⫽ λGC 8766h tdu(GC) ⫽ nf(GC) tdu(GC)
Defining ⬍tdu(WF)⬎ = ⬍tdu(OWEC)⬎ + ⬍tdu(GC)⬎ as the Annual Average
Unplanned Downtime per WEC we get for the net Availability related to the planned
uptime (= 8766h - ⬍tdp⬎) [6]:
⬍tdu(WF)⬎ ⬍ tdu(WF) ⬎
A ⫽ 1⫺ ⫽ 1⫺ (22)
8766h ⫺ ⬍tdp⬎ 8766h PF

3.2.2. Technical speed losses (Performance P)

Technical speed losses can be caused by

앫 Non-perfect Maximum Power Point (MPP-) tracking (see Fig. 6A)

앫 Deviation from the rated power curve (see Fig. 6B)
앫 Minor stoppages for quick repair of small faults

MPP-tracking as depicted in Fig. 6A is performed by adapting the rotational speed

ω of the rotor to the actual wind speed vw as to maintain a fixed tip speed ratio
λts = rω / vw which is optimal for the chosen rotor shape and radius r (normally
λts should be around 5–6). With this value of λts the maximum power is extracted
from the wind (Betz’ Law1 [7]). If by any reason the closed loop control of rotational
speed is disturbed then the output power of the WEC is sub-optimal, i.e. the differ-

1
Betz’ law says that you can only convert less than 16/27 (or 59%) of the kinetic energy in the wind
into mechanical energy.
 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178 2175

Fig. 6. (A) Speed losses due to non-perfect MPP tracking; (B) Speed losses due to deviation of the
actual power curve from the rated power curve. Both cases are quantified by a power reduction factor
f = Pac / Prc used for Performance P.

ence to the maximum power point (MPP) is lost power. Reasons for missing the
optimum tip speed ratio can be:

앫 Variable speed control is not possible or disturbed

앫 Pitch control is out of optimum
앫 Yaw error i.e. yaw angle is out of optimum (90{deg} to wind direction)

In all these cases the WEC’s (and hence the park’s) valuable electrical energy (the
valuable production time) is reduced as if the production speed was decreased
(therefore: speed losses).
Alternatively, the performance of the rotor or generator might be disturbed in
2176 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178

terms of deviation from the specified power curve (see Fig. 6B). Hence, although
running with optimum tip speed ratio the power produced at a given wind speed is
lower than specified during engineering. Hence, the WEC is performing as if it was
running with lower speed (therefore: speed losses). Reasons for deviations from the
rated power curve can be:

앫 Ice or salt or dirt layers on the blades

앫 Blade shape deviates from specified shape (deformation)
앫 Bearing faults that lead to braking of the rotation

Appropriate on-line condition monitoring and power curve modeling systems can
detect such deviations and initiate service and/or maintenance [8].
Minor stoppages comprise quick corrective maintenance actions to fix minor faults
identified by condition monitoring, i.e. those that did not lead to WEC shut-down.
In the most benign cases, the faults can be fixed during the next regular maintenance
period, but if the function or proper control of the WEC is disturbed a separate quick
repair is performed on one of the next maintenance tours, whereby a short switch-
off of the turbine is initiated to allow for safe access of servicemen. Since this switch-
off is only for minutes or 1 h the loss of produced electrical energy resembles more
the effect of a low-wind period rather than a downtime; therefore, it is listed as
technical speed loss. Averaging the technical speed loss factors over all WECs the
Performance becomes (with f = power reduction factor):
P ⫽ Tnet / Tgros ⫽ Enet / Egros ⫽ ⬍Pac⬎ / ⬍Prc⬎ (23)

⫽ (1 / N) 冘 fi ⫽ f

From our present point of view, the term technical quality losses is not relevant for
wind farms unless unplanned power quality deficiencies really occur at grid connec-
tion point and, additionally, lead to the need to derate the wind farm power level.
Hence, in a first approximation we can confine to OEE = A ∗ P ∗ Q = A ∗ f with
A according to (Eq. (22)).

