Energy 34 (2009) 1236–1245

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Reviewing optimisation criteria for energy systems analyses of renewable

energy integration
Poul Alberg Østergaard*
Department of Development and Planning, Aalborg University, Fibigerstræde 13, 9220 Aalborg, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: The utilisation of fluctuating renewable energy sources is increasing world-wide; however, so is the
Received 26 November 2008 concern about how to integrate these resources into the energy systems. The design of optimal energy
Received in revised form resource mixes in climate change mitigation actions is a challenge faced in many places. This optimi-
16 March 2009
sation may be implemented according to economic objectives or with a focus on techno-operational
Accepted 1 May 2009
aims and within these two main groupings, several different criteria may potentially be applied to the
Available online 13 June 2009
design process.
In this article, a series of optimisation criteria are reviewed and subsequently applied to an energy
Keywords:
Renewable energy integration system model of Western Denmark in an analysis of how to use heat pumps for the integration of wind
Optimisation criteria power.
Energy systems analyses The analyses demonstrate that the fact whether the system in question is modelled as operated in
Energy cities island mode or not has a large impact on the definition of the optimal wind power level. If energy savings
Energy islands and CO2 emission reductions beyond the system boundary are not included in the analysis, then it is
either not feasible to expand wind power to a high degree or it is conversely more feasible to install
relocation technologies that can utilise any excess production. The analyses also demonstrate that
different optimisation criteria render different optimal designs.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction The transition from fossil fuel-based energy supplies to

renewable energy supplies has a number of impacts on the energy
Increasing attention is given to the abatement of climate change, system. With a few exceptions, such as bio-fuels and hydropower,
to the societal costs of covering an ever growing energy need and to most renewable energy sources are of a fluctuating nature and are,
the security of supply in countries which, in many cases, rely on furthermore, of a ‘‘use it or lose it’’ character. In spite of these
politically volatile regions for their energy supply. Due to these constraints, systems relying on e.g. wind power – whether small-
circumstances, cities, regions and countries focus on harnessing scale wind-diesel hybrid systems or large-scale systems involving
locally available renewable energy sources. Many geographic more types of productions and demand – still need to have the
locations are hence in the process of making local energy plans same load following capabilities as conventionally fuelled energy
with the aim of becoming renewable energy cities or renewable systems.
energy islands. This includes cities such as Frederikshavn, At a general level, such systems may be designed from an
Denmark; Dardesheim, Germany, and Dong-Than in China, and economic perspective or from a techno-operational perspective;
islands like Samsø, Denmark, and Gotland, Sweden [1] but also, to but within these two pillars, several sub-divisions can be found.
a lesser extent, entire countries, of which, e.g., Denmark has had Economic optimisations criteria include e.g. total energy systems
ambitious renewable energy targets for a long time – see, e.g., [2– costs, capacity costs and societal costs. From a techno-operational
4]. However, questions remain open: What is a renewable energy perspective, optimisation criteria include fuel savings, CO2 emis-
city? When can an area claim to have a sustainable energy supply? sions, reserve/back-up capacity, required condensing mode power
And how does one determine the optimal resource mix? generation, minimisation of import/export, and elimination of
excess power generation. All of these criteria can be applied to
assess how well the system integrates renewable energy. In addi-
* Tel.: þ45 99408424; fax þ45 98153788. tion to these criteria, systems may be analysed in either island
E-mail address: [email protected] mode or as connected to surrounding areas. This opens up for new

0360-5442/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2009.05.004
 P.A. Østergaard / Energy 34 (2009) 1236–1245 1237

issues, such as whether or not CO2 emission reductions realized in renders a positive effect on the fuel use beyond a certain threshold
neighbouring areas, as a consequence of changes in the area in value.
focus, should be credited this area. However, while this would be an analytically quantifiable
Many scholars have treated the issue of integrating renewable approach, it would not result in an objective identification of an
energy sources into energy systems without compromising the optimal system design. A possibly better approach is to set up,
load following capability of the system (see e.g. [5–17]). However, analyse and compare alternatives. Methods such as the Diamond E
there is no generally accepted common design criterion according (see [20]) can then be used to help identifying which requirements
to which the systems are analysed, optimised and designed. This is the system must meet and thereby help establishing relevant
also the case of different political institutions, which use different criteria.
criteria when defining aims. While some articles focus on economic criteria and others focus
on technical criteria, it is in practice difficult to make a clear
2. Scope of the article distinction between these, as also deliberated in the following
review of the criteria.
This article reviews a number of possible optimisation criteria
for the design of energy systems with large shares of fluctuating
3.1. Reserve capacity requirement
renewable energy sources. A selection of these criteria is applied to
an analysis of Western Denmark. This is a region with a large
From a technical point of view, reserve capacity (or backup
penetration of wind power at approx. 26% in 2007 (based on hourly
capacity) requirement is a parameter used in, e.g., [21] to assess
production and consumption data from [18]) and where wind
how well a system integrates wind power. By using this parameter,
power is expected to increase even further. Denmark is also
the maximum required generating capacity should basically be
a country with a large share of the heat demand being covered by
minimised over a period of time, e.g., a year or more. The criteria
district heating; generally produced on cogeneration of heat and
may or may not take fault situations into consideration – so-called
power (CHP) plants giving rise to a district heat–tied electricity
n-1 or n-2 in which one or more production units are unavailable.
generation. A high wind penetration in combination with district
This, of course, adds to the required reserve capacity. However, this
heat–tied electricity generation makes the area very relevant for
added reserve capacity should not necessarily be attributed to the
studies of integration of wind power. In the article, a model is thus
circumstance that the analysed system is characterised by a large
set up detailing electricity and district heat demands, conventional
share of fluctuating renewable energy sources. The added reserve
thermal power plants, CHP plants, heat storages in connection with
capacity is irrespective of the type of energy system. In fact, the
the CHP plants, wind turbines and foreign transmission connec-
replacement of one 300-MW unit in a thermal power plant by 100
tions. Based on this model, it is analysed how wind power can best
3-MW wind turbines decreases the impact of a faulty unit.
be expanded from 20 to 40% of the demand, using heat pumps for
However, as n-1 or n-2 considerations deal with the most critical
integration. Heat pumps are not the only technology that may assist
units – and hence any reserve capacity in renewable energy-based
in the integration of wind power, however it is a moderately priced
systems – any addition which they provide to the reserve capacity
and energy efficient technology that already exist in large-scale
is typically irrespective of whether the system is an ordinary
applications such as the Stockholm District Heating system. This is
thermal system or a renewable energy-based system.
in contrast to costly electricity storage technologies based on, e.g.,
Explicitly or not, a strong relationship can be found between
vanadium redox batteries or based on hydrogen which in addition
reserve capacity considerations and economic costs. The capacity
have poor cycle efficiencies. Technologies such as compressed air
naturally has a fixed cost which must be covered by the electricity
energy storages (CAES) have also yet to prove economically
consumers, regardless of the fact whether or not this capacity is
attractive as demonstrated by, e.g., Lund & Salgi [19]. Heat pumps
used.
are a logical option due to the existence of CHP plants, district
heating grids, and heat storages in Denmark. They give a very
energy efficient downward regulation possibility by introducing an 3.2. Use of import and export
electricity demand while at the same time reducing the heat tied
production on CHP plants. Their upwards regulating ability is Related to reserve capacity considerations is the use of import or
restricted to when they are operating though. export in securing a system’s load following capability. Import and
reserve capacity play similar roles in the energy system. Export
3. Optimisation criteria from the system is not per se a positive or negative quality from
a technical perspective, and is hence not directly convertible to
A wide range of criteria exists for the design of optimal energy a design optimisation criterion. Critical import/export – i.e.,
systems configurations. In the following, a number of these are required import/export beyond the transmission line capacity –
presented and deliberated. Within some fields, it is possible to must of course be avoided. If not, then added transmission capacity
determine a global extreme. This may e.g. apply to per-unit costs as must be considered. An example of analyses aimed at limiting the
a function of the production volume in a manufacturing industry export of electricity is found in [2]. In the case of critical export,
where start-up costs are high and where additional labour or production units may also be shut down at the expense of the
additional machinery will, at some point, be required. In between system’s capability of exploiting available ‘‘use it or lose it’’
these two points, a global (or at least a local) minimum may be resources.
found. In contrast, when analysing e.g. the optimal expansion of From an economic perspective, non-critical export may provide
wind power and using, e.g., fuel use as an optimisation criteria, the an income. However, if the export takes place at times when the
fuel used is much more likely to approach a fixed level, more or less given energy system is forced to export due to, e.g., windy condi-
asymptotically. Without a clear extreme of the fuel use, in the case tions and a lack of local integration capability, then the seller is in
of installed wind capacity function, this cannot simply be used as an a poor bargaining position, as there is no alternative to transmitting
optimisation parameter. One could consider using the derivative the excess electricity beyond the system boundary. The same is
and setting a limit to this. In the physical world, this would mean naturally the case with import – though this is of course a potential
expanding wind power as long as the incremental expansion expense.
1238 P.A. Østergaard / Energy 34 (2009) 1236–1245

