Wind Power Integration With Heat Pumps, Heat Storages, and Electric Vehicles - Energy Systems Analysis and Modelling

Downloaded from orbit.dtu.
dk on: Nov 09, 2017
Wind power integration with heat pumps, heat storages, and electric vehicles – Energy
systems analysis and modelling
Hedegaard, Karsten; Morthorst, Poul Erik; Münster, Marie; Detlefsen, Nina
Publication date:
2013
Document Version
Publisher's PDF, also known as Version of record
Link back to DTU Orbit
Citation (APA):
Hedegaard, K., Morthorst, P. E., Münster, M., & Detlefsen, N. (2013). Wind power integration with heat pumps,
heat storages, and electric vehicles – Energy systems analysis and modelling. Department of Management
Engineering, Technical University of Denmark.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Wind power integration with heat pumps,
heat storages, and electric vehicles
– Energy systems analysis and modelling
PhD thesis 7.2013
DTU Management Engineering
Karsten Hedegaard
September 2013
- PhD thesis -
Wind power integration with heat pumps, heat storages,

and electric vehicles
by
Karsten Hedegaard
DTU Management Engineering, Risø campus
Energy Systems Analysis
In partial fulfillment of the requirements for the PhD degree

2013
f
Wind power integration with heat pumps, heat storages, and electric vehicles
PhD thesis
Author: Karsten Hedegaard
Main supervisor: Professor Poul Erik Morthorst

Co-supervisors: Senior researcher Marie Münster and former senior consultant at Energinet.dk, Nina Detlefsen

Department of Management Engineering
Technical University of Denmark
Risø Campus
Frederiksborgvej 399, P.O. Box 49, Building 130
DK-4000 Roskilde
Denmark
Tel. +45 46 77 51 00
Fax +45 46 77 51 99
www.man.dtu.dk
Print: Schultz Grafisk A/S
Published September, 2013

Handed in April 15, 2013. Defended July 2, 2013
ISBN: 97887-92706-32-4
Introduction
Table of Contents
1 Introduction .............................................................................................................................. 1
1.1 Challenges of wind power integration .............................................................................. 1
1.2 The role of heat pumps, heat storages, and electric vehicles ............................................ 2
1.3 Goals and research question ............................................................................................. 3
1.4 Contents of papers ............................................................................................................ 4
1.5 Previous studies ................................................................................................................ 6
1.5.1 Individual heat pumps and heat storages .................................................................. 6
1.5.2 Electric vehicles ........................................................................................................ 7
1.5.3 Other fields ................................................................................................................ 7
2 Technologies and power balancing .......................................................................................... 8
2.1 Demands for power balancing .......................................................................................... 8
2.2 Individual heat pumps and heat storages ........................................................................ 10
2.3 Large heat pumps, electric boilers, and heat storages in the district heating system ...... 14
2.4 Electric vehicles .............................................................................................................. 17
2.5 Overview of the technologies' power balancing potentials ............................................ 18
3 Methods.................................................................................................................................. 21
3.1 Choice of models ............................................................................................................ 21
3.2 EnergyPLAN .................................................................................................................. 22
3.3 Model development in EnergyPLAN ............................................................................. 22
3.4 Balmorel ......................................................................................................................... 25
3.5 Model development in Balmorel .................................................................................... 26
4 Energy systems analyses ........................................................................................................ 33
4.1 Overview of energy systems analyses ............................................................................ 33
4.2 Results............................................................................................................................. 34
4.2.1 System benefits of individual heat pumps and heat storages .................................. 34
4.2.2 Influence of individual heat pumps and heat storages on investments ................... 39
4.2.3 Comparison of heat storage options for individual heat pumps .............................. 42
4.2.4 System effects of electric vehicles .......................................................................... 44
4.2.5 District heating versus individual heating ............................................................... 46
4.3 Summary and comparison of results ............................................................................... 47
4.3.1 Overview of results ................................................................................................. 47
4.3.2 Comparison of system benefits ............................................................................... 48
4.3.3 The importance of flexible operation for individual heat pumps and electric
vehicles ................................................................................................................................. 49
5 Discussion and Conclusion .................................................................................................... 51
5.1 Answer to research question ........................................................................................... 51
5.2 Comparison with other studies ....................................................................................... 52
II
Introduction
5.3 Perspectives .................................................................................................................... 53

5.4 Fulfilment of goals .......................................................................................................... 55
5.5 Further research .............................................................................................................. 56
6 Bibliography .......................................................................................................................... 58
Appendix
 Paper I: Wind power impacts and electricity storage – A time scale perspective
 Paper II: Effects of electric vehicles on power systems in Northern Europe
 Paper III: Wind power integration using individual heat pumps – Analysis of different
heat storage options
 Paper IV: Energy system investment model incorporating heat pumps with thermal
storage in buildings and buffer tanks
 Paper V: Influence of individual heat pumps on wind power integration – Energy system
investments and operation
 Paper VI: District heating versus individual heating in a 100 % renewable energy system
by 2050
 Report chapter: Balmorel model results – EVs and power system investments.
 Supplemental data and model illustrations
III
Introduction
Abstract
Wind power is in many countries considered a key renewable energy technology in achieving the
goals of reducing greenhouse gas emissions and relieving the dependency on fossil fuels.
However, the fluctuating and only partly predictable nature of wind challenges an effective
integration of large wind penetrations. This PhD investigates to which extent heat pumps, heat
storages, and electric vehicles can support the integration of wind power. Considering the gaps in
existing research, the main focus is put on individual heat pumps in the residential sector (one-
family houses) and the possibilities for flexible operation, using the heat storage options
available. Several energy systems analyses are performed using the energy system models,
Balmorel, developed at the former TSO, ElkraftSystem, and, EnergyPLAN, developed at Aalborg
University. The Danish energy system towards 2030, with wind power penetrations of up to 60
%, is used as a case study in most of the analyses.
Both models have been developed further, resulting in an improved representation of individual
heat pumps and heat storages. An extensive model add-on for Balmorel renders it possible to
optimise investment and operation of individual heat pumps and different types of heat storages,
in integration with the energy system. Total costs of the energy system are minimised in the
optimisation. The add-on incorporates thermal building dynamics and covers various different
heat storage options: intelligent heat storage in the building structure for houses with radiator
heating and floor heating, respectively, heat accumulation tanks on the space heating circuit, as
well as hot water tanks. In EnergyPLAN, some of the heat storage options have been modelled in
a technical optimisation that minimises fuel consumption of the energy system and utilises as
much wind power as possible.
The energy systems analyses reveal that in terms of supporting wind power integration, the
installation of individual heat pumps is an important step, while adding heat storages to the heat
pumps is less influential. As such, the installation of individual heat pumps can contribute
significantly to facilitating larger wind power investments and reducing system costs, fuel
consumption, and CO2 emissions. This is first due to the high energy-efficiency and economic
competitiveness of the heat pumps. Moreover, their electricity demand profile is well suited for
integrating wind power, even when not operated intelligently. The political phase out of coal in
Denmark by 2030 furthermore creates particularly good conditions for utilising wind power in
meeting the electricity demand for the heat pumps.
When equipping the heat pumps with heat storages, only moderate system benefits can be gained.
Hereof, the main system benefit is that the need for peak/reserve capacity investments can be
reduced through peak load shaving; in Denmark by about 300-600 MW, corresponding to the size
of a large power plant. This can be achieved when investing in socio-economically feasible heat
storages complementing the heat pumps. The potential for reducing the required investments in
peak/reserve capacities is crucial for the feasibility of the heat storages.
Intelligent heat storage in the building structure is identified as socio-economically feasible in 20-
75 % of the houses with heat pump installations, depending on the cost of control equipment in
particular. Investment in control equipment, enabling utilisation of existing hot water tanks for
flexible heat pump operation, is found socio-economically feasible in about 20-70 % of the
IV
Introduction
houses. In contrast, heat accumulation tanks are not competitive, due to their higher investments
costs.
Further analyses investigate the system effects of a gradual large-scale implementation of battery
electric vehicles (BEVs) and plug-in hybrid electric vehicles (PHEVs) in Denmark, Finland,
Norway, Sweden, and Germany towards 2030. When charged/discharged intelligently, the
electric vehicles can, in the long term, facilitate larger wind power investments, while they in the
short term in many cases are likely to result in increased coal-based electricity generation. The
electric vehicles can contribute significantly to reducing CO2 emissions, while system costs are
generally increased, due to assumed investments in the costly BEVs. The need for peak/reserve
capacities can be reduced through the use of vehicle-to-grid capability.
Flexible operation will be more important for electric vehicles than for individual heat pumps.
The reason is that in the situation without flexible operation, the electricity demand for charging
of electric vehicles will typically be concentrated in the hours, where conventional electricity
demand peaks, while individual heat pumps will have a more distributed load profile.
Competing flexibility measures, such as large heat pumps, electric boilers, and thermal storages
in the district heating system, have also been included in the energy systems analyses. These
technologies can together facilitate increased wind power investments and reduce CO2 emissions
in the same order of magnitude as a large-scale implementation of electric vehicles. The
connection between large heat pumps/electric boilers and the large district heating storages
allows for storing electricity as heat during longer periods when needed. This is an advantage
compared to individual heat pumps and electric vehicles, which will mainly be able to provide
power balancing intra-day and intra-hour, due to smaller storage capacities.
Overall, it is concluded that individual heat pumps, flexibility measures in the district heating
system, and PHEVs, can provide significant contributions to a cost-effective integration of wind
power towards 2030. Heat storages complementing individual heat pumps can contribute only
moderately in this regard.
V
Introduction
Resume
Vindkraft betragtes i mange lande som en vigtig vedvarende energikilde i forhold til at nå
målsætningerne om at reducere drivhusgasemissioner og afhængigheden af fossile brændsler.
Imidlertid er vinden fluktuerende og kun delvis forudsigelig, hvilket vanskeliggør en effektiv
integration af store mængder vindkraft. Denne Ph.d.-afhandling undersøger i hvilket omfang
varmepumper, varmelagre og elbiler kan fremme integrationen af vindkraft. Hovedfokus er på
individuelle varmepumper i boligsektoren (enfamilieshuse) og deres muligheder for fleksibel drift
ved brug af varmelagring, da dette område hidtil har været mangelfuldt belyst i et
energisystemperspektiv. Adskillige energisystem-analyser er foretaget ved brug af
energimodellerne, Balmorel, udviklet ved den tidligere TSO ElkraftSystem, og EnergyPLAN,
udviklet ved Aalborg Universitet. Det danske energisystem mod 2030 med vindkraft-andele på op
mod 60 % er anvendt som case i de fleste af analyserne.
Begge modeller er blevet videreudviklet til at give en forbedret repræsentation af individuelle

varmepumper og varmelagre. Et omfattende model add-on til Balmorel gør det muligt at optimere
investeringer og drift af individuelle varmepumper og forskellige typer varmelagre i integration
med energisystemet. De samlede omkostninger for energisystemet minimeres i optimeringen.
Modellen inkorporerer bygningers termiske forhold og omfatter en bred vifte af forskellige
varmelagre: Intelligent varmelagring i bygningskonstruktionen for hhv. radiator- og
gulvvarmesystemer, varmeakkumuleringstanke på rumvarmekredsløbet samt beholdere til varmt
brugsvand. I EnergyPLAN er nogle af varmelagerteknologierne modelleret i en teknisk
optimering, der minimerer brændselsforbruget i energisystemet og udnytter så meget vindkraft
som muligt.
Energisystemanalyserne viser, at med hensyn til at understøtte integration af vindkraft, er

implementering af selve varmepumperne et vigtigt skridt, mens det har mindre betydning at koble
varmelagre til varmepumperne. Således kan varmepumperne bidrage betydeligt til at fremme
vindkraftinvesteringer og reducere systemomkostninger, brændselsforbrug og CO2-emissioner.
Det skyldes dels, at varmepumperne har en høj energieffektivitet og er økonomisk
konkurrencedygtige. Derudover har de selv uden fleksibel drift en elbehovsprofil, som er
velegnet til at integrere vindkraft. Den politiske udfasning af kul i Danmark mod 2030 giver
desuden særligt gode muligheder for at anvende vindkraft til at dække elforbruget til
varmepumperne.
At udstyre varmepumperne med varmelagre giver kun moderate systemgevinster. Heraf er den
største systemfordel, at behovet for spids- og reservelastkapacitet kan reduceres ved at udglatte
spidserne i elforbruget; i Danmark med 300-600 MW svarende til et stort kraftværk. Dette kan
opnås ved samfundsøkonomisk rentable investeringer i varmelagre til varmepumperne.
Potentialet for at reducere investeringer i spids- og reservelastkapacitet er afgørende for
rentabiliteten af varmelagrene.
Intelligent varmelagring i bygningskonstruktionen er identificeret samfundsøkonomisk rentabelt i

20-75 % af husene med varmepumper afhængig af især omkostningerne til kontroludstyr. Det er
desuden fundet samfundsøkonomisk rentabelt at investere i kontroludstyr, der understøtter
anvendelse af eksisterende varmtvandsbeholdere til fleksibel drift af varmepumperne i omkring
VI
Introduction
20-70 % af husene. Derimod er varmeakkumuleringstanke pga. deres højere

investeringsomkostninger ikke konkurrencedygtige.
Yderligere analyser undersøger systemeffekterne af en gradvis storskala-implementering af

batteri-elbiler og plug-in hybrid-elbiler i Denmark, Norge, Sverige, Finland og Tyskland mod
2030. Ved intelligent ladning/afladning kan elbilerne fremme større vindkraftinvesteringer, mens
de på kort sigt i mange tilfælde sandsynligvis vil øge den kulbaserede elproduktion. Elbilerne
bidrager betydeligt til at reducere CO2-udledningerne, mens de samlede systemomkostninger
øges pga. en antaget implementering af de investeringstunge batteri-elbiler. Behovet for spids- og
reservelastkapacitet kan reduceres ved at sørge for, at elbilerne kan levere elektricitet tilbage til
elsystemet.
Fleksibel drift vil være vigtigere for elbiler end for individuelle varmepumper. Det skyldes, at i en
situation uden fleksibel drift vil elforbruget til opladning af elbilerne typisk være koncentreret i
spidslasttimerne for det konventionelle elforbrug, mens individuelle varmepumper vil have en
mere fordelt elbehovsprofil.
Konkurrerende fleksible energiteknologier, så som store varmepumper, elkedler og varmelagre i

fjernvarmesystemet, er også inkluderet i energisystemanalyserne. Disse teknologier kan
tilsammen øge vindkraftinvesteringer og reducere CO2-emissioner i samme størrelsesorden som
en storskala-implementering af elbiler. Forbindelsen mellem de store varmepumper/elkeder og de
store fjernvarmelagre gør det muligt at lagre el som varme igennem længere perioder, når der er
behov for det. Dette er en fordel i forhold til individuelle varmepumper og elbiler, som pga.
mindre lagre primært vil kunne yde fleksibilitet inden for døgnet og timen.
Samlet set kan det konkluderes, at individuelle varmepumper, fleksible teknologier i

fjernvarmesystemet samt plug-in hybrid-elbiler kan bidrage betydeligt til en omkostningseffektiv
integration af vindkraft mod 2030. Varmelagre til individuelle varmepumper kan kun bidrage
moderat hertil.
VII
Introduction
Publications
Primary publications
Paper I
Wind power impacts and electricity storage – A time scale perspective.
Hedegaard, K., Meibom, P.
Published in Renewable Energy. vol. 36, issue 1, pp. 318-324. January, 2012.
Paper II
Effects of electric vehicles on power systems in Northern Europe.
Hedegaard, K., Ravn, H., Juul, N., Meibom, P.
Published in Energy. vol. 48, issue 1, pp. 356-368. December, 2012.
Paper III
Wind power integration using individual heat pumps – Analysis of different heat storage options.
Hedegaard, K., Mathiesen, BV, Lund, H., Heiselberg, P.
Published in Energy. vol. 47, issue 1, pp. 284-293. November, 2012.
Paper IV
Energy system investment model incorporating heat pumps with thermal storage in buildings and
buffer tanks.
Hedegaard, K., Balyk, O.
Accepted for publication in Energy. September, 2013.
Paper V
Influence of individual heat pumps on wind power integration – Energy system investments and
operation.
Hedegaard, K., Münster, M.
Published in Energy Conversion and Management. vol 75. pp. 673-684. November, 20131.
Paper VI
District heating versus individual heating in a 100 % renewable energy system by 2050.
Karlsson, K., Balyk, O., Zvingilaite, E., Hedegaard, K.
Peer reviewed conference proceeding published at the 6th Dubrovnik Conference on Sustainable
Development of Energy Water and Environment Systems, September 25-29, 2011.
Report chapter
Electricity for Road Transport, Flexible Power Systems, and Wind power.
Chapter 9: Balmorel model results – EVs and power system investments.
Risø-Report, Risø-R-1804 (EN), December 2011. Systems Analysis Department, Risø DTU
National Laboratory for Sustainable Energy
1
Published online: September 11, 2013.
VIII
Introduction
Secondary publications
Conference proceedings later expanded and submitted to international scientific journals
Effects of electric vehicles on power system investments and operation.
Peer reviewed conference proceeding. 6th Dubrovnik Conference on Sustainable Development of
Energy Water and Environment Systems, September 25-29, 2011.
Wind power impacts, electricity storage, and heat measures – A time scale perspective.
Hedegaard, K., Meibom, P.
Conference proceeding, Risø International Energy Conference, May 10-12, 2011.
IX
Introduction
Acknowledgments
I would like to thank... for...

Poul Erik Morthorst, my main supervisor constructive guidance and encouragement and
and Marie Münster, my co-supervisor commenting on thesis and articles.
Peter Meibom, Dansk Energi introducing me to optimisation models.
Nina Detlefsen, co-supervisor, Energinet.dk hosting me when visiting and arranging meetings with
relevant contact persons.
Olexandr Balyk everlasting high spirit and valuable assistance in the
modelling work.
Hans Ravn, Ramløse Edb discussing Balmorel model functionalities and
improvement potentials.
Sascha Schröder great companionship at the office and commenting on
parts of the thesis.
Kenneth Karlsson fruitful discussions on preconditions for the energy
systems analyses.
Nina Juul consulting in use of the transport add-on in Balmorel.
Helge V. Larsen assistance in the use of databases and queries.
The Systems Analysis Division hosting me.
My colleagues inspiration and companionship.
Brian Vad Mathiesen, Henrik Lund, Aalborg introducing me to the EnergyPLAN model and for
University beneficial collaboration.
Per Heiselberg, Aalborg University discussing concepts for modelling thermal storage in
buildings and valuable collaboration
Lars Olsen and Svend V. Pedersen, contributing with important knowledge on heating of
Technological Institute of Denmark, buildings, heat pumps, heat storage options, and
Jacob J. Andersen, Vølund varmeteknik, thermal building dynamics.
Kim Wittchen, Danish Building Research
Institute, Brian Elmegaard, DTU
Mechanical Engineering, Otto Hammer and
Helge Christensen, Danfoss and many others
Peder Bacher, DTU Informatics providing measurements for heat pump installations.
My lovely wife Marianne her great support and for taking care of so many
things along the way, especially in the final stages of
the project.
My two wonderful boys, Oliver and bringing so much joy into my life!
Jonathan
X
Introduction
Abbreviations
Energy and monetary units
DKK Danish kroner
€, EUR Euro
M€ Million euro
MW Mega watt
MW-e Mega watt electricity
MW-th Mega watt thermal
TWh Tera watt hours (3.6 Peta joule)
USD US dollar
Other abbreviations
BEV Battery electric vehicle
DH District heating
CO2 Carbon dioxide
CHP Combined heat and power
EV Electric vehicle
Flex Flexibility
HP Heat pump
iHP Individual heat pump
inv. Investment
O&M Operation & maintenance
PHEV Plug-in hybrid electric vehicle
TSO Transmission System Operator
XI
Introduction
Nomenclature
Indices
a, A heating area, set of heating areas in simulation
h, H time period (hour), set of time periods in simulation
g, G generation technology/storage, set of generation technologies/storages in simulation
GEl set of electricity generation technologies in simulation
r, R region, set of regions in simulation
Parameters
cc capacity credit of electricity generation unit ( )
CB effective heat capacity of building structure (Wh/°C/m2 floor area)
CI heat capacity of indoor air, furniture and for radiator heating systems, also water in
radiators (Wh/°C/m2 floor area)
CF heat capacity of concrete floor heating system, incl. concrete and water in tubes
(Wh/°C/m2 floor area)
CL effective heat capacity of walls/ceiling (Wh/°C/m2 floor area)
DElConv conventional electricity demand, i.e. excl. heat pumps, electric boilers, and electricity
storage (MW)
KE capacity of existing generation technologies (MW) and storages (MWh)
TO ambient (outdoor) air temperature (°C)
QPel heat contribution from persons and electrical appliances (W/m2 floor area), (MW)
Variables
KN capacity of new generation technologies (MW) and storages (MWh)
DElFlex electricity consumption for heat pumps/electric boilers and loading of electricity
storages (MW)
LDACT loading of heat accumulation tanks (MW)
TB temperature of building structure (°C)
TF temperature of floor (°C)
TI temperature of indoor air (°C)
TL temperature of walls/ceiling (°C)
QCool cooling (MW)
QSH space heating from existing and new heating installations and unloading from heat
accumulation tanks (MW)
QVen ventilation loss through building envelope (MW)
QIB heat transfer from indoor air to building structure (MW)
QBO heat transfer from building structure to ambient air (MW)
QFI heat transfer from floor to indoor air (MW)
QIL heat transfer from indoor air to walls/ceiling (MW)
QLO heat transfer from walls/ceiling to ambient air (MW)
QTLos heat loss from heat accumulation tanks contributing to space heating (MW)
XII
Introduction
1 Introduction
1.1 Challenges of wind power integration
Our energy systems face serious challenges of reducing greenhouse gas emissions and relieving
the dependency on fossil fuels [1]. Wind power is considered an important renewable energy
source, due to its high economic competitiveness [2]. This is reflected in the significant increases
in wind power capacities planned in many countries [3]. However, wind power is not
dispatchable like e.g. biomass fired power plants, but expresses significant fluctuations depending
on wind variations [4]. This challenges an efficient integration of large amounts of wind power.
As such, in high wind systems, wind generation will often exceed electricity demand, resulting in
forced electricity export (see Figure 1).
4500
4000
3500 Wind power
3000 generation
MW
2500 Conventional
2000 electricity demand
1500
1000
500
0
10
19
28
37
46
55
64
73
82
91
1
100
109
118
127
136
145
154
163
172
181
190
199
208
217
226
235
244
253
262
271
280
289
298
307
316
325
334
Hour
Figure 1. Wind power generation and conventional electricity demand in Western Denmark by 2020 for a wind
power share in Denmark of around 50 % of annual electricity demand [5]. A two week period in March is
illustrated.
High excess electricity production combined with bottlenecks in the transmission lines can have a
substantial effect on electricity prices [6] and even negative electricity prices can occur, as could
be observed for single instances during the last years [7]. The variability of wind moreover
creates a need for dispatchable power plants to back it up in low wind periods, as also illustrated
in Figure 1. Additionally, wind power is only partly predictable, which can create a costly need
for reserve capacity investment and operation, responding to forecasts errors [8].
Denmark is among the countries in the world with the highest share of wind power in the system,
currently around 30 % of annual electricity demand [9], and this share is planned to increase to 50
% by 2020 [10]. The Danish energy system thus forms an interesting case for analysing the
challenges of integrating large amounts of wind power. Moreover, the Danish energy system is
characterised by a large share of district heating (46 % [11]) and combined heat and power (CHP,
55% of thermal power production [12]). The high share of CHP increases the energy efficiency of
the system but can also contribute to generating excess electricity production, due to constraints
between heat and power generation [7, 13]. Analyses of the Danish system are therefore
particularly relevant for other countries, which strive to integrate large amounts of wind power
and at the same time ensure a high efficiency of the energy system.
In the context of wind power integration, Denmark benefits from the electrical interconnections
with Norway and Sweden, which give access to vast hydro power reservoirs. These reservoirs
1
Introduction
function as large scale electricity storages, since water intake to the reservoirs can be stored in
high wind periods and then used for hydro power generation in low wind periods. This
contributes to improving the possibilities of integrating large amounts of wind power in the
Danish energy system [14]. Furthermore, Denmark has benefited from the option of exporting
excess electricity to Germany [15].
1.2 The role of heat pumps, heat storages, and electric vehicles
As described above, electrical interconnections to neighbouring countries can contribute
significantly in meeting the challenges of integrating wind power. However, significant wind
power expansions are also expected in neighbouring regions, particularly German wind power
installations in the North Sea, and wind power production from these sites is highly correlated to
production from wind turbines in Western Denmark. This will reduce the benefit of exporting
electricity to Germany in high wind periods, in the future [14, 15]. Similar tendencies are
expected for other neighbouring regions to Denmark [14]. Flexible electricity demand
technologies are therefore considered a valuable contribution in supporting an effective
integration of wind power in Denmark [14]. In this regard, large heat pumps in the district heating
system, individual heat pumps in the residential sector, and electric vehicles in the transport
sector are expected to play an important role [6, 14, 16].
Due to the high efficiency of these technologies compared to the technologies they displace, e.g.
individual boilers, and combustion engines in conventional vehicles, they can reduce the fuel
consumption of the energy system [6, 11]. By displacing fossil fuels, particularly the use of oil
products in individual boilers and in the transport sector, they can moreover increase security of
supply [14]. The heat pumps and electric vehicles also contribute in making a larger part of
society's energy demands electricity based. This improves the possibilities for using wind power
and other renewable energy sources to cover the energy demands of the system. This can in itself
support the integration of renewable energy and increase security of supply. The use of heat
pumps for integrating wind power furthermore benefits from the fact that both wind power and
heat demand are typically high in the cold periods [14].
As illustrated in Figure 2, the Danish Transmission System Operator (TSO), Energinet.dk,

expects that individual heat pumps and large heat pumps will comprise a significant part of the
Danish electricity demand by 2030.
Danish electricity demand, 2030

