Applied Energy 101 (2013) 533540

Contents lists available at SciVerse ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Wind power design in isolated energy systems: Impacts of daily wind patterns
K. Suomalainen a,, C. Silva a, P. Ferro a, S. Connors b
a
b

IN+ Center for Innovation, Technology and Policy Research, Instituto Superior Tcnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisbon, Portugal
Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139-4307, USA

h i g h l i g h t s
" We apply a methodology for synthetic wind speed data including daily wind patterns.
" We study the impact of daily wind patterns on the energy system of two islands.
" We compare trade-offs between surplus energy and renewables penetration rate.
" We discover that placement of turbines can be used to minimise surplus energy.

a r t i c l e

i n f o

Article history:
Received 7 February 2012
Received in revised form 7 June 2012
Accepted 8 June 2012
Available online 21 July 2012
Keywords:
High wind penetration
Isolated energy systems
Daily wind patterns

a b s t r a c t
Increasing levels of intermittent renewables, especially wind power, in energy systems require accurate
temporal characterisation of the resources availability at seasonal, daily and hourly scales. This is crucial
for isolated energy systems, where increasing wind power penetration is limited due to costly backup
power generation requirements. In the case studies presented in this paper, the energy systems of two
islands are simulated using a new methodology for synthetic wind speed scenarios including daily wind
patterns. A trade-off analysis was conducted in terms of surplus wind power and renewables penetration
rate, with the objective of supplying decision support on wind turbine placement. Results show that there
may be a signicant advantage in locating future wind parks on sites where wind speed patterns better
match electricity demand patters, rather than just choosing a site with the highest mean wind speed, but
only if the annual mean wind speed is still sufciently high to make the investment economically
feasible.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
The Azores archipelago, an autonomous region of Portugal, has a
long history of using renewable energy to satisfy its energy needs.
The rst source of electricity on the islands was a hydro power station built in 1899 [1]. Also, the rst wind park of Portugal, consisting
of nine 30 kW turbines, was installed in the Azores in 1988 [2]. In
2010 wind power accounted for 4% of the Azores total electricity demand, whereas with current hydro and geothermal power the share
of renewable sources rises to 28% [3]. The goal of the regional government is to increase renewables penetration in the archipelago
to reach 75% of total electricity demand by 2018 [4] and consequently wind power is likely to play an important role also in the
future electricity mix of the archipelago.

Corresponding author. Present address: European Commission, Joint Research

Centre, Institute for Energy and Transport, P.O. Box 2, NL-1755 ZG Petten,
Netherlands. Tel.: +31 224 56 5097; fax: +31 224 56 5616.
E-mail address: kiti.suomalainen@ec.europa.eu (K. Suomalainen).
0306-2619/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.apenergy.2012.06.027

A key technical challenge in reaching such ambitious renewables penetration targets is the management of surpluses and
shortages of intermittent energy resources such as the wind, while
securing the quality of supply. For example by combining building
and transportation end-uses, a coherent solution for integrating
high levels of wind power is possible, even without centralised
storage [5]. However, for detailed economic evaluation, it is fundamental to accurately characterise the wind energy resource in
terms of wind power availability and the timing and scale of surpluses and shortages before the feasibility of options such as demand response, smart grids and electric vehicles can be evaluated.
Suomalainen et al. [6] studied the impact of daily wind patterns
on energy systems with high wind penetration and concluded that
daily wind patterns can have a signicant impact on the timing of
energy surpluses and shortages along the day. In the energy
systems analysis presented in this paper, the trade-offs between
fuel use and surplus energy resulting from the different wind characteristics of different locations on two islands, are evaluated. A
comparison is made with scenarios of various installed capacities
of wind power at a coastal and a mid-island location on both

