Eder Er 2015
Eder Er 2015
Eder Er 2015
art ic l e i nf o a b s t r a c t
Article history: An actual growth rate greater than 30% indicates that offshore wind is a reasonable alternative to other
Received 14 March 2014 energy sources. The industry today is faced with the challenge of becoming competitive and thus
Received in revised form signicantly reduce the cost of electricity from offshore wind. This situation implies that the evaluation of
2 May 2014
costs incurred during development, installation and operation is one of the most pressing issues in this
Accepted 22 October 2014
industry at the moment. Unfortunately, actual cost analyses suffer from less resilient input data and the
application of simple methodologies. Therefore, the objective of this study was to elevate the discussion,
Keywords: providing stakeholders with a sophisticated methodology and representative benchmark gures. The use
Offshore wind of Data Envelopment Analysis (DEA) allowed for plants to be modelled as entities and costs to be related
Data envelopment analysis (DEA) to the main specics, such as distance to shore and water depth, ensuring the necessary comparability.
Cost efciency
Moreover, a particularly reliable database was established using cost data from annual reports. Offshore
Best practice
wind capacity of 3.6 GW was benchmarked regarding capital and operating cost efciency, best-practice
Economies of scale
Learning-by-doing
cost frontiers were determined, and the effects of learning-by-doing and economies of scale were
investigated, ensuring that this article is of signicant interest for the offshore wind industry.
& 2014 Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035
2. Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1036
2.1. Characterisation of DMUs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1036
2.1.1. Capital cost efciency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1036
2.1.2. Operating cost efciency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1036
2.2. Mathematical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1038
2.2.1. DEA application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1038
2.2.2. Best-practice frontier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1039
2.2.3. Tobit regression and learning-by-doing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1039
3. Case study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1040
3.1. Capital cost efciency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1040
3.2. Operating cost efciency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1040
3.3. Cost data preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1041
4. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1041
4.1. Cost efciency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1041
4.2. Best-practice frontier. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1042
4.3. Tobit regression and learning-by-doing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1042
5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1042
6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044
Abbreviations: BCC, BankerCharnesCooper; CAPEX, capital expenditures; CCR, CharnesCooperRhodes; DEA, data envelopment analysis; OPEX, operating expenditures;
OWF, offshore wind farm; WEC, wind energy converter
n
Permanent address: Bandwirkerstrae 13, 22041 Hamburg, Germany. Tel.: 43 664 41 899 48.
E-mail address: nikolaus.ederer@gmail.com
http://dx.doi.org/10.1016/j.rser.2014.10.071
1364-0321/& 2014 Elsevier Ltd. All rights reserved.
N. Ederer / Renewable and Sustainable Energy Reviews 42 (2015) 10341046 1035
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044
participants the maintenance of competitive advantages and raises parameters inuence the DMU efciency is provided in the
barriers to entry. The capital cost estimation completed in [8] following.
shows which typical sources are used for gathering cost data.
Sources originate either from offshore wind farm owners' websites 2.1.1. Capital cost efciency
or reports, in which it is difcult to detect whether the gures In this case, the DMU is the entity that develops and installs the
were massaged, or from online databases, such as [26], and reports OWF. As shown in Fig. 1, this entity transforms the only input
of consulting companies, where the original source is often not specic capital costs into an OWF of a specic size ( installed
disclosed and it is not clear how data were processed (e.g., capacity) for a specic distance to shore and water depth.
deation). It is clear that these sources are not sufciently reliable As mentioned previously, capital costs (also referred to as capital
to generate signicant results. Therefore, all cost data used for the expenditures or CAPEX) are dened as all expenditures that occur
investigation originate directly from annual nancial statements until the OWF is commissioned, meaning that they comprise all
and are veried using only information from owners' websites and investments for OWF development, such as soil examinations,
reports, which is possible because offshore wind projects are environmental assessments and appraisals for certication as well
usually arranged through Special Purpose Vehicle (SPV) compa- as all investments for OWF deployment, such as the purchase and
niesseparate legal entities that are used to isolate the owner installation of components and project management [13]. Installed
from nancial risks [27,28]. The annual reports of these SPVs are capacity is often used as an indicator for the size of a wind power
ofcially deposited and accessible at the respective register of plant. It is the product of the wind energy converters' (WEC) rated
companies [2934]. This database is unique and ensures the power, dening the level of power for which the turbine and its
signicance and usefulness of the article's results for stakeholders components are designed, and the number of turbines of the plant.
of the offshore wind industry. The names and cost data of Distance to shore and water depth were included because they are
individual OWFs' SPVs are intentionally not quoted because the the main drivers for capital costs. Thus, the theoretical production
intention of this publication is to provide a reliable scientic function can be formulated: the greater the specic capital costs
analysis of offshore wind costs and not to compromise anybody invested for deploying the OWF, the larger (installed capacity) it
on any account. will become, the farther off shore the OWF will be situated and the
The next section presents how capital and operating cost greater the water depth of the OWF site will be.
efciency were modelled using DEA. Section 3 provides a descrip-
tion of the input data for the case study and their preparation.
2.1.2. Operating cost efciency
The results of the analysis are presented in Section 4 and subsequently
For the operating cost efciency analysis, the DMU is the entity
discussed in Section 5.
that is responsible for the OWF during its operational phase, which
might be the operation and maintenance department of an energy
utility or an external service provider. In the following, this DMU is
2. Modelling
referred to as the service agent. To be able to select parameters
that determine the performance of an offshore wind service agent,
Before the DMUs are characterised and the mathematical
it is necessary to understand what affects the efciency of a WEC,
model is explained, it must be noted that the methodology
which is discussed in the following. The author of [35] provides a
presented in the following represents the result of a comprehen-
good visualisation of a WEC's efciency, plotting the energy
sive analysis in which many different congurations of inputs
density as a function of wind speed. Fig. 2 shows an adapted
(outputs) and modications of DEA were investigated with respect
version of this graph using a generic 5 MW WEC as the basis and
to their reasonableness and applicability to reality.