40
Electric vehicles
35
30 Individual heat pumps
TWh-e
25
Large electric boilers
20
15 Large heat pumps
10
5 Conventional
electricity demand
0
Figure 2. Danish electricity demand in 2030 as expected by Energinet.dk (incl. grid losses) [5].
2
Introduction
Energinet.dk expects that electric vehicles will constitute a relatively low part of the electricity
demand towards 2030, based on the challenges of competing with conventional vehicles; but also
stresses the uncertainty associated with the projection. Large electric boilers are here estimated to
represent a very low electricity demand annually since they, due to their low efficiency, will
typically be operated in relatively few hours of the year, when electricity prices are sufficiently
low2.
Apart from contributing in electrifying the energy system and increasing its energy efficiency,
heat pumps and electric vehicles can provide flexibility to the system, if utilising the storage
options available. Large heat pumps can be connected to thermal storages used in the CHP
system, individual heat pumps can store heat in the building structure and/or in thermal storage
tanks, and electric vehicles can utilise their batteries. As a result, heat pumps and electric vehicles
can be integrated as a flexible electricity demand in the system. Electric vehicles can with
vehicle-to-grid capability moreover deliver power back to the system when needed [2, 17, 18].
A variety of dedicated electricity storage technologies also exists, e.g. pumped hydro storage,
compressed air energy storage (CAES), electrolysis combined with fuel cells, and batteries/flow
batteries, which can provide system flexibility. Hereof, pumped hydro is today the most
developed and cost-effective technology for large scale electricity storage [19, 20]. However, this
is not an immediate option to utilise locally in Denmark, due to the lack of suitable topography
[20]. The other electricity storage technologies are known to be expensive and some still require
significant technology development [2, 13, 20-23]. This is also reflected in the fact that
Energinet.dk, together with expansion of electrical interconnections, considers heat pumps and
electric vehicles (not electricity storage technologies) as the most important wind integration
measures in a near term perspective [14].
Overall, heat pumps, heat storages, and electric vehicles can potentially play an important role in
supporting the integration of wind power in near future. Analyses are needed to improve the
understanding of the potentials and limitations of these technologies. The technologies have
different technical and economic properties and will compete with each other in providing
flexibility to the system. Moreover, they will affect the operation of the energy system and can
also affect energy system investments. These complex interactions can be captured in the use of
energy system models.
1.3 Goals and research question

The above arguments form the background for the PhD project, which has the goals to:
1. Contribute to the national decision-making with regard to using heat pumps, heat
storages, and electric vehicles in supporting wind power integration
2. Improve the modelling of some of the wind integration technologies enabling a better
representation and understanding of their potentials and limitations.
2
Efficiency: 1 kWh-e converted to approx. 1 kWh-th. Full load hours: 200 full load hours here assumed,
representing typical magnitude as given in [118].
3
Introduction
The PhD project aims to yield a scientific contribution to answering the following general
research question:
 To which extent can heat pumps, heat storages, and electric vehicles contribute to
integrating wind power?
The results are intended as input to policy makers in prioritising efforts and incentives within the
field of wind power integration. The potential users of the model developments in the thesis are
researchers and other professionals using energy models.
The main focus is placed on wind power integration in the Danish energy system towards 2030.
System effects of heat pumps, heat storages, and electric vehicles, will be evaluated in terms
socio-economic costs, fuel consumption, and CO2 emissions. Moreover, effects on energy system
investments will be analysed, particularly to which extent the different technologies can facilitate
larger socio-economically optimal wind power investments and reduced need for dispatchable
power capacities. Considering the gaps in existing research within in the field, the main focus is
put on individual heat pumps in the residential sector and complementing heat storage options.
The thesis is structure in the following way. First, the content of each of the paper is briefly
described, including the connections between the papers. Next, an overview is given of relevant
previous studies in the field. In this regard, the gaps identified in the existing literature are
brought forward, thereby presenting the background for the focus of each paper. Chapter 2
describes the technologies analysed and expected demands for power balancing. The methods
applied in the thesis are described in Chapter 3, covering the choice of models and the model
developments performed. In Chapter 4, results of the energy systems analyses are presented and
compared. Finally, a discussion and conclusion is given in Chapter 5. All full papers can be found
in Appendix.
1.4 Contents of papers

The thesis includes six papers and a report chapter. An overview of the different publications is
given in Table 1, also indicating how model development and energy systems analysis, performed
by the author, is distributed among the publications.
Table 1. Overview of primary publications in the thesis.
Model Energy systems

Paper Main focus development analysis
Paper I Wind power impacts and power balancing
potentials of different flexibility measures
Paper II + Report Electric vehicles + X
Paper III Individual heat pumps and heat storages ++ X
Paper IV Individual heat pumps and heat storages +++++ (X)
Paper V Individual heat pumps and heat storages X
Paper VI District heating vs. individual heating + X
As shown, three of the papers focus on individual heat pumps and heat storages. One paper (and
report chapter) focuses on electric vehicles, while the remaining two papers focus on other
aspects related to wind power integration. Model development and/or energy systems analysis is
4
Introduction
performed in most of the papers; however, the main model development is undertaken in Paper
IV and Paper III. The content of each paper is described further in the following. Throughout the
thesis, references are made to the papers when relevant.
Paper I, "Wind power impacts and electricity storage – A time scale perspective", investigates
wind power impacts and the expected demands for power balancing on different time scales for a
future high wind system. The system of Western Denmark in 2025, with an expected wind power
share of 57 %, is used as a case study. Furthermore, the paper categorises electric vehicles and a
number of different electricity storage technologies, in terms of the time scales at which they are
suited in providing power balancing. As a follow up, an addition has been made in the thesis,
where heat pumps, electric boilers, and heat storages, are categorised in the same way.
In Paper II, "Effects of electric vehicles on power systems in Northern Europe", it is analysed
how a large-scale penetration of battery electric vehicles (BEVs) and plug-in hybrid electric
vehicles (PHEVs) in Denmark, Finland, Norway, Sweden, and Germany towards 2030 would
affect the energy systems. The paper investigates to which extent the electric vehicles can
incorporate larger amounts of wind power into the system and reduce the need for dispatchable
generation capacity, and how they will affect system costs and CO2 emissions. Various energy
systems analyses are performed using the Balmorel model and applying a socio-economic
optimisation.
Paper III, "Wind power integration using individual heat pumps – Analysis of different heat
storage options", focuses on the potentials of individual heat pumps and their different heat
storage options, in terms of ensuring a fuel- and cost-efficient utilisation of wind power. The
study covers model development and energy systems analysis, in EnergyPLAN, of the Danish
energy system in 2020 with around 50 % wind power. The model development enables the
representation of flexible heat pump operation utilising passive heat storage and heat
accumulation tanks. A technical optimisation is applied that minimises total fuel consumption of
the system and utilises as much wind power as possible.
Paper IV, "Energy system investment model incorporating heat pumps with thermal storage in
buildings and buffer tanks", presents extensive model development in Balmorel. A model add-on
is presented that facilitates optimising investments and operation of individual heat pumps and
different heat storage options, in integration with the energy system. It is in Paper III found that
intelligent heat storage in the building structure is the most fuel-efficient and cost-effective heat
storage option among those analysed. Therefore, particular emphasis is put on this option in the
further model development. The model developed in Paper IV thus improves the representation
of the heat dynamics of buildings, compared to the modelling approach in Paper III. The model
covers passive heat storage in the building structure via radiator heating systems, active heat
storage in concrete floors via floor heating systems, and the use of thermal storage tanks. Through
illustrative analyses and comparison with measured data, it is shown that the model is well
qualified for analysing the possibilities and system benefits of operating individual heat pumps
intelligently. This includes peak load shaving and prioritising heat pump operation for hours with
low marginal electricity production costs.
5
Introduction
In Paper V, "Influence of individual heat pumps on wind power integration – Energy system
investments and operation", the Balmorel model add-on presented in Paper IV is applied in
several energy systems analyses. Through socio-economic optimisations of the Danish energy
system by 2030 with optimised wind penetrations of about 50-60 %, it is investigated how
important individual heat pumps and their heat storage options are for the integration of wind
power. Effects on wind power investments, the need for peak/reserve capacities, fuel
consumption, CO2, and system costs etc. are quantified. The analyses also reveal to which extent
it will be socio-economically feasible to invest in different types of heat storages for the heat
pumps.
The background for Paper VI, "District heating versus individual heating in a 100 % renewable
energy system by 2050" is that district heating in the Danish energy system is determining for the
future potential of installing individual heat pumps and large heat pumps and thermal storages in
the CHP system. As such, the district heating grid sets boundaries for the possibilities of using
these technologies for supporting wind power integration. Against this background, it is
investigated whether there is a clear socio-economic advantage of building and operating a large-
scale district heating system, in a long term perspective, i.e. in a 100 % renewable energy system
by 2050. The analysis is most relevant for countries without existing district heating
infrastructure.
1.5 Previous studies

In this section, a brief overview is given of existing studies within the research fields covered in
the thesis. The main focus in the literature review is put on individual heat pumps, in line with the
main focus of the thesis as a whole. More detailed literature reviews can be found in the
introduction of the papers given in Appendix.
1.5.1 Individual heat pumps and heat storages

A number of studies have dealt with demand side management for increasing system flexibility in
a more broad sense, e.g. [24-28]. However, the analysis of individual heat pumps in an energy
system context is a relatively new research area. As such, very few studies within this area have
been identified at the early stages of this PhD project. This forms an important part of the
background for paying special attention to this research field in the project.
Recently, a significant amount of studies focusing on intelligent operation of individual heat

pumps have emerged. A large group of these, apply a single building level perspective, e.g. [29-
33] [34] and some apply a more large scale perspective, covering up to a few hundreds or
thousands of buildings, e.g. [35-40]. However, only few studies analyse individual heat pumps in
energy system models on national scale [8, 14, 41, 42].
All the studies identified apply models that cover only operation of the system, i.e. excluding
investments. The reports by the Danish TSO [14] and the Danish Energy Association [8] make
estimates on potential reductions in peak capacities, through flexible operation of the heat pumps.
However, these estimates are made separately and not as part of the optimisation. Furthermore,
the flexible operation is in these studies not modelled, but e.g. in [14] merely represented as a
modified fixed demand profile, assuming that operation can be optimised within the day.
6
Introduction
Individual heat pumps and complementing heat storages could have important effects on e.g.
wind power investments and the required investment in peak and reserve capacities. Moreover,
flexible operation of the heat pumps requires investment in control equipment and/or thermal
storage tanks on consumer level. Such investments could suffer from low economies of scale, and
will in some aspects compete with other flexibility measures in the system. These issues call for a
model that can analyse individual heat pumps and heat storage options in integration with the
energy system, when optimising both operation and investments. This forms the background for
the model development presented in Paper IV.
Most of the existing studies in the field focus on presenting methods and control concepts for
how individual heat pumps (or electric heating) can be operated flexibly, responding to the needs
of the power system: i.e. contribute in load shifting, peak load shaving, delivering ancillary
services, and in increasing wind power utilisation. Only few studies analyse how the heat pumps
will affect the energy system [8, 14, 41, 42], and none of these apply an energy system model
optimising both operation and investments. This is done in Paper V.
Several of the existing studies focus on using heat storage in the building structure for flexible
heat pump operation, e.g. [36, 37, 41, 43, 44], while fewer focus on using thermal storage tanks,
e.g. [35]. This indicates widespread acknowledgment of the potentials for utilising the thermal
storage capacity of buildings in supporting flexible heat pump operation.
1.5.2 Electric vehicles

The modelling of electric vehicles in integration with the power system has been a research area
for a longer period of time, comprising several studies, e.g. [2, 17, 45-53]. However, none of the
existing studies investigate how transition towards increased electrification of the transport sector
would affect power system investments and operation. This forms the background for Paper II,
where this is analysed for the cases of Denmark, Norway, Sweden, Finland, and Germany.
1.5.3 Other fields

No previous studies have been identified that deal with future wind power impacts and expected
demands for power balancing, by analysing the occurrence and length of low net load periods
(high wind power and low electricity demand) and high net load periods (low wind power and
high electricity demand). This is the point of departure for Paper I.
Previous studies analysing district heating versus individual heating in energy system models [11,
54] have used the existing district heating grid as a premise, and have analysed whether it is
feasible to expand the district heating grid further. As supplement, it is relevant to analyse the
socio-economic competitiveness of district heating versus individual heating, in a long term
perspective, for a situation without existing district heating infrastructure as a premise. This
perspective is however mainly relevant for countries without a district heating grid present. The
above forms the background for Paper VI. Another new contribution of this paper, compared to
the above mentioned studies [11, 54], is that possibilities to invest in heat savings in buildings are
modelled endogenously, as an alternative to heat generation. Additionally, human health
externalities related to local air pollution are internalised in the socio-economic cost optimisation.
Heat savings and externalities are represented using the model developed in [55].
7
Technologies and power balancing
2 Technologies and power balancing

This chapter first summarises the results of Paper I, in terms of the expected demands for power
balancing in a near term Danish high wind system. The term power balancing is here used in a
broad sense, covering balancing responding to imbalances in supply and demand intra-hour and
over longer periods, lasting up to several days or more. The future demands for power balancing
intra-hour have however not been analysed.
Subsequently, the flexibility measures covered in the thesis are described: 1) Individual heat
pumps and heat storages 2) Large heat pumps, electric boilers, and heat storages in the district
heating system and 3) Electric vehicles. Main emphasis is here put on the technologies' technical
potentials of providing power balancing, based on their storage possibilities and response times.
At the end of the chapter, the power balancing potentials of the different technologies are
compared and their magnitude illustrated.
2.1 Demands for power balancing

In Paper I, the case of Western Denmark in 2025 with an expected 57 % wind power penetration
is used for analysing the expected demands for power balancing on different time scales. The
system impacts of wind power have been analysed based on hourly data of net load (residual
load), defined as gross load (electricity demand) minus wind power and minus an assumed
minimum production at centralised power plants for grid stabilisation. The following two
challenging operational situations are treated, namely:
 Low net load periods i.e. hours with low electricity demand and high wind power,
representing hours with low electricity prices.
 High net load periods, i.e. hours with high electricity demand and low wind power,
representing hours with high electricity prices.
By analysing the length of low and high net load periods, the expected demands for power system
balancing at different time scales are investigated. Regarding the low net load periods, focus is
placed on negative net load periods, representing situations with expected excess electricity
production. The length of the low/high net load periods are analysed with the aim of investigating
the demands for power balancing, in the following time scales:
 Intra-day: High/low net periods of one hour to around half a day (12 hours),
corresponding to a time scale typically relevant for power balancing within the day, i.e.
e.g. flexible electricity demand of one to several hours or e.g. shifting of demand from
day to night.
 Intra-week: High/low net load periods approaching one day to several days,
corresponding to a time scale relevant for balancing intra-week
 Seasonal: High/low net loads periods representing imbalances in supply and demand over
months.
The results in Paper I indicate a demand for technologies capable of flexibly charging/activating
demand and discharging/inactivating demand in periods of one hour to around half a day,
providing intra-day balancing. Furthermore, the results indicates an expected demand for
8
technologies flexibly charging/activating demand in periods approaching one day to several days,
i.e. providing balancing intra-week, as well as a potential for power balancing on seasonal level.
The net load approach above is based on assuming perfect foresight on wind power and load. As
such, this approach only captures the variability and not the imperfect predictability of wind
power and load. In reality, forecast errors on wind power and load (as well operation problems at
power plants and in transmission lines) create an additional need for power balancing. This is in
the Danish system handled at the regulating power market and the reserve market (see Table 2).
Table 2. Current Danish electricity markets.
Market Description
Elspot Integrated part of the Nordic day-ahead spot market where power
production, based on market bids and forecasted electricity demand and
wind power, is planned for every hour of the next day, 12 to 36 hours before
the actual operation hour.
Elbas A continuous market operating after the end of the spot market where market
actors expecting deviations between realised production and day-ahead
production plans, have the possibility of trading with an official price towards
balance up until 1 hour before the operating hour.
Regulating Deviations between operation as planned after Elspot and Elbas and actual
power market operation are balanced intra-hour by the TSO, using power installations with
a response time of 15 minutes. The most important causes for imbalances
are forecast errors on wind power and demand and operation problems at
plants and in transmission lines, including outages.
Reserve The TSO buys system services, including primary and secondary reserves,
market which, in case of small imbalances of supply that have not been balanced by
the regulating market, re-establish the balance and stabilise the frequency.
 Primary reserves: frequency controlled, response time of a few to
30 seconds and deliver power of maximum 15 minutes (Western
Denmark) 3.
 Secondary reserves: Response time of 15 minutes and rarely
deliver power for longer than 15 minutes (Western Denmark)2.
Sources: [56-58].
The models used in the energy systems analyses in the thesis, Balmorel and EnergyPLAN, have
an hourly time resolution and are both deterministic, i.e. assuming perfect foresight of wind
power and load [59, 60]. As such, the operation of the regulating power market and the reserve
market is not integrated in the models. Nevertheless, the need for reserve capacities is
incorporated via a capacity balance restriction, in the analyses in Balmorel (see Section 3.5 for
the capacity balance restriction applied in the analyses of individual heat pumps). This ensures
and adequate production capacity and reserve margin in each power region, taking into account
the variable and only partly unpredictable nature of wind, and the risk of failure on transmission
lines. The term reserve capacity is used in this broad sense, in the modelling in this thesis.
3
In Eastern Denmark, Frequency controlled normal operation reserves and Frequency controlled operation
disturbance reserves are instead used [56].
9
2.2 Individual heat pumps and heat storages

Heat pumps transfer heat from a low-temperature location, i.e. the heat source, to a warmer
location, using energy input in the process. All individual heat pumps today are compression heat
pumps, using electricity as input (in contrast to absorption heat pumps using steam, hot water or
flue gas as input [61]). Individual heat pumps with a water-based central heating system, i.e.
ground source heat pumps and air-to-water heat pumps, can store heat in thermal storage tanks,
the building structure, and the central heating system. Thereby, they can provide a distributed
flexible electricity demand. Individual heat pumps represent a shiftable electricity demand, since
the heat pumps are restricted to satisfy the heat demand, while their operation can be shifted in
time [62]. The energy efficiency of heat pumps can be characterised by the coefficient of
performance (COP):
= ∙
For instance, a COP of 3 means that the heat pump converts 1 kWh electricity and 2 kWh of heat
collected from the heat source (typically the ground or ambient air) into 3 kWh of heat output
[63]. The COP in a given instance is closely related to the temperature lift, i.e. the difference
between the temperature of the heat source and the output temperature of the heat pump. The
higher the temperature lift required, the lower the COP [64, 65]. As such, the COP can vary from
hour to hour; more for air-water heat pumps than for ground source heat pumps, considering that
ambient air temperature variations are larger than ground temperature variations. It is however
complex to model hourly COP variations and it can be difficult to handle in linear models. In
energy systems analyses and technology assessments, an annual average COP is thus normally
applied, as also done in this thesis.
Ground source heat pumps and air/water heat pumps are typically supplemented by an electric
boiler to cover peak loads, in order to reduce heat pump investment costs. The heat pump is
typically dimensioned to 72%-82 % of the total capacity [66]. Thereby, the heat pump can cover
about 99 % of the annual heat demand4. The electric boiler has an efficiency of around 1, i.e. it
converts 1 kWh electricity to 1 kWh heat. The influence of the electric boiler on the electricity
consumption is illustrated in Figure 3. As shown, the electricity consumption increases
significantly, when the heat demand exceeds the heat pump capacity, thereby forcing use of the
less efficient electric boiler.
4
For a full year simulation in a thermal building model, using ambient temperatures in [119].
10
Heat pump Electric boiler

consumption (kW-e) 6
5
Electricity
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10
Heat output (kW-th)
Figure 3. Illustration of how the supplemental electric boiler influences the electricity consumption of
individual heat pumps. A total heat installation capacity of 10 kW-th, a heat pump capacity share of 77 %, and a
COP of 3.1 is applied.
Heat pumps can be regulated between cold and full load instantly [67] and can thus deliver
ancillary services intra-hour, i.e. up/down regulating power and secondary reserves through
up/down regulation of demand [14, 36, 38, 39, 68, 69]. Heat pumps can be frequency controlled
and thus also deliver primary reserves. [68].
It is hardly realistic to expect households to engage in individual trading on the electricity

markets. As such, it is more likely that e.g. a balancing responsible agent controls a large group of
heat pumps on aggregated level. This control concept is currently being demonstrated in a project
initiated by Energinet.dk [14, 69]. In fact, a load of minimum 10 MW is currently required in
order to be able to act on the regulating power market [56], which in itself necessitates the
pooling of a large amount of individual heat pumps.
Individual heat pumps are generally available in two main types of regulation: on/off regulation
and capacity regulation, continuously variable down to about 20 % of the maximum capacity. In
on/off regulation, the compressor will work full load and stop at intervals adapted to the heat
demand, while capacity regulation works through a variable-speed compressor where the amount
of refrigerant flow through the refrigerant cycle is adapted to the demand. Most air-water heat
pumps and all ground source heat pumps on the market today have on/off regulation [63].
The ability to shift demand from hour to hour depends on the available heat storage capacity.
Ground heat pumps and air-to-water heat pumps for one-family houses are today typically
installed in combination with a 150-200 litre tank for hot water (for showering, dish washing
etc.), and a small buffer tank, typically around 40-80 litres. The buffer tank ensures that the
number of heat pump start-ups can be minimised thereby enabling better operating conditions and
improving the technical life-time. The buffer tank is nevertheless not large enough to enable
flexible operation from hour to hour [18, 70]. Such flexibility can however be achieved if
investing in a heat accumulation tank connected to the space heating circuit. Given space
requirements, heat accumulation tanks of up to around 1000 litres can realistically be installed in
a one-family house [18]. The small hot water tank can to some extent also shift demand from
hour to hour; however, its storage capacity is limited, corresponding to less than one hour of full
heat pump load [71].
As an alternative to using thermal storage tanks, heat can be stored passively in the building
structure through radiation, thermal conductance, and convection, if allowing some variation in
11
the indoor temperature. The term passive refers to the fact that the heat transfer occurs without
the use of a heat transferring media going in and out of the storage [72]. Houses with a central
concrete floor heating system can moreover store heat in the thermal mass of the concrete floor,
via water tubes in the concrete, i.e. actively [72]. Concrete floor heating is the dominant type of
floor heating installed in Danish houses today due to its low cost [73, 74] and is characterised by
a large thermal capacity (in contrast to wooden floor heating systems, which have an insignificant
thermal storage capacity [75]). In connection with utilising the storage capacity of buildings, an
allowed indoor temperature interval of 20-23 °C during the day and 19-22 °C at night is
considered realistic in satisfying thermal comfort needs. This is based on indoor temperature
measurements in 28 Danish households [76] (see Appendix), as well as on indoor temperature
preferences of a number of Danish households [32], and on [31].
The heat storage capacity of buildings can be increased with the use of phase change materials,
thereby utilising latent heat storage. However, experiences show that the costs for such materials
are unreasonably high, compared to the benefits they can provide (both for implementation in
existing buildings and in new buildings) [75]. In the analyses of individual heat pumps in this
thesis, focus is therefore placed on the other heat storage options described.
The magnitude of the different thermal storage potentials on house level is estimated in Figure 4a
for a typical Danish one-family house. It should be noted that passive/active heat storage
potentials will vary significantly from house to house, depending on the heat capacity of the
building. The assumptions applied in the illustration are given in Table 3. Expected investment
costs required for enabling the different thermal storage options are also illustrated in Figure 4b,
given in annualised socio-economic costs.
70 100
Hot water tank
Heat storage (kWh/house)
60 (180 litres)
Inv. cost (€/house/yr)
80 Digital
50 thermostats
Heat accumulation tank 60
40 Tank
(1000 litres)
30 40
Central
20 Active heat storage in concrete controller
floor (and passive heat storage 20
10 in other structures)
0 0
Radiator Floor Passive heat storage in Utilising Passive/active Heat
heating heating building structure existing hot heat storage accumulation
water tank in the building tank, new
Figure 4. a) Magnitude of heat storage potentials for a typical Danish one-family house, depending on the
central heating system. b) Investment costs for the different heat storage options are given in annualised socio-
economic costs (source: [77, 78], discount rate of 3 % applied). Central controller cost represents expected cost,
incl. installation [77].
12
Table 3. Assumptions applied in illustrating the magnitude of heat storage potentials for a typical Danish one-
family house (151 m2).
Temp. interval utilised in