534

K. Suomalainen et al. / Applied Energy 101 (2013) 533540

islands, and a common offshore location. This paper also demonstrates the impact that daily wind speed patterns can have on results from energy systems planning tools and the error resulting
from neglecting these patterns in locations where these patterns
are frequently present.
The following section gives a background to wind power in
isolated energy systems. Section 3 presents the case study and
summarises the methodology used for generating synthetic wind
speed scenarios with daily patterns. Section 4 presents the results
of applying these scenarios to the case studies followed by a
discussion and conclusions in Section 5.
2. Wind power in isolated energy systems
In this section a quick glance is given to the existing studies of
wind power in isolated energy systems and some of the lessons
learned. Also some of the needs in wind power characterisation
for isolated systems and the potential benets of a more detailed
assessment of daily variability of the wind resource are identied.
2.1. Wind variability and characteristics in isolated energy systems
Numerous studies and cases of isolated energy systems with
wind power have been documented [7,8]. In a wind energy potential assessment study Bekele and Palm [9] identied a diurnal effect of afternoon winds in their wind data from Ethiopia. In
terms of wind energy availability this becomes relevant especially
in climates where electricity demand peaks in the afternoons, e.g.
due to increased air-conditioning.
Wind characteristics in Saudi Arabia were identied for ve
geographically and climatologically different locations by Al-Abbadi [10]. Daily, monthly and frequency proles of the wind speed
showed that two sites, Dhulum and Arar, have higher wind energy
potential than the other sites due to the higher annual mean wind
speeds. However, at the coastal sites, although the annual mean
wind speed is lower than at Dhulum and Arar, the wind speed increases signicantly during afternoon hours. They concluded that
grid-connected wind power at these sites provides an advantage
in satisfying demand during peak hours.
Bowen et al. [11] studied the performance of an isolated winddiesel system for one household. They report that while almost a
fth of the wind power had to be dumped to water heating, over
one quarter of total electricity supply still came from diesel. The
winds generally reached maximum in the mid afternoon showing
a strong inuence of the coastal sea breeze especially in the summer months. The household electrical load showed regular peaks
around lunchtime and early evening, most activity ceased by
22:30 and resumed again in the early morning. They conclude that
the fact that the primary energy source, the wind, and the load
have similar proles across a typical day was a very positive feature and an excellent characteristic for a remote wind energy
power system. They agreed with Ineld and Scotney [12] saying
that even a small amount of energy storage can have a signicant
impact on the system performance. In addition, signicant
improvements in system efciency could be achieved if the energy
fed into the batteries was minimised by arranging the energy demand timing to t generation more closely.
In a more detailed study on the effects of temporal wind patterns
on the power system Fripp and Wiser [13] concluded that temporal
patterns have substantial impact on the capacity factor of wind
parks during peak hours. Locations that were best correlated with
demand peak periods produced 3040% more power during the
top 10% peak hours than on the annual average. The worst correlated
sites produced 3060% less power than on average. The relationship
between wind resource patterns and electricity demand was also

studied by Sinden [14]; it was shown that during peak demand periods the capacity factor of wind power in the UK was approximately
30% higher than on average, in this case showing a positive correlation between temporal wind patterns and demand.
For many island energy systems, due to a typically high dependence on imported fuels, increasing the penetration of local renewable energy sources to the energy mix has become an objective and
a challenge. Chen et al. [15] looked at several islands with various
renewable energy resources, including wind, and concluded that
the deployment of renewable energy in islands is a particularly
interesting opportunity for testing new technologies, in circumstances where conventional technologies are costly, and new solutions are more efcient.
Also Duc et al. [16] has conducted a number of case studies on
increasing renewables penetration on islands and concludes that
by adding a suitable storage solution it is possible to signicantly increase the penetration of local energy resources, and thus increase
security of supply and decrease dependence on imported fuels.
What remains unanswered is the question of optimising the energy
system design for islands with signicant wind resources, and this
requires including daily wind patterns in the resource
characterisation.
It is clear that daily wind patterns are frequently found in both
general climatic studies as well as in wind resource characterisation studies for energy systems modelling purposes. However, daily wind patterns are frequently omitted in energy system studies,
including studies with high wind penetration levels in relatively
small energy systems, such as islands. For example Katsaprakakis
et al. [17] conducted a study on a wind powered pumped hydro
storage system for an isolated power system, using wind speed
measurements for dimensioning and siting the wind park and estimating the annual electricity production. No daily wind patterns or
alternative locations were evaluated. In fact, due to the uncertainty
of the wind power availability the system was designed to not allow the wind power enter directly into the grid at any time. This is
one example of where including daily wind patterns, likely to occur
on islands, could give valuable additional information on the availability of wind power during peak demand hours. Thus, a higher
amount of the thermal production could possibly be offset directly
with wind power and possibly a smaller reservoir could be constructed for the pumped hydro station, if the daily wind patterns
could be characterised.
According to Zhou et al. [18], who reviewed the current status
of research on optimum sizing of stand-alone hybrid solarwind
power generation systems, unavailable meteorological wind speed
data is typically obtained by synthetically generating data from
monthly average values. This paper provides an example of a
methodology to generate such wind speed data but including daily
patterns, giving thus a higher level of characterisation of the wind
resource.
Khan and Iqbal [19] analysed a small wind-hydrogen standalone hybrid energy system noted that site specic resource and
load variations, and optimum sizing issues contribute to a challenge in such system development. The presented methodology
of the current paper supports the next step of such studies where
local specic dynamics between the wind resource variation and
power demand must be addressed. Whereas their wind model consisting of the sum of a mean term and a noise term is sufcient for
the time-span of their study, a higher level of variations needs to be
considered for energy systems modelling purposes, i.e. including
hourly, daily, seasonal and annual variations.
Kaldellis et al. [20] developed a stochastic model for estimating
the wind energy contribution in remote island electrical networks.
They characterise the wind resource by a wind speed probability
distribution and conclude that wind energy cannot play a key role
in solving the electrication issues in many Greek island regions