3.3. Island mode, connected mode or connected island mode systems is not unlike Kant’s Categorical Imperative [22] or Egner’s
Cardamom Law [23].
The discussion of import/export opens up for the much wider If the area in question is literally islanded, then this approach is
discussion of whether or not the system shall (or shall be able to) of course also the relevant one.
operate in island mode, i.e., without relying on transmission The connected approach resembles the current Danish situation
capacity to and from the outside world. There are arguments in in which imbalances to some extent are remedied with the assis-
favour of the island mode approach, the connected approach and tance of the balancing capabilities of surrounding areas. Or in other
what may be coined ‘‘connected island mode’’, where the latter words, the problems are shared with the neighbours. As only few
refers to a system which is connected to the outside world but in areas have high penetrations of fluctuating renewable energy
which the use of the inter-connector is avoided, if possible. sources, this approach is not a problem in most of today’s systems.
In favour of the island mode approach is the circumstance that However, there may be economic issues to address, as noted under
while it is indeed possible to have large shares of, e.g., fluctuating import/export.
wind power in an energy system already with the present tech- Harnessing fluctuating renewable energy sources over a larger
nologies, this is to a high degree due to the fact that the system can geographic area does offer some synergies in terms of evening out
rely on the outside world in terms of balancing supply and demand. natural variability through spatial distribution, but the effects of
Analysing an area in island mode thus reveals more about the this are limited [21].
system’s dynamics. Designing a system that may function in island Finally, the Connected island mode attempts to bridge the two
mode may thus enable the system operators to make voluntary former approaches by giving priority to the ability to operate in
decisions on when to import and export, rather than being forced island mode – but also by including a certain possibility of exchange
through external and non-controllable circumstances. with the surroundings. The issue remains of course how to estab-
Whether or not a system is modelled as operated in island mode lish a limit of permissible exchange; is it only to be permitted in
is also related to how credits are treated across boundaries. If contingency situations; is it also permitted under non-contingency
energy saving and CO2 emission reductions beyond the system but still abnormal conditions, or is it even permitted as a daily
boundary are not credited in the system – which is the case in, e.g., occurrence.
the Nordic power pool Nord Pool – then it may either not be
feasible to expand wind power to a high degree or be conversely
3.4. Condensing mode operation
more feasible to install relocation technologies that can utilise any
excess production.
In energy systems with CHP plants, electricity generation on
For Western Denmark, the duration curve in Fig. 1 demonstrates
condensing mode power plants is usually avoided to the highest
the high reliance its wind power intensive system has on the ability
extent possible. An example of an analysis applying this method-
to trade electricity with neighbouring countries. Wind power alone
ology is [15]. While having higher electric efficiencies than back-
accounted for more than 100% of the electricity demand in this area
pressure or extraction CHP plants, condensing mode power plants
for more than 50 h of 2007, and during 1574 h (18% of the time), the
have far lower total efficiencies, as substantial amounts of waste
contribution was larger than 50%. With a large amount of district
heat are discarded. Regardless of whether the system is exploiting
heating tied CHP electricity production, there are further restric-
CHP plants or not, fuel use is to be minimised. In non-CHP systems,
tions on the energy system emphasizing the fact that surrounding
however, a more immediate relationship is found between fluctu-
areas are used for balancing purposes in the current situation.
ating renewable energy input and fuel savings on condensing mode
In a future situation in which such neighbouring regions may
power plants. In systems with CHP, the optimisation process offers
also exploit fluctuating energy sources, these regions will probably
more possibilities, such as the appropriate use of heat storages and
not have the flexibility to assist other regions in load balancing.
the scheduling of the CHP plants. There is, hence, not a simple
Hence, in a future situation with extensive use of fluctuating
correlation between fluctuating renewable energy input and fuel
renewable energy sources, the system may need to be, if not
savings in such systems. The impact is determined by the system’s
physically then virtually, split up into a number of self-reliant
configuration and the choice of regulation strategy, making
subsystems each with appropriate load following capabilities. The
condensing mode operation an interesting performance indicator.
line of reasoning for opting to model and design self-reliant energy

3.5. Primary energy consumption (PEC)/Fuel use

120 Rather than focusing on condensing mode power generation

and the fuel savings that may be achieved through the mini-
Wind share of demand [%]

100 misation of this, a wider approach is simply to look at the PEC of the
energy system in question. The issue of import/export plays a role
80 here. Should the system in question be credited fuel savings beyond
the system boundary caused by the export of, e.g., electricity,
60 district heating or other energy carriers produced within the
system? Should the system be debited import of these – and how
40 should the fuel equivalence be assessed? Focusing on electricity,
one way would be to assume that the alternative to import/export
20 would be running the marginal production facility more or less.
This would typically be a condensing mode power plant. It may be
0 argued that the marginal production in complex systems with
0 20 40 60 80 100
many production technologies, energy carriers and interdepen-
Share of the year [%]
dencies is not always clearly identified as it may neither be the
Fig. 1. Duration curve of wind power’s relative contribution to the electricity demand same in all hours of the year nor the same under all operating
in Western Denmark 2007. Compiled on the basis of data from [18]. conditions.
 P.A. Østergaard / Energy 34 (2009) 1236–1245 1239