depicting the annual energy yield per rotor swept area as a
function of wind speed. The main input for a WEC is, of course,
2.1. Characterisation of DMUs the wind or, more specically, the kinetic energy content of the
wind. This kinetic energy is given exogenously by nature and can
In the course of researching the literature for this article, it was be calculated using the following equation:
particularly noticeable that most authors attach importance to a
Ekin_wind 1=2Av3 t 1
detailed description of the chosen DEA model, whereas there is
often a lack of reasoning why specic input (output) parameters where A designates the swept area of the rotor, is the air density,
were selected, why specic parameters were used as input rather v is the wind speed and t is time. According to Betz's momentum
than as output or vice versa and how they are related to the theory, there exists a physical limit for the amount of mechanical
efciency of the DMU. It is clear that the selection of parameters energy that can be extracted from a free-stream airow by an
and how they are included in the model are as signicant as the energy converter under ideal conditions, which is 16/27 59.3% of
model itself with regard to the quality of the results. Thus, a Ekin_wind . Whereas this loss cannot be avoided, the next sectional
detailed description of the selected DMUs and how the selected area shown in Fig. 2 represents the aerodynamic losses that occur
because a real wind rotor has an efciency below the theoretical availability is dened as the fraction of a given operating period
limit. Due to economic and technological constraints, a WEC is in which a wind turbine generating system is performing its
designed for a specic rated power, which causes another major intended services within the design specication. Therefore,
loss. Finally, the remaining area, after subtracting mechanical and 95% time-based availability means that the WEC does not perform
electrical losses, represents the gross annual energy yield. This its intended service (i.e., it is not able to convert energy due to
yield can be calculated using the power curve of the WEC and the malfunction, maintenance, etc., for 5% of the operating period).
wind resource at the wind farm site. Thus far, all losses are a result However, this gure is not actually tantamount to an energy loss of
of physical constraints or the wind turbine's design. 5% because it depends on the prevailing wind speed during the
The last portion indicates the loss due to the WEC being downtime of the turbine. Because the operator is remunerated for
unavailable, which is in principle the only loss that can be the energy fed into the grid, the time-based availability is only an
inuenced by the service agent.2 This is why availability is a indicator but not an optimal performance metric. Refs. [3941]
common performance metric for the operation and maintenance provide empirical studies and reviews of this topic.
of wind power plants [36]. Unfortunately, the distinction between The amount of energy lost during the period of turbine
time-based and production-based (also referred to as energy- or unavailability depends on the technical quality of the wind
yield-based) availability is often overlooked, although the differ- turbine and the performance of the service agent. The quality of
ence between the two is signicant for evaluating operational a wind turbine can be specied in terms of reliability (i.e., the
performance. One reason for this oversight might be that an probability that the WEC satisfactorily performs its intended
international standard for dening the latter is still under devel- function under given circumstances for a specied period of time)
opment [37]. According to IEC/TS 61400-26-1 [38], time-based [38]. In terms of energy loss, it is important that the turbine is
reliable, particularly during periods of high wind, during which
the WEC would generate high yield. The reliability issue is even
2
This is not entirely true because wear and tear, blade degradation, etc., lead to more important offshore because rough weather conditions do
an efciency loss during the WEC's lifetime, which can be limited up to a certain
point through qualied maintenance by the service agent. Because this effect is
not allow access to the WEC; therefore, the turbine remains
usually rather small, efciency losses that cannot be eliminated by the service unavailable until troubleshooting by a maintenance team is
agent were neglected. possible [42]. Finally, the energy loss due to unavailability reects
1038 N. Ederer / Renewable and Sustainable Energy Reviews 42 (2015) 10341046
the performance of the service agent because he or she is units and identication of benchmarks against which they can be
responsible for keeping the WEC in good condition, organising compared [44]. The main advantage of DEA is that prior assump-
spare parts and scheduling maintenance. Therefore, in a bench- tions regarding the underlying functional relationship between
marking analysis of service agents, both time-based availability inputs and outputs are not required [45]. Because DEA is a well-
and a metric that reects the loss of energy during unavailability established method for relative efciency evaluation and exten-
should be considered. sively applied in the literature, only a brief description is provided
Fig. 3 shows the selected parameters for the operating cost in the following. The application presented in this article, and
efciency analysis. The only inputs are the specic operating costs therefore the description provided, is mainly based on [44], which
(also referred to as the operating expenditures or OPEX), which provides a focused overview, and [46], which is a comprehensive
comprise all expenditures that occur during operation, such as reference book for DEA.
maintenance, insurance and administrative expenses [13]. In The two basic DEA models are the CharnesCooperRhodes
contrast to the capital costs, which are one-time investment costs, (CCR) model [47], which assumes constant returns to scale, and
operating costs are usually assessed on an annual basis. Therefore, the BankerCharnesCooper (BCC) model [48], which assumes
this operating cost efciency analysis assesses the performance of variable returns to scale. Both models use a radial efciency
different OWFs in different operational years. Installed capacity is measure, which means that all inputs and outputs are adjusted
again an output, which also appears to be a good measure in this proportionally. Depending on the selected orientation, either
case for the service agent's scale effort because it combines the inputs are proportionally reduced while outputs remain xed
number and size of WECs. In addition, the distance to shore has (input-oriented) or outputs are proportionally increased while
again been identied to be signicant for these investigations [43]. inputs are held constant (output-oriented). These adjustments
Because there is usually only one port used as the base for the are made until the efciency frontier, which is determined by
operational phase, which is not always the case for the construc- the production possibility set (i.e., the set of feasible activities or, in
tion phase, the distance to the operating port is used. Following other words, the combinations of inputs and outputs) is reached.
the argument regarding a service agent's inuence on the tur- Using the input-oriented model appeared to be appropriate for
bine's energy efciency, a metric reecting energy performance is this application because it could ensure that the outputs would
needed. The optimal way to quantify energy performance might be not be maximised to an implausible level. According to [44], the
to determine the energy lost during downtimes using the mea- CCR model in the envelopment form can be formulated assuming
sured wind speed and the power curve of the wind turbine a set of n DMUs, with each DMU j j 1; ; m using m inputs xij
[39,41]. However, this calculation would require detailed opera- (i 1; ; m) and generating s outputs yrj (r 1; ; s), in the
tional data, which was not available for this study; therefore, the following manner:
output energy performance was dened as follows: !
m s
AEP AEP min o U si sr
p 2 i1 r1
Ekin_wind N 1=2Av3 tN
n
where AEP designates the annual energy production of the OWF subject to xij U j si o U xio ; i A 1; ; m
j1
and N is the number of turbines. As shown in Fig. 2, the kinetic n
energy of wind would be calculated precisely using the wind yrj U j sr yro ; - r A 1; ; s
speed frequency distribution, which was also not available for this j1
study. Thus, the annual average wind speed at hub height was j ; si ; sr Z 0; 8 i; r; j
used for v. In addition, was assumed to be 1.225 kg/m3, t was 0 unrestricted 3
assumed to be 8760 h and for A, the rotor swept area of the
respective plant's WEC was used. Therefore, the output energy For the BCC model, the following constraint must be added:
performance reects how efciently the available kinetic energy n
according to [46], scale efciency is dened as follows: shows a simplied visualisation of this procedure for the single-
n
input, single-output case.