K. Suomalainen et al. / Applied Energy 101 (2013) 533540

without energy storage systems. However, since a distribution

gives no information on the timing of the wind speeds, time-series
data could be helpful to better understand the match or mismatch
between the resource and the demand pattern. This is an example
where studying the system dynamics at an hour-of-the-day level
may give additional insights than when using only the distribution.
The following subsection investigates how such patterns are
accounted for in energy systems models.
2.2. Wind resource characterisation in energy systems modelling
Based on the comparative report on existing energy system
modelling tools by Morch [21], it is evident that existing energy
modelling tools are mainly designed for large centralised energy
infrastructures and typically overlook the local, short-term variability in renewable energy resources at hourly, seasonal and
inter-annual scales. Of the 25 software tools assessed, the only
ones that take the variability of renewables into account are RETScreen [22] and HOMER [23].
RETScreen allows the use of locally monitored ground data or
NASAs satellite data to include monthly variability of meteorological parameters. HOMER simulates the energy system based on
hourly energy resource and demand data. The wind resource can
be introduced either by giving monthly averages, from which
HOMER simulates a set of hourly data. A daily pattern can be introduced with a factor calculated from the hourly mean values of a
historical data set, which is the most detailed approach among
the compared software tools for introducing daily patterns of
renewable energy resources.
More recently, Connolly et al. [24] have made an extensive
review of existing energy systems modelling tools focusing on
how the various tools deal with renewable resource dynamics.
According to the study, many of the tools designed to model integration of renewable energy sources with the electricity system,
such as HOMER [23], SimREN [25], EnergyPLAN [26], SIVAEL [27],
H2RES [28] and MARKAL/TIMES [29], require hourly data, i.e. wind
speed, wind energy or capacity factor data, as input for wind power
simulation. However, only one of those tools, HOMER, has the feature of generating different wind speed scenarios.
Lund [30], for example, has used EnergyPLAN for several studies
with hourly input data, with the objective to study how the energy
system can benet from the different patterns in the uctuations of
renewable energy sources. With the purpose to identify optimal
technical combinations, he concludes that only together with other
measures such as investing in exibility on both the energy supply
and demand side, and integration of the transport sector, can a
solution be found to large-scale integration of uctuating energy
sources such as wind power.
Another study using EnergyPLAN was conducted by Connolly
et al. [31] for designing scenarios for a 100% renewable energy system for Ireland. All scenarios are characterised by high levels of
wind power penetration, and are an example of where wind speed
scenarios including daily patterns could be used as input for their
more for a more detailed characterisation of the wind characteristics are different locations.
Pina et al. [32] developed a model in TIMES with an hourly
time-step to include the variability of renewable energy sources
at the seasonal, daily and hourly levels. They conclude that the
inclusion of such dynamics contributes to a better assessment of
the renewable energy potential, which is valuable for designing
effective policy measures under high renewable penetration energy systems.
The island of Malta was modelled by Busuttil et al. [33] using
the H2RES model. It uses the Meteonorm simulation tool [34] to
generate 1 year of hourly wind speeds based on interpolation models and data from the local meteorological institute. Other

535

examples of studies based on the H2RES model and using hourly

wind speed data are given in [15,16,35,36].
Considering the high variability of the wind, simulating scenarios that accurately and realistically capture the characteristics of
the wind at the modelled location is crucial for a reliable evaluation of the wind power surpluses and shortages at different levels
of wind power penetration. This links directly to the objective of
the presented research: developing a methodology for wind
resource characterisation that includes daily wind speed patterns,
which can be used as input for existing energy systems models.
This complements the advanced existing tools with a valuable input from the resource side, while avoiding the construction of an
entirely new energy model.
The generated wind scenarios are not a prediction of the wind
speed at a given hour, but a simulation of a possible chain of
events. They convey valuable information on the wind resource,
with impact on energy availability, surplus energy and potential
shortages. This new knowledge may be used in energy systems
modelling to optimise system dimensioning, in terms of both wind
power installed capacity and turbine size.
The presented research is contributes to energy systems planning
methodologies by using a method for generating synthetic wind
speed data for energy systems modelling, which captures daily wind
speed patterns more accurately than conventional methods. Most
importantly, this new methodology provides an approach to answer
to the needs of future energy systems, where intermittent energy
sources represent a signicant share of the energy mix and their
stochastic characteristics need to be well understood.
The applied methodology for wind energy resource characterisation and simulation contributes to a better understanding of
the temporal availability of wind energy and how it may contribute
to satisfying local demand. The islands studied in the current paper
are isolated systems with vast renewable energy resources and thus
a broader application of the proposed methodology to numerous
other islands and remote areas is also possible.
3. Wind power in the Azores: Faial and Pico case study
Faial and Pico, neighbouring islands in the central group of the
Azores archipelago in the Atlantic Ocean are separated by an
approximately 10 km wide strip of sea, reaching a depth of approximately 100 m. Faial is slightly more populated with approximately
15700 inhabitants, compared to Picos 14900 [37]. This is also reected in energy demand, as the islands had an electricity consumption of 55.3 GWh and 47.9 GWh, respectively, in 2010 [3].
Electricity is produced mainly using diesel generators, although
both islands have 6  300 kW installed wind capacity. In addition
Faial has a 320 kW hydro power station. In 2010 renewable energy
accounted for 5.2% and 11.6% on Faial and Pico, respectively. The
higher renewables penetration rate on Pico is explained by the
excellent wind conditions at the Pico wind park site and the fact
that on Faial the wind park is turned off for several hours every
night due to noise pollution issues that inuence nearby housing.
3.1. Scenarios
In order to reach the 2018 renewable energy targets, wind
power capacity will be increased on most of the islands. The total
number, capacity and placement of the turbines, however, has not
been dened. In this paper several combinations of the number of
turbines and their placement is studied. The scenarios are all based
on one turbine type and hub-height, with the objective to clearly
show differences in results due to the daily patterns of the wind
resource in the different locations. Thus a 330 kW turbine model already used on the islands is taken as a reference turbine for the current study. This does not imply that it is the optimum turbine