Thus, the applicability of using a marginal production facility to described OECD, IEA and Eurostat methodology makes energy
compensate for import/export has its limits. Export is typically only systems with substantial renewable energy sources, like wind
relevant when the condensing mode power plants cannot be power, appear more favourable compared to conventionally fuelled
down-regulated any further, in which case the condensing mode energy systems; however, the methodology tends to hide such
power plants are not an option and thus do not represent the energy sources when renewable energy shares are calculated. Table
marginal production facility. Likewise, import would also typically 1 shows the effect of applying the OECD methodology to a simple
be given second priority to production within the system, which system with 20 TWh of wind power and 80 TWh of nuclear power.
means that import would only occur when the required production Clearly, the PEC is lower with the OECD methodology, but the
exceeds the available capacity of the system. The condensing mode renewable energy fraction appears more favourable when wind
power plant would not be the correct marginal facility in this case, power is modelled as, e.g., nuclear power.
either. A different application of the methodology would be to use To complicate matters further, renewable energy shares may
the marginal production facility in the neighbouring country/area also be calculated in terms of final energy consumption. Using the
in the same way as the system-internal marginal production facility same efficiencies as in Table 1 in a system with only electricity
was used. consumption in fact corresponds to using final energy consump-
Other energy carriers may be treated in a similar way as elec- tion. However, this is an exceptional case.
tricity, with, e.g., a boiler for district heating, an electrolyser for A newly defined European Union goal of achieving 20% renewable
hydrogen, etc. Of course, some of the same issues as outlined for energy coverage of the energy demand relates to final energy
electricity are relevant here, as well. consumption [25], whereas e.g. national Danish statistics include the
Contemplating fuel use, it may be considered to focus a) solely losses in the energy transformation sector [26] and thus refer to PEC.
on actual physical fuels, as suggested by the term ‘‘fuel use’’; b) The burden of meeting a certain requirement is generally
solely on non-renewable fuels, or c) on all energy sources including reduced when applied to final energy consumption rather than to
wind, solar, wave, etc. This is demonstrated in Fig. 2, in which it is PEC. If the renewable energy sources have higher losses through the
noticed that waste is indicated as a renewable as well as a non- transformation system, the burden is not reduced. If the fuel
renewable source. It could be argued that waste is renewable on the equivalence of solar cells is calculated using an actual physical
grounds that, despite the fact that a large proportion derives from efficiency of, e.g., 10%, they will constitute a much higher produc-
fossil fuels, the marginal effect of combusting, e.g., plastic produced tion in terms of PEC than in terms of final energy consumption.
on the basis of hydrocarbons is nil if decomposition in nature is the However, as the fuel equivalence of wind power, solar cells and
alternative. This process is very slow, though. some other electricity producing technologies is typically modelled
According to Organisation for Economic Co-operation and as being identical to the electricity production, according to OECD,
Development (OECD), International Energy Agency (IEA) and IEA and Eurostat methodology, this is most often not the case.
Eurostat methodology, waste is considered partly renewable and, The burden of meeting specific targets will therefore also be
as such, should be distributed according to the biodegradable/non- reduced more for countries with inefficient transformation systems
biodegradable fractions, see [24]. than for countries with more efficient transformation systems,
Also according to OECD, IEA and Eurostat methodology, nuclear when adapting a goal related to final rather than PEC.
is accounted for in terms of fuel equivalence, assuming that it is
produced at a power plant with an electric efficiency of 33%,
whereas the fuel equivalence of, e.g., wind power is calculated 3.7. Carbon dioxide emissions
using an efficiency of 100%. Using this methodology, thus gives
a lower PEC for systems with wind power than for systems with Optimising the design by quantifying – and minimising – carbon
nuclear power. dioxide emissions is a success parameter used in many national
energy plans, as a consequence of the work in the United Nations
3.6. Renewable energy shares Framework Convention on Climate Change and the country-specific
targets lined up in the Kyoto Protocol [27]. Carbon dioxide emissions
Measuring the renewable energy share of the energy are closely linked to the fuel use considered previously with the
consumption is yet a fuel-accounting criterion. Again, it must be distinction that different fuels have different carbon dioxide emission
settled how to account for import/export of energy carriers. It is factors. Thus, where PEC was neutral to the fact whether fuels were
also important how non-fuel renewable energy sources are con- high emission fuels, like lignite, or low emission fuels, like natural gas,
verted to fuel equivalence. One may argue that the previously biomass or even nuclear, the distribution of fuels is important here.
The considerations pertaining to import and export are clearly
relevant for carbon dioxide emissions, too. This has, in fact, proved
Energy sources: to be an obstacle in the internal European Union re-distribution of
Kyoto Protocol requirements. Countries like Denmark have unsuc-
Fuel use: cessfully advocated that the base-year should be corrected for
international electricity trade and for climatic deviations from the
Wind long-term average, which affects the demand for space heating.
Solar Non-RE fuel use:
Biogas Related to carbon dioxide emissions are newer concepts such as
Wave Biomass Fossil oil
Geothermal Fossil gas carbon footprints – see, e.g., [28]. In the simplest form, the carbon
Waste footprint merely corresponds to the carbon dioxide emissions of
Hydro Fissile material
Tidal Waste
Ocean
Table 1
Renewable energy share determination.

OECD methodology Wind modelled as nuclear

Renewable energy share 7.7% 20.0%
PEC 260 TWh 300 TWh
Fig. 2. Energy source vs fuel use vs non-RE fuel use.
1240 P.A. Østergaard / Energy 34 (2009) 1236–1245