CCR
nscale 6
nBCC 2.2.3. Tobit regression and learning-by-doing
and because the CCR score is always lower than the BCC score, the To investigate the relationship between certain factorsfor
value of scale is also between zero and unity.
n example, the year of commissioningand the efciency scores
Finally, the denitions of the output availability and energy obtained with DEA, the Tobit model proposed in [49] was applied.
performance for the operating cost analysis imply that the This model is a regression model that can address censored data
measures can never exceed 100%; this restriction had to be and is needed because DEA efciency scores are between zero and
incorporated into the model by modifying the standard DEA unity. Therefore, the Tobit model is a common second-stage
models. According to the so-called bounded variable model analysis that has been frequently applied in the literature in
described in [46], this limit can be considered by adding the conjunction with DEA [50]. Nevertheless, it is important to men-
following constraint: tion that this approach is not uncontroversial. For example, in [51],
it is criticised that contextual variables used in Tobit models are
n
probably correlated with the efciency scores calculated pre-
ybj j r lu ; b A 1; ; s 7
j1 viously, which leads to the inconsistency problem of estimators;
therefore, the authors propose the use of bootstrapping. However,
where lu designates the upper bound for the bounded output yb . [5254] conclude after their investigations that the application of
the Tobit model as a second-stage analysis is valid.
In contrast to similar applications in this eld, every factor
2.2.2. Best-practice frontier under investigation is analysed on its own instead of performing a
The idea of calculating and providing the best-practice frontier multiple regression. In this manner, the relationship between the
arose from the policy of not publishing cost data for individual efciency scores and every individual factor was determined.
OWFs in this article. Otherwise, it would have been possible to According to [50], the Tobit model applied in a DEA second-
simply provide the projection of the input for each OWF. However, stage analysis can be described as follows:
to determine the best-practice frontier, a generic DMU was added.
The rst step was to choose a set of values for the generic DMU's nj UC j j with j N0; 2
8
if j r0
n
outputs (e.g., 100 MW installed capacity, 10 m water depth, 10 km >
>
>
< 0
distance to shore) in the range between the minimum and n
j nj if 0 o j o 1 8
maximum values of the respective output of the existing DMUs. >
>
>
: 1 if j Z 1
n
The value for the generic DMU's input was chosen to equal the
maximum value of the input of the existing DMUs. This ensured
that the efciency frontier was not altered. Calculating the where j designates the CCR/BCC/scale efciency score, the set
projection of the generic DMU's input reveals the point on the of parameters to be estimated, C j the respective factor under
efciency frontier for the chosen set of output values. Thus, it was investigation and j the error term. Tobit regression was used to
possible to vary the set of output values and trace the projection of determine the relationship with respect to the factors YEAR (year
the generic DMU's input to scan the best-practice frontier. Fig. 4 of commissioning), CAP (installed capacity), CUMCAP (cumulative
installed capacity), RATPOW (rated power of WECs) and NUMWEC
(number of WECs). In addition, the factor OPYEAR (year of
operation) was included for the operating cost analysis.
As described in the introduction, another objective of applying
the Tobit regression model was to analyse cost development as a
function of technological change, which is commonly described by
learning curves. Therefore, the obtained relationship between cost
efciency and cumulative installed capacity was used to determine
the so-called learning-by-doing rate. The concept of measuring
learning through cumulative production or capacity was rst
introduced by [55] and describes the process of gaining produc-
tivity increases and cost reductions by the accumulation of
experience [56]. A common approach is to investigate the
learning-by-doing effect based on specic capital costs using the
cumulative installed capacity as a proxy for the accumulation of
experience. Following this approach, using the efciency scores
instead of specic capital costs and the equations provided by [57],
the learning-by-doing effect can be calculated using the following
equations:
CUMCAP t
t t 0
CUMCAP t 0
LBD 1 2 9
where t0 /t is the efciency score at time zero/t,
CUMCAP t0 /CUMCAP t is the cumulative installed capacity at time
zero/t, is the learning coefcient and LBD is the learning-by-
doing rate. LBD is typically expected to be positive because a
positive value indicates a reduction in unit cost. Because the cost
Fig. 4. Determination of best-practice frontier. efciency should increase, it is important to note that for this
1040 N. Ederer / Renewable and Sustainable Energy Reviews 42 (2015) 10341046
Table 1
Data for capital cost efciency analysis.
# OWF Ref. Country Commissioning Number of Rated power Installed capacity Distance to shore Water depth
turbines (kW) (MW) (km) (m)
application, LBD is negative in the case of a positive learning effect. directly attributable to bringing the asset to the location, the
Thus, a LBD of 10% indicates that a doubling of cumulative conditions necessary for operation and the initial cost estimate for
experience has led to an increase of 10% in cost efciency. dismantling and removing the item and restoring the site on
which it is located. Using this approach ensured a high quality of
input data for the relative analysis and the generation of signi-
3. Case study cant results. On average, the specic capital cost of these 22 OWFs
was 2992.90 EUR/kW (Median 2520.51 EUR/kW).
Although the data collection was challenging, the DEA conven-
tion indicating that the minimum number of DMUs has to be 3.2. Operating cost efciency
greater than three times the number of inputs plus outputs [58]
was adhered to. For the analysis of the capital cost efciency, Unfortunately, access to operational data of OWFs is very limited.
22Z 3 (1 3) observations were included, and for the operating Nevertheless, Table 2 provides an overview of the OWFs included in
cost efciency, 26Z 3 (1 4) observations were included. the analysis. One data source was the operational reports of the four
UK round 1 OWFsScroby Sands [8082], Kentish Flats [6870],
3.1. Capital cost efciency North Hoyle [7375] and Barrow [62,63]that were obliged to
publish these data due to the UK Offshore Wind Capital Grants
For the capital cost efciency analysis, the data for 3.6 GW of Scheme [87]. These reports contain data about wind speed, avail-
offshore wind capacity was used. The inclusion of plants in the ability, annual energy production, operation and maintenance costs,
analysis was based on the principles that they have a distance to etc. [88] provides a good summary and rst analysis of these
shore greater than 3 km, an installed capacity greater than 30 MW projects' data. The operation and maintenance costs mentioned in
and that the nancial statements of the commissioning year be those reports were compared with the annual nancial statements
available. The data provided in Table 1 were selected very care- of the projects' SPVs, and the annual energy production was
fully, which means that only data from the owner's ofcial compared with data provided by [89] to ensure comparability. In
websites were used. addition, the offshore wind farms alpha ventus [59,9092], Egmond
To ensure comparability, it was necessary to apply the same aan Zee [93,94] and Middelgrunden [9,95,96] were included in the
method for assessing the specic capital costs to every OWF. The analysis. Operating costs for Middelgrunden were obtained from
method was applied using the highest value for tangible assets [95], whereas the data for alpha ventus and Egmond aan Zee
stated in the respective SPVs' annual nancial statements, which required an assessment of the SPVs' annual reports.