536

K. Suomalainen et al. / Applied Energy 101 (2013) 533540

Table 1
Considered scenarios for installed wind capacity (kW) in different locations on the two islands. Offshore wind is considered for Faial and Pico separately for method verication
purposes only and included in the results only in scenarios that assume a grid-connection between the two islands.
kW

Hydro, existing

Wind, existing

Wind, additional mid-island

Wind, additional coastal

Wind, additional offshore

Faial

320

1800

0  330
1  330
2  330
3  330
6  330
10  330

0  330
1  330
2  330
3  330
6  330
10  330

0
(1  2000)

1800

0  330
1  330
2  330
4  330
6  330
10  330

0  330
1  330
2  330
4  330
6  330
10  330

0
(1  2000)

1800

0  330
3  330
6  330
10  330
0  330
2  330
4  330
10  330

0  330
3  330
6  330
10  330
0  330
2  330
4  330
10  330

0
1  2000

Pico

Faial & Pico

320

1800

Table 2
Growth rate scenarios.
GWh

2010

2018
1%

2018
3%

Faial
Pico
Faial & Pico

53.9
39.5
93.4

58.4
42.7
101.1

68.3
50.0
118.3

choice, indeed other turbine models could also be used to compare

their performance in the different wind regimes. However, in order
to eliminate differences in results due to technology choice, only
the reference turbine was used in this study. The considered scenarios regarding installed wind capacity are given in Table 1.
Scenarios with offshore wind are considered for method validation, when comparing results with original data and a reference
method for synthetic wind speed scenarios. Due to the difference
in rated capacity between the offshore and onshore turbines, in
all other results the offshore option is included only in the joint
scenarios of Faial and Pico i.e. when a grid connection is assumed
between the two islands. As a modelling decision a minimum rate
of 30% of demand must be provided from the diesel generators at
any given hour, in order to guarantee adequate frequency control
and power quality.
The electricity demand has grown at an average annual rate of
2.6% in the Azores between 2004 and 2010 [3]. For Faial and Pico
the equivalent gures are 1.1% and 3.1%, respectively [3]. Due to
the uncertainty in estimating the future growth rate, and for the
sake on consistency, two annual growth rates, 1% and 3%, are applied for both islands to estimate two scenarios of annual electricity demand for 2018, as given in Table 2.
3.2. Wind data generation
The methodology used for simulating wind speed scenarios for
1 year of hourly data is based on a new methodology developed by
Suomalainen et al. [6] specically designed to capture the daily
wind speed patterns. In this methodology, the patterns identied
in measured wind speed data are simulated when producing synthetic wind speed scenarios for the energy systems model. Five different day types are dened as:
 No peak days
1. Days with no signicant variability along the day.

 Peak days
1. Days with maximum
and 06:00.
2. Days with maximum
and 12:00.
3. Days with maximum
and 18:00.
4. Days with maximum
and 24:00.

energy available between 00:00

energy available between 06:00
energy available between 12:00
energy available between 18:00

A probability distribution matrix is determined for each location based on the identied daily patterns and the mean wind
speed of each day. This matrix shows the probability distribution
for day type as a function of daily mean wind speed. Fig. 1 shows
these matrices as determined from the historical data for the ve
locations. The coastal locations on both islands show a clear diurnal effect of afternoon winds, day type 4, on days with low mean
wind speeds. However, as the daily mean wind speed increases
no peak days, day type 1, become the most probable day type.
In the mid-island locations the diurnal effect is not quite that
clear. On Faial there is a tendency to have afternoon winds on days
with low mean wind speeds, but for example on days with a mean
wind speed between 5 and 15 m/s, there is no signicant difference
in their probabilities. At Picos mid-island location there is no evidence of an afternoon wind effect. On days with the lowest mean
wind speed, no peak days are most probable, but at other daily
mean wind speeds all day types are occur at similar frequency.
In the offshore location, at low mean wind speeds, day types 2
and 5, i.e. decreasing or increasing wind speed throughout the day,
are most common whereas no peak days become common with
higher daily mean wind speeds. In addition, day types 3 and 4,
peaking in the morning and afternoon, respectively, are least frequent at the offshore location.
From these matrices a day type is sampled for each day of the
year. A reference day for each day type is scaled to the mean wind
speed of the day. An AR-model is used to add hourly variability to
the scaled reference days. For methodological details on the generation of the wind speed scenarios please see [6].
4. Trade-off analysis for different wind power scenarios
In this section the impacts of wind turbine placement, each site
option having a different wind regime, are studied through a trade-

537

K. Suomalainen et al. / Applied Energy 101 (2013) 533540

Pico, coastal
Probability

Probability

Faial, coastal

0.5
0
12

34

Day type

10

0.5
0
12

20

15

Daily mean
[m/s]

Day type

0.5

Day type

34

10

10

20

15

Daily mean
[m/s]

Pico, midisland
Probability

Probability

Faial, midisland

0
12

34

15

20

Daily mean
[m/s]

0.5
0
12

34

Day type

10

15

20

Daily mean
[m/s]

Probability

Offshore

0.5
0
12

Day type

34

10

15

20

Daily mean [m/s]

Fig. 1. Day type distributions by daily mean wind speed per location.

off analysis for the different scenarios of installed wind capacity at

each location. All scenario simulations were carried out using

MATLAB software [38].