Heat capacity building, floor, or tank (°C) Ref.
Passive heat storage in
building structure 100 Wh/°C/m2 floor areaa 2d [18]
Active heat storage in
concrete floor 67 Wh/°C/m2 floor areab 3d [73, 74]
Heat accumulation tank [18, 70,
(1000 litres) 17.4 Wh/litre 15c 79]
Hot water tank (180 litres) 17.4 Wh/litre 15 [18, 71]
a
Median of typical range for the effective heat capacity of Danish one-family houses 60-140 Wh°/C/m2 floor
area (the heat capacity of the building that can be utilised for passive heat storage diurnally) [79, 80]
b
10 cm concrete assumed, representing the typical application [35, 39].
c
Up to 15 °C difference between forward and return temperature in the central heating system, representing a
typical configuration of heat pumps in radiator heating, systems prioritising a reasonable COP [18, 70, 79].
d
Simulations in a thermal building model have verified that this temperature difference can be utilised when
allowing an indoor temperature variation of 20-23 °C at day and 19-22 °C at night and optimising heat pump
operation diurnally.
Figure 4a shows that the heat storage potential offered by the hot water tank installed in
connection with the heat pumps is relatively small, while heat storage in the building structure
(passive/active heat storage) or in a large heat accumulation (1000 litres) can provide
significantly larger storage capacities.
As shown, the total achievable heat storage capacity for a typical house is estimated to be in the
magnitude of 50-60 kWh-th. When considering a typical heat pump capacity of around 6-8 kW-th
(excl. supplemental electric boiler) [63, 81], the heat storage options can thus enable flexibility
corresponding to several hours of full load heat pump operation. This illustrates that individual
heat pumps through up/down regulation of demand can provide power balancing of several hours
intra-day; if the required heat storage investments are made. Nevertheless, the storage potentials
will typically not be sufficient to provide power balancing intra-week. This assessment is in
agreement with the view in a recent report by Energinet.dk [82].
It can furthermore be noted that during the non-heating season, typically May 15 to September 15
in Denmark (33 % of the year), the heat storage potential of individual heat pumps will be limited
to the small storage capacity of the hot water tank. Moreover, the capacity margin available for
flexible operation is low in very cold periods, where the heat pump will be forced to operate at its
capacity limit in several hours (illustrated in Section 4.2.2).
As shown in Figure 4b, flexible operation of individual heat pumps generally requires investment
in control equipment communicating with the power system [77, 83]. In addition, intelligent heat
storage in the building structure requires investment in digital thermostats, while the use of a heat
accumulation tank requires investment in the tank [44, 83]. It can be seen that utilising the
existing hot water tank is associated with the lowest investment costs; but on the other hand also
offers limited storage potential. Among the two other heat storage options, intelligent heat storage
in the building structure has lower investment costs than heat accumulation tanks. On the other
13
hand, heat accumulation tanks have the advantage of not requiring any indoor temperature
variation.
As indicated in Figure 4a, houses with concrete floor heating will typically have higher storage
potential than houses with radiator heating. Furthermore, concrete floor heating systems deliver
heat directly to the thermal storage in the concrete, which means that the heat pump can to some
extent be operated flexibly, with only little indoor temperature variations, e.g. +/- 0.25 °C [30,
31]. In contrast, changes in the heat output from radiator heating systems will have a more direct
influence on indoor temperatures. However, radiators have for a long period of time been the
typical type of central heating system installed in connection with construction of new houses in
Denmark. Furthermore, installation of floor heating in existing houses is expensive. As a result,
radiator-based central heating is expected to comprise the bulk of the individual heat demand of
Danish one-family houses by 2030 (see Figure 5). Therefore, when considering a large scale
installation of individual heat pumps within this time frame, the storage possibilities offered by
radiator heating systems are most important, while floor heating system are less influential.
160 16
Heated floor area (Mio. m2)
Floor heating, Floor heating, new

140 14houses
Heat demand (TWh)
new houses (constructed

120 (constructed
12 2005-2030)
2005-2030)
100 10 heating,
Floor Floor heating, existing
80 existing
8 houses houses (constructed up
(constructed up till 2005)
60 6
till 2005)
40 Radiator
4 heating
Radiator heating
20 2
0 0
Figure 5. Individually heated one-family houses in Denmark by 2030 distributed on expected central heating
system. Floor heating here represents central floor heating, i.e. not only in bathrooms/kitchen.
Sources: [74, 83-86].
Taking the distribution in Figure 5 into account, the total heat storage potential in individually
heated one-family houses can be estimated to around 45 GWh-th. This corresponds to an
electricity consumption for individual heat pumps of around 22 GWh-e, corresponding to some
hours of high excess electricity production (negative net load)5.
2.3 Large heat pumps, electric boilers, and heat storages in the
district heating system
Large compression heat pumps and electric boilers can be connected to thermal storages in the
district heating system and thereby represent a flexible electricity demand. Large heat pumps can
moreover relieve CHP plants from heat bound power production and thereby increase system
flexibility and support wind power integration. The thermal storages increase the flexibility of
heat pumps and electric boilers and improve the integration of wind power into CHP systems:
When wind power is high, possibilities for reducing power production from CHP plants are
improved since heat demand can be satisfied from the heat storage, and when wind power is low,
5
Assuming a COP of 3.1 and a heat pump capacity share of 77 % (the remaining covered by electric boilers).
14
feasibility of increasing power production at CHP plants is improved, since surplus heat
production can be stored.
Large heat pumps and electric boilers represent a convertible electricity demand, when integrated
in CHP plants, since the electricity demand can be converted to a fuel demand, using the fuel-
based CHP units (or boilers) to satisfy the heat demand [62]. This creates good possibilities for
turning off the heat pumps/electric boilers in longer periods with high electricity prices and in
peak load hours, determining for the need for peak/reserve capacities. Large heat pumps can
utilise various different heat sources to drive the process, i.e. waste-heat from industrial
processes, heat in waste water, ambient air, ground, lakes or the sea [61]. Large heat pumps have
a relatively high efficiency, but on the other hand also relatively high investment costs, compared
to electric boilers (see Figure 6).
Efficiency Investment cost

3.5 0.8
Inv. cost (M€/MW-th)
3.0
Efficiency ( )
2.5 0.6
2.0
0.4
1.5
1.0 0.2
0.5
0.0 0.0
Heat pump, Electric boiler, Heat pump, Electric boiler,
large large large large
Figure 6. Efficiency and investment costs for large heat pumps (heat source: waste water) and electric boilers in
the district heating system. Sources: [14, 87-89].
Therefore, investment in large heat pumps is only feasible to the extent that a relatively high
number of full load hours can be ensured (4600-5200 full load hours identified for large heat
pumps in Paper V). It will be economically attractive to operate large heat pumps in all hours
where the ratio between the electricity price (in the given power region) and the district heating
price (in the given area) is below the COP of the heat pump. For at heat pump with a reasonable
COP, i.e. around 3, this can occur many hours of the year. In contrast, the low efficiency of
electric boilers means that it is only economically attractive to operate them when the electricity
price (incl. variable O&M cost) drops below the district heating price.
Large heat pumps and electric boilers can be regulated instantly and can thus deliver regulating
power, and primary and secondary reserves [14, 67, 68, 90]6. The possibility of using large heat
pumps and electric boilers to flexibly activate demand in periods of several hours or days depends
on the available thermal storage capacity. In Denmark, CHP plants typically have thermal
storage tank capacity corresponding to two days to a full week of heat demand during the cold
season [61, 91]. These heat storages can also be utilised for large heat pumps/electric boilers.
Heat storage capacities can furthermore be increased in connection with installing heat
pumps/electric boilers, and large scale thermal storage is relatively cheap, i.e. around 3 €/kWh-th
(steel tank) [61] (and even lower for large seasonal storages) compared to 40-70 €/kWh-e for e.g.
electricity storage in batteries [82]. Overall, large heat pumps and electric boilers are therefore
6
However, among the electric boilers, only the type using electrical resistance have sufficiently low response
time for delivering primary reserves [67].
15
suited for flexibly activating demand in periods of up to several days. As such, they can provide,
provide intra-week balancing, in addition to intra-day and intra-hour balancing.
Due to storage capacity limitations and storage losses (typically corresponding to around 5 % of
the stored content over a week [61]), conventional heat storages in the district heating system are
not suited for storing heat across months [92]. However, the relative heat losses and investments
costs per m3 can be reduced significantly when up-scaling to very large heat storages and a few
large seasonal thermal storage demonstration plants already exist in Denmark. The use of
seasonal thermal storages have traditionally been focusing on improved utilisation of solar
thermal, but can also be relevant for large heat pumps/electric boilers (or heat from waste and
industrial waste heat) [92].
The economies of scale for thermal storages are illustrated in Figure 7, where also individual heat
accumulation tanks are shown for comparison.
3000
Investment cost (€/m3)
2500
2000
1500
1000
500
0
1,000 m3
5,000 m3
0.3 M m3
0.3 m3
0.5 m3
1 m3
10,000 m3
10,000 m3
75,000 m3
1 M m3
5 M m3
Indiv. heat Large steel Large water

accum. tanks tanks bassins
Figure 7. Investment costs per volume for different sizes of thermal storages. Large thermal storages: [92].
Individual heat accumulation tanks: [93].
It can be seen that the cost per storage volume is considerably lower for steel tanks in the district
heating system compared to for individual heat accumulation tanks and even lower for seasonal
thermal storages of large water basins. In addition, the relative heat losses are, as mentioned,
lower the higher the heat storage (due to a reduced surface area in relation to storage volume)
[92].
16
2.4 Electric vehicles

Plug-in electric vehicles can be divided in two overall categories:
 Battery electric vehicles (BEVs) driving on electricity only
 Plug-in hybrid electric vehicles (PHEVs) with a battery supporting short driving distances
and a combustion engine, that can be used as a range extender, allowing for longer
distance driving
With implementation of intelligent charging, electric vehicles can form a distributed flexible
electricity demand. BEVs represent a shiftable electricity demand, as the demand can be shifted
in time. The electricity demand of PHEVs can be characterised as a convertible electricity
demand, since the combustion engine can be used as an alternative.
If intelligent vehicle-to-grid power capability is additionally implemented, the batteries can form
a large-scale distributed electricity storage controlled by the needs of the electric system [50].
Electric vehicles can deliver high power to the grid within seconds and are well suited for
delivering regulating power, and primary and secondary reserves [14, 15, 82]. Delivery of such
ancillary services will very likely be the most beneficial are of the vehicle-to-grid capability [53].
Power flows between vehicle and grid can only take place when the vehicle is grid-connected. A
typical weekday driving pattern can be assumed to express a cycle of driving for work in the
morning and returning home in late afternoon. Electric vehicles will during this period only be
grid connected to the extent that charging spots at employer lots and e.g. mass transit stations are
set up and used. Moreover, with current battery technology, a typical battery electric vehicle has a
driving range up to around 150 km per full charge [23] corresponding to a few hours of driving.
Therefore, charging of BEVs on a daily basis will likely be pursued in most cases. If the expected
larger driving ranges of BEVs, of up to 350 km towards 2030 [53], are achieved, daily charging
would not necessarily be needed. However, even for a battery size supporting such a long driving
range, 50 kWh-e [94], the electricity storage would not correspond to more than 7 hours of full
load charging7. This indicates that the storage capacity of electric vehicles will, even in the longer
term, mainly be suited for power balancing intra-day power and intra-hour (rather than intra-week
balancing). This assessment is in agreement with the view in [82].
PHEVs have a shorter range in full electric mode, typically around 20-80 km per full charge [95].
Daily charging of PHEVs can be expected to be typical, due to the short battery range, and the
fact that the lowest driving costs are achieved in electric mode [17, 53]. As such, also PHEVs can
mainly be expected to provide power balancing intra-day.
A very ambitious large-scale implementation of electric vehicles corresponding to 53 % of the

private passenger vehicle fleet, hereof 2/3 PHEVs and 1/3 BEVs (in total 1.7 million electric
vehicles as assumed by 2030 in Paper II) corresponds to a total electric storage capacity of around
40 GWh-e. This corresponds to some hours of high excess electricity production. As such, even
7
Assuming a grid-connection capacity of 6.9 kW, corresponding to a standard 230V connection with 3 phases
10 Amps each [15].
17
with a very large implementation, the electric vehicles cannot store several days of high excess
electricity production.
If electric vehicles are not charged intelligently they will typically be charged at around 17:00-
21:00 o'clock when people return from home, i.e. at the same time as the conventional electricity
demand peaks. With intelligent charging, the battery can instead be charged over night when
electricity prices are typically low and wind power high [14]. If utilising vehicle-to-grid
capability, electric vehicles can deliver power back to the system, when conventional electricity
consumption peaks and electricity prices are high, typically at around 9:00-13:00 o'clock and
16:00-20:00 [14], and/or when it is affordable to provide ancillary services.
2.5 Overview of the technologies' power balancing potentials

Figure 8 summarises the categorisation of the technologies, with respect to the time scales they
will typically be suited for in supporting wind power integration.
Intra-hour Intra-day Intra-week Seasonal

Electric vehicles
Individual heat pum ps
Large heat pumps/electric boilers*
Figure 8. Categorisation of a number of flexibility measures in terms of the time scale, at which they will
typically be suited for, in providing power balancing. *Seasonal balancing requires connection to seasonal
thermal storages.
As shown, individual heat pumps and electric vehicles can provide power balancing intra-day and
can also contribute in delivering ancillary services intra-hour. The connection to the large thermal
storages in the district heating system makes large heat pumps and electric boilers capable of also
providing intra-week balancing. In addition, they can contribute to seasonal power balancing, if
combining them with seasonal thermal storages. Seasonal power balancing can however currently
largely be handled through the electrical interconnections with Norway and Sweden, giving
access to vast hydro power reservoirs [14]. As mentioned, the models used, Balmorel and
EnergyPLAN, do not include operation of ancillary services intra-hour; however, the need for
reserve capacities is included in the analyses in Balmorel (see Section 3.5).
In absolute terms (GWh-e), the power balancing potentials offered by individual heat pumps and
electric vehicles will depend on the scale at which they are deployed in the system. However, the
large scale implementations analysed in this thesis indicate that their balancing potential is in the
same order of magnitude, i.e. 20-40 GWh-e, corresponding to some hours of high excess
electricity production. According to data in the models, Balmorel, EnergyPLAN, and Ramses, the
Danish district heating storages are in the same order of magnitude, i.e. 30-50 GWh-th. However,
Energinet.dk roughly estimates the district heating storage to be around 300-500 GWh-th [82].
The magnitude of the different balancing potentials is illustrated in Figure 9. A district heating
storage of 50 GWh-th, as used in Balmorel and EnergyPLAN, is here illustrated in parallel with
the median of Energinet.dks high estimate, 400 GWh-th. As shown, the power balancing
potential, represented by the district heating storages, depends on whether large heat pumps or
electric boilers are used (due to differences in the electricity to heat conversion efficiency). The
18
balancing potentials are purely technical, not economic, and are only used to illustrate
magnitudes.
Individual HPs
EVs
Large HPs, DH
Large EBs, DH
Large HPs, DH, High estimate
Large EBs, DH, High estimate
0 50 100 150 200 250 300 350 400 450

GWh-e
Figure 9. Magnitude of technical power balancing potentials (GWh-e) provided by individual heat pumps,
electric vehicles, and the use of large heat pumps or electric boilers in Denmark. HP: heat pump. EV: electric
vehicle. EB: electric boiler. DH: district heating.
Figure 9 shows that that in absolute terms, a large scale implementation of individual heat pumps
and electric vehicles can provide power balancing in the same order of magnitude as the existing
Danish district heating storages, according to the district heating data in the three energy system
models. According to the higher estimates by Energinet.dk, the district heating system can
however provide considerably larger balancing potentials, corresponding to up to several days of
high excess electricity production. The Danish District Heating Association is currently
investigating the size of the district heating storage potential [91].
In total, large heat pumps/electric boilers in connection with district heating storages, individual
heat pumps and electric vehicles, can potentially provide a power balancing potential in the
magnitude of 80-460 GWh-e, depending on the assumed district heating storage potential and
whether large electric boilers or large heat pumps are used. In comparison, around 2000 GWh-
e/year of excess electricity production have in a technical optimisation in the EnergyPLAN model
been estimated, for the Danish energy system by 2020 with around 50 % wind power (in a
situation without utilising these flexibility measures). Considering that the storages can be filled
and emptied many times over the year, this indicates that heat pumps/electric boilers and electric
vehicles can contribute significantly in increasing wind power utilisation and reducing excess
electricity production, in the near term.
In a long term perspective, for a Danish system in 2050 independent on fossil fuels and with
fluctuating renewable energy sources (mainly wind) comprising up to 80 % of annual electricity
generation, Energiet.dk assess that heat pumps, heat storages, electric vehicles, and electrical
interconnections, are not enough to handle the power fluctuations and to ensure security of supply
[82, 96]. Moreover, the scarcity of biomass resources and the need for liquid/gaseous fuels for
transport could create a need for converting electricity via electrolysis and hydrogen into various
forms of renewable gases and liquids [13]. The vast gas storage capacities in salt caverns and
underground formations could play an important role for storing such fuels [13, 82, 96]. The gas
storages yield considerable storage potentials, around 11 TWh if stored as e.g. methane and 3-4
TWh if stored as hydrogen [82].
19
Overall, the technical power balancing potentials offered by heat pumps/electric boilers, heat
storages, and electric vehicles, are of a magnitude facilitating a significant contribution to
integrating wind power, in the near term. In the long term, towards a 100 % renewable energy
system by 2050, additional energy conversion and storage technologies will likely become
increasingly valuable.
20
Methods
3 Methods
Energy systems modelling and analyses forms the backbone of the methods applied in the thesis.
As such, energy models are used in five of the six papers in the thesis. The Balmorel model is
used in most of the studies, while EnergyPLAN is used in one of the papers. In this chapter, the
background for selecting these two models is described. After short descriptions of the models,
the model developments are then presented.
In the process of modelling and analysing individual heat pumps, several research institutions and
producers of heat pumps have been contacted, in order to provide a solid ground for the work (see
Acknowledgements). For model validation purposes, comparisons have been made with
measured data for heat pump installations in Danish one-family houses [76].
3.1 Choice of models

The models used in the analyses have been chosen based on the following criteria. The models
should:
 Have a sufficiently detailed representation of energy conversion and storage technologies
(bottom-up models).
 Include hour-by-hour variations in wind power generation (and electricity and heat
demand), in order to fully capture the influence of wind power variability.
 Represent the Danish CHP system, including back-pressure and extraction units, heat
only boilers, large heat pumps, electric boilers, and thermal storages.
 Be able to cover individual heating (relevant for analyses focusing on individual heat
pumps) and private passenger road transport (relevant for analyses focusing on electric
vehicles)8.
 Include a valid data set for the existing Danish heat & power plants and transmission and
storage capacities.
 Be validated in scientific international journals.
Balmorel and EnergyPLAN both satisfy the criteria above and have therefore been selected. The
reason for using EnergyPLAN in the first analysis of individual heat pumps is that such heat
pumps have as a starting point been well represented in this model. Furthermore, EnergyPLAN
includes the possibility of applying a technical regulation strategy. This facilitates analysing the
potentials of individual heat pumps and heat storages, with respect to reducing excess electricity
production and fuel consumption of the system, utilisation as much wind power as possible. Such
technical potential cannot directly be revealed in economic optimisations.
The background for putting main emphasis on Balmorel in the PhD project is the following. First
of all, Balmorel has the capability of optimising both investments and operation of all units in the
energy system. This renders it possible to analyse the economic competitiveness of the different
flexibility measures and to investigate whether they can facilitate increased wind power
investments. Moreover, it allows for analysing to which extent flexible operation of e.g.
individual heat pumps can reduce the need for investing in dispatchable power capacities. Finally,
8
In Balmorel, only private passenger vehicles are currently represented via the transport add-on. EnergyPLAN
also covers the other parts of the transport sector.
21
Methods
Balmorel provides direct access to model code, which is a necessity for allowing larger
independent model developments.
3.2 EnergyPLAN
The EnergyPLAN model is developed and maintained at Aalborg University in Denmark.
EnergyPLAN is a deterministic model, based on analytical programming and a selection of
regulation strategies [97]. The model optimises the operation of an energy system over a full year
with hourly time resolution. Inputs are demands and demand distributions, capacities of
technologies included, distributions of fluctuating renewable energy sources, fuel potentials, fuel
and CO2 costs etc. The model outputs comprise energy balances, energy productions, fuel
consumptions, CO2 emissions, and costs etc. The model covers the whole energy system, i.e.
individual heating and district heating, the electricity, transport, and industry sector [98].
EnergyPLAN is further described in [98, 99] and in [59], where previous applications and
comparison with other models can also be found. The model allows the use of a technical
regulation strategy, minimising fuel consumption of the system, or a market-economic regulation
strategy, identifying the least-cost solution based on the business-economic costs of
the individual plant owners, assuming perfect competition. In EnergyPLAN, a full year
simulation with hourly time resolution takes only a few seconds [98].
Individual heating is in EnergyPLAN represented as the total of space heating and hot water,
using an aggregated heat demand profile. The electric boilers, which supplement individual heat
pumps in covering peak loads, are included. The assumed capacity share of the heat pump is
given as input by the user as a fraction (from 0 to 1). A heat storage capacity can be given as
input (expressed in days of average heat demand).
3.3 Model development in EnergyPLAN