usually occurs during the year of commissioning because depre- For this case study, operating costs were dened to be all
ciation is applied afterwards. Tangible assets are all assets that expenditures needed to keep the OWF operating over its lifetime.
have a physical form, such as property, plant and equipment. In other words, all costs stated in the prot and loss account of a
According to the international standard [86], the cost of a SPV, such as administrative expenses, operation and maintenance
corresponding item is composed of the purchase price, any costs costs for WECs and grid connection facilities, insurance costs,
N. Ederer / Renewable and Sustainable Energy Reviews 42 (2015) 10341046 1041
Table 2
Data for operating cost efciency analysis.
# OWF Ref. Installed Distance to Annual wind Annual energy Energy Availability Years of
capacity operating port speeda (m/s) productiona (GWh) performancea (time-based)a available
(MW) (km) (%) (%) data
a
Average value over all operational years included in the analysis of the respective OWF.
b
Only operational data of the 10 southern WECs owned by Middelgrunden Wind Turbine Cooperative were available and therefore considered in the analysis.
rent for port facilities, etc., were summed. Depreciations were Table 3
not included, and governmental grant credits were subtracted. Results of capital cost efciency analysis.
The specic operating costs of these OWFs account for 63.49 EUR/
kW (Median 46.88 EUR/kW) per year on average. Finally, the OWF nCCR (%) nBCC (%) nscale (%)
output energy performance was calculated using formula (2).
alpha ventus 100.00 100.00 100.00
Baltic 1 54.97 54.99 99.95
3.3. Cost data preparation Belwind Phase 1 94.59 96.43 98.09
Barrow 72.56 81.85 88.66
Burbo Bank 31.82 66.72 47.69
For the relative cost analysis comparing OWFs that were
Egmond aan Zee 100.00 100.00 100.00
commissioned (operated) in different years, it was particularly Greater Gabbard 100.00 100.00 100.00
important to deate costs well considered. Unfortunately, studies Guneet Sands 1 & 2 55.08 83.35 66.09
regarding costs of offshore wind often do not contain a description Horns Rev 2 100.00 100.00 100.00
Kentish Flats 44.20 85.29 51.82
of how the costs were deated, which seriously calls into question
Lillgrund 55.22 100.00 55.22
their quality, especially when OWFs that were commissioned more Lynn and Inner Dowsing 59.35 84.66 70.11
than 10 years ago are included. Moreover, it is very likely that the Middelgrunden 41.23 100.00 41.23
costs of offshore wind did not increase with normal ination. For North Hoyle 41.55 66.88 62.13
example, [8] showed that there is a relationship between capital Ormonde 73.21 77.67 94.26
Rhyl Flats 50.10 75.33 66.50
costs and steel price, and [25] references the recent cost increase
Robin Rigg 48.44 68.77 70.44
in offshore wind to a surge in prices of commodities, such as Rodsand 2 68.69 95.08 72.25
copper and steel. Considering the value breakdown of costs Scroby Sands 69.88 78.82 88.66
presented in [11], this is reasonable. According to [11], the CAPEX Sheringham Shoal 53.54 56.45 94.85
Thanet 72.19 73.65 98.02
of a conventional OWF consists of 35% labour, 34% material and
Walney 1 & 2 84.18 88.39 95.24
31% other costs, and the OPEX consists of 35% labour, 14% material
and 52% other costs. Labour costs are dened to include direct and Minimum 31.82 54.99 41.23
indirect labour; material costs include all raw materials and Maximum 100.00 100.00 100.00
components, consumables, equipment, plant and buildings, and Average 66.85 83.38 80.05
Median 64.02 84.00 88.66
other costs comprise services (e.g., vessels, cranes), insurances and
other overheads. Ref. [97] also provides an illustrative breakdown
of material in a 500 MW offshore wind farm with 100 turbines,
which shows that a large proportion of the used material is indeed
steel, copper, aluminium, etc., whose price increase was beyond indicate the level of cost inefciency. When interpreting these
the normal ination rate. To incorporate this fact, all costs used in results it should be kept in mind that DEA is a relative analysis
this analysis were deated using a rate based on the cost split method. Thus, one reason for the full capital and high operating cost
presented in [11] and using the deator for labour costs, a efciency of alpha ventus might be a lack of peers, i.e., OWFs that
commodity price index for the material costs and the GDP deator have specics in the same range, which would ensure better
for other costs provided by [98]. Therefore, all costs mentioned in comparability. Furthermore, the fact that the operating cost ef-
this article refer to a price level in the European Union in 2012. ciency scores of each OWF are on fairly the same levelapart from a
few outliers, which might be due to a year with good wind
conditions or exceptional damageindicate that the model gener-
4. Results ated reasonable results. Because the output installed capacity and
distance to operating port do not change with operating year, this
4.1. Cost efciency characteristic result is expected. Another part of the results would
be the projections of all parameters (i.e., the target value assuming
Tables 3 and 4 present the results of the capital (operating) cost that the DMU operates efciently). Because the provision of the
efciency analysis in the form of efciency scores. An efciency projection would show the costs of individual OWFs instead, the
score of 100% indicates full cost efciency, and values below best-practice frontiers are presented in the next section.
1042 N. Ederer / Renewable and Sustainable Energy Reviews 42 (2015) 10341046
4.2. Best-practice frontier water depths than those of the OWFs that were included in the
analysis. However, the validity of such projections is questionable,
In examinations of the best-practice frontier, it is important to and because the objective of this article was to provide reliable
keep in mind that DEA is a relative efciency analysis method (i.e., results, the best-practice frontier was calculated over a range
the signicance of the results strongly depends on the disposa- where it was ensured that a sufcient number benchmarks were
bility of benchmarks for the DMU under investigation). DEA available. Therefore, the capital cost best-practice frontier shown
would, for example, also allow for the projection of the specic in Fig. 5 was assessed for installed capacities of 100, 200 and
costs of OWFs that are farther offshore and at sites with greater 300 MW, for distances to shore of up to 25 km and water depths of
up to 25 m. Due to the limited database, the operating cost best-
practice frontier was assessed for installed capacities of 60, 80 and
100 MW and distances to operating port between 5 and 20 km.
Table 4
The output energy performance was set to 45%, and the time-
Results of operating cost efciency analysis.
based availability was set to 95%. The results are shown in Fig. 6.