Fig. 2 shows the trade-off between surplus energy and the energy generated by diesel generators for each simulated scenario
for Faial and Pico, using historical wind speed data, synthetic wind
speed scenarios from the new method, and synthetic wind speed

data using a commonly used energy systems simulation tool,

HOMER [23], used as a reference method to compare results.
The energy generated in diesel generators is inversely related to
the penetration level of renewable energy sources, in this case
the number of installed wind turbines. In other words, the lower
the level of energy generated in diesel generators, the higher the
penetration of renewable energy sources, and thus the higher the
level of surplus energy.
The results using the new method for synthetic wind speed data
are similar to the results of using 1 year of historical data, both
showing a clear separation of scenarios with additional wind
power only at the coastal location, marked with a star in Fig. 2,
from scenarios with additional wind power only in the mid-island
location, marked with a plus-sign. In contrast, the reference method does not capture the various daily patterns as accurately as the
new method, and thus shows only little difference between scenarios with the same additional installed capacity, but in different
locations. Consequently, the reference method estimates signicantly higher levels of surplus wind power for the same level of
diesel power, indicating that it is underestimating the rate at
which the daily wind speed patterns coincide with demand.
Figs. 35 show the results of the trade-off analysis between the
share of wind energy that is surplus and the share of renewable energy sources in electricity demand. Again, all scenarios are marked
with a dot, whereas scenarios with additional wind capacity only
in the coastal location are marked with a star and scenarios with
additional wind capacity only in the mid-island location are
marked with a plus-sign. The left-side graphs show all scenarios
and the right-side graphs give a closer look at selected scenarios
within the dashed rectangle of the left-side graphs.
The two numbers given for selected scenarios for Faial and Pico
separately in the right-side graphs (Figs. 3 and 4) give the number of
additional turbines installed at the coastal (rst number) and midisland (second number) location. In Fig. 5 selected scenarios are
identied with ve numbers indicating from left to right the number of additional turbines at the coastal location of Faial, the coastal
location of Pico, the mid-island location of Faial, the mid-island

REF.

Surplus (GWh)
Surplus (GWh)

NEW

Surplus (GWh)

FAIAL
ORIG.

PICO

20

10
10
5
0

30

40

50

0
15
20

20

25

30

35

20

25

30

35

20

25

30

35

10
10
5
0

30

40

50

0
15
20

10
10
5
0

30

40

50

0
15

Diesel power (GWh)

All scenarios

Diesel power (GWh)

Coastal

Midisland

Fig. 2. Surplus renewable energy versus diesel generated power, evaluated using a year of original data, synthetic wind speed data from the new method and synthetic wind
speed data from the reference method. Each scenario is plotted with a grey dot; scenarios with additional wind power only at the coastal location or the mid-island location
are plotted with a star and plus-sign, respectively.

538

K. Suomalainen et al. / Applied Energy 101 (2013) 533540

2018
3%

Share
wind surplus

2018
1%

Share
wind surplus

2010

Share
wind surplus

FAIAL

ZOOMEDIN
0.1
0,10

0.2
0.05

0.1

0,6

2,3

0.1

0.2

0.3

0.4

3,6 10,0

2,6

1,6

6,3

3,2 6,0 6,1 6,2

0
0.5
0.22 0.24 0.26 0.28 0.3 0.32 0.34
0.08

0.2

0,10

0.04

0.1
0

0.1

0.2

0.3

0.4

2,6 3,6
0,6 1,6
2,3
6,3 10,0
3,2 6,0 6,1 6,2

0
0.2 0.22 0.24 0.26 0.28 0.3 0.32
0.04

0.5

0.2

0,10

0.02

0.1
0

2,3
3,2

0.1

0.2

0.3

0.4

0
0.18

0.5

6,0

6,1 6,2

6,3 10,0

0.22 0.24 0.26 0.28

Renewable share of demand

Renewable share of demand

All scenarios

1,6

0.2

3,6

2,6

0,6

Coastal

Midisland

Fig. 3. Share of surplus wind energy versus renewable energy share of electricity demand in Faial. The different scenarios are marked with a dot; scenarios with additional
wind power only in the mid-island location are marked with a plus-sign (+) and scenarios with additional wind power only in the coastal location are marked with a star ().
The two numbers given for selected scenarios in the right-side graphs give the number of additional turbines installed at the coastal (rst) and mid-island (second) location.