a given area or activity, but it may also be based on a life cycle externalities may e.g. be added and the analysis may be conducted
assessment (LCA), thus including emissions throughout the life using import values. This is particularly the case in situations of
cycle of the energy system in question, involving e.g. construction, unemployment, where it may be argued that domestic labour costs
demolishment, growing of energy crops, and the production of for the society are zero if the labourers in question would not
fertilisers for energy crops. LCA is also used in its own right to assess otherwise be contributing to the wealth of the society.
energy technologies as exemplified by [29]. The LCA Carbon foot-
print methodology is also applied to non-energy products – e.g.,
[30], thereby providing a common frame for environmental impact 3.10. Cost of energy (CoE), utility costs and the rate impact
assessments. The Ecological footprint (see, e.g., [31–33]) adopts
another approach and determines the land area required to sustain Utility costs adopt, as the term suggests, a much more confined
a certain activity. Thus, it gives an indication of sustainability business economic approach. Apart from the case of formal IRP, this
through a comparison of the required land area to the actual land term is not applied broadly. Related to the utility costs is, however,
use or the available land area. the CoE, which is used broadly by, e.g., [36–38]. This parameter
determines the cost per unit of energy whether being electricity,
district heating, natural gas, hydrogen, etc. It is, thus, also closely
3.8. Economic costs
related to the Rate impact, with the main difference lying in the
application. The CoE is often applied to simple systems – e.g., the
Economic cost evaluation is a main parameter for assessing the
cost of producing 1 kWh on a given solar cell panel, on a given fuel
feasibility of a given energy system configuration. However,
cell or in a given wind diesel hybrid system. The rate impact is
‘‘economic costs’’ is also a term with ample latitude for interpre-
typically more holistic as it determines how the energy price for
tation. As demonstrated by Integrated Resource Planning (IRP) (see
consumers is affected. As consumers are supplied by systems of
[34] for one of the first journal references to this planning meth-
varying complexity, more factors and a system-oriented approach
odology), several types of economic costs can be defined, including
are hence required to determine the rate impact than in the case of
the simple CoE of a single technology.
 Societal costs –also including, e.g., external costs and benefits,
The CoE is also referred to as the Levelised Cost of Energy (LCE)
 utility costs – determining the costs incurred by the energy
or Levelised Unit Electricity Cost (LUEC), as in [39]. Both of these
companies,
terms emphasize the fact that the cost is determined over a certain
 rate impact – determining the cost of each unit of, e.g., elec-
time horizon – e.g., the life time of the specific technology. In
tricity, and
practise, the objective it is a matter of ‘‘finding the price that sets the
 total resources – determining all incurred costs.
sum of all future discounted cash flows (net present value, or NPV) to
zero’’, as phrased by Ayres et al [39].
For the sake of completeness, a last parameter in IRP is Partici-
pant Costs, which quantifies benefits for consumers engaged in
Demand Side Management (DSM) projects. However, this param- 3.11. Total resources
eter is too specific for the aim of this article.
One common feature for most economic criteria is the fact that Total resources in IRP terminology refer to resources spent by
the costs can be determined as net present values, annual costs, or consumers, energy companies and government bodies (e.g., as
levelised annual costs. subsidies) alike. They are not to be confused with societal costs, as
societal costs also include costs without a direct market value that
3.9. Societal costs hence need to be monetized beyond the market system.

An important element in IRP is the analysis of how different

actors are influenced by given measures. This is also seen in the list 3.12. Marginal costs
of costs, in which each cost criterion focuses on the costs for
a distinct target group. Hence, in IRP, a given measure should prove Closely related to optimisation criteria are long-term marginal
beneficial according to all of the above parameters, though in effect, costs and short-term marginal costs (sometimes referred to as
this is rarely possible. For this reason, societal costs are typically short-run marginal cost and long-run marginal cost). Here, the cost
used as the measuring gauge in IRP work as exemplified by [35]. of producing (or consuming) one extra unit of energy is calculated
However, societal costs are also used beyond formal IRP, as exem- (short-term marginal cost) or the cost of producing (or consuming)
plified by [2] who applies these costs as a parameter for the opti- one extra unit of energy including capacity expansion/renewal is
misation of energy systems. While the term Societal costs relatively calculated (long-term marginal cost). But as applied in, e.g., [40],
easily may be defined as ‘‘all cost to the society’’, this is, in practise, this may rather be used to check for economic sustainability by
more complicated for boundary reasons. From the simplest point of comparing these unit costs to the unit cost of energy for consumers.
departure, the societal cost of a given good might be approximated
as market price excluding taxes, but progressively more ambitious Table 3
definitions may include more and more elements. Environmental Efficiencies of the modelled system. The efficiencies for dwellings not connected to
district heating networks vary depending on fuel; with the lowest efficiency for
biomass boilers and the highest for natural gas boilers.
Table 2
Energy system scenario parameters. Electric efficiency Heat efficiency
Individual heat – 70–90%
Consumption [TWh] Generating capacity [MW] Boiler DH – 88%
24.87 Electricity 1450 Small-scale CHP Small CHP 38% 49%
21.21 CHP District heating 1300 Large-scale CHP Boiler – 90%
3.50 Boiler district heating heat Unlimited Power plants in condensing mode Large CHP 39% 47%
2500 Wind onshore Boiler – 90%
1445 Wind offshore Condensing 50% –
 P.A. Østergaard / Energy 34 (2009) 1236–1245 1241

Table 4
Distribution of fuels used in the energy system. Biomass is given as a fixed available quantity per year, whereas the other fuels are variable and are given as fractions of the non-
biomass fuel use. The only exception is individual and industrial use, which is included as fixed amounts of coal, oil and natural gas.

Coal [% of fossil or TWh] Oil [% of fossil or TWh] Natural Gas [% of fossil or TWh] Biomass [TWh/year]
Industry 1.97 TWh 8.86 TWh 5.19 TWh 0.56
Individual heat 0 3.64 TWh 4.99 TWh 3.10
Boiler DH 0 100% 0 1.46
Small CHP 0 0 100% 4.64
Boiler 0 100% 0 0
Large CHP 45% 0 55% 1.85
Boiler 0 100% 0 0
Condensing 45% 0 55% 0.62