The model development performed in EnergyPLAN renders it possible to represent the following
heat storage options:
 Intelligent passive heat storage: utilising the heat storage capacity of the building
structure for flexible heat pump operation.
 Heat accumulation tanks: using a thermal storage tank operating on the space heating
circuit for flexible heat pump operation.
The modelling of passive heat storage is focused on representing radiator heating, as this will be
the dominating central heating system in Danish houses for many years to come (see Section 2.2).
The modelling of heat accumulation tanks is generally applicable to both radiator heating and
floor heating systems. The principle of the two different types of heat storage is illustrated in
Figure 10.
22
Methods
Acc.
Heat Heat
tank
pump pump
Radiators Radiators
Passive heat storage Storage in heat accumulation tank
Figure 10. a) Passive heat storage in the building structure. b) Heat accumulation tank inserted on the space
heating circuit (both illustrated for a radiator heating system, air-water heat pump).
As shown, in the passive heat storage concept, the radiators transfer heat to the indoor air (and to
directly to the building structure through radiation) and heat is then exchanged between indoor air
and building structure, via radiation, thermal conductivity, and convection [83]. The heat capacity
of the building structure can thereby be utilised for flexible heat pump operation. The storage
capacity of the building is mainly comprised by internal constructions in walls, ceiling, and floor,
while windows, doors, and furniture have minor influence [80]. The term intelligent heat storage
in this context refers to the fact that the storage capacity of the building is utilised for flexible
heat pump operation, based on what is optimal for the energy system. As shown in Figure 10b,
the heat accumulation tank interacts with the heat pump and the central heating system
The heat storage options have been integrated in the technical optimisation strategy in
EnergyPLAN, minimising excess electricity production and fuel consumption of the energy
system and utilising as much wind power as possible. The model development ensures that heat
stored in the building structure or in heat accumulation tank can be used to satisfy space heating
demand only. This is implemented by restricting the heat pump to satisfy the hot water demand in
each hour, as a minimum (the hot water tank is not modelled).
Intelligent passive heat storage has been modelled based on considerations of heat capacities and
temperature states of the relevant thermal masses. Due to the low heat capacity of air (0.8
Wh/°C/m2 floor area [40]) and the fact that a radiator heating system is considered, it is assumed
that a desired indoor air temperate can quickly be reached. Due to the large thermal mass of the
building structure (60-140 Wh/°C/m2 floor area, see Section 2.2), the temperature of this mass is
assumed to vary from hour to hour, depending on the amount of heat stored in it. The possible
loading/unloading to/from the passive heat storage in a given hour is assumed constrained by the
temperature difference between indoor air and the building structure. As a result, the following
loading/unloading restrictions have been implemented for the passive heat storage:
 When the passive heat storage is empty, the possible loading is highest, while the possible
unloading is zero.
 When the passive heat storage is full, the possible unloading is highest, while the possible
loading is zero.
23
Methods
 In the stages between empty and full passive heat storage, loading and unloading is
constrained, based on assuming a linear relation between the situation at empty and full
storage.
As a result, the development in the heat storage level during a full loading or unloading period is
not linear. Rather, the possible marginal increase in the storage level is reduced during a loading
period, and the possible marginal decrease in the storage level is reduced during an unloading
period. This is illustrated in Figure 11 for a simulated loading and subsequent unloading period.
2500
2000 Excess electricity
MW
1500 production
1000
Condensing power
500
production
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64
Hour
2000
Heat demand
1500
MW
1000
Heat pump
500
production
0
Loading(-)
-500
/unloading(+)
-1000
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64
Hour
100%
Storage level
80%
60% Stor(t)/Storcap
40%
20%
0%
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64
Hour
Figure 11. Optimisation of a large-scale installation of heat pumps using passive heat storage, modelled in
integration with the energy system (allowing ∓ 1°C indoor air temp. var. and assuming a passive heat storage
capacity of 120 W/m2 floor area/K). A loading and subsequent unloading period is shown, together representing
around 2-3 days in mid January with relatively mild weather. Excess electricity production and condensing
power production is shown after the optimised heat pump operation minimising their occurrence [83].
In the model, loading of the passive heat storage corresponds to a situation where the heat transfer
from indoor air to building structure is higher than the transmission loss from building structure
to ambient air (outdoor environment). Correspondingly, unloading of the storage represents a
situation where the heat transfer from indoor air to building structure is lower than the
transmission loss from building structure to ambient air. The model is further described in Paper
III, where the equations are also presented.
The model development makes it possible to analyse the potentials of individual heat pumps and
the given heat storage options, in terms of supporting a fuel- and cost-efficient utilisation of wind
power.
24
Methods
3.4 Balmorel
The Balmorel model was originally developed in a project financed mainly by the Danish Energy
Agency and hosted by the former Danish TSO, ElkraftSystem and is today developed and
distributed under open source ideals. Balmorel is currently used and developed at the Technical
University of Denmark, at RAM-løse, at Ea Energy Analyses, and at the Danish Energy
Association, among other institutions. Balmorel is an optimisation model and is deterministic,
assuming perfect foresight. Furthermore, Balmorel assumes perfect competition, i.e. without
monopolies or skewed market power and where all actors act economic rationally seeking to
maximise their profit [2, 17, 60] [100]. A linear version of the Balmorel model is applied in this
thesis.
The model optimises investments in power/heat production, storage, and transmission capacities,
and the operation of the system. This is done minimising total costs in the energy system over a
given year, covering annualised investment costs, operation and maintenance costs of existing
and new units, and fuel and CO2 quota costs. The optimisation is performed subject to a number
of constraints, including satisfaction of demands for electricity and heat in each time period, fuel
potentials, and technical unit restrictions on units in the system. The model covers the electricity
and district heating sector, and additionally, individual heating, industry, and part of the road
transport sector, depending on the use of model add-ons and data.
Inputs to the model comprise demands and demand distributions and capacities, technical
lifetimes, and efficiencies of existing technologies as well as costs, technical lifetimes, and
efficiencies of technologies available for investment, and distributions of fluctuating renewable
energy sources, fuel potentials, and fuel and CO2 costs etc. The model outputs comprise
investments, electricity/heat generation, fuel consumptions, CO2 emissions, and costs etc.
The assumed perfect foresight means that the model does not capture forecast errors on wind and
electricity demand and the resulting need for operation of reserves. However, the need for reserve
capacity can be represented in Balmorel, e.g. in the form of a capacity balance restriction, as used
in this thesis (see Section 3.5). Furthermore, the linear version of Balmorel applied does not
include start-up costs, minimum load requirements, or part load efficiencies.
Balmorel operates with three geographical entities: countries, regions, and areas. Countries are
divided into regions connected with transmission lines and regions are further divided into areas.
Electricity supply and demand is balanced on regional level, whereas heating is balanced on area
level. The optimisation is performed with a yearly time horizon and the year is divided into
seasons, which may be used to represent weeks, and into time periods, which may represent
hours. Model runs can take minutes to days depending on the size of the problem.
Individual heating can be represented on aggregated level, typically with two areas representing
Eastern Denmark and Western Denmark, respectively. Individual heating is represented as
demand for space heating and hot water in total, using an aggregated demand profile. Within each
individually heated area, different types of heating installations can be combined, e.g. heat
25
Methods
pumps, fuel boilers, solar thermal, and electric boilers as well as thermal storage tanks. Electric
boilers are not handled as an integrated part of the heat pump unit.
3.5 Model development in Balmorel

Intelligent heat storage in the building structure is in the analysis in EnergyPLAN (Paper III)
identified as the most promising option, among those analysed. Therefore, particular emphasis is
put on this heat storage option, in the further model development. In this regard, it is considered
important to obtain a better representation of the thermal building dynamics. Furthermore,
EnergyPLAN is not capable of optimising investments, and the technical optimisation applied
does not account for electricity import/export. Finally, the modelling approach in Paper III does
not facilitate analysing the use of thermal storage in concrete floor heating systems or in hot
water tanks. Therefore, further model development is made in order to capture the above aspects.
A thermal building model add-on has thus been developed for the investment model Balmorel
covering the following heat storage options:
 Intelligent heat storage in the building structure:
o For houses with radiator heating: Passive heat storage in walls, ceiling, and floor9
o For houses with (concrete) floor heating: Active heat storage in the floor (and
passive heat storage in walls and ceiling).
 Heat accumulation tanks for space heating.
 Hot water tanks for domestic hot water (showering, dish washing etc.)
The scientific contribution of the model development is that it covers a wide range of different
heat storage options for individual heat pumps and at the same time optimises both operation and
investments in integration with the energy system. Such model capability not been observed in
previous studies.
The basic principles of passive heat storage and heat accumulation tanks are here the same as
previously illustrated in Figure 10. However, the thermal building dynamics are as mentioned
better represented. In addition, heat losses from the thermal storage tanks are included as heat
contributions to the indoor air. The hot water tank is modelled as a traditional thermal storage
operating on the hot water circuit only.
Floor heating systems here represent floor heating systems used as the central heating system of
the house, i.e. not only in bathrooms/kitchen. The modelling of floor heating is generally
applicable to all floor heating systems. However, concrete floor heating systems are most
relevant, due to their current market dominance and large thermal capacity. In concrete floor
heating systems, heat is transferred to the concrete floor via water tubes, and then from the floor
to the indoor air via thermal conductance, convection, and radiation [72] (see Figure 12).
9
Windows, doors, and furniture have minor influence [80].
26
Methods
Heat
pump
Figure 12. Active heat storage in concrete floor heating systems.
In connection with flexible heat pump operation, concrete floor heating systems can exploit that
the heat supply (and thus the electricity consumption) can be changed rapidly, while the indoor
temperature will change slowly [31]. This is due to the large heat capacity of the concrete floor,
and that heat can be delivered directly to this thermal mass.
The thermal building model is based on physical definitions and relations of heat transfers and
heat capacities and can be characterised as a linear state space model. The modelling of radiator
heating systems and floor heating systems, respectively, is illustrated in Figure 13 and Figure 14.
For simplicity, only the space heating circuit is here shown, i.e. excl. the hot water circuit.
CI , TI QIB QBO
QSH
QTLos CB
TO
HP QPel TB
ACT
QSH QVen
LDACT
CF, T F
Figure 13. Thermal building model of houses with radiator heating (only the space heating circuit is illustrated).
HP: Heat pump (or other heating installation). ACT: Heat accumulation tank. For further nomenclature, see the
beginning of the thesis.
27
Methods
CI, T I QIL QLO
QTLos CL
TO
HP QPel TL
ACT QVen
LDACT QFI
QSH QSH CF, TF
Figure 14. Thermal building model of houses with floor heating (only the space heating circuit is illustrated).
HP: Heat pump (or other heating installation). ACT: Heat accumulation tank. For further nomenclature, see the
beginning of the thesis.
As shown, the model includes the ambient temperature (TO) and the heat capacity and
temperature states of the indoor air (CI, TI) and building structure (CB, TB for houses with radiator
heating, and CL, TL and CF, TF for houses with floor heating). In houses with radiator heating, the
heating installation (e.g. a heat pump) delivers heat (QSH) directly to the indoor air, while in
houses with floor heating, heat is delivered to the concrete floor. In a given hour, heat will be
exchanged between the masses depending on the temperature difference between them, and the
heat transfer coefficients applied (U values). The heat losses through the building envelope will
create a need for space heating in order to ensure a given indoor temperature. Ventilation losses
(QVen) occur from indoor air to ambient air, through the small cracks in the building envelope etc.
Transmission losses occur in two steps: 1) From indoor air to building structure (QIB and QIL,
respectively) and 2) From building structure to ambient air (QBO and QLO, respectively). When
allowing some indoor temperature variation, the heat storage capacity of the building can be
utilised for shifting space heating in time. Alternatively or as a supplement, if investing in a heat
accumulation tank, this can be used as buffer on the space heating circuit. The model is described
in detail in Paper IV, where all equations are also presented.
Figure 15 shows in a very simplified manner how the thermal building model add-on is integrated
into the energy system model Balmorel.
28
Methods
Electricity
Flexible elec. demand Thermal building model
Heat
- Indiv. HPs incl. EBs Individual heat
Fuel
- Indiv. fuel boilers demand
Import/export
- Heat storages
Wind power +
other RES
Conventional
PP electricity
demand
Fuels
- Natural gas District
- Coal CHP heating Large HPs
- Biomass storages EBs
- Waste
Boilers District heat
demand
Solar thermal
Figure 15. The integration of the thermal building model add-on into the energy system model Balmorel. The
illustration is very simplified and does not reveal all technologies or functionalities covered in Balmorel.
PP: Power production, CHP: Combined heat and power production, HP: Heat pump, EB: Electric boiler. RES:
Renewable energy sources.
As shown in Figure 15, the thermal building model represents the individual heating sector
including, heat demands, heat supply technologies, and heat storages including the building
structure. When combined with heat storages, the electricity consumption for the heat pumps
becomes flexible. As illustrated, the electricity consumption for large heat pumps and electric
boilers in the district heating system also becomes flexible, when connected to the district heating
storages. The competition between these flexibility measures is captured in the model.
For validation purposes, the thermal building model has been compared against measured data for
heat pump installations in Danish one-family houses. This comparison is illustrated in Figure 16.
As indicated, the thermal building model gives a reasonable representation of actual space
heating profiles.
Nov 1 - Dec 31 Mar 15 - May 15, Sep 15 - Dec 31

0.07 0.07
Space heating, normalised
Measured av. Measured av.

0.06 profile
0.06 profile
0.05 0.05
0.04 Modelled
0.04 profile, Modelled profile,
estimation A estimation A
0.03 0.03
0.02 0.02
Modelled profile, Modelled profile,
0.01 0.01
estimation B estimation B
0.00 0.00
1 3 5 7 9 11 13 15 17 19 21 23 1 3 5 7 9 11 13 15 17 19 21 23
Hour Hour
Figure 16. Model generated average diurnal space heating profile, compared to a corresponding measured
average profile for heat pumps in Danish one-family houses [76]. Radiator heating systems are illustrated.
Estimation A: UBO for monitored houses estimated based on construction year. Estimation B: UBO for monitored
houses estimated based on measured space heating over the given period. a) Comparison covering November 1
to December 31 (data available for 24 houses). b) Comparison covering March 15 to May 1 and September 15
to December 31 (data available for 5 houses).
29
Methods
The situation without flexible heat pump operation is modelled by restricting the indoor
temperature to a fixed level (in addition to deactivating the option of investing in thermal storage
tanks). This reflects the typical constant temperature set-point applied in individual heat pump
installations in Danish houses today [70, 76]. In this situation, the heat pump will be forced to
supply the amount of heat needed to cover the net heat demand in the given hour, i.e. the
transmission and ventilation minus the heat contribution from other sources (persons and
electrical appliances10). As shown in Figure 17, the space heating supply will in this situation be
high when the ambient temperature is low, and vice versa.
4500 20 Space heating

4000 15
3500 10
Ambient temp. ( C)
3000 Transmission loss
5 from indoor air to
MW-th
2500
0 construction
2000
-5 Heat from persons
1500
and appliances
1000 -10
500 -15
Ventilation loss
0 -20
T113
T117
T001
T021
T041
T061
T081
T101
T121
T141
T161
T013
T033
T053
T073
T093
T133
T153
T005
T025
T045
T065
T085
T105
T125
T145
T165
T017
T037
T057
T077
T097
T137
T157
Ambient
S02 S12 S40 S51 temperature ( C)
Week, hour
Figure 17. Space heating from individual heat pumps, other heat contributions, heat losses, and ambient
temperatures, for a situation without heat storage investments, and with a fixed indoor temperature requirement
of 21.5 °C (the simulated heating season is illustrated focusing on houses with radiator heating).
When allowing the indoor temperature to vary within a certain interval, the heat pumps can be
operated flexibly, utilising intelligent heat storage in the building structure. In this situation, the
heat pump operation will be prioritised for hours with low marginal electricity production costs,
which can be interpreted as electricity prices. This is illustrated in Figure 18. Additionally, it
enables using the heat pumps for peak load shaving, as described later in this section, and
illustrated in Section 4.2.2.
1600 140
1400 120
1200 Electricity
100 consumption for
€/MWh-e
1000 individual heat

MW-e
80
800 pumps (MW-e)
60
600
400 40
Electricity price
200 20
(€/MWh-e)
0 0
T001
T021
T041
T061
T081
T101
T121
T141
T161
T013
T033
T053
T073
T093
T113
T133
T153
T005
T025
T045
T065
T085
T105
T125
T145
T165
T017
T037
T057
T077
T097
T117
T137
T157
S02 S12 S40 S51

Week, hour
Figure 18. Modelled flexible operation of individual heat pumps in the heating season, shown for Western
Denmark by 2030.
10
Heat contribution from solar transmission has been excluded due to lack of data. Considerations on this issue
are given in Paper IV.
30
Methods
Apart from incorporating thermal building dynamics, the model development generally improves
the representation of individual heat pumps and other individual heating technologies. As such,
the following investment constraints are implemented for individual heating installations:
1. It is ensured that each primary heating installation (heat pump or fuel-based boiler)
complies with building regulations, in terms of being able to cover the dimensioning heat
loss, at an ambient temperature of -12 °C and an indoor temperature of 20 °C, and satisfy
the average hot water demand, at the same time [101, 102].
2. For individual heat pumps, the use of a supplemental electric boiler to cover peak loads is
modelled. This is implemented by defining the capacity share covered by the heat pump
(typically 72-82 % [66]).
3. In reality, residents in individually heated areas will invest in only one type of primary
heating technology (in Denmark typically a fuel boiler or heat pump). Therefore, the use
of one new primary heating installation (e.g. heat pump) for base load operation,
supplemented by another new installation (e.g. natural gas or wood pellet boiler) for peak
load operation, should be avoided in the model. This is taken into account, by restricting a
minimum number of full load hours for new primary heating technologies in each area.
4. Based on space requirements, the possible investment in heat accumulation tanks is
constrained to maximum 1000 litres per house [18].
The first of the above constraints ensures that the required capacity of individual heating
installations cannot be reduced through flexible operation of these, as in reality. Furthremore,
the constraint ensures a more correct dimensioning of the individual heating installations by
avoiding an underestimation of the required capacity. The reason is that the need for individual
heating capacities would in Balmorel normally be based on aggregated heat demand profiles for
several thousand of households. For such an aggregated heat demand, the peak demand would be
relatively low due to levelling effects from the aggregation. In reality, each household must be
able to cover their peak heat demand. This is reflected in the constraint implemented.
Alternatively, the underestimation problem could to some extent be handled by setting a
maximum number of full load hours in the model. However, the constraint described above is
more accurate, as it is based on actual building regulations.
The inclusion of the electric boiler (constraint no. 2) is mainly important due to its high impact on
the peaks in the electricity consumption of the total heat pump unit. This can be determining for
the need for peak and reserve capacity investments. Moreover, the model will generally seek to
minimise the use of the supplemental electric boiler, due to its low efficiency.
In Balmorel, sufficient dispatchable power capacity in each power region can to some extent be
ensured by including the period of the year with highest net loads in the optimisation (residual
loads, i.e. gross load minus wind power). This would however likely underestimate the need for
peak and reserve capacity due to the assumed perfect foresight on wind power and load.
Furthermore, electricity import would in the optimisation contribute in covering the high net
loads, while in reality, risk of failures on transmission lines should be taken into account.
31
Methods
A capacity balance restriction has therefore been developed, inspired by [2] to ensure adequate
production capacity and reserve margin in each power region (for full nomenclature, see
beginning of the thesis):
, + , ∙ ≥ , , + , , ,
∈ ( ) ∈ ∈ ( ) ∈
∀ ∈ ; ∈ ;ℎ ∈
The term capacity credit (cc) is here applied, which represents the share of total installed capacity
that is available for electricity generation at a certain level of confidence. Based on [2], the
capacity credit of conventional units is set to 0.99, while the capacity credit for wind power is set
to 0.14. The capacity balance restriction takes into account that the need for new production
capacity and reserves can be reduced through peak load shaving. As the constraint is applied for
each hour and week, it allows for lowering the heat pump operation in peak load hours, taking the
hourly pattern of conventional (DElConv) and flexible electricity demand (DElFlex) into account. This
is new compared to the capacity balance restriction in [2] (also applied in Paper II), where the
right hand side of the equation is a constant peak load set for each country. The peak load shaving
capability is illustrated in Section 4.2.2.
The possibilities for exploiting intelligent heat storage in the building structure depend on the
characteristics of the given house. This is taken into account in the modelling of the stock of one-
family houses. As such, individually heated houses can in the model be divided into a number of
different areas depending on: the type of central heating system, space heating demands
(represented by heat transfer coefficients), heat capacity of the building, hot water demand, and
house size. In the applications of the model, the individually heated (existing) Danish one-family
houses have been divided into 10 different areas in Eastern and Western Denmark, respectively.
The use of an aggregated energy system model means that the heat storage capacity within each
heating area can be shared among the heat pumps within the area. This implies best-case
conditions for the heat storages, since the flexibility they offer can thus be distributed among the
houses. Ideally, each of the many thousands of individually heated houses should be modelled
separately (716,000 in this case). However, it would hardly be realistic to run such a model.
Due to the linearity of the model versions applied, the modelled heat pumps will be able to
operate continuously between zero and full load. The observant reader might wonder how this fits
with the fact that today’s heat pumps are typically on/off regulated (as mentioned in see Section
2.2). However, with the hourly time resolution applied, the continuously variable operation
corresponds to on/off regulation, where it merely varies how large a share of the given hour the
heat pump is operating.
32
Energy systems analyses
4 Energy systems analyses

In this chapter, an overview is first given of the energy systems analyses performed, in terms of
preconditions and focus. Subsequently, the most important results are presented. Finally, the
results are summarised and compared.
4.1 Overview of energy systems analyses

Table 4 gives an overview of the flexibility measures included in the different energy systems
analyses. As shown, large heat pumps, electric boilers, and thermal storages in the district heating
system are included in all the analyses. The reason is that these technologies compete with
individual heat pumps and electric vehicles, in providing flexibility to the system. For instance,
the large heat pumps/electric boilers will compete with individual heat pumps, in terms of
prioritising operation for low electricity price periods. Furthermore, the system effects of the
flexibility measures in the district heating system are in Paper II compared to the effects of
electric vehicles. Possibilities to invest in transmission capacity expansions, further than planned,
have moreover been included in Paper II and Paper V, representing other competing flexibility
measures.
Table 4. Overview of flexibility measures included in the different energy systems analyses performed in the
PhD project.
Paper II +
Sector Flexibility measure Report Paper III Paper Va Paper VI
District heating Large heat pumps, electric boilers, thermal storages X X X X
Individual heating Individual heat pumps X X X
Electric boiler supplementing in covering peak loads X X (X) b
Intelligent passive heat storage in the building X X
structure via radiator heating
Intelligent active heat storage in concrete floors via X
floor heating
Heat accumulation tanks X X (X)c
Hot water tanks X (X)c
Transport Battery electric vehicles and plug-in hybrid electric X
vehicles
a
The same applies to the analyses performed in Paper IV, illustrating model functionality.
b
Electric boiler is included along with other individual heating installations but not as an integrated part of the heat pump
unit.
c
Individual thermal storage tanks included for total heating (space heating and hot water).
An overview of the energy systems analyses is given in Table 5, in terms of focus, preconditions,
and type of model and optimisation applied. As shown, the analyses differ with regard to whether
Danish policy goals concerning renewable energy and fossil fuels have been implemented as
boundary conditions. This is due to fact that the Danish energy policy has changed during the
course of the PhD project and due to differences in focus of the studies.
Paper II pays special attention to the development in the portfolio of power & heat plants over a
period of years towards 2030, when assuming a gradually increasing penetration of electric
vehicles. Therefore simulations are made with five year intervals, where optimal investments
33
identified in previous years are included in the optimisations of subsequent years. In the
remaining papers of the thesis, a single year is optimised.
The technical optimisation in EnergyPLAN (Paper III) is applied in order to investigate the
potentials of individual heat pumps and heat storages in terms of increasing wind power
utilisation and minimising fuel consumption of the system. In the analyses performed in
Balmorel, the aim is to achieve the lowest cost solution for society incl. investment costs.
Therefore, an optimisation minimising socio-economic investment and operation costs, i.e.
excluding taxes, subsidies, and tariffs, has been applied.
Table 5. Overview of focus and preconditions for energy systems analyses performed during the PhD project.
Paper II + Report Paper III Paper V Paper VI

Year 2015, 2020, 2025, 2020 2030 2050
2030
Technology focus Electric vehicles Individual heat pumps Individual heat pumps District heating vs.
individual heating
Model Balmorel EnergyPLAN Balmorel Balmorel
Optimisation Min. socio-economic Min. fuel consumption, Min. socio-economic Min. socio-economic
costs incl. investments operation only costs incl. investments costs incl. investments
and operation and operation and operation
Wind power in DK 20-50 % (resulting 50% set (and 0-100 % 50-60% (resulting from 75% (resulting from
(pct. of annual elec. from optimisation) in sensitivity analyses) optimisation) optimisation)
demand)
Danish energy Expected wind power 50 % wind power by 50 % wind power by 100 % renewable
policies implemented capacities 2020 2020. Phase out of energy for electricity,
implemented in coal and individual oil heat and transport by
sensitivity analyses boilers by 2030. 2050
Investment restrictions
based on goal of 100
% renewable
electricity & heat
supply by 2035. The
biomass agreement.
Countries in focus DK, NO, SE, FI, GE DK (excl. electricity DK (incl. electricity DK (incl. electricity
trade) trade) trade)
Countries modelled DK, NO, SE, FI, GE DK DK, NO, SE, FI, GE DK, NO, SE, FI, GE
Sectors modelled Electricity, district Electricity, district Electricity, district Electricity, district
heating, transport* heating, individual heating, individual heating, individual
heating, transport, heating heating, industry
industry,
DK: Denmark. NO: Norway. SE: Sweden. FI: Finland. GE: Germany.
*Private passenger vehicles for road transport.
4.2 Results
4.2.1 System benefits of individual heat pumps and heat storages