OWF Operating year nCCR (%) nBCC (%) nscale (%) As described previously, the CCR score describes the global
technical efciency of a DMU, and the BCC score describes the local
alpha ventus 1. 98.73 100.00 98.73 pure technical efciency. Therefore, an OWF that achieves a level
Barrow 1. 37.81 37.81 99.98
of costs lying on the BCC frontier is considered to be locally
2. 40.97 40.98 99.98
Egmond aan Zee 1. 68.70 68.71 99.99 efcient but not globally efcient, which would require reaching
2. 76.56 76.57 99.99 the CCR frontier as well. The point (segment) where the two
3. 86.30 86.30 100.00 frontiers overlap is also interesting because it indicates the most
4. 99.99 100.00 99.99 productive scale size.
5. 97.32 100.00 97.32
6. 61.37 100.00 61.37
Kentish Flats 1. 69.29 69.29 100.00
2. 69.30 69.30 100.00
4.3. Tobit regression and learning-by-doing
3. 100.00 100.00 100.00
Middelgrunden 1. 63.99 69.64 91.88 Table 5 provides the results of the Tobit regression analysis
2. 100.00 100.00 100.00 performed during a second stage and the learning-by-doing rate
3. 81.22 100.00 81.22
for the different efciency scores. For the factor CUMCAP, the
4. 62.94 63.31 99.41
9. 43.28 79.72 54.28 cumulative installed offshore wind capacity in Europe provided in
10. 50.44 100.00 50.44 [1] was used.
11. 39.05 39.31 99.34
12. 42.74 43.65 97.91
North Hoyle 1. 48.06 48.07 99.97
2. 58.71 58.73 99.97 5. Discussion
3. 53.72 53.74 99.97
Scroby Sands 1. 56.90 59.41 95.78
In general, the presented methodology based on the concept of
2. 50.86 55.93 90.94
3. 55.32 57.87 95.59 DEA appears to provide a practicable approach for benchmarking
OWFs. It enables stakeholders to evaluate the reasonableness of
Minimum 37.81 37.81 50.44 the capital and operating cost levels relative to those of other
Maximum 100.00 100.00 100.00 OWFs and to estimate the target costs that would be optimal
Average 65.91 72.24 92.85
Median 62.15 69.30 99.69
corresponding to the specics (operating parameters) of the OWF
under investigation.
3,500 EUR/kW
3,346 3,346 3,353
3,303
3,000 EUR/kW 3,181 3,198
2,821
2,687 2,689
specific capital costs
2,616 2,706
2,562 2,600
2,500 EUR/kW 2,545
2,300
2,132 2,318
2,068 2,273
2,000 EUR/kW 1,818 2,003
1,910
1,300 CCR
BCC
distance to shore 10 km 15 km 20 km 25 km 10 km 15 km 20 km 25 km 10 km 15 km 20 km 25 km
water depth 10 m 15 m 20 m 25 m 10 m 15 m 20 m 25 m 10 m 15 m 20 m 25 m
60 EUR/kW
51.9
51.9 51.9 49.2 49.2 49.2
specific operating costs p.a.
41.5
41.5 40.7 40.7
40 EUR/kW
38.8 38.8 38.8
32.2 32.2
33.9 33.9
30 EUR/kW
28.5 29.6 29.6
CCR
25.9
BCC
installed capacity 60000 6000060 MW60000 60000 80000 8000080 MW80000 80000 100000 100000 100000
100 MW 100000
distance to port 5 km 10 km 15 km 20 km 5 km 10 km 15 km 20 km 5 km 10 km 15 km 20 km
Table 5
Results of Tobit regression analysis and learning-by-doing effect.
At rst glance, the capital cost best-practice frontier indicates a the presented frontiers reect the best practices and not the
strong dependence of the specic capital cost on distance to shore overall trend. Furthermore, the evaluation of economies of scale
and water depth. This nding underlines the importance of in offshore wind should be well considered because installed
quoting OWF specic capital costs as a function of the main capacity, which is commonly used to specify the scale of an
specics and the signicance of this analysis. Furthermore, com- OWF, does not refer to the number of produced and installed
paring the values with the scale factors provided by [15], the units (e.g., WECs, foundations). Thus, it is important to distinguish
impact of these cost drivers appears to have been underestimated between dimension (installed capacity) and quantity (number of
in the past. A closer look at the capital cost efciency frontier WECs). Considering the results of the Tobit regression analysis in
reveals three implications. First, the most productive scale size this context reveals that for every 100 MW increase in OWF size,
suggested by the model (i.e., where the CCR and BCC frontiers the CCR and scale efciency improved by 9.2% and 9.4%, respec-
touch) shows that smaller wind farms should be installed closer to tively, and by approximately 0.3% per additional WEC. In conclu-
shore and in shallower water. Second, at least for the range under sion, although the best-practice frontier reveals that it might be
investigation, it is possible to build a smaller OWF that is possible to develop and install a smaller OWF at a comparatively
comparatively less expensive than a larger one at the same lower cost than a larger one, the overall trend shows that the
distance to shore and water depth. Third, the size of the OWF larger the OWF is, the higher the capital cost efciency will be. The
has a weaker impact on costs the farther away from shore and the best-practice frontier for the operating cost efciency reveals
deeper the water at the OWF site. These implications suggest that similar implications, and the Tobit regression analysis also reveals
the best-practice OWFs have been developed and installed with the positive effect of installed capacity on cost efciency. However,
negative economies of scale and that this effect levels off the the database used for the operating cost efciency analysis and the
farther offshore and the deeper the water at the OWF site. OWFs considered were rather small (the largest is Egmond aan Zee
Although the presence of negative economies of scale is also OWF, with an installed capacity of 108 MW). Therefore, it is
reported by other sources (e.g., [25]), it must be kept in mind that questionable to what extent the results of the operating cost
1044 N. Ederer / Renewable and Sustainable Energy Reviews 42 (2015) 10341046
Fig. 7. Specic CAPEX and OPEX estimations by PF [4]; TCE [5] (adjusted from 2011) and RBC [6] (adjusted from 2013).
efciency are applicable to large-scale projects that have recently drivers that were not included in the model, such as the tightness
become operational. in the market of wind turbine manufacturers and installation
In general, the Tobit regression analysis generated the expected service providers [25]. However, the developed methodology
results. Nevertheless, it is remarkable that the expected relation- generated plausible results, verifying the method's practicality
ships were veried partly even with high statistical signicance. for offshore wind cost analysis.