4. Application of the criteria to a test case The only investment cost that is included in the analyses is that
of heat pumps. Costs of large-scale heat pumps using ambient
The various criteria detailed in the previous sections are tested temperature heat sources are, according to [45], in the range from
in an analysis with the objective of determining the appropriate 0.6 to 1.3 MV per MWth. In this article, a cost of 1 MV per MWth is
level of heat pumps needed to assist the integration of wind power used, corresponding to 3 MV per MWe.
in the Western Danish energy system. The heat pumps are modelled with a COP (coefficient of
performance) of 3.0 and they are assumed to have this COP at the
4.1. Energy system scenario for Western Denmark required flow temperature level. In other words, no boilers are
applied to raise the flow temperature.
Denmark is separated into two non-connected electricity The costs of the remainder of the system are not included. This
systems (many more if counting in Greenland and the Faeroe of course means that it is not possible to determine the production
Islands). These analyses focus on the continental Western Danish costs of electricity and heat; only the marginal effect of heat pumps
system. The analyses are based on an energy system scenario for the on the economy is assessed.
year 2020 created by a working group established by the Danish All the parameters in Tables 2–5 remain constant throughout all
Energy Authority [41,42]. The same scenario has been used in analyses, and only heat pumps are varied in size in order to analyse
modified forms in, e.g., [15,17] for analyses of the integration of wind how they affect the integration of wind power into such a system.
power. The main parameters of this scenario are presented in Table 2.
In addition to the production plants, the CHP plants are also
4.2. The energy systems analyses model
combined with heat storage with a capacity of 10 GWh, corre-
sponding to roughly 4 h of average heat demand.
The system is modelled using the EnergyPLAN model, which is
The conversion efficiencies – relevant for the analyses including
a model developed particularly with the intention to enable hour-
fuel usage – are as listed in Table 3.
by-hour analyses of energy systems with many interdependencies
The fuels for the system are a mixture of coal, natural gas, oil and
(see Fig. 3) and with many fluctuating energy sources. EnergyPLAN
biomass (See Table 4). Biomass is used on all electricity producing
is hence appropriate for analysing future energy systems in which
plants and in non-CHP district heating. Oil is used only for boilers.
the integration of fluctuating energy sources becomes a main issue.
Note that boilers are also used at the CHP plants during peak load
The model is a deterministic model based on hourly distribu-
periods or when electricity demand/prices are too low to warrant
tions of energy sources and energy demands – mainly electricity
CHP operation.
and heat demands – for a one-year-period. The model applies
Note that energy consumption for transport is not included in
a sequence of priority to the production technologies. Top priority
the analyses.
is given ‘‘use it or lose it’’ – productions like wind power, photo
Fuel costs are as listed in Table 5. Notice that compared to 2008
voltaic cells, wave power, solar collectors and to some extent
world market prices, costs are fairly low. A world market crude oil
geothermal power production. Then comes technologies that
price of, e.g., 100 US$ per barrel corresponds to approximately 11 V/
traditionally follow a set schedule such as industrial cogeneration
GJ; however, in order to make the analyses consistent with the
or waste incineration, where industrial cogeneration follow the
scenario, the low fuel price is used.
requirements of the industry and waste incineration typically is
For the analyses in which electricity trade to the outside world is
operated at nominal load to ensure optimal combustion for envi-
permitted, a synthetic electricity spot market price variation with
ronmental reasons. Then come CHP plants for district heating,
an average price of 225 DKK/MWh (30.2 V/MWh) is used. This has
where the model determines their operation within the limits
been the average Nord Pool spot market price for Western Denmark
given by the heat demand and the contents of heat storages. Lastly
for the years 2000–2007, according to Nord Pool [43].
come condensing mode power plants for electricity generation and
Besides fuel costs, some fuel handling, operation and mainte-
boilers for district heat generation that are added as the last and
nance costs are included in the analyses. Taxes are also applied
most energy inefficient resort.
where required. These are based on scenario work by Lund &
In addition to heat pumps and heat storages, the model also
Mathiesen [44]. Taxes are also included to determine Total
handles electricity storages such as pumped hydro storages, battery
Resources, as this is a business economic cost.
storages, CAES, electric vehicles, and hydrogen systems. If stipu-
lated, these and others may add flexibility to the system and
Table 5
thereby provide a possibility for ensuring the balance between
Fuel costs used in the analyses. A Euro – DKK (Danish Kroner) exchange rate of 7.46 is
used.
electricity production and demand.
One of the cornerstones of the model is its ability to model
Coal Oil Natural Gas Biomass
different regulation strategies. In actual systems, CHP plants may,
DKK/GJ 14 26 30 22 e.g., be operated with the single purpose of following the heat load,
V/GJ 1.88 3.49 4.02 –
according to a triple electricity tariff, or in order to facilitate the
1242 P.A. Østergaard / Energy 34 (2009) 1236–1245

Fig. 3. Energy system outline in the EnergyPLAN model. Front page view of the model downloadable from energy.plan.aau.dk

integration of fluctuating renewable energy sources by utilising modelled to analyse their impact on the various optimisation
heat storage and heat pumps for temporal load shifting. In these criteria. Heat pumps are modelled in levels of 200, 400 and
analyses, the latter of these regulation strategies is applied, as focus 600 MWe installed capacity, respectively. In addition to these levels,
is placed on the best possible integration of renewable energy. a reference case is included with no heat pump capacity installed.
The EnergyPLAN model ensures that the operation of the elec- This reference is mainly used to determine the Rate Impact.
tricity system is within the certain boundaries in order to ensure As deliberated in the previous section, it is of large importance
system stability. It will not permit production based on asynchro- whether or not the system is permitted to interact with the
nous generators to cover too much of the hourly electricity demand, surroundings – both from a technical perspective regarding the
as these are not able to supply ancillary services – see, e.g,. [15]. utilisation of import and export for balancing purposes, and from
Apart from technical analyses of how to design systems capable an economic level, e.g., regarding the value of export to the system.
of integrating fluctuating energy sources, the model is also able to The analyses are thus separated into these two main cases.
perform certain economic calculations involving e.g. income from Total resources are calculated as socioeconomic costs with the
selling electricity beyond the system boundary. addition of taxes on fuels and electricity for district heating, power
The model is further described by its creator Henrik Lund in, e.g. plant and individual use. Non-utility commercial consumers of
[8,46]. energy are not taxed. Hence, only the domestic use of electricity is
taxed. This is estimated at 31% of the total electricity demand, based
4.3. Energy systems analyses on 2006 data [47]. Electricity for heat pumps is taxed separately.
VAT is added to all costs, irrespective of the type of consumer, at the
As mentioned previously, it is not always possible to find Danish flat rate of 25%.
a specific mathematical optimum when designing energy systems. In the specific Scandinavian context, a so-called bottleneck cost
Thus, three different levels of installed heat pump capacities are or income is introduced, when required interconnection capacity

Table 6
Optimisation criteria for the energy system modelled in island mode. The shaded fields indicate the optimal installed heat pump capacity for each optimisation criteria. Rate
impacts are relative to the reference (0 MW HP).

Criteria 0 MWe HP 200 MWe HP 400 MWe HP 600 MWe HP

Reserve capacity requirement 2635 MW 2709 MW 2709 MW 2709 MW
Use of import and export - peak / / / /
Use of import and export – average / / / /
Condensing mode operation 4.00 TWh/year 4.22 TWh/year 4.27 TWh/Year 4.28 TWh/year
PEC 88.78 TWh/year 87.37 TWh/year 86.66 TWh/year 86.24 TWh/year
PEC corrected for import/export 88.78 TWh/year 87.37 TWh/year 86.66 TWh/year 86.24 TWh/year
Renewable energy shares (of PEC) 29.2% 29.6% 29.9% 30.0%
Carbon dioxide emissions (CO2) 15.87 MT 15.45 MT 15.23 MT 15.10 MT
CO2 corrected for import/export 15.87 MT 15.45 MT 15.23 MT 15.10 MT
Societal costs (annual) 11748 MDKK 11607 MDKK 11604 MDKK 11663 MDKK
1575 MV 1556 MV 1555 MV 1563 MV
Rate Impact (Societal) – 5.66 DKK/MWh 5.79 DKK/MWh 3.42 DKK/MWh
Total Resources (business economic) 24503 MDKK 24756 MDKK 24990 MDKK 25476 MDKK
3285 MV 3319 MV 3350 MV 3415 MV
Rate impact (business economic) – 10.05 DKK/MWh 19.46 DKK/MWh 39.00 DKK/MWh
 P.A. Østergaard / Energy 34 (2009) 1236–1245 1243

Table 7
Optimisation criteria for the energy system modelled in connected mode with 1700 MW interconnection capacity. The shaded fields indicate the optimal installed heat pump
capacity for each optimisation criterion. Rate impacts are relative to the reference (0 MW HP)