The energy systems analyses in Paper III and Paper V both show that the installation of
individual heat pumps can significantly support the integration of wind power, while
complementing heat storages can contribute only moderately in this regard. This overall finding
is illustrated in the following.
In Paper III, individual heat pumps and different heat storage options are investigated, in terms of
their potentials in increasing wind power utilisation and reducing excess electricity production
34
and fuel consumption of the system. This is facilitated by applying a technical optimisation in the
EnergyPLAN model. The potentials are analysed in the context of a large-scale installation of
individual heat pumps in the Danish energy system by 2020 with 50 % wind power. Individual
heat pumps are assumed installed where they are considered particularly relevant in the short
term, i.e. in houses presently heated with oil boilers and/or electric heating (detached houses are
here considered). The resulting electricity consumption for individual heat pumps constitutes 1.9
TWh-e.
In Figure 19, the changes in fuel consumption of the system are shown, compared to a situation
without individual heat pumps.
Scenario De scription
HP-NOSTOR abbre viations
NOHP No heat pumps in the houses.
Tank300
HP-NOSTOR Heat pumps installed in detached
Tank1000 houses w ith oil boilers and/or electric
heating, but w ithout investment in
1C-PasLow heat storages.
Tank300/500/1000 Inv. in 300/500/1000 litre heat
1C-PasHigh accumulation tanks in all houses
w here heat pumps are installed.
2C-PasLow PasLow /High Inv. in passive heat storage in all
houses w ith heat pumps, assuming
2C-PasHigh
low /high end passive heat storage
cap. (60 Wh/K/m2 and 120 Wh/K/m2,
-5 -4 -3 -2 -1 0 1 2
respectively)
Change in fuel consumption excl. wind (TWh) 1C/2C Allow ing an indoor temp var. of
Oil Coal Natural gas ∓1°C/∓2°C w hen utilising passive
heat storage
Figure 19. Changes in fuel consumption when installing individual heat pumps and when further adding heat
storages to the heat pumps.
It can be seen that the installation of heat pumps results in large oil savings, resulting from the
displacement of individual oil boilers (HP-NOSTOR scenario). Due to the high efficiency of the
heat pumps11 combined with the high efficiency of the Danish CHP system (and displacement of
the less efficient electric heating), only moderate increases in coal and natural gas consumption
for electricity generation are observed. In total, a significant net reduction in the fuel consumption
of the system is achieved.
When equipping the heat pumps with heat storages (Tank300/1000, 1C/2C-PasLow/High
scenarios), the flexible heat pump operation achieved means that more wind power production is
utilised and more condensing power production displaced. This results in a lower net
consumption of coal and natural gas. However, compared to the fuel savings achieved by
installing the heat pumps, the fuel savings provided by the heat storages are moderate.
The explanation for this is that the efficiency advantage of the heat pumps has high impact on the
fuel consumption, due to the many operating hours of the heat pumps throughout the year. In
11
Average COP of 2.7 applied, representing heat pumps in radiator heating systems [11], hereof 75 % air-water
heat pumps and 25 % ground source heat pumps [120].
35
comparison, the potential for further reducing fuel consumption by increasing wind power
utilisation is limited by the fact that excess electricity production only occurs in some periods of
the year; in around 20 % of the year for the case of the Danish energy system by 2020. Overall, it
is thus found that in terms of reducing fuel consumption of the system, the installation of
individual heat pumps is the most important step, while the heat storages are less influential.
If applying a system boundary including not only Denmark but also neighbouring countries, the
fuel savings provided by enabling flexible heat pump operation would be even smaller. The
reason is that the excess electricity production, utilised with flexible heat pump operation, would
otherwise have been exported, displacing fuels in other countries (to the extent that the excess
electricity production it is within the transmission capacity). When taking this lost opportunity
into account, the fuel savings provided by the heat storages would be reduced further.
In Paper V, the system benefits of individual heat pumps are analysed in Balmorel, taking into
account effects on investments and applying a socio-economic cost optimisation. Moreover, the
influence of electricity import/export is covered by including the energy systems of Sweden,
Norway, Germany and Finland in the optimisation. The system in 2030 is analysed, including
Danish energy policies of e.g. phasing out individual oil boilers and the use of coal for
power/district heat generation. Individually heated one-family houses are modelled, representing
the bulk (78 % [103]) of individual heating in Denmark. Existing houses are considered, which
by 2030 will still comprise the large majority of the individual heat demand of Danish one-family
houses (see Figure 5). In total, about 716,000 individually heated houses are covered.
When allowing investment in individual heat pumps, the optimisation yields heat pump
investments in all the individually heated houses. This confirms the socio-economic
competitiveness of the heat pumps found in previous studies [11, 54]. As a result, the individual
heat pumps represent a significant electricity demand, i.e. 4.3 TWh. In comparison the expected
conventional Danish electricity demand in the same year is 30.1 TWh [84].
The results of the analysis show that the installation of the heat pumps can bring significant
benefits to the energy system, in terms of facilitating larger wind power investments and reducing
fuel consumption. This is illustrated in Figure 20 (iHP scenario vs. NOiHP scenario).
30
Scenario Description
25 NOiHP No individual heat pumps.
20 NOiHP iHP Investment in individual heat
pumps allowed
TWh
iHP iHP-Flex Investment in individual heat

15
Flex
iHP-Flex pumps allowed as well as in
10 thermal storages enabling
flexible operation of the heat
5 pumps
0
Wind Wood Munic. Straw Natural Wood
chips waste gas pellets
Figure 20. Primary energy consumption in the Danish energy system in 2030 for different scenarios.
36
As shown, the consumption of wood pellets and natural gas is reduced significantly when
installing individual heat pumps, displacing natural gas boilers and wood pellet boilers (23%). As
a result, the CO2 emissions of the Danish system are also reduced considerably (37 %). The
percentagewise CO2 reduction obtained is however relatively large, due to the implemented
political goal of phasing out coal, resulting in generally low absolute CO2 emissions. The large
reduction in the wood pellet consumption shows, that the individual heat pumps can contribute
significantly to reducing the pressure on the limited biomass resources. The national security of
supply is in this case furthermore improved since wood pellets are in Denmark typically
imported.
The increase in electricity demand resulting from the installation of individual heat pumps is
largely met by wind power (97 %, the remaining amount is supplied by natural gas power plants).
This is partly explained by the fact that wind power does not compete with coal power plants, due
to the implemented political phase out of coal in the system. Moreover, the other low cost fuels,
wood chips and municipal waste, are constrained by the available national resources, resulting in
consumption up to this limit in all scenarios (the more costly wood pellets, which constitute an
international trading good, are assumed unconstrained). Furthermore, even when individual heat
pumps are operated without flexibility, their electricity demand is distributed over many hours in
the day, and both heat demand and wind power is typically high in cold periods (see Figure 21).
This creates good possibilities for utilising wind generation in covering the electricity demand for
the heat pumps, while using dispatchable power generation to back it up. In contrast, e.g. dumb
charging of electric vehicles would typically be concentrated in a few hours of the day, when
people return from work, making it difficult to utilise wind power. The use of dispatchable power
capacities as back up for wind power, in meeting the electricity demand increase, is illustrated in
Figure 21.
2000
Wind power
Increase in elec. generation (MW-e)
1500
Wood chips
1000
500 Straw.
0
Wood pellets
T116
T118
T111
T001
T024
T047
T070
T093
T139
T162
T017
T040
T063
T086
T109
T132
T155
T010
T033
T056
T079
T102
T125
T148
T003
T026
T049
T072
T095
T141
T164
T019
T042
T065
T088
T134
T157
-500
S02 S12 S31 S40 S51 Natural gas
-1000
-1500 Elec. demand

increase due to
-2000 indiv. heat pumps
Week, Hour
Figure 21. Increase in electricity demand due to installation of individual heat pumps, for a situation without
flexible operation, shown together with the resulting increases/decreases in electricity generation.
As shown, the electricity generation on biomass and natural fired plants is increased, when
needed to supplement the wind generation. Due to the limited national biomass resources, the
electricity generation on the biomass fired plants is reduced accordingly in other periods.
Electricity import/export is also changed on hourly level but is for simplicity not shown on the
figure. The fluctuations in the increase in electricity demand in week S02 represent changes in the
37
operation of large heat pumps in the system, resulting from the installation of the individual heat
pumps.
When enabling flexible operation of the individual heat pumps, wind power generation is
practically unchanged, as shown in Figure 20 (iHP-Flex vs. iHP scenario). Natural gas
consumption is slightly decreased, due to an avoided use of open cycle gas turbines for peak load
operation. As a result, only moderate reductions in fuel consumption and CO2 emissions are
achieved (0.3 % and 2 %, respectively).
The effects on system costs are illustrated in Figure 22.
1800
1600
1400
NOiHP
1200
M€/yr
1000 iHP
800 iHP-Flex
600
400
200
0
Figure 22. Annualised costs for the modelled Danish heat and power system in 2030 for different scenarios.
As shown, fuel costs are reduced considerably with the installation of individual heat pumps (iHP
scenario). This is partly explained by the high efficiency of the heat pumps, and that the increase
in electricity demand is largely covered by wind power. Furthermore, the displaced technologies,
natural gas boilers and wood pellet boilers, have relatively high fuel costs. The increase in
investment costs is caused by: 1) larger investment costs of individual heat pumps compared to
individual gas boilers, and 2) the increased investments in wind power and dispatchable power
capacities, needed to supply electricity for the heat pumps. The total system costs for the Danish
heat and power system are reduced by 12 % with the installation of individual heat pumps. In
comparison, system costs are reduced by only 0.9 %, when investing in heat storages for the heat
pumps (iHP-Flex scenario).
Overall, the results show that the installation of individual heat pumps can provide significant
system benefits, in terms of facilitating larger wind power investments and reducing fuel
consumption, CO2 emissions, and system costs. Flexible operation of the heat pumps, enabled by
heat storages, can yield only moderate benefits in these regards. This result has been confirmed in
various sensitivity analyses, covering e.g. variation in investment costs of control equipment
required for the flexible heat pump operation, COP, investment costs, and capacity share of
individual heat pumps, discount rate, as well as fuel and CO2 prices. A full description of the
sensitivity analyses can be found in Paper V.
38
4.2.2 Influence of individual heat pumps and heat storages on investments

This section focuses on how individual heat pumps and heat storages affect energy system
investments. The section mainly draws upon the analysis in Paper V, applying the Balmorel
model and covering socio-economic optimisations of investments and operation of the Danish
energy system by 2030. Moreover, result figures in Paper IV are used to illustrate the peak load
shaving ability of the heat pumps (the preconditions and input data applied in Paper IV
correspond to those applied in the main scenarios in Paper V).
In Figure 23, the investments in electricity generation capacities in the Danish energy system are
illustrated for the three scenarios analysed: the scenario without individual heat pumps (NOiHP),
the scenario with individual heat pumps operated without flexibility (iHP), and the scenario with
allowed investments in heat storages for the heat pumps (iHP-Flex). In the iHP-Flex scenario,
investments in intelligent heat storage have been observed in 34 % of the houses with heat pump
installations. The heat storage investments occur within all 20 defined categories of individually
heated houses. However, the investments are highest in buildings with high heat capacities,
typically corresponding to buildings with a high share of concrete, and in buildings with high heat
demands (cf. Paper IV).
6000
5000
4000 NOiHP
MW-e
3000 iHP
2000 iHP-Flex
1000
Figure 23. Investments in electricity generation capacities for different scenarios (NOiHP, iHP, and iHP-Flex).
On: onshore, Off: offshore. WoodC: wood chips/waste, OC-GasT: open cycle gas turbines.
As shown, the installation of individual heat pumps facilitates significantly increased wind power
investments (iHP vs. NOiHP scenario). This forms the background for the increased wind power
generation, presented previously in Figure 20. Flexible operation of the heat pumps can be seen to
facilitate reduced wind power investments onshore and increased investments offshore (iHP-Flex
vs. iHP scenario). This is due to the fact that 1) flexibility added to a system generally improves
the conditions for technologies with relatively high investment costs and low variable costs, and
2) that offshore wind power has higher investment costs, but on the other hand higher obtainable
full load hours and a different variation profile. However, the total wind power generation is
practically unchanged, as evident from Figure 20.
As shown in Figure 23, the installation of individual heat pumps leads to increased investments in
wood chip CHP and open cycle gas turbines. This is caused by a need to supplement the
39
fluctuating wind power, in covering the electricity demand for the individual heat pumps. In this
regard, the wood chip and natural gas fired units also cover an increased need for peak and
reserve capacity, imposed by the heat pumps.
The required investments in dispatchable power capacities are reduced, when operating the heat
pumps flexibly, i.e. around 440 MW in the main scenario. The reason is that the flexible heat
pump operation allows for shaving down the peaks in the total load of the system (via the
implemented capacity balance restriction). This reduces the need for peak/reserve capacities. The
peak load shaving in the Danish energy system is illustrated in Figure 24. The week with the
highest peak loads of the year is illustrated, representing the week determining for the need for
peak/reserve capacities in the optimisation.
Non-flexible operation of indiv. heat pumps Flexible operation of indiv. heat pumps
7000 7000
Electricity consumption Electricity consumption
6000 for6000
individual heat pumps for individual heat pumps
(MW-e) (MW-e)*
5000 5000
MW-e
MW-e
4000 4000 consumption

Electricity Electricity consumption
for large heat pumps and for large heat pumps and
3000 3000 boilers (MW-e)
electric electric boilers (MW-e)
2000 2000
Conventional electricity Conventional electricity
1000 1000 (MW-e)
demand demand (MW-e)
0 0
T001
T011
T021
T031
T041
T051
T061
T071
T081
T091
T101
T111
T121
T131
T141
T151
T161
T001
T011
T021
T031
T041
T051
T061
T071
T081
T091
T101
T111
T121
T131
T141
T151
T161
Hour Hour
Figure 24. Peak load shaving achieved due to flexible operation of individual heat pumps. The week with the
highest peak loads of the year is illustrated (week no. 51). *Incl. supplemental electric boilers.
a) HP scenario b) HP-Flex scenario.
The figure clearly shows that peaks in total load are shaved down, when the individual heat
pumps are operated flexibly. This is facilitated by using the thermal storage capacity of the
buildings for shifting the load of the individual heat pumps a few hours within the day: from the
peak load hours in the morning and evening, to the load valleys in the afternoon, and to some
extent also to the night.
Figure 24 illustrates that if peak loads were to be shaved down further, it would require lowering
the individual heat pump load further, in longer consecutive periods lasting up to 15 hours (hour
T056 to T070). This is difficult, since ambient temperatures are very low in this week (see Figure
25). The individual heat pumps are therefore forced to operate at their capacity limit in longer
periods, and the electric boilers needed for covering peak loads. This leaves a limited capacity
margin available for load shifting, regardless of the heat storage available.
40
2000 4
Electricity
1800 2 consumption for indiv.
1600 0 electric boilers
Ambient temp. ( C)
1400 -2 (MW-e)
MW-e
1200 -4
1000 -6 Electricity
800 -8 consumption for indiv.
600 -10 heat pumps
(MW-e)
400 -12
200 -14
0 -16 Ambient temperature
( C)
T111
T001
T012
T023
T034
T045
T056
T067
T078
T089
T100
T122
T133
T144
T155
T166
Hour
Figure 25. Electricity consumption for individual heat pumps and supplemental electric boilers, shown for the
week with the highest peak loads of the year (week no. 51). The scenario without flexible heat pump operation
is illustrated (HP scenario).
The above conditions explain why the need for peak/reserve capacity is not reduced further, when
operating the heat pumps flexibly. Technically, further peak load shaving would be possible, to
some extent, if investing in larger heat storage capacities. However, this is not economically
feasible due to the challenges described above, and when accounting for the investment costs of
the thermal storages.
It is found that the achievable reduction in peak/reserve capacity is sensitive to e.g. the
dimensioning of the individual heat pumps and their efficiency. As such, with a heat pump
capacity share in the low end (72 %), the impact of the electric boiler in peak load hours becomes
more critical. This results in larger reduction in peak/reserve capacities (560 MW), when
operating the heat pumps flexibly. Correspondingly, if applying a heat pump capacity share in the
high end (82 %), the achieved reduction in peak/reserve capacity is lower (290 MW). A higher
COP of the individual heat pumps results in a lower impact on peak loads and thus a lower
reduction in peak/reserve capacities (320 MW). In all sensitivity analyses, the reduction in
peak/reserve capacities is within the approximate interval 300-600 MW, corresponding to the size
of a large power plant.
Investment in intelligent heat storage in the building structure generally brings a socio-economic
cost reduction of around 21-38 M€ per year, corresponding to 0.6-1.1 % of system costs. When
expressed in system benefit achieved per house investing in heat storage capability, a range of
approximately 60-200 €/house/yr has been identified. The low end of this interval (60
€/house/year) is found when assuming low control equipment costs, yielding investment in heat
storages in around 75 % of the houses, but resulting in lower benefit per house on average. The
high end of the interval (200 €/house/year) represents a scenario with high control equipment
costs, yielding investments in heat storages in only 20 % of the houses, but resulting in higher
benefit per house on average. The feasibility of investing in intelligent heat storage in the
building structure is thus sensitive to the control equipment costs.
The system benefit achieved with flexible heat pump operation mainly constitutes savings on
energy system investment costs (mainly for peak/reserve capacities). This is illustrated in Figure
26 for the main scenario.
41
Energy system inv.

and fixed O&M
Fuel
CO2 emission
Variable O&M
-200 -150 -100 -50 0 50 100

Inv. in intelligent
Change in socio-economic costs heat storage in
(€/house/yr) building structure
Figure 26. Change in annualised system costs due to flexible operation of individual heat pumps; given as
average costs per house investing in intelligent heat storage in the building structure.
Figure 26 illustrates that the potential for reducing peak/reserve capacity investments is important
for the feasibility of the thermal storages. This has been confirmed in various sensitivity
scenarios. For instance, if forcing through investments in heat storages in all houses with heat
pumps, instead of optimising the heat storage investments, practically the same reduction in
peak/reserve capacity is achieved, while the average system benefit per house drops substantially
(to 8 €/house/yr compared to 120 €/house/yr for the case with optimised heat storage
investments). This illustrates that the optimised heat storage investments are largely made to the
extent, that they can cost-effectively reduce the need for peak/reserve capacity in the system.
Overall, it has been found that when utilising the storage capacity of buildings, individual heat
pumps can contribute to peak load shaving and thereby reduce the need for peak/reserve capacity;
in Denmark with 300-600 MW, corresponding to the size of a large power plant. This is
identified as the main system of equipping the heat pumps with heat storages, and is found to be
crucial for the feasibility of the heat storage investments.
4.2.3 Comparison of heat storage options for individual heat pumps

In Paper III, two different heat storage options for individual heat pumps, namely heat
accumulation tanks and passive heat storage in the building structure, have been compared with
regard to their potentials in facilitating a fuel-efficient and cost-effectiveness integration of wind
power. In Figure 27, it is shown how system costs are affected by the different heat storages.
15
Investment
10
Costs (M€/yr)
Fuel
5
CO2
0
-5 Variable
O&M
-10
Figure 27. Changes in socio-economic system cost when enabling different heat storages for individual heat
pumps (for scenario descriptions, see Figure 19).
42
As shown, passive heat storage (1C/2C-PasLow/High) has lower investment costs than heat
accumulation tanks (Tank300/500/1000) and furthermore equivalent to higher benefits in terms of
reduced costs for fuel and CO2 (and variable O&M). This shows that passive heat storage is more
cost-effective, as also confirmed in various sensitivity analyses. Furthermore, it is found that
passive heat storage can bring equivalent to larger reductions in fuel consumption and excess
electricity consumption compared to the heat accumulation tanks.
The socio-economic feasibility of different heat storage options has been analysed further in
Paper V, optimising both investments and operation, and including electricity import/export.
Furthermore, the representation of intelligent heat storage in the building structure is improved,
and the possibility of utilising existing hot water tanks is also covered. As mentioned, this
analysis focusing on the Danish energy system by 2030 reveals socio-economically feasibility of
investing in intelligent heat storage in the building structure, in 20-75 % of the houses with heat
pumps. In line with the analyses performed in Paper III, heat accumulation tanks have generally
not been found competitive. This result has been confirmed by in various sensitivity analyses.
It is found that a reduction in peak/reserve capacity can also be obtained if merely utilising the
existing hot water tank for flexible heat pump operation. This only requires investment in a
central controller enabling communication with the power system. This option is identified as
socio-economically feasible in about 20-70 % of the houses.
Tests have shown that forcing through investments in intelligent heat storage in the building
structure in all houses, gives the same reduction in peak/reserve capacity, as forcing through
investments in heat accumulation tanks of 1000 litres in all houses. The two types of thermal
storages can thus provide equal flexibility in this regard. The reason why heat accumulation tanks
are not competitive is their higher investment costs.
However, heat accumulation tanks have the advantage of not requiring any indoor temperature
variations. In comparison, some indoor air temperature variation (here assumed 20-23 °C during
the day and 19-22 °C at night) will particularly for houses with radiator heating, i.e. the bulk of
one-family houses towards 2030, be required in order to utilise intelligent heat storage in the
building structure. If excluding the option of investing in intelligent heat storage in the building
structure, investments in heat accumulation tanks (1000 litres per house) are observed in around 4
% of the houses. However, when at the same time assuming intelligent use of the existing hot
water tanks for flexible heat pump operation, investment in heat accumulation tanks becomes
marginal, occurring in only 0.8 % of the houses12. This indicates that socio-economic feasibility
of heat accumulation tanks is generally doubtful.
12
In contrast, intelligent heat storage in the building structure has been found to be even more feasible when
assuming that the hot water tank is already utilised intelligently. This is due to the fact that investment in the
common control equipment has in this case already been made (see Section 2.2).
43
4.2.4 System effects of electric vehicles

In Paper II, it is analysed how a large-scale implementation of plug-in hybrid electric vehicles
(PHEVs) and battery electric vehicles (BEVs) towards 2030 would influence the energy systems
of Denmark, Finland, Germany, Norway, and Sweden. Increasing shares of electric vehicles are
assumed in all five countries; comprising 2.5 %, 15 %, 34 %, and 53 % of the private passenger
vehicle fleet in 2015, 2020, 2025, and 2030, respectively (see Figure 28). The resulting electricity
demand for electric vehicles in Denmark in 2025-2030 is 2.0-3.4 TWh-e (1.0-1.7 million electric
vehicles). This represents a very ambitious penetration of electric vehicles compared to
Energinet.dk's central projection of the electricity demand for electric vehicles in 2030 of 0.8
TWh [5].
100% 100%
New vehicle shares
80% 80%
Vehicle fleet
60% ICE
60%
BEV
40% 40%
PHEV
20% 20%
0% 0%
Figure 28. Assumed development in the fleet of private passenger vehicles distributed on internal combustion
Engine vehicles (ICEs), battery electric vehicles (BEVs), and plug-in hybrid electric vehicles (PHEVs).
The Balmorel model is applied and road transport has been modelled, using the add-on presented
by Juul and Meibom in [17]. However, further model development has been made, in order to
handle the gradual implementation of different vehicle vintages in the vehicle fleet towards 2030.
Intelligent charging/discharging is assumed in all electric vehicle scenarios. As in [17], plug-in
patterns are based on a national investigation of transport habits, and intelligent
charging/discharging is identified as part of the system optimisation. Socio-economic cost
minimisation is applied.
The results show that the electric vehicles can facilitate significantly increased wind power
investments in all of the countries analysed, particularly in the long term. As such, the study
indicates that wind power will likely provide a large share of the electricity for electric vehicles in
several of the countries (see Figure 29). However, a large part of the electricity generation
increase, induced by the electric vehicles, is in several cases coal-based, particularly in the short
term. This is thus is a likely outcome, if electric vehicles are not followed up by economic
support for renewable energy technologies (other than CO2 quotas13).
13
In this study, in 2015-2030 assumed to be 20-39 €/tonne in main scenario [121] and 15-30 €/tonne and 26-60
€/tonne, respectively, in sensitivity analyses (given in €2008).
44
DENMARK GERMANY NORWAY

8 60
WOOD 4
WOOD
50
6 WIND 3
40 WIND
4 2
TWh
TWh
TWh
30
STRAW WATER
2 20 1
NAT_GAS
10 STRAW
0 0
0
MUNI_WAST NUCLEAR
2015 2020 2025 2030 2015 2020 2025 2030
-2 -10
E 2015 2020 2025 2030 -1
FINLAND SWEDEN TOTAL
15 15 80 Coal
WOOD_WASTE
WOOD
10 60 Wind
10 WIND Lignite
WATER
TWh
5 40
TWh
TWh
5 STRAW Wood
PEAT
0
NUCLEAR
20 Natural gas
0 2015 2020
NAT_GAS 2025 2030 0 Straw
-5
LIGHTOIL
2015 2020 2025 2030 FUELOIL Electricity
2015 2020for2025
EDVs 2030
-5 -10
COAL -20
Coal Wind Wood Natural gas Straw Lignite Electricity for EVs
Figure 29. Changes in annual electricity generation due to implementation of electric vehicles. Generated power
increases in each year will not necessarily correspond to the electricity demand for electric vehicles. This is
mainly due to the influence of electricity import/export.
As illustrated, the effects of electric vehicles vary significantly from country to country.
Moreover, the effects of electric vehicles on the system are found to be sensitive to e.g. fuel and
CO2 price variations. It is therefore difficult to draw more general conclusions regarding how
electricity demand for electric vehicles would be met.
When charged/discharged intelligently, the electric vehicles can due to vehicle-to-grid

capabilities contribute in covering peak loads and thereby reduce the need for new dispatchable
power capacities in several of the countries, including Denmark. As such, the investment in
coal/natural gas fired power capacities is in Denmark, in some cases, significantly reduced with
the implementation of the intelligently charged/discharged electric vehicles (a reduction of 300-
2200 MW by 2020-2030, mainly open cycle gas turbines and secondarily coal CHP). However, in
some sensitivity scenarios, the introduction of electric vehicles results in net increases in the
investment in dispatchable power capacities in Denmark (600-1700 MW).
For the five countries as a whole, the electric vehicles bring significant CO2 reductions in the long
term, 1-6 % in 2025 and 3-28 % in 2030; considering all sensitivity scenarios. Total system costs
for the five countries are increased with around 0.8-3.9 %, depending on the year (see Figure 30).
It is assessed that the cost increases are caused by the modelled forced implementation of BEVs,
having relatively high investment costs; since PHEVs alone have previously been shown to
provide system cost reductions [17, 53] and BEVs alone in system cost increases [53].
45
200 600
180
CO2 (Mtonnes)
160 Bas
500
e Base
140 Hea
Bill. €/yr
400
120 t Heat
100 EV300
EV
80 EVH
60 200
eat EVHeat
40 100
20
0 0
2015 2020 2025 2030 2015 2020 2025 2030
Figure 30. a) Total annualised system costs for the simulated countries, for scenarios with/without electric
vehicles (EV) and with/without possibility to invest in large heat pumps, electric boiler, and thermal storages in
the district heating system (Heat). b) Total CO2 emissions for the simulated countries.
In relation to the cost effects, it can be noted that infrastructure costs, covering charging spots and
potential local grid enhancements, are difficult to quantify and are in [104] assessed to be
moderate compared to vehicle costs. Against this background, infrastructure costs are not
included.
Average CO2 reduction costs for the assumed implementation of electric vehicles are reduced
considerably towards 2030; from very high levels of 7100 €/tonne in 2015 and 1500 €/tonne in
2020, to 140 €/tonne in 2025 and 80 €/tonne in 2030. This is firstly due to increased use of wind
power to cover the electricity demand for electric vehicles over the period. This leads to higher
CO2 reductions, and hence reduced costs per ton CO 2 reduction achieved. Secondly, the technical
and economic improvements of electric vehicles assumed towards 2030 also play a role.
4.2.5 District heating versus individual heating