The results show that the global capital cost efciency (CCR)
improved by 3.9% per year, and the scale efciency improved by
4.8% since the commissioning of the Middelgrunden OWF in 2001. 6. Conclusions
Furthermore, the results reveal a negative relationship ( 2.1% p.a.)
between the global operating cost efciency (CCR) and the year of In this article, the functionality of the operations research tool
operation, which is reasonable because wear and tear induces DEA was exploited with the aim of providing a useful methodol-
increased maintenance effort and thus higher operating costs the ogy that enables the evaluation of the relative capital and operat-
older an OWF becomes. Finally, an interesting issue to explore may ing cost efciency of OWFs based on their main specics.
be the role of WEC size because all OWFs considered have been Furthermore, best-practice frontiers were determined that over-
developed, installed and operated during a period of rapid come the difculties regarding the appraisal of capital and
increase in WEC size that is still on-going. Although it is not operating costs by providing offshore wind stakeholders bench-
veried with high statistical signicance, the results indicate the mark gures. The results revealed that more sophisticated cost
expected increase in capital and operating cost efciency with assessments are needed and how meaningless the declaration of
increasing WEC size. averaged specic cost gures for OWFs is. Finally, a Tobit regres-
The learning-by-doing rate for capital cost efciency shows sion analysis veried and quantied the expected relationships
that global technical and scale efciency have increased signi- between the efciency scores calculated by DEA and certain
cantly ( 9.6% and 10.3%, respectively) with accumulated experi- factors of interest, such as an increasing capital cost efciency as
ence. These values contradict the low values of 3% and 5%, a function of time, a decreasing operating cost efciency as a
respectively, presented in [25] and thus demonstrate how impor- function of the operating year and the presence of economies of
tant it is to incorporate the distance to shore and water depth in scale and learning-by-doing. Gathering cost inputs mainly from
these investigations, which was also noted by the authors of [25]. annual reports ensured a particularly reliable database and high
However, it must be kept in mind that the concept of learning-by- scientic quality, which stands out and will hopefully contribute to
doing usually relates specic costs to cumulative capacity, and the the development of an industry that is challenged to improve cost
basis of this analysis were efciency scores limited to 100%. efciency to gain competitiveness.
Fig. 7 shows a visualisation of specic CAPEX and OPEX
estimations provided by key references in this eld for the base
Acknowledgements
year 2012 (partly adjusted). These estimations deviate signicantly
from each other and also from the gures processed and calcu-
lated in this analysis. Given that these reports are inter alia the The author would like to thank the German Federal Ministry for
basis of decision making on energy policy issues such as remu- the Environment, Nature Conservation and Nuclear Safety (BMU)
neration schemes and expansion targets again demonstrates the and Projekttrger Llich (PTJ) for the permission to use wind data
importance of reliable input data and the application of compre- measured at the FINO1 research platform, as well as Hans
hensive and advanced cost analysis methods. However, the rela- Christian Srensen and Bart Hoefakker for providing operational
tive cost reduction targets presented in these publications appear data from the Middelgrunden and Egmoond aan Zee offshore wind
to be realistic. The calculated learning-by-doing rate suggests that farms, respectively.
the benchmarks presented previously should be reached with an
accumulated experience of approximately 12 GW, which corre- References
sponds to the year 2016 according to recent projections.
Finally, it should be noted, which might also be the main point [1] European wind energy association (EWEA). The European offshore wind
industry key trends and statistics 2013; 2014.
of criticism and uncertainty, that the results depend highly on the
[2] European wind energy association (EWEA). Policy issues offshore wind.
input data. It is clear that the more data included in the analysis, URL: http://www.ewea.org/policy-issues/offshore/ [accessed 10.02.14].
the more signicant the results will be. Admittedly, the database [3] Bilgili M, Yasar A, Simsek E. Offshore wind power development in Europe and
for the operating cost efciency was not excessively large, and the its comparison with onshore counterpart. Renew Sustain Energy Rev
2011;15:90515.
selection of input/output congurations was also inuenced by the [4] Prognos AG, Fichtner GmbH & Co. KG. Cost reduction potentials of offshore
availability of data. In addition, there may also be other cost wind power in Germany short version; 2013.
N. Ederer / Renewable and Sustainable Energy Reviews 42 (2015) 10341046 1045
[5] The crown estate. Offshore wind cost reduction pathways study; 2012. [39] Conroy N, Deane JP, Gallachir BP. Wind turbine availability: should it be
[6] Roland Berger Consultants GmbH. Offshore wind toward 2020 on the time or energy based? a case study in Ireland Renew Energy
pathway to cost competitiveness; 2013. URL: http://www.rolandberger.com/ 2011;36:296771.
media/pdf/Roland_Berger_Offshore_Wind_Study_20130506.pdf [accessed [40] Faulstich S, Lyding P, Tavner PJ. Effects of wind speed on wind turbine
28.02.14]. availability. In: Proceedings of European wind energy conference and exhibi-
[7] National renewable energy laboratory (NREL). Energy analysis of offshore tion 2011 (EWEC 2011). Brussels; 2011.
systems. URL: http://www.nrel.gov/wind/offshore_analysis.html [accessed [41] Chaves-Schwinteck P, Mnnich K. Temporal & energetic downtime losses and
12.02.14]. its inuence on wind farm economics. DEWI Mag 2013;42:514.
[8] Kaiser MJ, Snyder BF. Offshore wind energy cost modeling installation and [42] Scheu M, Matha D, Hofmann M, Muskulus M. Maintenance strategies for large
decommissioning. London: Springer-Verlag; 2012. offshore wind farms. Energy Procedia 2012;24:2818.
[9] Larsen JHM, Soerensen HC, Christiansen E, Naef S, Vlund P. Experiences from [43] Halvorsen-Weare EE, Gundegjerde C, Halvorsen IB, Hvattum LM, Nons LM.
Middelgrunden 40 MW offshore wind farm. Copenhagen offshore wind; Vessel eet analysis for maintenance operations at offshore wind farms.
2628 October 2005. Energy Procedia 2013;35:16776.
[10] RWE AG. Greater Gabbard Offshore wind farm. URL: http://www.rwe.com/ [44] Cook WD, Seiford LM. Data envelopment analysis (DEA) thirty years on. Eur J
web/cms/en/310132/rwe-innogy/sites/wind-offshore/in-operation/greater- Oper Res 2009;192:117.
gabbard/ [accessed 25.09.13]. [45] Seiford LM, Thrall RM. Recent developments in DEA: the mathematical
[11] BVG Associates. Value breakdown for the offshore wind sector; 2010. programming approach to frontier analysis. J Econom 1990;46:738.