PICO

2010

Share
wind surplus

0.4

ZOOMEDIN

0.2

0.2

0.1
0

0.2

Share
wind surplus
Share
wind surplus

2018
3%

6,0

0
0.3
0.3

0.4

0.4

2018
1%

0,10

0.3

0,6 2,6 4,6

2,4
4,2
4,4 6,4
10,0 6,2

0.35

0.4

0.45

0,10

0.2
0.2
2,4

0.1
0
0.4

0
0.25
0.2

0.2

0.1

0.2

0.4

4,6
0,6 2,6

6,4
6,0 4,2
4,4
10,06,2

0.3

0.35

0,6 2,6

4,4 6,4
6,0 4,2
10,0 6,2

0.2

0
0.25

0.4

Renewable share of demand

All scenarios

0.4
0,10

2,4

0.5

0.3

0.35

4,6

0.4

Renewable share of demand

Coastal

Midisland

Fig. 4. Share of surplus wind energy versus renewable energy share of electricity demand in Pico. The different scenarios are marked with a dot; scenarios with additional
wind power only in the mid-island location are marked with a plus-sign (+) and scenarios with additional wind power only in the coastal location are marked with a star ().
The two numbers given for selected scenarios give the number of additional turbines installed at the coastal (rst) and mid-island (second) location.

location of Pico and the offshore location. In addition, in order to

evaluate whether there is an advantage or disadvantage of joining
the grids of Faial and Pico, the combined scenarios of Faial and Pico,
without a common grid, are marked as black circles in the graphs in
the left-side graphs of Fig. 5.
As can be expected, both the share of surplus energy and the
renewables penetration rate decrease going from the 2010 reference scenario to the higher demand scenarios of 2018, in all three
energy systems. Also, a general characteristic for all energy

systems is that the share of surplus energy decreases with a

decreasing penetration rate of renewable energy sources. This
can also be expected since at lower installed wind capacities a
higher percentage of the generated renewable energy can be inserted into the grid without compromising the minimum share
of back-up energy required at all times.
It is also noticeable that none of the simulated scenarios reach
the 75% target penetration level of renewable energy sources of
electricity demand. Thus these scenarios can be seen as somewhat

539

K. Suomalainen et al. / Applied Energy 101 (2013) 533540

FAIAL & PICO

2018
3%

Share
wind surplus
Share
wind surplus

2018
1%

0.1

0,0,10,10,0

0,0,10,10,1

0.2
0.05 10,10,0,0,0
0

0.2

0.4

0.6

0.4

0
0.35

10,10,0,0,1

0.4

0.1

0.2

0.45

0.5

0,0,10,10,1
0,0,10,10,0

0.05
10,10,0,0,0

0.2

0.4

0.6

0.4

Share
wind surplus

2010

ZOOMEDIN
0.15

0.4

10,10,0,0,1

0
0.3
0.06

0.4

0,0,10,10,1
0,0,10,10,0

0.04
0.2
0.02
0

0.2

0.4

0.6

Renewable share of demand

0.5

10,10,0,0,0

0
0.25

0.3

10,10,0,0,1
0.35
0.4

Renewable share of demand

All scenarios
Coastal
Midisland
All scenarios, separated grids
Fig. 5. Share of surplus wind energy versus renewable energy share of electricity demand in Faial and Pico combined. In the left-side graphs the scenarios assuming a grid
connection between the two islands are marked with a dot and selected scenarios representing the sum of individual scenarios of Faial and Pico, without a common grid
connection, are marked with a circle (). Of the scenarios with a grid connection, scenarios with additional wind power only in the mid-island location are marked with a plussign (+) and scenarios with additional wind power only in the mid-island location are marked with a star (). The right-side graphs show only scenarios assuming a grid
connection. The ve numbers given for selected scenarios, from left to right, give the number of additional turbines installed at the coastal location of Faial, the coastal
location of Pico, the mid-island location of Faial, the mid-island location of Pico and the offshore location.

conservative in relation to the magnitudes of installed capacities.

However, for the scenarios with the highest number of wind turbines the share of surplus wind energy ranges from 20% to almost
40%, which are generally considered as very high levels of surplus
energy and would need additional, and not necessarily inexpensive, measures to be managed in a technically feasible manner.
4.1. Results for Faial
For Faial, Fig. 3, the right-side graphs show that a wind park of
ten wind turbines installed at the coastal location actually leads
to both less surplus energy and a higher rate of renewabless penetration than if the same size park was installed at the mid-island
location. This can be explained by the similar mean wind speeds
of the two locations and the diurnal effect of stronger afternoon
winds present at the coastal location, leading to a higher renewable
penetration rate than at the mid-island location. The same characteristic can be seen at a smaller scale for the indicated scenarios of a
total of six, seven, eight and nine turbines distributed at different
ratios between the two locations.
4.2. Results for Pico
In Pico, Fig. 4, this is not the case. Although the coastal location
shows a clear diurnal effect of afternoon winds, the mean wind
speed is signicantly lower than at the mid-island location. This
means that with the same installed capacity, for example ten wind
turbines, at the coastal location will lead to less surplus energy
since less wind energy is produced in total, and, for the same reason, a lower renewables penetration rate, than if placed at the
mid-island location. Another example of the advantage of locating
wind turbines at the mid-island site in Pico can be seen comparing
the scenario where two wind turbines are installed at the coastal

site and four at the mid-island site, identied as 2,4 in the rst
graph on the right-side, to the scenario where six wind turbines
are installed at the coastal site and two at the mid-island site, identied as 6,2 in the same plot. The scenarios are similar to each
other both in terms of surplus energy and renewables penetration,
yet one scenarios consists of six installed turbines and the other of
eight installed turbines in total. Clearly the scenario with a total of
six installed wind turbines is economically more feasible.