The point of departure for Paper VI is that district heating forms as an important boundary
condition for the market potentials of the different flexibility measures in the heating sector: large
heat pumps, electric boilers, and thermal storages in the district heating system, and small heat
pumps and heat storages for individual heating. The aim of this paper is to investigate whether
there is a socio-economic advantage of having a large scale district heating system in a long term
perspective, when not including existing district heating infrastructure as a premise, as is the case
in many countries. Total system costs are compared for a 100 % renewable energy system by
2050, for the following two cases:
 A case where district heating infrastructure is built and operated (i.e. including all
investment and operation costs).
 A case without district heating infrastructure, i.e. only individual heating.
The Danish energy system is applied as case study, and the Balmorel model is used to configure a
socio-economic optimal energy system, for each of the two cases. The results reveal only small
differences, 1.5-2.0 %, in total annualised system costs, between the case without district heating
and the case with district heating (lowest cost for the system without district heating). As such,
the results do not indicate a clear socio-economic advantage of building or not building a district
heating system in the applied long term perspective.
46
However, there is a massive net import of electricity in the optimisation, corresponding to about
30-40 % of annual national electricity demand. This is due to the 100 % renewable energy
condition implemented for Denmark, while no corresponding policies have been modelled for the
neighbouring countries. This questions how realistic the optimised system configuration is from a
self-sufficiency perspective. Furthermore, this dependency on net import of fossil fuel-based
electricity from other countries would make it difficult to justify the system politically as being
100 % based on renewable energy.
For Denmark, and other countries already having a large scale district heating grid, the most
relevant approach is to include the existing infrastructure as a premise, since this represents sunk
costs. Studies applying this approach [11, 54] have found that it is socio-economically feasible to
expand the Danish district heating grid further, and the Danish Energy Agency also projects a
moderate increase in the district heating share towards 2030 [105].
4.3 Summary and comparison of results
4.3.1 Overview of results

The most important findings, concerning the flexibility measures' ability to support the
integration of wind power, are summarised below:
Individual heat pumps and heat storages

 The installation of individual heat pumps can contribute significantly to facilitating larger
wind power investments and reducing system costs, fuel consumption, and CO2
emissions.
 Flexible operation of the heat pumps, enabled by heat storages, can contribute only
moderately in these respects.
 The main benefit of adding heat storages to individual heat pumps is that the required
investments in peak/reserve capacities can be reduced through peak load shaving (300-
600 MW in Denmark).
 Investment in intelligent heat storage in the building structure is found socio-
economically feasible in about 20-75 % of the houses.
 It is also found feasible to invest in control equipment, enabling the use of existing hot
water tanks for flexible heat pump operation, in about 20-70 % of the houses.
 Investments in heat accumulation tanks for space heating are not competitive.
Electric vehicles
 Electric vehicles will, when charged/discharged intelligently, in the long term facilitate
larger wind power investments, while they in the short term in many cases are likely to
result in increased coal-based electricity generation.
 Electric vehicles can significantly contribute to reducing CO2 emissions, while system
costs are generally increased due to the modelled forced implementation of the costly
BEVs.
 Average CO2 reduction costs for electric vehicles are reduced considerably towards 2030
as a result of improved conditions for wind power and improved technical and economic
properties of electric vehicles.
47
 The electric vehicles can through vehicle-to-grid capability reduce the need for
dispatchable power capacity.
District heating vs. individual heating
 No clear socio-economic benefit is identified of building or not building a district heating
system in a 100 % renewable energy system by 2050.
Table 6 summarises which flexibility measures have been found socio-economically feasible. The
optimal investments identified are also given.
Table 6. Overview of which flexibility measures have been found socio-economically feasible.
Sector Flexibility measure Socio- Optimal investments Papers including this

economic identified, technology
feasibility Denmark by 2030
District heating Large heat pumps + 700-2300 MW-th Paper II, III, IV, V, VI
Large electric boilers + 0-70 MW-th Paper II, III, IV, V, VI
Thermal storages + 7-15 GWh-th Paper II, III, IV, V, VI
Individual heating Individual heat pumps + 4700 MW-th Paper III, IV, V, VI
Intelligent passive heat storage in the + 20-75 % of houses b Paper III, IV, V
building structure via radiator heating
Intelligent active heat storage in concrete + 20-75 % of houses b Paper IV, V
floors via floor heating
Heat accumulation tanks ÷ Paper III, IV, V
Intelligent utilisation of existing hot water + 20-70 % of houses b Paper V
tanksa
Transport Battery electric vehiclesa ÷ Investments not optimised Paper II
Plug-in hybrid electric vehiclesa + Investments not optimised Paper II
a
Based on supplementing the analysis in Paper II with results of other studies [17, 53].
b
Percentages refer to shares of all modeled one-family houses, covering houses with radiator heating as well as
houses with floor heating.
As shown, socio-economic feasibility has been identified for individual heat pumps and some
heat storage options, PHEVs, and for large heat pumps, electric boilers, and thermal storages in
the district heating system.
4.3.2 Comparison of system benefits

It is found that individual heat pumps, through peak load shaving, and electric vehicles, through
vehicle-to-grid capability, can reduce the need for investing in dispatchable power capacity.
Moreover, both individual heat pumps and electric vehicles can facilitate increased wind power
investments and reductions in fuel consumption, and CO2 emissions. The system benefits of
individual heat pumps and electric vehicles cannot be compared further, based on the analyses
performed. The reason is that the analyses vary significantly in the approach and boundary
conditions applied (see Table 5).
In Paper II, it is shown that large heat pumps, electric boilers, and thermal storages in the district
heating system can also facilitate increased wind power investments in several countries,
including Denmark. It is found that in terms of increasing socio-economically optimal wind
power investments, and reducing CO2 emissions, these heat measures can together provide
48
benefits in the same order of magnitude as a large-scale implementation of electric vehicles.

However, while the heat measures reduce the total system costs of the five countries marginally
(0.004%-0.9 %); the electric vehicles increase the system costs (0.8-3.9 %), due to the high
investments costs of BEVs. The heat measures are in the following briefly compared to PHEVs
alone, based on other studies.
Results in [53] indicate that PHEVs can bring larger cost reductions than the heat measures in the
district heating system. On the other hand, large heat pumps in the district heating system can be
seen to facilitate significantly larger increases in wind power generation, compared to PHEVs, for
the Danish (and Finnish) energy system. This is in agreement with [2], which in analyses of the
Finnish energy system, finds that heat measures in the district heating system can facilitate
significantly larger increases in wind power investments. No clear indication can thus directly be
extracted regarding the cost-effectiveness of PHEVs versus the heat measures, in terms of
supporting wind power integration.
4.3.3 The importance of flexible operation for individual heat pumps and
electric vehicles
The model developed to represent individual heat pumps, renders it possible to analyse scenarios
with/without flexible operation. This allows for investigating the system effects of installing the
heat pumps and of ensuring flexible operation, respectively. A corresponding approach is not
possible with the transport add-on applied in the analyses of electric vehicles. As such, the
transport add-on can only reveal the differences between not having electric vehicles in the
system, and then having electric vehicles that are intelligently charged (intelligent discharging,
i.e. vehicle-to-grid capability, can however be turned on/off in the model). The situation
with/without flexible charging of the electric vehicles is assessed in the following.
As mentioned in Section 2.4, if electric vehicles are not charged intelligently they will typically
be charged over the few hours, around 17:00-21:00 o'clock, when people return from home [14].
This is the time of the day when electricity demand is typically highest. As a result, natural gas
and coal power plants will typically cover the electricity demand for the electric vehicles in this
situation [106]. In contrast, the energy systems analyses indicate that for the case of individual
heat pumps, wind power will by 2030 largely supply the electricity for the heat pumps, even
when they are not operated flexibly. As mentioned, this is possible since the operation of the
individual heat pumps is distributed over many hours of the day. This improves the possibilities
for utilising wind power in satisfying the electricity demand of the heat pumps (while using
dispatchable power plants as back up). Based on the above, it is assessed that in terms of
facilitating larger wind power investments, flexible operation is more important for electric
vehicles than for individual heat pumps.
Moreover, since dumb charging of electric vehicles would be concentrated in the peak load hours,
this would likely have a relatively high impact on the needs for investing in peak/reserve
capacity, and potentially also on enhancements in the distribution grids. In contrast, the electricity
demand of individual heat pumps will be more distributed over the day, in the situation without
flexible operation. This results in a lower impact on peak loads. As such, it is assessed that also in
49
terms of reducing investment costs in the system, flexible operation is more important for electric
vehicles than for individual heat pumps.
50
Discussion and Conclusion
5 Discussion and Conclusion

This chapter begins with answering the general research question of the thesis. Subsequently, the
most important results are compared with other studies and next, perspectives of the findings are
presented. It is then described how the goals of the PhD have been fulfilled, including
contributions to national decision making and model development. Finally, suggestions for
further research are given.
5.1 Answer to research question

The thesis yields a scientific contribution to answering the following research question:
 To which extent can heat pumps, heat storages, and electric vehicles contribute to
integrating wind power?
This question has been investigated in several energy systems analyses, mainly using the Danish
energy system towards 2030, with up to around 60 % wind power, as a case. It is found that
individual heat pumps can contribute significantly to the integration of wind power, even without
flexible operation. As such, the heat pumps can contribute significantly to incorporating larger
amounts of wind power into the system and to reducing total costs, fuel consumption, and CO2
emissions. Furthermore, the heat pumps can reduce the pressure on the limited biomass resources
by reducing the use of individual wood pellet boilers. The system benefits of the heat pumps are
first explained by their high energy efficiency and socio-economic competitiveness, compared to
the individual boilers they displace. Secondly, their load profile is distributed over many hours in
the day, and heating demand is typically high in the same periods as when wind power is high.
This creates good conditions for utilising wind power in covering the electricity demand for the
heat pumps. In the case analysed, the political phase out of coal in Denmark by 2030 and the
constrained biomass resource availability creates particularly good conditions for wind power.
When investing in heat storages complementing the heat pumps, the operation of the heat pumps
can be shifted a few hours within the day. This renders it possible to prioritise the heat pump
operation for hours with low electricity prices, and shave down the peaks in total electricity
demand of the system. However, the flexible operation of the heat pumps can provide only
moderate system benefits, in terms of facilitating larger wind power investments and reduce
system costs, fuel consumption, and CO2 emissions. The individual heat storages are thus less
influential on the integration of wind power. The use of an aggregated energy system model
means that the heat storage capacity in a given heating area is shared among the heat pumps
within the area. Even under these conditions, only moderate system benefits have been identified
for the heat storages.
The main system benefit of investing in heat storages for the heat pumps is that the need for
peak/reserve capacities can be reduced; by about 300-600 MW-e, corresponding to the size of a
large power plant, for the case of Denmark by 2030. This can be achieved with investments in
socio-economically feasible heat storages complementing the heat pumps. This effect is crucial
for the feasibility of the heat storages.
51
Investment in intelligent heat storage in the building structure is identified as socio-economically

feasible in 20-75 % of the modelled houses with heat pumps by 2030, depending on particularly
the investment cost of control equipment. This provides a socio-economic net benefit of around
60-200 € per year per house investing in this storage option. Hereof, savings on energy system
investments generally constitute the largest share. In addition, it is found socio-economically
feasible to invest in control equipment, facilitating use of existing hot water tanks for flexible
heat pump operation, in about 20-70 % of the houses. In contrast, investments in heat
accumulation tanks for space heating are not identified as competitive, due to their larger
investment costs.
Energy systems analyses of electric vehicles show that when charged/discharged intelligently,
they can in the longer term, 2025-2030, facilitate significantly increased wind power investments
and reduced CO2 emissions. Moreover, the electric vehicles can through vehicle-to-grid
capability reduce the need for peak/reserve capacities. However, the results indicate that only
PHEVs and not BEVs are socio-economically competitive towards 2030. Furthermore, electric
vehicles are likely to be supplied by coal-based power in the short term.
Large heat pumps, electric boilers, and thermal storages in the district heating system can also
contribute significantly to supporting wind power integration. As such, these heat measures can
together provide system benefits in the same order of magnitude as a large-scale implementation
of electric vehicles, in terms of facilitating larger wind power investments and reducing CO2
emissions. The connection between the large heat pumps/electric boilers and the large district
heating storages allows for storing electricity as heat during longer periods when needed. This is
an advantage compared to individual heat pumps and electric vehicles, which will mainly be able
to provide power balancing intra-day and intra-hour, due to smaller storage capacities.
In conclusion, individual heat pumps, flexibility measures in the district heating system, and
PHEVs, can provide significant contributions to integrating wind power cost-effectively towards
2030. Heat storages for individual heat pumps can contribute only moderately in this regard.
5.2 Comparison with other studies

The identified potential for reducing investments in dispatchable power capacities (300-600 MW)
is on level with the identified peak shaving potential in [49] for smart charging of electric
vehicles in the Finnish system (540 MW). It is however lower than the peak shaving potential
estimated by Energinet.dk, of 900 MW for flexible operation of individual heat pumps in Western
Denmark alone [43]. Part of the reason is that the interaction between heat pump and thermal
storages is in [43] not modelled but merely represented as a modified fixed demand profile,
assuming that operation can be distributed freely within the day. Furthermore, the result in [43]
represents a scenario where a biomass or gas boiler is assumed available to supplement the heat
pump, instead of an electric boiler. This makes it possible to turn off the heat pump in longer
periods. However, it will hardly be the typical situation that a household invests in a boiler in
addition to a heat pump.
Flexible operation of the heat pumps is found to bring a system cost reduction of around 0.6-1.1
%, which is on level with the system cost reduction of 0.45-0.95 % found in [41], for flexible
52
operation of a large scale installation of individual heat pumps in Germany. The identified system
benefit of approximately 60-200 €/house/year, is not comparable with the system benefits per
house estimated in [41] and [8, 14]. The reason is that in this thesis, investments in thermal
storages have been optimised, in contrast to the other studies, where all heat pumps are assumed
operated flexibly. Additionally, investment costs for control equipment have been included in this
thesis, in contrast to at least [41]14. If forcing through investments in intelligent heat storage in
building structures and in 1000 litres heat accumulation tanks in all houses with heat pump
installations, and excluding investment costs for the storages, an average system benefit of around
50 €/house/year is obtained. This is some-what higher than the system benefit of 25-40
€/house/year found in [41]. This fits well with the fact that only operation cost savings are
included in [41], while the model applied in this thesis also includes savings on investment costs.
The identified system benefit is however significantly lower than the benefit of 80-130
€/house/year estimated in [8, 14]. Part of the explanation for this is considered to be the identified
lower reduction in dispatchable power capacity, compared to the estimate in [43]. Furthermore,
[8, 14] include benefits in terms of reduced costs for delivering ancillary services, while the
model in this thesis only covers impacts on reserve capacities. Finally, [8, 14] include expected
cost reductions for enhancing distribution grids as well as an assumed socio-economic benefit of
a smart grid, in terms of reduced costs for alternative fulfilment of energy savings obligations [8].
Regarding electric vehicles, the assessment of coal-based power being likely to deliver electricity
for the electric vehicles in the short term is in line with findings of energy systems analyses in
[106] for an electric vehicle implementation in the Nordic power system towards 2020. The
assessment of wind power as being likely to supply the electricity for electric vehicles in the
longer term, 2025-2030, is in agreement with results of energy systems analyses of electric
vehicles [53], where it is found that electric vehicles drive on wind power in Denmark and
Norway by 2030.
5.3 Perspectives
The energy systems analyses performed show that individual heat pumps are highly socio-
economically competitive and can bring significant system benefits. However, whether the heat
pumps will in fact be installed in the households, depends on whether they are attractive from a
private economic perspective, i.e. considering existing taxes, tariffs, and subsidies. In [55], it is
concluded that wood pellet boilers are in fact more attractive than individual heat pumps from a
private economic perspective. This indicates that stronger economic incentives might be needed
to ensure a large-scale implementation of individual heat pumps.
As shown, investments in heat storages complementing the heat pumps can through peak load
shaving reduce the need for dispatchable power plants. This can be beneficial from a socio-
economic perspective, in a time where the competitiveness of Danish thermal power plants is
decreasing (among other things due to the increasing amounts of renewable energy in the Nordic
system, pressing the electricity prices downwards) [107].
14
It is not clear whether control equipment costs are included in the estimates in [8, 14].
53
However, whether investments in heat storages for individual heat pumps will in fact be realised,
again depends on whether they are attractive from a private economic perspective. According to
estimates by Energinet.dk, a household would today only be able to gain a benefit of around 40 €
per/year, before investment in control equipment; even when gaining access to the regulating
power market, in addition to the spot market. This is also expected to be valid in a longer time
perspective, with larger price variations on the spot market and regulating power market [62]. As
such, Energinet.dk does not assess investment in flexible operation of individual heat pumps to be
attractive from a private economic perspective, under current regulation [62]. This assessment is
further confirmed when taking into account that control equipment costs (in this thesis estimated
to be around 11-37 €/house/year15) would further reduce the net benefit for the household.
The analyses performed nevertheless show that when including savings on energy system
investments, the socio-economic benefit of enabling flexible heat pump operation is significantly
higher, i.e. 60-200 €/house/yr (even when accounting for control equipment costs). The largest
benefit of flexible heat pump operation, namely savings on investment costs in the system, will
however not be visible on the electricity bill paid by the heat pump owner. It is therefore
important to transfer some of this benefit to the consumers, if investments in flexible heat pump
operation are to become attractive from a private economic perspective.
Currently, the electricity price faced by a household is typically set based on statistics concerning
average electricity consumption patterns. As such, the households are not exposed to electricity
price variations and do not have any economic incentive for operating heat pumps flexibly [69].
Such economic incentives require that access is increased for consumers to hourly accounted
electricity consumption, possibly with the option of turning off electricity demand in peak load
hours [14]. In practice, flexible operation of the heat pumps would e.g. be handled by a balancing
responsible agent controlling a large group of heat pumps on an aggregated level. It is
furthermore important that the consumers gain access to the regulating power market, since the
price variations on this market are larger [62]. Moreover, a large share of the electricity price for
Danish households is today comprised by fixed tariffs and taxes, while only a moderate share,
around 20 % [108], follows the electricity price variations on the market [69, 109]. If making the
taxes/tariffs dynamic, e.g. following electricity price fluctuations, it would increase the incentive
for households to engage in flexible electricity demand options [108, 110].
Concerning electric vehicles, the results of this thesis and other studies [17, 53], indicate that
PHEVs will be socio-economically attractive towards 2030, while BEVs will have difficulties in
competing. In a private economic perspective, the current consumer price of BEVs compared to
conventional vehicles is moreover significantly higher than for a conventional vehicle [111, 112].
Furthermore, the question is whether consumers will accept the limitations of BEVs, in terms of
shorter driving range and longer charging time etc. compared to conventional vehicles. So far, the
sale of BEVs in Denmark has been much slower than expected [113], which illustrates the
difficulties of BEVs in competing on the market (only around 800 electric vehicles on the road in
Denmark as of 2012). The PHEVs do not have the same disadvantages as the BEVs, due to their
fuel flexibility, making them more competitive in this regard. As supplemental measures in
relieving the transport sectors' dependency on oil, the use of other fuels such as compressed
15
Applying a discount rate of 5 %, cf. Paper V, Table 7.
54
natural gas or biogas in combustion engines could also become relevant and is currently
envisaged by the Danish Government [114].
5.4 Fulfilment of goals

Two overall goals of the PhD project have been outlined in the introduction. In the following, it is
shortly evaluated how these goals have been fulfilled. The first goal is to:
1. Contribute to the national decision-making with regard to using heat pumps, heat
storages, and electric vehicles in supporting wind power integration
In the course of this PhD project, several energy systems analyses have been performed, focusing
on how and to which extent heat pumps, heat storages, and electric vehicles can support the
integration of wind power. The results form basis for a number of policy recommendations.
These are oulined below:
The results show that the system benefits of installing individual heat pumps are considerable,
while the benefits of enabling flexible operation are moderate. When prioritising efforts and
public funds for creating incentives within this field, first priority should therefore be given to
ensure a large-scale deployment of the heat pumps. The enabling of flexible operation should be
given second priority.
The main system benefit of operating individual heat pumps flexibly, namely the reduced
peak/reserve capacity requirement, will not be visible on the electricity bills paid by the heat
pump owners. It is therefore important to transfer some of this benefit to the heat pump owners, if
investments in flexible operation are to become private economically attractive.
If electric vehicles are not operated intelligently, their charging will typically be concentrated in
the hours, where conventional electricity demand peaks. In contrast, individual heat pumps will
have a more distributed load profile in the situation with non-flexible operation. In terms of
supporting wind power integration and avoiding significant impacts on the need for peak/reserve
capacity investments (and potentially also distribution grid enhancements), it is therefore more
important to ensure intelligent operation of electric vehicles than of individual heat pumps.
Expansion of renewable electricity generation technologies should be ensured along with

deployment of individual heat pumps and electric vehicles. Otherwise, the resulting increase in
electricity demand risks being met by coal based electricity generation in the short term.
The second goal of the PhD project is to:

2. Improve the modelling of some of these wind integration technologies enabling a
better representation and understanding of their potentials and limitations
The energy system models, Balmorel and EnergyPLAN, have both been further developed,
thereby improving the representation of individual heat pumps and the interaction with different
types of heat storages. This renders it possible to analyse the possibilities and system effects of
operating the heat pumps flexibly. The model development in Balmorel facilitates economic
optimisation of investments and operation, covering the whole energy system, including
55
individual heat pumps and various thermal storage options. This modelling functionality has not
been found in previous studies. Furthermore, intelligent heat storage in the building structure is
here represented by integrating a thermal building model into Balmorel, representing the hour by
hour temperature states of the relevant thermal masses, and the heat exchange between them.
Thereby, the thermal dynamics of buildings are best captured. The incorporation of the thermal
behaviour of buildings in an energy system model has only been identified in one other study
[41], which did not include energy system investments.
The model development in EnergyPLAN enables representing flexible operation of individual

heat pumps, using heat storage in the building structure and in heat accumulation tanks, applying
a technical optimisation. The technical optimisation facilitates analysing the potentials of
individual heat pumps and complementing heat storages, in terms of minimising excess electricity
production and fuel consumption of the system, and utilising as much wind power as possible.
This is a new functionality compared to existing models in the field. The technical optimisation
can be a valuable supplement to economic optimisations, since these cannot directly reveal the
technical potentials described above. The model development made in EnergyPLAN has the
advantage of being compatible with heat demand profiles often used in energy system models. It
is therefore easier to integrate into the typical structure of energy system models and requires less
input data and calibration.
Overall, the two model developments thus provide an important methodological contribution in
the representation of individual heat pumps and heat storages and supplement each other well.
Finally, both model developments have the advantage of facilitating analyses of scenarios
with/without flexible operation of the heat pumps.
The thermal building model developed has been used for analysing the potentials of individual
heat pumps in supporting wind power integration. However, the model is also qualified for other
applications, e.g. analysing possibilities for involving individual heat pumps in the integration of
other fluctuating renewable energy sources or in the optimisation of thermal/nuclear power
dominated systems (peak load shaving, optimised power plant portfolio and operation etc.).
In warmer climates, there would typically be a demand for cooling of buildings rather than space
heating. Such cooling demand would in many cases be covered by an air-condition unit or a heat
pump, i.e. also an electricity driven unit. The thermal building model could with only minor
modifications also be used for analysing the system benefits of operating such cooling units
flexibly, by utilising the thermal capacity of the building. Finally, the stand-alone version of the
thermal building model16 could be used for private-economic analyses on household level;
investigating e.g. required incentives, comparing heating installations, or identifying optimal heat
pump/electric boiler capacity shares.
5.5 Further research