[12] Smit T, Junginger M, Smits R. Technological learning in offshore wind energy: [46] Cooper WW, Seiford LM, Tone K. Data envelopment analysis a comprehen-
different roles of the government. Energy Policy 2007;35:643144. sive text with models, applications, references. 2nd ed.. New York: Springer
[13] KPMG AG Wirtschaftsprfungsgesellschaft. Offshore wind in Europe 2010 Science Business Media, LLC; 2007.
market report; 2010. [47] Charnes A, Cooper WW, Rhodes E. Measuring the efciency of decision
[14] Dicorato M, Forte G, Pisani M, Trovato M. Guidelines for assessment of making units. Eur J Oper Res 1978;2:42944.
investment cost for offshore wind generation. Renew Energy 2011;36: [48] Banker RD, Charnes A, Cooper WW. Some models for estimating technical and
204351. scale inefciencies in data envelopment analysis. Manag Sci 1984;30:107892.
[15] Eerens H, de Visser E. Wind-energy potential in Europe 20202030. Bilthoven; [49] Tobin J. Estimation of relationships for limited dependent variables. Econo-
2008. metrica 1958;26:2436.
[16] Zhou P, Ang BW, Poh KL. A survey of data envelopment analysis in energy and [50] Bogetoft P, Otto L. Benchmarking with DEA, SFA, and R. New York: Springer
environmental studies. Eur J Oper Res 2008;189:118. Science Business Media, LLC; 2011.
[17] Liu CH, Lin SJ, Lewis C. Evaluation of thermal power plant operational [51] Simar L, Wilson PW. Estimation and inference in two-stage, semi-parametric
performance in Taiwan by data envelopment analysis. Energy Policy models of production processes. J Econom 2007;136:3164.
2010;38:104958. [52] Hoff A. Second stage DEA: comparison of approaches for modelling the DEA
[18] Iglesias G, Castellanos P, Seijas A. Measurement of productive efciency with score. Eur J Oper Res 2007;181:42535.
frontier methods: a case study for wind farms. Energy Econ 2010;32: [53] McDonald J. Using least squares and tobit in second stage DEA efciency
1199208. analyses. Eur J Oper Res 2009;197:7928.
[19] Sarca K, Or I. Efciency assessment of Turkish power plants using data [54] Banker RD, Natarajan R. Evaluating contextual variables affecting productivity
using data envelopment analysis. Oper Res 2008;56:4858.
envelopment analysis. Energy 2007;32:148499.
[55] Arrow KJ. The economic implications of learning by doing. Rev Econ Stud
[20] Iribarren D, Martn-Gamboa M, Dufour J. Environmental benchmarking of
1962;29:15573.
wind farms according to their operational performance. Energy
[56] Bodde DL. Riding the experience curve. Technol Rev 1976;78:538.
2013;61:58997.
[57] Qiu Y, Anadon LD. The price of wind power in China during its expansion:
[21] Iribarren D, Vzquez-Rowe I, Rugani B, Benetto E. On the feasibility of using
technology adoption, learning-by-doing, economies of scale, and manufactur-
emergy analysis as a source of benchmarking criteria through data envelop-
ing localization. Energy Econ 2012;34:77285.
ment analysis: a case study for wind energy. Energy 2014;67:52737.
[58] Raab RL, Lichty RW. Identifying subareas that comprise a greater metropolitan
[22] European Environment Agency (EEA). Europe's onshore and offshore wind
area: the criterion of county relative efciency. J Reg Sci 2002;42:57994.
energy potential an assessment of environmental and economic constraints.
[59] Deutsche Offshore-Testfeld und Infrastruktur GmbH & Co.KG. FACT-SHEET
Copenhagen 2009.
alpha ventus. Oldenburg; 2012.
[23] Green R, Vasilakos N. The economics of offshore wind. Energy Policy
[60] EnBW Erneuerbare und Konventionelle Erzeugung AG. Windpark EnBW Baltic
2011;39:496502.
1 Der erste kommerzielle Offshore-Windpark Deutschlands. URL: http://
[24] Junginger M, Faaij A, Turkenburg WC. Cost reduction prospects for offshore
www.enbw.com/media/konzern/images/energieerzeugung/baltic/eee_bal
wind farms. Wind Eng 2004;28:97118.
tic1_8s_d_0413_lay1a.pdf [accessed 28.02.14].
[25] Van der Zwaan B, Rivera-Tinoco R, Lensink S, van den Oosterkamp P. Cost
[61] Belwind Offshore Energy. URL: http://www.belwind.eu/en [accessed
reductions for offshore wind power: exploring the balance between scaling,
25.09.13].
learning and R&D. Renew Energy 2012;41:38993. [62] DONG Energy A/S. Barrow offshore wind farm 1st annual report: July 2006
[26] 4C Offshore Limited. 4C Offshore. URL: http://www.4coffshore.com/
June 2007; 2008.
[accessed 26.10.13]. [63] DONG Energy A/S. Barrow offshore wind farm 2nd annual report: July 2007
[27] Morthorst PE, Auer H, Garrad A, Blanco I. Wind energy the facts Part III: the June 2008; 2009.
economics of wind power; 2009. [64] DONG Energy A/S. Burbo bank offshore wind farm (leaet). URL: http://www.
[28] European Wind Energy Association (EWEA). Where's the money coming dongenergy.com/SiteCollectionDocuments/NEW Corporate/Burbo/BurboBank
from? nancing offshore wind farms; 2013. Leaet.pdf [accessed 05.12.13].
[29] Bolagsverket. Bolagsverket (Swedish companies registration ofce). URL: [65] Van Pijkeren L, Hoefakker B. Offshore windfarm Egmond aan Zee 5 years of
http://www.bolagsverket.se/en [accessed 12.02.14]. operation (presentation). Ijmujiden 2012.
[30] Bundesanzeiger Verlag GmbH. Unternehmensregister (German company [66] Guneet Sands Ltd. Guneet Sands 1 & 2 offshore wind farm (leaet). URL:
register). URL: https://www.unternehmensregister.de/ureg/?submitaction= http://www.dongenergy.com/SiteCollectionDocuments/wind/guneet_
language&language=en [accessed 12.02.14]. sands/Guneet_Sands_Brochure_MAY-2010.pdf [accessed 05.12.13].
[31] Companies House. WebCHeck (UK company register). URL: http://wck2. [67] DONG Energy A/S. Horns Rev 2 offshore wind farm. URL: http://www.
companieshouse.gov.uk [accessed 12.02.14]. dongenergy.com/Hornsrev2/EN/Pages/index.aspx [accessed 26.09.13].