4.3. Results for joined grids of Faial and Pico

The combined scenarios, Fig. 5, shows that in terms of surplus
energy and the penetration rate of renewable energy sources, there
is an advantage in joining the two grids, as it will allow for higher
renewable energy penetration and lower shares of surplus wind
energy.
In addition, the impact of including one offshore turbine can be
seen e.g. by comparing the scenarios 0,0,10,10,0 and 0,0,10,
10,1; the 2000 kW turbine increases signicantly the share of surplus wind energy from approximately 7% to almost 10% for the
2018 scenario with a 1% growth rate, while the penetration rate
of renewable energy sources increases from approximately 41% to
approximately 47%.
An interesting trade-off can also be seen in the same graph between the scenarios 0,0,10,10,0 and 10,10,0,0,1 i.e. installing
ten wind turbines at the mid-island locations of both islands or
installing ten wind turbines on the coastal locations of both islands
and one offshore wind turbine. The renewable energy source penetration level is similar, approximately 41%, while the share of surplus
drops from approximately 7% to approximately 3%. Ofcourse the
additional costs of an offshore wind turbine would hardly justify
choosing the 10,10,0,0,1 scenario. Instead, a combination of wind
turbines at the different location should be sought that satises both

540

K. Suomalainen et al. / Applied Energy 101 (2013) 533540

the target for the renewable energy penetration level and the
acceptable level of surplus wind energy, taking into account also
the economic constraints of the system.

5. Discussion and conclusions

With ambitious renewable energy targets, the Azores are facing
the challenges of signicantly increasing renewable energy penetration in the power system with as economically sound choices
as possible. Wind power has proven to be an excellent local energy
resource in the Azores, and applying a strategic energy policy plan
to increase wind power will be an important part of the solution.
However, with limited options for locating new wind parks on
the islands and the different characteristics of the wind at different
parts of the islands, it requires careful energy systems planning,
taking into account both the resource availability and island-specic demand patterns, to design a pathway to a future energy system that reaches the set targets in an economic, secure and reliable
manner. This requires that the wind data used for planning purposes in any energy systems modelling tool describes all the relevant dynamics including daily patterns. In many cases measured
wind data is not available, and when synthetic data that does not
contain these dynamics is used, the results may be misleading.
Further, this paper has presented some of the trade-offs
involved in choosing the location for a future wind park, each with
different daily wind speed patterns. It was shown that in certain
cases, as in the case of Faial, a location with clear diurnal effects
may lead to better penetration rates of wind power and lower
shares of surplus wind energy than a location with no diurnal
effects. In other cases, namely if the annual mean wind speed is
signicantly lower than at the alternative site with no diurnal effects, as in the case of Pico, the positive effect of producing more
energy during peak hours may be out-weighed by the negative effect of not producing enough energy in total.
It was also shown that from a pure wind resource perspective,
there is an advantage of joining the two grids of Faial and Pico,
as this increases the total penetration rate of wind power compared to if the two grids are considered separately. The problem
of high shares of surplus wind energy also raise the concern of
additional technical solutions that may be necessary, if the 75% target for renewable energy resource penetration is to be reached for
these islands individually.
It is concluded that there may be a signicant advantage of locating future wind parks in locations with higher wind speeds during
periods of the day when electricity demand is high, but only if the
annual mean wind speed is also sufciently high to make the investment economically feasible. For future energy systems where technological solutions such as smart grids, electric vehicles and
demand response allow higher penetration rates of wind power, it
becomes crucial to be able to simulate hourly wind speed scenarios
that include the natural variability of the wind at interannual, seasonal, daily and hourly levels.

References
[1] Expresso das Nove. Renovveis com 75% de penetrao nos Aores newspaper
article. <http://www.expressodasnove.pt/interiores.php?id=3664> [accessed
06.11].
[2] Regio Autnoma dos Aores. Estudo do potencial elico. Elaborada para:
ARENA, Agncia Regional de Energia dos Aores, Freely translated:
Autonomous Region of the Azores, Study of wind potential, Conducted for:
ARENA, Regional Energy Agency of the Azores, inegi Instituto de Engenharia
Mecnica e Gesto Industrial; 2004.
[3] Informao Estatstica. Dezembro 2004Dezembro 2010. Freely translated:
Statistical Information. December 20042010. EDA Electricidade dos Aores;
2011.