Both energy system models used, Balmorel (linear version) and EnergyPLAN are linear and do
not include start-up costs, minimum load requirements, or part load efficiencies. Moreover, the
models assume perfect foresight and hence do not include power balancing responding to errors
16
The model version not integrated in Balmorel but using electricity price profiles as input.
56
in forecasts of wind power and load; only effects on required reserve capacities are modelled in
Balmorel. Start-up costs and minimum load requirements for thermal power plants can have a
significantly influence on the operation cost savings obtained, when operating individual heat
pumps flexibly, as illustrated in [41]. As a result, the savings on operation costs are
conservatively represented in the models applied. It would therefore be interesting to integrate the
thermal building model into a stochastic unit commitment model, such as Wilmar [115]. This
would yield improved estimates of the operation costs savings, including provision of ancillary
services. However, effects on energy system investments would not be captured in such a model,
and these effects have been identified as crucial. The ideal approach would therefore be a two
step approach, where 1) effects on energy system investments are first revealed in e.g. the
Balmorel model, and 2) where the optimised energy system configuration is then used as input in
e.g. the Wilmar model, for analysing the effects on system operation, including ancillary services.
Such an approach has previously been applied for the case of electric vehicles in [49].
The models applied in analysing the system effects of individual heat pumps (and electric
vehicles) do not cover potential needs for distribution grid enhancements. A significant expansion
with individual heat pumps might require enhancing the distribution grid in certain areas [14]. In
[14], it is roughly estimated that savings from distribution grid enhancements are of the same
magnitude as savings on the spot market and the regulating power market, respectively [69]. On
the other hand, overloading of the distribution grids could be amplified, if synchronising the heat
pumps to react to the same price signals [116]. It would therefore be interesting to include the
aspect of distribution grid enhancements in further modelling work. Inclusion of hourly COP
variations and the heat contribution from solar transmission also constitute relevant subjects for
further research.
The feasibility of individual heat pumps and electric vehicles are mutually influential as they both
represent an electricity demand. Furthermore, intelligent solutions for individual heat pumps will
in some aspects compete with those for electric vehicles. For instance, it is found that the need for
peak/reserve capacities can be reduced if either investing in heat storages for the heat pumps, or if
utilising vehicle-to-grid capability for the electric vehicles. Competition can also arise in terms of
placing operation in hours with low electricity prices. It would be interesting to analyse this
competition in further analyses.
In economic optimisations, it is generally assumed that all agents behave economically rational.
However, this assumption does not always hold; perhaps particularly not for individual
consumers, where other factors such as comfort, convenience, safety, ease of use, social identity,
and habits also play a role [117]. Therefore, it would be relevant to supplement the economic
optimisations with social behaviour studies, emphasising the influence of such matters. It would
be interesting to perform socio-economic valuation and internalisation of relevant factors, and
include them in the optimisation. In this regard, it would be relevant to include a valuation of the
thermal comfort, as influenced by the indoor temperature variations experienced, when utilising
intelligent heat storage in the building structure.
Nevertheless, the model development performed in this PhD project is considered a big step
forward, in providing the methodology needed for analysing individual heat pumps and heat
storages in an energy system context.
57
Bibliography
6 Bibliography
[1] International Energy Agency, "World energy outlook 2012," International Energy Agency,
Paris, France, 2012.
[2] J. Kiviluoma and P. Meibom, "Influence of wind power, plug-in electric vehicles, and heat
storages on power system investments," Energy, vol. 35, pp. 1244-1255, 3, 2010.
[3] L. W. M. Beurskens, M. Hekkenberg and P. Vethman, "Renewable energy projections as

published in the national renewable energy action plans of the European member states," ECN
and European Environment Agency, 2011, 2011.
[4] H. Holttinen, "Hourly wind power variations in the Nordic countries," Wind Energy, vol. 8,
pp. 173-195, 2005.
[5] Energinet.dk, "Energinet.dks analysis preconditions 2012-2035, July 2012 [in Danish]
[Energinet.dks analyseforudsætninger 2012-2035, juli 2012]," Energinet.dk, Fredericia, Denmark,
2012.
[6] B. V. Mathiesen and H. Lund, "Comparative analyses of seven technologies to facilitate the
integration of fluctuating renewable energy sources," IET Renewable Power Generation, vol. 3,
pp. 190-204, 2009.
[7] F. H. Ruhne, "Excess electricity resulted in negative electricity prices over Christmas
[Overskudsstrøm gav negative elpriser i Julen] [in Danish]," Ingeniøren, Valby, Denmark.
Available at: http://ing.dk/artikel/overskudsstrom-gav-negative-elpriser-i-julen-135324, January
4, 2013.
[8] Danish Energy Association and Energinet.dk, "Smart grid in Denmark [in Danish] [Smart grid
i danmark]," Danish Energy Association, Copenhagen; Energinet.dk, Fredericia. Denmark, 2010.
[9] Danish Energy Agency, "Energy statistics 2011 [in Danish] [Energistatistik 2011]," Danish
Energy Agency, Copenhagen, Denmark. Available at: http://www.ens.dk/, 2012.
[10] Danish Ministry of Climate, Energy and Buildings, "The Governments' energy and climate
political goals and the results of the energy agreement in 2020 [in Danish] [Regeringens energi-
og klimapolitiske mål - og resultaterne af energiaftalen i 2020]," Danish Ministry of Climate,
Energy and Buildings, Copenhagen, Denmark, 2012.
[11] H. Lund, B. Möller, B. V. Mathiesen and A. Dyrelund, "The role of district heating in future
renewable energy systems," Energy, vol. 35, pp. 1381-1390, 3, 2010.
[12] Danish Energy Agency, "Energy statistics 2008 [Energistatistik 2008] [in Danish]," Danish
Energy Agency, Copenhagen, Denmark. Available at: http://www.ens.dk. 2009.
[13] H. Lund, A. N. Andersen, P. A. Østergaard, B. V. Mathiesen and D. Connolly, "From

electricity smart grids to smart energy systems – A market operation based approach and
understanding," Energy, vol. 42, pp. 96-102, 6, 2012.
58
Bibliography
[14] Energinet.dk, "Efficient utilisation of wind power in Denmark [Effektiv anvendelse af

vindkraftbaseret el i danmark] [in Danish],"
. Energinet.dk, Fredericia, Denmark. Available at: www.energinet.dk. 2009.
[15] N. Juul, "Modelling and analysis of distributed energy systems with respect to sustainable
energy - focus on electric drive vehicles. PhD thesis," DTU Management Engineering, Technical
University of Denmark, Kgs. Lyngby, Denmark, 2011.
[16] Energinet.dk, "Energy in 2050 - development tracks [in Danish] [Energi i 2050 -
udviklingsspor]," Energinet.dk, Fredericia, Denmark. Available at: www.energinet.dk, 2010.
[17] N. Juul and P. Meibom, "Optimal configuration of an integrated power and transport
system," Energy, vol. 36, pp. 3523-3530, 5, 2011.
[18] J. B. Jensen, C. M. Hvenegaard and S. V. Pedersen, "The little blue about heat pumps [in
Danish] [Den lille blå om varmepumper]," Danish Energy Association, Copenhagen, Denmark.
Available at:
http://www.tekniq.dk/Medlemsraadgivning/Energi/Vedvarende%20energi/Varmepumper/Den-
lille-blaa-om-varmepumper.aspx, 2011.
[19] I. Hadjipaschalis, A. Poullikkas and V. Efthimiou, "Overview of current and future energy
storage technologies for electric power applications," Renewable and Sustainable Energy
Reviews, vol. 13, pp. 1513-1522, 9, 2009.
[20] C. K. Ekman and S. H. Jensen, "Prospects for large scale electricity storage in Denmark,"
Energy Conversion and Managment, vol. 51, pp. 1140-1147, 6, 2010.
[21] H. Lund and G. Salgi, "The role of compressed air energy storage (CAES) in future
sustainable energy systems," Energy Conversion and Management, vol. 50, pp. 1172-1179, 5,
2009.
[22] Risø DTU National Laboratory for Sustainable Energy, "Risø energy report 8: The
intelligent energy system infrastructure for the future," Risø DTU, Denmark, 2009.
[23] B. V. Mathiesen, "Fuel cells and electrolysers in future energy system. PhD thesis,"
Department of Development and Planning, Aalborg University, Denmark, 2008.
[24] G. Strbac, "Demand side management: Benefits and challenges," Energy Policy, vol. 36, pp.
4419-4426, 12, 2008.
[25] International Energy Agency, "IEA demand side management programme, task XVII:
Integration of demand side management, distributed generation, renewable energy sources and
energy storages," International Energy Agency. Available at: http://www.ieadsm.org, 2008.
[26] A. Keane, A. Tuohy, P. Meibom, E. Denny, D. Flynn, A. Mullane and M. O’Malley,

"Demand side resource operation on the Irish power system with high wind power penetration,"
Energy Policy, vol. 39, pp. 2925-2934, 5, 2011.
[27] G. Papagiannis, A. Dagoumas, N. Lettas and P. Dokopoulos, "Economic and environmental

impacts from the implementation of an intelligent demand side management system at the
European level," Energy Policy, vol. 36, pp. 163-180, 1, 2008.
59
Bibliography
[28] P. S. Moura and A. T. de Almeida, "The role of demand-side management in the grid
integration of wind power," Appl. Energy, vol. 87, pp. 2581-2588, 8, 2010.
[29] J. Hong, C. Johnstone, J. Torriti and M. Leach, "Discrete demand side control performance
under dynamic building simulation: A heat pump application," Renewable Energy, vol. 39, pp.
85-95, 3, 2012.
[30] F. Tahersima, J. Stoustrup, M. A. Soroush and H. Rasmussen, "Contribution of domestic

heating systems to smart grid control," in 6th Dubrovnik Conference on Sustainable Development
of Energy Water and Environment Systems, Dubrovnik, Croatia, 2011.
[31] R. Halvgaard, N. K. Kjølstad, H. Madsen and J. B. Jørgensen, "Economic model predictive

control for climate control of buildings in a smart energy system," in 50th IEEE Conference on
Decision and Control and European Control Conference (CDC-ECC), Orlando, USA, 2011.
[32] T. S. Pedersen, P. Andersen, K. M. Nielsen, H. L. Staermose and P. D. Pedersen, "Using

Heat Pump Energy Storages in the Power Grid," 2011 Ieee International Conference on Control
Applications (Cca), pp. 1106-1111, 2011.
[33] F. Oldewurtel, A. Ulbig, A. Parisio, G. Andersson and M. Morari, "Reducing peak electricity
demand in building climate control using real-time pricing and model predictive control," in
Decision and Control (CDC), 2010 49th IEEE Conference on, 2010, pp. 1927-1932.
[34] F. Tahersima, J. Stoustrup and H. Rasmussen, "Optimal power consumption in a central

heating system with geothermal heat pump," in IFAC World Congress, Milano, Italy, 2011.
[35] P. MacDougall, C. Warmer and K. Kok, "Raising the accomodation ceiling for wind power
by intelligent response of demand and distributed generation," in 10th International Workshop on
Large-Scale Integration of Wind Power into Power Systems as Well as on Transmission
Networks for Offshore Wind Power Plants, Aarhus, Denmark, 2011.
[36] D. Wang, S. Parkinson, W. Miao, H. Jia, C. Crawford and N. Djilali, "Online voltage
security assessment considering comfort-constrained demand response control of distributed heat
pump systems," Appl. Energy, vol. 96, pp. 104-114, 8, 2012.
[37] S. Parkinson, D. Wang, C. Crawford and N. Djilali, "Comfort-Constrained Distributed Heat

Pump Management," Energy Procedia, vol. 12, pp. 849-855, 2011.
[38] T. Williams, D. Wang, C. Crawford and N. Djilali, "Integrating renewable energy using a
smart distribution system: Potential of self-regulating demand response," Renewable Energy, vol.
52, pp. 46-56, 4, 2013.
[39] D. Wang, S. Parkinson, W. Miao, H. Jia, C. Crawford and N. Djilali, "Hierarchical market
integration of responsive loads as spinning reserve," Appl. Energy, vol. 104, pp. 229-238, 4,
2013.
[40] P. Nyeng, "System integration of distributed energy resources - ICT, ancillary services and
markets. PhD thesis. Technical university of Denmark, department of electrical engineering,
electric energy systems, Kgs. Lyngby, Denmark," 2010.
60
Bibliography
[41] G. Papaefthymiou, B. Hasche and C. Nabe, "Potential of Heat Pumps for Demand Side
Management and Wind Power Integration in the German Electricity Market," Sustainable
Energy, IEEE Transactions on, vol. 3, pp. 636-642, 2012.
[42] M. Klobasa, "Analysis of demand response and wind integration in Germany's electricity
market," Renewable Power Generation, IET, vol. 4, pp. 55-63, 2010.
[43] D. S. Callaway, "Tapping the energy storage potential in electric loads to deliver load
following and regulation, with application to wind energy," Energy Conversion and Management,
vol. 50, pp. 1389-1400, 5, 2009.
[44] J. E. Braun, "Load control using building thermal mass," J. Solar Energy Eng., vol. 125, pp.
292-301, 2003.
[45] R. W. McCarthy, C. Yang and J. Ogden, "Impacts of electric-drive vehicles on California’s

energy system," Tech. Rep. UCD-ITS-RP-08-24, 2008.
[46] R. Sioshansi and P. Denholm, "The Value of Plug-In Hybrid Electric Vehicles as Grid
Resources," Energy J., vol. 31, pp. 1-23, 2010.
[47] B. Soares M.C. Borba, A. Szklo and R. Schaeffer, "Plug-in hybrid electric vehicles as a way
to maximize the integration of variable renewable energy in power systems: The case of wind
generation in northeastern Brazil," Energy, vol. 37, pp. 469-481, 1, 2012.
[48] P. Denholm and W. Short, "An evaluation of utility system impacts and benefits of optimally
dispatched plug-in hybrid electric vehicles," NREL, Tech. Rep. NREL/TP-620-40293, 2006.
[49] J. Kiviluoma and P. Meibom, "Methodology for modelling plug-in electric vehicles in the
power system and cost estimates for a system with either smart or dumb electric vehicles,"
Energy, vol. 36, pp. 1758-1767, 3, 2011.
[50] H. Lund and W. Kempton, "Integration of renewable energy into the transport and electricity
sectors through V2G," Energy Policy, vol. 36, pp. 3578-3587, 2008.
[51] T. K. Kristoffersen, K. Capion and P. Meibom, "Optimal charging of electric drive vehicles
in a market environment," Appl. Energy, vol. 88, pp. 1940-1948, 5, 2011.
[52] E. Ibáñez, J. McCalley, D. Aliprantis, R. Brown, K. Gkritza, A. Somani and L. Wang,

"National energy and transportation systems: Interdependencies within a long term planning
model," in Energy 2030 Conference, 2008. ENERGY 2008. IEEE.
[53] N. Juul and P. Meibom, "Road transport and power system scenarios for Northern Europe in
2030," Appl. Energy, vol. 92, pp. 573-582, 4, 2012.
[54] M. Münster, P. E. Morthorst, H. V. Larsen, L. Bregnbæk, J. Werling, H. H. Lindboe and H.

Ravn, "The role of district heating in the future Danish energy system," Energy, vol. 48, pp. 47-
55, 12, 2012.
[55] E. Zvingilaite, "Health externalitites and heat savings in energy system modelling," PhD
thesis. DTU Management Engineering, Technical University of Denmark, Denmark, 2012.
61
Bibliography
[56] Energinet.dk, "Delivery of ancillary services in Denmark - supply conditions [in Danish]
[Systemydelser til levering i Danmark - udbudsbetingelser]," Energinet.dk, Fredericia, Denmark.
Available at: http://energinet.dk/DA/El/Saadan-driver-vi-elsystemet/Systemydelser-for-
el/Sider/Systemydelserforel.aspx, 2012.
[57] H. Holttinen, "The impact of large scale wind power production on the Nordic electricity
system," PhD thesis. Department of Engineering Physics and Mathematics, Helsinki University
of Technology, Finland, 2004.
[58] The Nordel Market Development Group, "Description of balance regulation in the Nordic
countries," Organisation for the Nordic Transmission System Operators. 2008.
[59] D. Connolly, H. Lund, B. V. Mathiesen and M. Leahy, "A review of computer tools for
analysing the integration of renewable energy into various energy systems," Appl. Energy, vol.
87, pp. 1059-1082, 4, 2010.
[60] H. Ravn, M. Hindsberger, M. Petersen, R. Schmidt, R. Bøg and P. E. Grohnheit, "Balmorel:

A model for analyses of the electricity and CHP markets in the Baltic Sea region," Available at:
http://www.balmorel.com. 2001.
[61] Danish Energy Agency and Energinet.dk, "Technology data for energy plants - generation of
electricity and district heating, energy storage and carrier generation and conversion," Danish
Energy Agency, Energinet.dk. Copenhagen and Fredericia, Denmark, 2012.
[62] Energinet.dk, "Potential and possibilities for flexible electricity demand with particular focus
on individual heat pumps [in Danish] [Potentiale og muligheder for fleksibelt elforbrug med
særligt fokus på individuelle varmepumper]," Energinet.dk, Fredericia, Denmark. Available at:
http://www.energinet.dk, 2011.
[63] COWI, Danish Technological Institute and Danish Gas Technology Centre, "Technology
data for energy plants - individual heating plants and energy transport," Danish Energy Agency
and Energinet.dk. Copenhagen and Fredericia, Denmark, 2012.
[64] International Energy Agency, "Heat pump performance," International Energy Agency,
Paris, France. Available at:
http://www.heatpumpcentre.org/en/aboutheatpumps/heatpumpperformance/Sidor/default.aspx,
January 25, 2012.
[65] M. J. Moran and H. N. Shapiro, "Fundamentals of Engineering Thermodynamics," John

Wiley & Sons Ltd., West Sussex, England, 2000.
[66] K. A. Jensen, "Dimensioning of heat pumps most important [Dimensionering af

varmepumper vigtigst] [in Danish]," TEKNIQ The Electricians' Organisation. Available at:
http://www.tekniq.dk/PresseOgNyheder/Nyheder/2010/Marts/Dimensionering-af-varmepumper-
er-vigtigst.aspx. 2010.
[67] Technical University of Denmark, Aalborg University, ECON, Energynaut, Ea

Energianalyse and Paul Fredrik Bach et al, "EcoGrid.dk - steps towards a Danish power system
with 50 % wind energy, phase 1, WP4: New measures for integration of large scale renewable
energy," Danish Technological Institute, Aarhus, Denmark. Available at: www.energinet.dk.
2007.
62
Bibliography
[68] Steen Kramer. Energinet.dk, "Personal communication," March 2011.
[69] Energinet.dk, "The interaction between wind power and heat pumps - background note [in
Danish] [Samspillet mellem vindkraft og varmepumper - baggrundsnotat]," Energinet.dk,
Fredericia, Denmark. Available at: http://www.energinet.dk. 2010.
[70] Jacob J. Andersen. Vølund Varmeteknik, "Personal communication," August 2010, January,
March 2012.
[71] Otto Hammer. Danfoss: section for heat pumps, "Personal communication," October 12 and
31, 2012.
[72] K. Duer, T. Dalsgaard, K. Ellehauge and P. Ladekarl, "Status for active heat storage in
buildings [in Danish] [status for aktiv varmelagring i bygninger]," DTU Civil Engineering,
Technical University of Denmark, Lyngby, Denmark. 2001.
[73] Helge Chrisensen. Danfoss: floor heating department., "Personal communication,"

November 6, 2012.
[74] Knud Mortensen. Saint Gobain Weber (producer of construction materials), "Personal
communication," November 19, 2012.
[75] Per Heiselberg. Aalborg University, "Personal communication," June, 2011 and November
8, 2012.
[76] Energinet.dk, "Measured data of heat pump installations in Danish households, from the
project control your heat pump," Energinet.dk, Fredericia, Denmark, Available at:
www.styrdinvarmepumpe.dk (accessed March 2, 2012), 2012.
[77] Danish Energy Association, "Heat pumps as part of the flexible electricity demand -
background note for the report: The little blue about heat pumps [in Danish] [Notat:
Varmepumper som en del af det fleksible elforbrug]," Danish Energy Association, 2010.
[78] Danfoss, "Danfoss living connect, product prices," Danfoss, Hasselager, Denmark. Available
at: http://www.billigvvs.dk/Varmesystemer-Radiatorer-Radiator-tilbehoer-Danfoss-Living-
Connect.html (accessed December 14, 2011). 2011.
[79] Lars Olsen. Danish Technological Institute, "Personal communication," June 30, September
8, 2010, April 2011, January 2012, and November 2012.
[80] S. Aggerholm and K. Grau, "SBi instruction 213, buildings' energy demands, PC programme
and calculation instruction [SBi-anvisning 213, Bygningers energibehov, pc-program og
beregningsvejledning] [in Danish]," Danish Building Research Institute, Hørsholm, Denmark.
Available at: www.sbi.dk, 2008.
[81] Rambøll Denmark A/S and Aalborg University, "Heat plan Denmark, appendix [Varmeplan
Danmark, bilagsrapport] [in Danish]," Danish District Heating Association, Kolding, Denmark,
2008.
[82] Energinet.dk, "Energy in 2050 - Wind track [in Danish] [Energi i 2050 - Vindspor],"
Energinet.dk, Fredericia, Denmark. Available at: http://energinet.dk, 2011.
63
Bibliography
[83] K. Hedegaard, B. V. Mathiesen, H. Lund and P. Heiselberg, "Wind power integration using
individual heat pumps - Analysis of different heat storage options," Energy, vol. 47, pp. 284-293,
2012.
[84] Danish Commission on Climate Change Policy, "Documentation for the report of the Danish
commission on climate change policy [in Danish] [Dokumentationsdelen til klimakommissionens
samlede rapport. Grøn energi - vejen mod et dansk energisystem uden fossile brændsler]," Danish
Commission on Climate Change Policy, Copenhagen, Denmark. Available at:
http://www.klimakommissionen.dk/daDK/OmKlimakommissionen/Klimakommissionensrapport/
Sider/Forside.aspx. 2010.
[85] Christian H. Christiansen. Danish Technological Institute, "Personal communication," March

27, 2012.
[86] Bolius Knowledge center for house owners, "Floor heating in older houses [in Danish]
[Gulvvarme i ældre huse]," Bolius knowledge center for house owners, Ballerup, Denmark.
Available at: http://www.bolius.dk/alt-om/energi/artikel/gulvvarme-i-aeldre-huse/. (accessed
November 7, 2012), November 5, 2006.
[87] Danish Energy Agency, "Technology data for energy plants," Danish Energy Agency,
Copenhagen, Denmark. Available at: http://www.ens.dk/Documents/Netboghandel%20-
%20publikationer/2010/Technology_data_for_energy_plants.pdf, 2010.
[88] P. E. Morthorst, "Heat pumps and low temperature district heating [in Danish]
[Varmepumper og lavtemperaturfjernvarme. rapportering fra to workshops]," Delrapportering fra
EFP-projekt Effektiv fjernvarme i fremtidens energisystem. Risø DTU National Laboratory for
Sustainable Energy, Roskilde, Denmark, Tech. Rep. Risø-R-1234(DA), 2008.
[89] Thomas E. Pedersen. COWI: Energy department, "Personal communication on experiences

from the 1 MW heat pump installed in Frederikshavn, Denmark," April 5, 2011.
[90] Energinet.dk, "System plan 2007," Energinet.dk, Fredericia, Denmark. Available at:
www.energinet.dk. 2007.
[91] John Tang. Danish District Heating Association, "Personal communication," March 6, 2013.
[92] A. Dyrelund, H. Lund, K. Fafner, B. V. Mathiesen, F. Ulbjerg and F. Hvelplund, "Heat plan
Denmark 2010 - main report [Varmeplan danmark 2010 - hovedrapport] [in Danish]," Danish
District Heating Association, Kolding, Denmark. Available at:
http://www.fjernvarmen.dk/Faneblade/ForskningFANE6/FogU/Projektrapporter.aspx, 2010.
[93] Vølund Varmeteknik, "Pricelist - recommended sale prices april 2010 [Prisliste - vejledende
udsalgspriser april 2010] [in Danish]
," Vølund Varmeteknik, Videbæk, Denmark. Available at:
http://www.volundvt.dk/Volund/Priseliste/ (accessed November 1, 2011), 2010.
[94] K. Hedegaard, H. Ravn, N. Juul and P. Meibom, "Effects of electric vehicles on power
systems in Northern Europe," Energy, vol. 48, pp. 356-368, 12, 2012.
[95] International Energy Agency, "Technology roadmap - electric and plug-in hybrid electric
vehicles," International Energy Agency, Paris, France. Available at:
http://www.iea.org/papers/2009/EV_PHEV_Roadmap.pdf. 2009.
64
Bibliography
[96] S. Wittrup, "Energinet.dk: Store excess wind power in a green gas system [in Danish]
[Energinet.dk: Gem overskydende vindmøllestrøm i et grønt gassystem]," Ingeniøren, Valby,
Denmark. Available at: http://ing.dk/artikel/energinetdk-gem-overskydende-vindmollestrom-i-et-
gront-gassystem-116015, January 29, 2011.
[97] H. Lund, "Renewable Energy Systems: The Choice and Modelling of 100% Renewable
Solutions," Doctoral Thesis. Academic Press, 2010.
[98] H. Lund, "EnergyPLAN - advanced energy systems analysis computer model, February
2010, documentation version 8.0," Aalborg University, Aalborg, Denmark. Available at:
http://energy.plan.aau.dk/manual.php, 2010.
[99] H. Lund and E. Münster, "Modelling of energy systems with a high percentage of CHP and
wind power," Renewable Energy, vol. 28, pp. 2179-2193, 11, 2003.
[100] M. Münster, "Energy system analysis of waste-to-energy technologies. PhD thesis."