[32] Erhvervsstyrelsen. Det centrale virksomhedsregister (Danish company register). [68] Vattenfall Wind Power. Kentish ats offshore wind farm 1st annual report:
URL: http://www.cvr.dk [accessed 12.02.14]. January 2006December 2006; 2007.
[33] Kamer van Koophandel. Handelsregister (Dutch company register). URL: [69] Vattenfall Wind Power. Kentish ats offshore wind farm 2nd annual report:
www.kvk.nl [accessed 12.02.14]. January 2007December 2007; 2008.
[34] FOD Economie K.M.O. Middenstand en Energie. BCE Public Search (Belgian [70] Vattenfall Wind Power. Kentish ats offshore wind farm 3rd annual report:
company register). URL: http://kbopub.economie.fgov.be/kbopub/zoekwoor January 2008December 2008; 2009.
denform.html [accessed 12.02.14]. [71] Jeppsson J, Larsen PE, Larsson . Technical description Lillgrund wind power
[35] Hau E. Wind Turbines Fundamentals, Technologies, Application, Economics. plant; 2008.
2nd ed.. Berlin Heidelberg: Springer-Verlag; 2006. [72] Centrica. Lynn and Inner Dowsing wind farms project background. URL:
[36] Kaldellis JK, Zarakis D. The inuence of technical availability on the energy http://www.centrica.com/les/pdf/centrica_energy/lynn_and_inner_dow
performance of wind farms: overview of critical factors and development of a sing_background.pdf [accessed 05.12.13].
proxy prediction model. J Wind Eng Ind Aerodyn 2013;115:6581. [73] npower Renewables Limited. North Hoyle offshore wind farm 1st annual
[37] International Electrotechnical Commission (IEC). TC 88 Dashboard Project: IEC/ report: July 2004June 2005; 2005.
TS 61400-26-2 Ed. 1.0. URL: http://www.iec.ch/dyn/www/f?p 103:38:0:::: [74] npower Renewables Limited. North Hoyle offshore wind farm 2nd annual
FSP_ORG_ID,FSP_APEX_PAGE,FSP_LANG_ID,FSP_PROJECT:1282,23,25,IEC/TS report: July 2005June 2006; 2006.
61400-26-2 Ed. 1.0 [accessed 27.10.13]. [75] npower Renewables Limited. North Hoyle offshore wind farm 3rd annual
[38] International Electrotechnical Commission (IEC). Wind turbines Part 26-1: report: July 2006June 2007; 2007.
Time-based availability for wind turbine generating systems (IEC/TS 61400- [76] NWP Offshore Limited. North Hoyle offshore wind farm environmental
26-1); 2011. statement non-technical summary; 2002.
1046 N. Ederer / Renewable and Sustainable Energy Reviews 42 (2015) 10341046
[77] Vattenfall AB. Ormonde offshore wind farm. URL: http://www.vattenfall.co. [88] Feng Y, Tavner PJ, Long H. Early experiences with UK round 1 offshore wind
uk/en/ormonde.htm [accessed 26.09.13]. farms. Proc Inst Civ Eng - Energy 2010;163:16781.
[78] RWE AG. Rhyl ats offshore wind farm. URL: http://www.rwe.com/web/cms/ [89] Ofce for Gas and Electricity Markets (ofgem). ofgem E-Serve. URL: https://
en/310584/rwe-innogy/sites/wind-offshore/in-operation/rhyl-ats/summary/ www.renewablesandchp.ofgem.gov.uk/Public/ReportManager.aspx?ReportVi
[accessed 26.09.13]. sibility=1&ReportCategory=0 [accessed 18.12.13].
[79] E.ON Climate & Renewables GmbH. E . ON offshore wind energy factbook. [90] Deutsche Offshore-Testfeld und Infrastruktur GmbH & Co. KG. alpha ventus.
Dsseldorf; 2011. URL: http://www.alpha-ventus.de [accessed 28.02.14].
[80] E.ON UK Climate and Renewables. Scroby sands offshore wind farm 1st annual [91] Bundesamt fr Seeschifffahrt und Hydrographie (BSH). FINO Database.
report: January 2005December 2005; 2006. URL: http://no.bsh.de/ [accessed 18.12.13].
[81] E.ON UK Climate and Renewables. Scroby sands offshore wind farm 2nd [92] Fraunhofer Institut fr Windenergie und Energiesystemtechnik (IWES).
annual report: January 2006December 2006; 2007. Research at alpha ventus (RAVE) 2013. URL: http://rave.iwes.fraunhofer.de/
[82] E.ON UK Climate and Renewables. Scroby sands offshore wind farm 3rd
rave/pages/welcome [accessed 01.06.13].
annual report: January 2007December 2007; 2008.
[93] Noordzee Wind. Egmond aan Zee offshore wind farm. URL: http://www.
[83] Scira Offshore Energy Limited. Sheringham shoal offshore wind farm. URL:
noordzeewind.nl/en/ [accessed 01.06.13].
http://www.scira.co.uk/index.php# [accessed 26.09.13].
[94] Hoefakker B. Email communication data of Egmond aan Zee offshore wind
[84] Vattenfall AB. Thanet offshore wind farm. URL: http://www.vattenfall.co.uk/
farm; 2013.
en/thanet-offshore-wind-farm.htm [accessed 26.09.13].
[95] Middelgrundens Vindmllelaug I/S. Middelgrunden offshore wind farm.
[85] Walney (UK) offshore windfarms limited. Walney offshore windfarms. URL:
http://www.dongenergy.com/Walney/Pages/Index.aspx [accessed 26.09.13]. URL: http://www.middelgrunden.dk/middelgrunden/?q=en [accessed 20.12.13].
[86] International nancial reporting standards foundation (IFRS Foundation). [96] Srensen HC. Email communication data of Middelgrunden offshore wind
International Accounting standard 16 property, plant and equipment. URL: farm; 2013.
http://eifrs.ifrs.org/eifrs/bnstandards/en/2013/ias16.pdf [accessed 09.12.13]. [97] BVG Associates. Offshore wind: opportunities for the composites industry;
[87] Department of Trade and Industry. Offshore wind capital grant scheme 2011.
guidance notes 2013. URL: http://www.berr.gov.uk/les/le23956.pdf [98] European Central Bank (ECB). Statistical Data Warehouse. 2013 2013. URL:
[accessed 01.06.13]. http://sdw.ecb.europa.eu/home.do [accessed 06.12.13].