Criteria 0 MWe HP 200 MWe HP 400 MWe HP 600 MWe HP

Reserve capacity requirement 2707 MW 2988 MW 2988 MW 2988 MW
Use of import/export – peak 0/1700 MW 7/1700 MW 7/1700 MW 7/1700 MW
Use of import/export – average 0/307 MW 0/267 MW 0/238 MW 0/230 MW
Condensing mode operation 5.15 TWh/year 6.19 TWh/year 6.78 TWh/year 7.47 TWh/year
PEC 90.47 TWh/year 88.35 TWh/year 86.89 TWh/year 86.20 TWh/year
PEC corrected for import/export 85.05 TWh/year 83.60 TWh/year 82.62 TWh/year 82.05 TWh/year
Renewable energy shares (of PEC) 27.0% 27.6% 28.01% 28.3%
Carbon dioxide emissions (CO2) 16.73 MT 16.20 MT 15.85 MT 15.71 MT
CO2 corrected for import/export 15.42 MT 15.05 MT 14.82 MT 14.71 MT
Societal costs (annual) 12242 MDKK 11909 MDKK 11738 MDKK 11685 MDKK
1641 MV 1596 MV 1573 MV 1566 MV
Rate Impact (societal) – 13.38 DKK/MWh 20.26 DKK/MWh 22.40 DKK/MWh
1.79 V/MWh 2.71 V/MWh 3.00 V/MWh
Total Resources (Business economic) 25330 MDKK 25499 MDKK 25664 MDKK 25871 MDKK
3395 MV 3418 MV 3440 MV 3468 MV
Rate impact (business economic) – 6.79 DKK/MWh 13.42 DKK/MWh 21.75 DKK/MWh
0.91 V/MWh 1.80 V/MWh 2.92 V/MWh

supersedes installed interconnection capacity. This is not included process but also induces a complexity that needs to be handled in
in the analyses. an appropriate way.
Energy systems analyses of the system run in island mode Multi-criteria decision analysis (MCDA) is a method for incor-
generate the results listed in Table 6; and for the connected mode, porating multiple decision criteria into a framework for decision-
the results are listed in Table 7 for a number of the main optimi- making, and an extensive literature on the matter exists [48–52]. As
sation criteria deliberated. phrased by Cavallaro and Ciraolo, ‘‘The strength of the multi-criteria
The most favourable technical design varies according to the analysis lies in its ability to simultaneously evaluate a number of
optimisation criteria applied, as indicated by the shading in Tables 6 alternatives in relation to a multiplicity of viewpoints and to produce
and 7, both in the interconnected mode and in the island mode. It results that take into consideration any eventual tradeoffs between the
also varies with the same criterion, when comparing island mode values examined.’’ [48]. MCDA is not a uniquely defined method-
with connected mode. Hence, no unequivocal best design can be ology, though. There is a multitude of MCDA methodologies [49,51]
defined; i.e., one design that is optimal according to all the opti- which may generally be classified as
misation criteria. In addition to this, a few comments should be
attached to some of the results.  Value measurement methods
In general, the higher the share of heat pumps, the better within  Goal, aspiration and reference level models,
the range analysed. Some optimisation criteria show a different  Outranking models [51,53]
profile, though. Condensing mode operation increases, e.g., with
the share of heat pumps. This could be attributed to the circum- In fact, as Løken concludes, ‘‘Choosing among all the MCDA
stance that added heat pump capacity increases the utilisation of methods that exist can be said to be a multicriteria problem.’’ [51].
wind turbines. However, present wind turbines are not able to Only the first – the value measurements class and more specifically
supply ancillary services [15]; which means that added wind uti- the multi attribute value theory (MAVT) – will be treated here, as an
lisation requires additional power plant operation in order to example of how to apply MCDA to energy planning optimisation
secure that ancillary services are supplied in a correct proportion. problems. For an extensive literature review of other MCDA
Societal costs are lowered through the expansion of heat pumps, methods applicable in energy planning, see [51].
while total resource costs are increasing with higher heat pump In MAVT, a series of decision criteria are chosen. For each
capacity. The business economic decline is due to high taxes criterion, the potential numerical range of the results must be
imposed on electricity for heat production. Thus, the results can normalised to a common scale in order to become comparable to
also be used to show that there is a distorted tax incentive giving the other criteria. Each criterion is assigned a subjectively estab-
suboptimal solutions seen from a socioeconomic perspective. lished weight. The overall score of a given alternative is simply the
In the island mode, there is an optimal sizing of heat pumps sum of the products of decision criteria weights and normalised
within the 200–600 MWe range analysed for one of the criteria, decision criteria values. Different alternatives may then be
whereas this is not the case for the connected mode. compared using this sum or value score.

4.4. Multi-criteria decision analysis

Table 8
As the analysed case demonstrates, no uniquely best option can Applied normalisation and criteria weights for multi-criteria decision analysis.
be identified, as the various optimisation criteria to some extent Range of criteria values Criteria weight
have conflicting outcomes. Decision-makers facing the task of
Best Worst Unit [%]
choosing one energy system rather than another are thus faced
Reserve capacity requirement 0 3000 MW 0
with the choice between the different optimisation criteria.
Condensing mode operation 0 5 TWh 0
Through the choice of one optimisation criterion, the decision- PEC 50 100 TWh 20
maker sets the stage and thereby favours some options rather than RE Share 50 0 % 20
others. In actual decision processes, one optimisation criterion is Carbon dioxide emissions 0 20 MT 30
seldom used as a single decision criterion. Introducing more deci- Societal costs 0 2000 MV 20
Total resources 0 4000 MV 10
sion criteria gives a better understanding of the actual decision
1244 P.A. Østergaard / Energy 34 (2009) 1236–1245

Table 9
MAVT scores for the three alternatives and the reference analysed in island mode. ‘‘Value’’ is the actual criteria value. ‘‘Normal.’’ is the criteria value normalised to a range from
0 to 100. ‘‘N*W’’ is the normalised value multiplied by the criteria weight.