3.3. Revenue losses

In the present paper we confine the application of the model developed in sections
2 and 3 to the calculation of average annual revenue losses. The derivation of sea-
sonal variations of revenue losses and of maintenance cost, especially crew sizes
and labour cost, will be published elsewhere [3].
The calculation scheme is based on the average annual downtimes for scheduled
maintenance and unplanned service and repair as well as on external and technical
speed losses. From this we finally get the valuable production time and calculate the
revenue losses ⌬R(losses) due to lost production time:
⌬R(losses) ⫽ [Tvalu⫺Ttheo] N Prated e ⬍ 0
(24)
where e ⫽ energy price (USD / kWh)
 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178 2177

Since Tvalue = Total OEE Ttheo, we get Tvalu - Ttheo = Ttheo (Tvalu / Ttheo – 1) =
Ttheo (TotalOEE – 1) and, with Etheo(park) = N Prated Ttheo from (Eq. (6)), (Eq. (24))
transforms into:
⌬R(losses) ⫽ Etheo e ( TotalOEE ⫺ 1 ) ⫽ Rtheo(TotalOEE⫺1) ⬍ 0 (25)
where Rtheo = Etheo e are the theoretical wind farm revenues. Hence, the relative
annual revenue losses are:
⌬R / Rtheo ⫽ TotalOEE ⫺ 1 ⫽ PF OEE ⫺ 1 ⬍ 0 (26)

4. Summary and conclusions

Application of production technology terms to wind farms lead us to a concise

and comprehensive description of the transformation of produced electrical energy
at the terminals of the individual windmills to the total delivered electrical energy
at the grid connection point(s). Definition of a Layout Factor (LF) and adoption of
the Planning Factor (PF) and the Overall Equipment Effectiveness (OEE) rep-
resenting the external losses and the technical losses of wind farms, respectively,
enabled us to systematically describe and quantify the losses reducing the overall
capacity of the wind farm. The annual wind farm revenues are determined by the
efficiency (LF∗PF) of the wind farm as result of the development work and by the
effectiveness of the operation and maintenance concept. The scheme introduced thus
clearly defines the responsibility and influence areas of the Engineering & Operation
Planning Phase and the Operation & Maintenance Phase.
An O&M strategy for wind farms must strive for maximum TotalOEE (= PF ∗
OEE) in order to minimize the revenue losses proportional to TotalOEE - 1. This
requires minimization of downtimes arising from service and repair which (at least
in the case of offshore wind farms) strongly depend on the accessibility of the WECs
[6]. For the offshore case we will show [3] how the O&M-cost contributions can
be described within this framework by systematically calculating the required
amounts of time for regular maintenance, major overhauls and unplanned service
and repair and the related wind farm downtimes.

References

[1] Wauters J, Mathot J. OEE—Overall Equipment Effectiveness. ABB White Paper.

[2] Hansen RC. Overall equipment effectiveness. Industrial Press, 2001.
[3] Krokoszinski H-J, Voll B. Service an maintenance of offshore wind farms. To be presented at the
European Wind Energy Conference 2003, Madrid, Spain.
[4] Van Bussel GJW, Schöntag C. Operation and maintenance aspects of large offshore wind farms.
Proceedings of the European Wind Energy Conference, Dublin, Ireland, 1997.
[5] Rademakers l, Braam H, Robberegt M, Katz P. Risk based maintenance management to reduce the
operational costs of wind energy. 5. Wilhelmshaven, Germany: Deutsche Windenergiekonferenz,
WEK 2000, 7. –8.6.2000, p.49.
2178 H.-J. Krokoszinski / Renewable Energy 28 (2003) 2165–2178

[6] Krokoszinski H-J, Voll B. Availability of offshore wind farms. Submitted to All-Energy Opportunities
Conference 2003, Aberdeen, UK.
[7] www.windpower.org/stat/betzpro.htm
[8] Waldl H-P, Pahlke T. Wind farm surveillance and video monitoring. Proceedings of the European
Wind Energy Conference, Copenhagen, Denmark, 2001, p. 1110.