[4] Csar C. Apresentao do plano estratgico para energia nos Aores. Freely
translated: presentation of the strategic energy plan for the Azores. Regional
Government of the Azores; 2008.
[5] Lund H, Kempton W. Interation of renewable energy into the transport and
electricity sectors through V2G. Energy Policy 2008;36:357887.
[6] Suomalainen K, Silva CA, Ferro P, Connors S. Synthetic wind speed scenarios
including diurnal effects: implications for wind power dimensioning. Energy
2012;37:4150.
[7] Hansen LH, Lundsager P. Review of relevant studies of isolated systems.
Roskilde: Riso National Laboratory; 2000.
[8] Lundsager P, Bindner H, Clausen N-E, Frandsen S, Hansen LH, Hansen JC.
Isolated systems with wind power. Main report. Roskilde: Riso National
Laboratory; 2001.
[9] Bekele G, Palm B. Wind energy potential assessment at four typical locations in
Ethiopia. Appl Energy 2009;86:38896.
[10] Al-Abbadi NM. Wind energy resource assessment for ve locations in Saudi
Arabia. Renew Energy 2005;30:148999.
[11] Bowen AJ, Cowie M, Zakay N. The performance of a remote wind-diesel power
system. Renew Energy 2001;22:42945.
[12] Ineld DG, Scotney LPA. Wind-diesel systems design assessment and future
potential. Energy research unit. Oxfordshire, UK: Rutherford Appleton
Laboratory; 1992.
[13] Fripp M, Wiser RH. Effects of temporal wind patterns on the value of windgenerated electricity in California and the Northwest. IEEE Trans Power Syst
2008;23(2):47785.
[14] Sinden G. Characteristics of the UK wind resource: Long-term patterns and
relationship to electricity demand. Energy Policy 2007;35:11227.
[15] Chen F, Duc N, Alves LM, da Graa Carvalho M. Renewislands renewable
energy solutions for islands. Renew Sustain Energy Rev 2007;11:1888902.
[16] Duc N, Krajacic G, da Graa Carvalho M. RenewIslands methodology for
sustainable energy and resource planning for islands. Renew Sustain Energy
Rev 2008;12:103262.
[17] Katsaprakakis DA, Christakis DG, Pavlopoylos K, Stamataki S, Dimitrelou I,
Stefanakis I, et al. Introduction of a wind powered pumped storage system in
the isolated insular power system of KarpathosKasos. Appl Energy 2011.
[18] Zhou W, Lou C, Li Z, Lu L, Yang H. Current status of research on optimum sizing
of stand-alone hybrid solarwind power generation systems. Appl Energy
2010;87:3809.
[19] Khan MJ, Iqbal MT. Analysis of a small wind-hydrogen stand-alone hybrid
energy system. Appl Energy 2009;86:242942.
[20] Kaldellis JK, Kapsali M, Tiligadas D. Presentation of a stochastic model
estimating the wind energy contribution in remote island electrical
networks. Appl Energy 2011.
[21] Morch AZ. Software tools for energy planning: overview and comparison.
SINTEF energy research. Energy systems, project memo; 2005.

[22] RETScreen international. Empowering cleaner energy decisions. Natural

resources Canada. <http:www.retscreen.net/ang/home.php> [accessed 07.11].
[23] HOMER Energy, USA. The micropower optimization model, Ver. 2.68 Beta,
http://www.homerenergy.com [accessed 06.11].
[24] Connolly D, Lund H, Mathiesen BV, Leahy M. A review of computer tools for
analysing the integration of renewable energy into various energy systems.
Appl Energy 2010;87:105982.
[25] Institute for sustainable solutions and innovations. Energy rich Japan. <http://
www.energyrichjapan.info> [accessed 06.11].
[26] Aalborg University. EnergyPLAN: advanced energy system analysis computer
model. <http://energy.plan.aau.dk> [accessed 06.11].
[27] Energinet.dk,
SIVAEL.
<https://selvbetjening.preprod.energinet.dk/
www.energinet.dk/en/menu/Planning/Analysis+models/Sivael/SIVAEL.htm>
[accessed 06.11].
[28] H2RES energy planning of Islands and isolated regions. Faculty of Mechanical
Engineering and Naval Architecture, and University of Zagreb. <http://
www.powerlab.fsb.hr/h2RES/> [accessed 06.11].
[29] International Energy Agency. Energy technology systems analysis programme.
<http://www.etsap.org> [accessed 06.11].
[30] Lund H. Large-scale integration of optimal combinations of PV, wind and wave
power into the electricity supply. Renew Energy 2006;31:50315.
[31] Connolly D, Lund H, Mathiesen BV, Leahy M. The rst step towards a 100%
renewable energy-system for Ireland. Appl Energy 2011;88:5027.
[32] Pina A, Silva C, Ferro P. Modeling hourly electricity dynamics for policy
making in long-term scenarios. Energy Policy 2011;39:4692702.
[33] Busuttil A, Krajacic G, Duc N. Energy scenarios for Malta. Int J Hydrogen
Energy 2008;33:423546.
[34] METEOTEST. METEONORM global meteorological database for engineers,
planners
and
education.
<http://www.meteonorm.com/pages/en/
meteonorm.php> [accessed 06.11].
[35] Krajacic G, Duc N, da Graa Carvalho M. H2RES, Energy planning tool for island
systems the case of the Island of Mljet. Int J Hydrogen Energy
2009;34:701526.
[36] Duc N, da Graa Carvalho M. Increasing renewable energy sources in island
energy supply: case study Porto Santo. Renew Sustain Energy Rev
2004;8:38399.
[37] Statistics Azores. Portugal. Demography. <http://estatistica.azores.gov.pt>
[accessed 06.11].

[38] MATLAB the language of mathematical computing. MathWorks. <http://

www.mathworks.com/products/matlab/index.html> [accessed 06.11].