Reference 200 MW 400 MW 600 MW

Value Normal. N*W Value Normal. N*W Value Normal. N*W Value Normal. N*W
[actual] [0–100] [actual] [0–100] [actual] [0–100] [actual] [0–100]
Reserve capacity 2635 12.2 0.00 2709 9.7 0.00 2709 9.7 0.00 2709 9.7 0.00
requirement
Condensing mode operation 4 20.0 0.00 4.22 15.6 0.00 4.27 14.6 0.00 4.28 14.4 0.00
PEC 88.78 22.4 4.49 87.37 25.3 5.05 86.66 26.7 5.34 86.24 27.5 5.50
RE Share 29.2 58.4 11.68 29.6 59.2 11.84 29.9 59.8 11.96 30 60.0 12.00
Carbon dioxide emissions 15.87 20.7 6.20 15.45 22.8 6.83 15.23 23.9 7.16 15.1 24.5 7.35
Societal costs 1575 21.3 4.25 1556 22.2 4.44 1555 22.3 4.45 1563 21.9 4.37
Total resources 3285 17.9 1.79 3319 17.0 1.70 3350 16.3 1.63 3415 14.6 1.46

MAVT Score 28.40 29.86 30.53 30.69

However, while simple in math and principle, there is ample numerical results only stresses the importance of defining success
latitude for affecting the outcome of MAVT through the assignment criteria very explicitly.
of weights and through the normalisation of the decision criteria.
Both are based on subjective decisions, but it may be argued that 5. Conclusion
MAVT is a system that, at least, structures the subjectivity. Clearly,
the weight assigned to, e.g., carbon dioxide emissions is a choice, Many different optimisation criteria might be applied to the
but so is the normalisation. Is the optimal value zero or perhaps design of environmentally benign energy systems. This article has
a more realistically attainable target and how is the other end of the deliberated a variety of these and has applied these to a case. The
range established? The same issues are faced in LCAs, but here, case analyses are based on a comprehensive energy system
international norms and standards for normalisation have been description. The system has been modelled in a tool tailor-made to
established. It cannot be expected, however, that decision-makers investigate how energy systems may be designed to achieve the
would resort to internationally established standards for decision- optimal integration of fluctuating energy sources. The question
making; a matter that by its very political nature is subjective. asked, however, was a fairly simple question regarding the selec-
tion between three well-defined cases; and, in spite of the
simplicity of the question asked, the different optimisation criteria
4.5. Applying the MAVT to the analysed case
rendered different results. While multi-criteria analyses may better
reflect the diversity of considerations faced by decision-makers by
This example is based on the analysis of the islanded system.
taking several criteria into account, the methodology still requires
Priority has been given to PEC, renewable energy shares, carbon
the user to be able to quantify preferences by assigning weights to
dioxide emission as well as to economic costs; whereas operational
different criteria and establishing a procedure for the
parameters, such as condensing mode operation and reserve
normalisation.
capacity requirement, are disregarded. Condensing mode operation
This circumstance underlines the fact that no unequivocal
is indirectly addressed through the PEC. The criteria weights and
answer can be found to the question of how to design an optimal
the parameters used in the normalisation are listed in Table 8.
energy system. Furthermore, the analysis shows that, when refer-
Results of the analyses are listed in Table 9, in which it is seen
ences are being made to specific renewable energy targets or to
that the 600 MWe heat pump alternative has the highest MAVT
cities or areas with policy ambitions of changing to renewable
score, indicating that this alternative fulfils the objectives better
energy or becoming carbon dioxide neutral, the optimisation
than the lower heat pump alternatives. However, while the MAVT
criteria need to be clearly defined.
score points at the 600 MWe heat pump alternative, it is also very
sensitive to the assignment of criteria weights and to the normal-
Acknowledgements
isation parameters. In general, the more the technical criteria are
emphasized, the more will the MAVT score point at the 600 MWe
This article is a substantially enlarged and revised version of
heat pump alternative. Conversely, if more emphasis is placed on
a three-page paper presented at Renewable Energy 2008, Busan,
economic criteria, then the 200 and 400 MWe heat pump alterna-
South Korea, October 12–17th 2008. The said paper is published
tives may reach the highest MAVT scores.
through the conference. The work is supported by the Danish
Council for Strategic Research through the research project
4.6. Error analysis and validation of results Coherent Energy and Environmental System Analysis (CEESA).

The numerical modelling has been performed using the Ener- References
gyPLAN model, which has been used in a number of peer-reviewed
[1] Østergaard I. Renewable energy islands in Europe. Copenhagen, Denmark:
articles including [5,8,11,15,21,44,54]. The model is thus well Danish Energy Agency; 1998.
documented in literature and is furthermore under continuous [2] Lund H, Münster E. Management of surplus electricity-production from
development based on feedback from the energy planning a fluctuating renewable-energy source. Applied Energy 2003;76(1–3):65–74.
[3] Danish Ministry of Energy. Energy 2000. A plan of action for sustainable
community applying the model.
development. Copenhagen, Denmark: Danish Ministry of Energy; 1990.
The objective of this article is to review a series of optimisation [4] Danish Ministry of Energy. Energy 2000. Follow up. Responsible and forward-
criteria and apply these to a case in order to demonstrate that looking energy policy. Copenhagen, Denmark: Danish Ministry of Energy;
different design criteria will give differing results. The numerical 1993.
[5] Salgi G, Lund H. System behaviour of compressed-air energy-storage in
results of the actual system modelling are not the primary Denmark with a high penetration of renewable energy sources. Applied
outcomes of the article and, in fact, any divergence between these Energy 2008;85(4):182–9.
 P.A. Østergaard / Energy 34 (2009) 1236–1245 1245

[6] Salgi G, Donslund B, Alberg Østergaard P. Energy system analysis of utilizing [33] Singh RK, Murty HR, Gupta SK, Dikshit AK. An overview of sustainability
hydrogen as an energy carrier for wind power in the transportation sector in assessment methodologies. Ecological Indicators 2009;9(2):189–212.
Western Denmark. Utilities Policy 2008;16(2):99–106. [34] Hirst E, Brace M, Knutsen C. An electric-utility integrated resource plan:
[7] Lund H, Østergaard PA. Electric grid and heat planning scenarios with cen- a Pacific Northwest example. Energy 1989;14(3):131–9.
tralised and distributed sources of conventional, CHP and wind generation. [35] Alnatheer O. The potential contribution of renewable energy to electricity
Energy 2000;25(4):299–312. supply in Saudi Arabia. Energy Policy 2005;33(18):2298–312.
[8] Lund H, Duić N, Krajačić G, Graça Carvalho Md. Two energy system analysis [36] Beccali M, Brunone S, Cellura M, Franzitta V. Energy, economic and environ-
models: A comparison of methodologies and results. Energy 2007;32(6):948–54. mental analysis on RET-hydrogen systems in residential buildings. Renewable
[9] Andersen AN, Lund H. New CHP partnerships offering balancing of fluctuating Energy 2008;33(3):366–82.
renewable electricity productions. Journal of Cleaner Production [37] Diaf S, Notton G, Belhamel M, Haddadi M, Louche A. Design and techno-
2007;15(3):288–93. economical optimization for hybrid PV/wind system under various meteoro-
[10] Lund H. Large-scale integration of wind power into different energy systems. logical conditions. Applied Energy 2008;85(10):968–87.
Energy 2005;30(13):2402–12. [38] Bilgen E, Bakeka BJD. Solar collector systems to provide hot air in rural
[11] Lund H, Clark WW. Management of fluctuations in wind power and CHP applications. Renewable Energy 2008;33(7):1461–8.
comparing two possible Danish strategies. Energy 2002;27(5):471–83. [39] Ayres M, MacRae M, Stogran M. Levelised unit electricity cost comparison of
[12] Blarke MB, Lund H. The effectiveness of storage and relocation options in alternate technologies for baseload generation in Ontario. Calgary, Canada:
renewable energy systems. Renewable Energy 2008;33(7):1499–507. Canadian Energy Research Institute; 2004.
[13] Chen C, Lee T, Jan R. Optimal wind–thermal coordination dispatch in isolated [40] Urban F, Benders RMJ, Moll HC. Modelling energy systems for developing
power systems with large integration of wind capacity. Energy Conversion and countries. Energy Policy 2007;35(6):3473–82.
Management 2006;47(18–19):3456–72. [41] Danish Energy Agency. Report from the workgroup on electricity production
[14] Kaldellis JK, Zafirakis D. Optimum energy storage techniques for the from CHP and RES (Rapport fra arbejdsgruppen om kraftvarme og VE-elec-
improvement of renewable energy sources-based electricity generation tricitet) [In Danish]. Copenhagen, Denmark: Danish Energy Agency; 2001. See
economic efficiency. Energy 2007;32(12):2295–305. also:http://www.ens.dk/graphics/Publikationer/Forsyning/Eloverlobsrapport_
[15] Østergaard PA. Ancillary services and the integration of substantial quantities 11-10-01.pdf.
of wind power. Applied Energy 2006;83(5):451–63. [42] Danish Energy Agency. Attachments to the report from the working group on
[16] Østergaard PA. Transmission-grid requirements with scattered and fluctuating electricity production from CHP and RES, Attachment 6 (Rapport fra
renewable electricity-sources. Applied Energy 2003;76(1–3):247–55. arbejdsgruppen om kraftvarme og VE-elektricitet. Bilagsrapport) [In Danish].
[17] Østergaard PA. Modelling grid losses and the geographic distribution of Copenhagen, Denmark: Danish Energy Agency; 2001. See also: http://www.
electricity generation. Renewable Energy 2005;30(7):977–87. ens.dk/graphics/Publikationer/Forsyning/Bilag_eloverlob_16-10-01.pdf.
[18] Energinet.dk. Download of market data. Fredericia, Denmark; 2008. See [43] Nord Pool. Elspot monthly prices. Oslo, Norway: Nord Pool; 2008. See also:
also: http://www.energinet.dk/en/menu/Market/DownloadþofþMarketþData/ http://www.nordpool.dk/marketinfo/elspot/area/elspot.cgi.
DownloadþofþMarketþData.htm. [44] Lund H, Mathiesen BV. Energy system analysis of 100 percent renewable
[19] Lund H, Salgi G. The role of compressed air energy storage (CAES) in future energy systems: the case of Denmark year 2030 and 2050. Energy system
sustainable energy systems. Energy Conversion and Management 2009;50(5): analysis of 100 percent renewable energy systems: the case of Denmark year
1172–1179. 2030 and 2050. In: Dubrovnik Conference on Sustainable Development of
[20] Mæng H, Lund H, Hvelplund F. Biogas plants in Denmark: technological and Energy, Water and Environment Systems; Dubrovnik, Croatia. Dubrovnik,
economic developments. Applied Energy 1999;64(1–4):195–206. Croatia, 2007.
[21] Østergaard PA. Geographic aggregation and wind power output variance in [45] Danish Energy Authority. Elkraft System. Eltra. Technology data for heat and
Denmark. Energy 2008;33(9):1453–60. electricity generating plants. Copenhagen, Denmark: Danish Energy
[22] Kant E, Reath A. Critique of practical reason/Immanuel Kant/intr. by Andrews Authority; 2005. See also: http://www.ens.dk/graphics/Publikationer/
Reath. Cambridge, United Kingdom: Cambridge University Press; 1997. Forsyning_UK/Technology_Data_for_Electricity_and_Heat_Generating_Plants/
[23] Egner T. People and robbers of Cardamom town. Anchorage, USA: Anchorage pdf/Technology_Data_March05.pdf.
Press; 1976. [46] Lund H, Münster E, Tambjerg L.H. EnergyPlan. Computer model for energy
[24] Oecd I, EUROSTAT. Energy statistics manual. Paris Cedex, France: OECD/IEA; 2005. system analysis version 6.0. Aalborg, Denmark: Department of Development
[25] European Commission. Communication from the Commission to the Council and Planning, Aalborg University;2004. See also: http://www.plan.aau.dk/
and the European Parliament – renewable energy road map – renewable tms/publikationer/WorkingPaper12004.pdf.
energies in the 21st century: building a more sustainable future. Bruxelles, [47] Danish Energy Authority. Energy Statistics 2006. Copenhagen, Denmark:
Belgium: European Commission; 2007. Danish Energy Authority; 2007. See also: http://www.ens.dk/graphics/
[26] Danish Ministry of Climate and Energy. EUs Klima- og energipakke (The EU Publikationer/Statistik_UK/Energy Statistics_2006_a.pdf.
Climate and Energy package). Copenhagen, Denmark: Danish Ministry of [48] Cavallaro F, Ciraolo L. A multicriteria approach to evaluate wind energy plants
Climate and Energy; 2008 [In Danish]. on an Italian island. Energy Policy 2005;33(2):235–44.
[27] United Nations. Kyoto protocol to the United Nations framework con- [49] Hobbs BF, Horn GT. Building public confidence in energy planning: a multi-
vencention on climate change. Kyoto, Japan: United Nations; 1998. method MCDM approach to demand-side planning at BC gas. Energy Policy
[28] Perry S, Klemeš J, Bulatov I. Integrating waste and renewable energy to reduce 1997;25(3):357–75.
the carbon footprint of locally integrated energy sectors. Energy [50] Kaldellis JK, Zafirakis D, Kavadias K. Techno-economic comparison of energy
2008;33(10):1489–97. storage systems for island autonomous electrical networks. Renewable and
[29] Varun, Bhat IK, Prakash R. LCA of renewable energy for electricity generation Sustainable Energy Reviews 2009;13(2):378–92.
systemsdA review. Renewable and Sustainable Energy Reviews, 2009;13(5): [51] Løken E. Use of multicriteria decision analysis methods for energy planning
1067–1073. problems. Renewable and Sustainable Energy Reviews 2007;11(7):1584–95.
[30] Thrane M, Ziegler F, Sonesson U. Eco-labelling of wild-caught seafood prod- [52] Pohekar SD, Ramachandran M. Application of multi-criteria decision making
ucts. Journal of Cleaner Production 2009;17(3):416–23. to sustainable energy planning – A review. Renewable and Sustainable Energy
[31] Rees WE. Ecological footprints and appropriated carrying capacity: what Reviews 2004;8(4):365–81.
urban economics leaves out. Environment & Urbanization 1992;4(2):121–30. [53] Belton V, Stewart TJ. Multiple criteria decision analysis: an integrated
[32] Rees W, Wackernagel M. Urban ecological footprints: Why cities cannot be approach. Boston: Kluwer Academic Publishers; 2002.
sustainable – and why they are a key to sustainability. Environmental Impact [54] Mathiesen BV, Lund H, Nørgaard P. Integrated transport and renewable energy
Assessment Review 1996;16(4–6):223–48. systems. Utilities Policy 2008;16(2):107–16.