Department of Development and Planning, Aalborg University, Aalborg, Denmark, 2009.
[101] Danish Energy Agency, "The Danish building regulation [in Danish]
[Bygningsreglementet]," Danish Energy Agency, Copenhagen, Denmark. Available at:
http://www.bygningsreglementet.dk/br10_02_id123/0/42, 2011.
[102] Svend V. Pedersen. Danish Technological Institute, "Personal communication," January 11,
2013.
[103] Danish Energy Agency, "Energy statistics 2010 [in Danish] [Energistatistik 2010]," Danish
Energy Agency, Copenhagen, Denmark. Available at: http://www.ens.dk, 2011.
[104] L. H. Nielsen and H. Ravn, "Electricity for road transport, flexible power systems and wind
power," Risø DTU National Laboratory for Sustainable Energy, Tech. Rep. Risø-R-1804 (EN),
2011.
[105] Danish Energy Agency, "Denmarks' energy projection 2012 [in Danish] [Danmarks
energifremskrivning 2012]," Danish Energy Agency, Copenhagen, Denmark, 2012.
[106] Ea Energy Analyses, "CO2 emissions from future passenger cars in the Nordic countries [in
Danish] [CO2-udledning fra fremtidens personbiler i norden]," Ea Energy Analyses,
Copenhagen, Denmark. Available at: http://www.ea-energianalyse.dk, 2011.
[107] S. Wittrup, "Danish coal power plants gasp for breath [in Danish] [Danske kulkraftværker
gisper efter vejret]," Ingeniøren, Valby, Denmark. Available at: http://ing.dk/artikel/135659-
danske-kraftvaerker-gisper-efter
vejret?utc_medium=mobile&utc_source=mobile&utf_campaign=mobile&force_non_mobile,
January 18, 2013.
[108] Danish Energy Agency, "Report on possibilities for use of price flexible electricity demand
in the Danish power system [in Danish] [Redegørelse om mulighederne for anvendelse af
prisfleksibelt elforbrug i det danske elsystem]," Danish Energy Agency, Copenhagen, Denmark.
65
Bibliography
[109] Danish Energy Agency, "Consumer prices on electricity [in Danish] [Forbrugerpriser på
el]," Danish Energy Agency, Copenhagen, Denmark. Available at: http://www.ens.dk/da-
DK/UndergrundOgForsyning/ElOgVarmeForsyning/Elforsyning/Liberaliseringafelmarkedet/elpri
ser/Sider/Forside.aspx (accessed February 27, 2013).
[110] Det Energipolitiske Udvalg, "Report on possibilities for and the effects of modified tariffs
on electricity particularly in terms of improving integration of renewable energy (dynamic
tariffs)," Tax Ministry of Denmark, Copenhagen, Denmark, 2010.
[111] B. K. Sovacool and R. F. Hirsh, "Beyond batteries: An examination of the benefits and
barriers to plug-in hybrid electric vehicles (PHEVs) and a vehicle-to-grid (V2G) transition,"
Energy Policy, vol. 37, pp. 1095-1103, 3, 2009.
[112] M. K. Hidrue, G. R. Parsons, W. Kempton and M. P. Gardner, "Willingness to pay for

electric vehicles and their attributes," Resource and Energy Economics, vol. 33, pp. 686-705, 9,
2011.
[113] N. Østergaard, "The failure of electric cars makes the Government concentrate on gas in
Danish cars [in Danish] [Elbilernes fiasko får regeringen til at satse på gas i danskernes biler],"
Ingeniøren, Valby, Denmark. Available at: http://ing.dk/artikel/133557-elbilernes-fiasko-faar-
regeringen-til-at-satse-paa-gas-i-danskernes-biler, October 26, 2012.
[114] N. Østergaard, "Biogas from manure is to addrress the climate balance of cars [in Danish]
[Biogas af gylle skal rette op på bilers klimaregnskab]," Ingeniøren, Valby, Denmark. Available
at: http://ing.dk/artikel/biogas-af-gylle-skal-rette-op-pa-bilers-klimaregnskab-133565, October
27, 2012.
[115] P. Meibom, R. Barth, B. Hasche, H. Brand, C. Weber and M. O'Malley, "Stochastic

Optimization Model to Study the Operational Impacts of High Wind Penetrations in Ireland,"
IEEE Trans. Power Syst., vol. 26, pp. 1367-1379, 2011.
[116] L. H. Rasmussen, C. Bang and M. Togeby, "Managing congestion in distribution grids -

market design consideration." Ea Energy Analyses, Copenhagen, Denmark, 2012.
[117] J. M. Entwistle, "Energy consumption in private homes - user driven innovation [in Danish]
[Brugerdreven innovation i hjemmet - brugerdreven innovation]. presentation at heat pump
seminar at energinet.dk, march 12, 2010." Tech. Rep. The Allexandra Institute, Aarhus,
Denmark, 2010.
[118] Danish Energy Association, "Guide to grid connection of electric boilers [in Danish]
[vejledning om nettilslutning af elkedler]," Danish Energy Association, Copenhagen, Denmark.
Available at:
http://www.danskenergi.dk/~/media/Bogbutik_gratis_materiale/Nettilslutning_af_elkedler_april_
2011.ashx, 2011.
[119] P. R. Wang, M. Scharling and K. P. Nielsen, "2001 - 2010 design reference year for
Denmark," Danish Meteorological Institute, Copenhagen, Denmark. Available at:
http://www.dmi.dk/dmi/index/viden/dmi-publikationer/tekniskerapporter.htm, 2012.
[120] L. Olsen, C. S. Poulsen and M. H. Vorre, "Description of potentials - individual heat pumps
[Potentialebeskrivelse - individuelle varmepumper] [in Danish]," Danish Technological Institute,
Taastrup, Denmark, 2010.
66
Bibliography
[121] Danish Energy Agency. "Preconditions for socio-economic analyses in the field of energy
[Forudsætninger for samfundsøkonomiske analyser på energiområdet] [in Danish]," Danish
Energy Agency, Copenhagen, Denmark. 2010.
67
Appendix
Paper I: Wind power impacts and electricity storage – A time scale perspective
Paper II: Effects of electric vehicles on power systems in Northern Europe
Paper III: Wind power integration using individual heat pumps – Analysis of
different heat storage options
Paper IV: Energy system investment model incorporating heat pumps with
thermal storage in buildings and buffer tanks
Paper V: Influence of individual heat pumps on wind power integration – Energy

system investments and operation
Paper VI: District heating versus individual heating in a 100 % renewable energy
system by 2050
Report chapter: Balmorel model results – EVs and power system investments.
Supplemental data and model illustrations

Paper I
Wind power impacts and electricity storage – A time scale

perspective
Karsten Hedegaard & Peter Meibom.
Published in Renewable Energy. vol. 36, issue 1, pp. 318-324. January, 2012.
Renewable Energy 37 (2012) 318e324
Contents lists available at ScienceDirect
Renewable Energy
journal homepage: www.elsevier.com/locate/renene
Wind power impacts and electricity storage e A time scale perspective

K. Hedegaard a, *, P. Meibom b
a
Systems Analysis Division, Risø DTU National Laboratory for Sustainable Energy, Technical University of Denmark, Frederiksborgvej 399, Build. 130, P.O. 49, 4000 Roskilde, Denmark
b
Intelligent Energy Systems Programme, Risø DTU National Laboratory for Sustainable Energy, Technical University of Denmark, Frederiksborgvej 399, Build. 776,
P.O. 49, 4000 Roskilde, Denmark
a r t i c l e i n f o a b s t r a c t
Article history: Integrating large amounts of wind power in energy systems poses balancing challenges due to the
Received 29 October 2010 variable and only partly predictable nature of wind. The challenges cover different time scales from intra-
Accepted 10 June 2011 hour, intra-day/day-ahead to several days and seasonal level. Along with flexible electricity demand
Available online 20 July 2011
options, various electricity storage technologies are being discussed as candidates for contributing to
large-scale wind power integration and these also differ in terms of the time scales at which they can
Keywords:
operate. In this paper, using the case of Western Denmark in 2025 with an expected 57% wind power
Wind power
penetration, wind power impacts on different time scales are analysed. Results show consecutive
Net load
Power system balancing
negative and high net load period lengths indicating a significant potential for flexibility measures
Electricity storage capable of charging/activating demand and discharging/inactivating demand in periods of 1 h to one day.
The analysis suggests a lower but also significant potential for flexibility measures charging/activating
demand in periods of several days. In addition, the results indicate a physical potential for seasonal
electricity storage. In the study, a number of large-scale electricity storage technologies e batteries, flow
batteries, compressed air energy storage, electrolysis combined with fuel cells, and electric vehicles e are
moreover categorised with respect to the time scales at which they are suited to support wind power
integration. While all of these technologies are assessed suitable for intra-hour and intra-day/day-ahead
power balancing only some are found suited for responding to several days with high/low net loads and
even fewer for seasonal balancing.
Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction term goal of the Danish Government of phasing out the use of fossil
fuels an efficient utilisation of the wind power is important.
Several countries expect large wind power penetrations in the In addition to enhancing and expanding the existing power grid
future and this will pose system balancing challenges due to the and interconnections with neighbouring countries and to activate
variable and only partly predictable nature of wind. The Danish flexible electricity demand options such as heat pumps or electric
energy system forms an interesting case for analysing these chal- boilers, a broad range of electricity storage technologies can
lenges. This is firstly due to a high amount of wind power, corre- become relevant, e.g. batteries, flow batteries, compressed air
sponding to almost 20% of annual electricity consumption. energy storage (CAES), electrolysis combined with fuel cells, and
Secondly, around 50% of the power production is based on electric vehicles used as a distributed energy storage.
combined heat and power (CHP), resulting in a large amount of The challenges of balancing power production and demand
heat bound power production [1]. As a result, in periods with high cover different time scales from e.g. intra-hour, intra-day to
wind power production, high heat demand and low electricity seasonal level. Due to differences in properties such as response
demand, forced electricity export occurs. Furthermore, in order to time, storage efficiency, power related costs and storage related
ensure stable system operation and meeting electricity demand costs, storage technologies differ with regard to the time scales at
over the year, there is a significant need for regulating power and which they are suited to support wind power integration. The
ancillary services. The target of 50% wind power in Denmark in above calls for a time scale perspective on the system impacts of
2025 points to significantly increasing challenges of wind power wind power and the potentials of different electricity storage
integration in the near future [2]. Moreover, considering the long technologies.
A number of studies, e.g. [3e5] analyse wind variations and their
* Corresponding author. Tel.: þ45 26 81 82 98. impacts on the Nordic electricity system and Jónsson et al. [6]
E-mail address: [email protected] (K. Hedegaard). analyse how day-ahead wind forecasts affect electricity spot
0960-1481/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2011.06.034
K. Hedegaard, P. Meibom / Renewable Energy 37 (2012) 318e324 319
prices at the day-ahead spot market. Another study [7] evaluates 4000
Net load
the current occurrence of low and high wind power shares (as pct.
of electricity demand) in Western Denmark on an overall level as 3000
well as the impact on electricity prices. In this paper, using the case 2000 Centralised thermal
of Western Denmark in 2025 with an expected 57% wind power power cap.
penetration, wind power impacts on different time scales are 1000 (dispatchable*)
MW
analysed. The power system of Western Denmark is in focus as this Decentralised
will continue to comprise the majority of the Danish wind power 0 thermal power cap.
production, 75% in 2025 (the remaining in Eastern Denmark -1000
forming the second part of the power system). Import cap.
The system impacts of wind power are analysed based on hourly -2000
data on net load, defined as gross load (electricity demand) minus
wind power1 (see Section 2). Net loads provide a better indication -3000
Export cap.
of wind power impacts than do wind power shares. The reason is -4000
that net loads capture variations in wind power as well as in Hours 26304
electricity demand expressing what have to be covered by other
Fig. 1. Net load duration curve for 2006e2008 vs. thermal power capacities and
units in the power system. Overall, there are two possible chal-
import/export capacities in 2009 for Western Denmark (Net load [16], Capacities [17]).
lenging operational situations, namely high net load and low net
load, and these are both treated in this study. By analysing the
length of high and low net load periods, the expected demand for variations based on wind speed measurements at different loca-
power system balancing at different time scales is investigated. This tions. In total, 2800 MW wind power on-shore and 1760 MW off-
approach has not been found in previous studies and brings shore is assumed in 2025 covering 57% of the electricity demand
interesting perspectives to discussing wind power integration. on annual basis. The estimates are made to represent a normal
Regarding electricity storage technologies, a group of studies, wind year.
e.g. [8e10] focus on technical and economic properties of different An interconnection between the power system of Western and
technologies while other studies, e.g. [11e14] analyse different Eastern Denmark established in 2010 (600 MW) reduces the need
wind power integration measures based on energy system anal- for minimum power production at thermal plants in Western
yses. In this study, based on existing knowledge on technology Denmark. Moreover, in a future energy system with wind power
characteristics, a number of electricity storage technologies are penetration as high as 50% or more, wind turbines may be able to
categorised with respect to the time scales at which they are suited contribute to grid stabilising system services [18]. Overall, some-
to support wind power integration (see Section 3). This intends to what less, around 300 MW of minimum centralised power
provide a qualitative foundation for dealing with electricity storage capacity for grid stabilisation is therefore assumed in Western
in the context of wind power integration. Denmark in 2025. As shown in Fig. 2, the number of hours with
surplus wind power, i.e. negative net load, can be expected to be
considerable in 2025; corresponding to around 22% of the year.
2. System impacts of wind power
In addition, the dispatchable centralised thermal power capacity
will be lower; around 2000 MW in 2025 compared to around
2.1. The system today and in 2025
3000 MW currently. As a result, compared to the present situation
less abundant thermal capacity will be available to cover electricity
A certain minimum power production is needed to maintain
demand in high net load periods. Part of the high/negative net load
voltage and frequency stability of the grid, today normally supplied
periods in Western Denmark can be balanced through the inter-
from the large power plants. In Western Denmark, this minimum
connection with Eastern Denmark. When also including intercon-
power production is currently considered to be in the neighbour-
nections with neighboring countries, sufficient import and export
hood of 400 MW [15]. This constraints the system’s ability to
capacity, 3300e6000 MW and 3800e6000 MW, respectively, is
respond to wind power variations and is therefore taken into
expected to be available in 2025 for handling high and negative net
consideration. Thus, the net loads used in this study are defined as
load periods [19]. However, considerable wind power expansions
gross load minus wind power and minus the minimum power
production provided by the centralised thermal power plants.
In 2006e2008, in Western Denmark, wind power and minimum 4000
Centralised thermal
power production together exceeded electricity demand, resulting power cap.
3000 (dispatchable*)
in negative net load; in relatively few hours corresponding to
around 2% of the year on average and the negative net loads were 2000 Decentralised
well within the export capacity of 3840 MW (see Fig. 1). Further- thermal power cap.
more, as shown there is currently plenty of dispatchable thermal 1000
power capacity and import capacity to backup periods with high
MW
0 Net load
net loads, i.e. low wind power and high electricity demand.
Based on data from the Danish TSO regarding expected wind -1000
power production, electricity demand and thermal power capac-
Export cap. (min)
ities, a corresponding figure has been set up for 2025 (see Fig. 2). -2000
The estimated wind power production for Western Denmark in
2025 is based on expected on-shore and off-shore wind power -3000
Import cap. (min)
capacities at different locations and wind power production
-4000
Hours 8760
1
Danish power generation from other variable renewable energy sources such as Fig. 2. Estimated net load duration curve and expected thermal power capacities and
photo voltaic and wave power is negligible. minimum import/export capacities in 2025 for Western Denmark (Capacities [19]).
320 K. Hedegaard, P. Meibom / Renewable Energy 37 (2012) 318e324
are expected in the neighbouring areas, and wind power variations 25% 2.5-3
in these areas typically show patterns similar to the wind variations days
in Denmark. As a consequence, future electricity prices in neigh- 20% 2-2.5
bouring areas can often be expected to be low when domestic days
Pct. of the year

electricity prices are low leading to low value of exported wind 1.5-2
15%
power [20]. Correspondingly, electricity import in high net load days
periods could be costly. The higher the ability to obtain balance in 1-1.5
the energy system without having to rely on electricity import/ 10% days
export, the better possibilities will be for using external electricity 0.5-1
trade only when it is profitable. Against this background, electricity 5% days
storage and flexible electricity demand options can therefore 0-0.5
become relevant. 0% days
Setting a general threshold level for what should be considered Net load Net load Net load Net load Net load
high net loads in Western Denmark in 2025 is very difficult as the below below below below below
power production activated to cover the net load is based on an 0 MW -500 MW -1000 MW -1500 MW -2000 MW
economic market optimisation and as the situation will vary from
Fig. 3. Consecutive negative net load periods expected for Western Denmark in 2025.
hour to hour. Acknowledging the difficulties in setting such
a threshold limit, high net loads above 2000 MW, 2500 MW and
3000 MW, respectively, are investigated. Correspondingly, different Several days’ impacts: Impacts resulting in several days with
degrees of negative net loads are analysed, i.e. net loads below negative or high net loads
0 MW, 500 MW, 1000 MW, 1500 MW and 2000 MW, Seasonal impacts: Impacts due to seasonal wind variations
respectively. creating net load variations across months
2.2. Categorisation of system impacts Data for Danish wind power production and electricity demand
on second or minute level have not been available and hence, intra-
Apart from the financial market, the Danish power market hour impacts cannot be evaluated. However, wind power variations
includes four market places: Elspot, Elbas, the regulating power on second to a few minutes basis are smoothened by different gusts
market and the reserve market. Elspot is an integrated part of the for the individual turbines, inertia of the large rotors as wells as the
Nordic day-ahead spot market where power production based on variable speed turbines absorbing the variations and there is no
market bids and forecasted electricity demand and wind power is correlation between the variations of geographically dispersed
planned for every hour of the next day (24 h), 12e36 h before the wind farms. As a result, wind power variations within seconds or
actual operation hour. Elbas is a continuous market operating after few minutes, in the order of the activation notice of primary
the end of the spot market where market actors expecting devia- reserves, have very small effects on system operation even at
tions between realised production and day-ahead production plans considerable penetration [3,4]. For regulating power and secondary
have the possibility of trading with an official price towards balance reserves with an activation time of maximum 15 min, the impact of
closer to the operating hour. On the regulating market, deviations wind power forecast errors is significant but nevertheless lower
between operation as planned after Elspot and Elbas and actual than the impact of hourly wind power variations [23].
operation are balanced intra-hour by the Transmission System The remaining wind power impacts are in the following ana-
Operator (TSO) using power installations with a response time of lysed using projected hourly net load variations for Western
15 min [21]. The most important causes for imbalances are Denmark in 2025. The length of negative and high net load periods2
consumption and wind forecast errors and operation problems at is investigated based on two different approaches:
plants and in transmission lines, including outages [5]. Finally, at
the reserve market, the TSO buys system services, including 1) Length of consecutive negative/high net periods and
primary and secondary reserves, which in case of small imbalances 2) Rolling averages indicating periods with negative/high net
of supply that have not been balanced by the regulating market re- loads on average.
establish the balance and stabilise the frequency. Primary reserves
have an activation time of a few to 30 s and deliver power of It can be noted that the net load approach captures the vari-
maximum 15 min, while secondary reserves have a response time ability and not the imperfect predictability of wind power and load.
of 15 min and rarely deliver power for longer than 15 min [22]. On
an aggregated regional level, i.e. for Western Denmark in the
present case, the system impacts of wind power can be categorised 2.3. Negative and high net load periods
into the following types:
The results show that consecutive negative net load periods in
Intra-hour impacts: Impacts due to the imperfect predictability 2025 will have a length of maximum three days (see Fig. 3). Hereof,
of wind power requiring intra-hour balancing at the regulating the majority will cover periods with a length of up to one day, i.e.
market and at the reserve market. A subdivision can be made 61%, 71%, 91%, 100% and 100%, of the periods with net loads below
into impacts requiring activation of primary reserves with an 0 MW, 500 MW, 1000 MW, 1500 MW and 2000 MW,
activation time of a few to 30 s and impacts requiring regu- respectively. Periods of one to two days will comprise a lower but
lating power and activation of secondary reserves within also significant part; 32%, 25% and 9%, of the periods with net loads
15 min. below 0 MW, 500 MW and 1000 MW, respectively. As illus-
Intra-day/day-ahead impacts: Impacts due to the hourly and trated in Fig. 4, consecutive high net load periods will all have
daily variability and imperfect predictability of wind power a length below one day.
creating negative or high net load periods of 1 to 24 h affecting
unit commitment and economic dispatch of units at Elspot
2
day-ahead and at Elbas intra-day. After subtracting of 300 MW minimum power production on large plants.
25% 24%
22%
Net load above
20% 0.5-1 20% 2000 MW
days 18%
Pct. of the year
16%
Frequency
15%
14%
12% Net load above
10% 2500 MW
10%
8%
5% 0-0.5 6%
days
4% Net load above
2% 3000 MW
0%
Net load Net load Net load 0%
above above above 1 2 3 4 5 6 7 8 9
2000 MW 2500 MW 3000 MW
Days of rolling average
Fig. 4. Consecutive high net load periods expected for Western Denmark in 2025.
Fig. 6. Periods in Western Denmark 2025 with average high net loads indicated by
rolling averages.
The use of rolling averages indicates that periods where net
loads are on average negative will have a length of up to several
days (see Fig. 5). As such, e.g. periods with net loads below 0 MW on Electricity demand in the region also expresses variations across
average, when seen over 2e7 days, have a frequency of around months; presumably due to the influence of electricity based
9e20%. However, the frequency of periods with more critical heating and seasonal holidays (see Fig. 7). As shown, the projected
degrees of excess electricity, e.g. net loads below 500 MW on net load variations for 2025 also show significant seasonal
average, is more modest. Periods with high net loads on average variations.
when seen over 2e3 days have a moderate frequency (5e9%) while Fig. 8 illustrates that the amount of surplus wind power (net
average high net load periods lasting 4 days or more are rare loads below 0 MW) exceeds the amount of high net loads (above
(frequency below 2%) (see Fig. 6). 2000 MW) in some months, while in other months; the amount of
The estimated length of consecutive negative and high net load high net loads exceeds surplus wind power. This indicates a phys-
periods in Western Denmark 2025 indicate a significant physical ical potential for electricity storage from month to month.
potential for electricity storage or flexible electricity demand being It should be stressed that the analysis in this study only reveals
able to charge/activate demand and discharge/inactivate demand physical balancing potentials while economic potentials of elec-
in periods from 1 h up to one day, i.e. responding to intra-day/day- tricity storage and other flexibility measures will be determined by
ahead impacts of wind power. The occurrence of consecutive electricity price variations and market conditions.
negative net load periods with lengths of one to three days and of
average negative net load periods with lengths of several days
indicate a lower but also significant potential for flexibility 3. Electricity storage technologies in a time scale perspective
measures capable of charging/activating demand in periods of
several days. The following subsections cover different technologies that
could be relevant for large-scale electricity storage putting
emphasis on the time scales, at which the technologies are suited
2.4. Seasonal variations for power system balancing. Electricity storage technologies such as
flywheels, superconducting magnets and super capacitors are not
Wind power production in Western Denmark expresses signif- included, as these are mainly suited for special applications and are
icant seasonal variations with typically highest wind power not likely to play a substantial role in the future energy system [24].
production in the cold months of the year (October to February) Due to the dependence on revenue potentials at the market and
and lowest production in the warmer months (April to August). energy system in question, economic feasibility of a storage
24% 100
Net load below
22% 0 MW Wind
20% 80 power
18% Net load below
16% -500 MW
GWh/day
Frequency
14% 60
12% Net load below Gross
-1000 MW load
10% 40
8% Net load below
6% -1500 MW
20
4% Net
2% Net load below load
0% -2000 MW 0
1 3 5 7 9 11 13 15 17 19 21 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Days of rolling average Month
Fig. 5. Periods in Western Denmark 2025 with average negative net load indicated by Fig. 7. Average wind power production, gross load and net load per day in each month
rolling averages. estimated for Western Denmark in 2025.
400 durations. In addition, depending on the battery type, storage los-

ses can be high (see Table 1).
350 High net
loads
300 (above 3.2. Flow batteries
2000 MW)
250
GW h
Flow batteries are an emerging technology but only few MW

200 systems have been installed so far [25]. Flow batteries do not have
150 Negative any self-discharge, i.e. loss of energy over time of storage, as the
net loads electrolytes cannot react when they are stored separately. This
100 (below 0 makes them more suitable than conventional batteries for appli-
MW) cations that require long duration storages [26]. The size of the
50
electrolyte reservoirs used in the flow batteries determines the
0 storage capacity while the power capacity depends on the rates of
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec the electrode reactions occurring at the anode and cathode. In
Month effect, the power capacity can be designed independently from the
energy storage volume, which is an advantage compared to
Fig. 8. Monthly distribution of expected negative net loads and high net loads (above
2500 MW) in Western Denmark in 2025.
conventional batteries. This should reduce costs and could allow
flow batteries to provide power for many hours or even days [24].
For economic reasons, the storage volume is however currently
technology in different time scales should ideally be evaluated case typically limited to maximum one day [27] and storage related
by case. However, based on technology properties such as response costs for flow batteries are also relatively high (see Table 1), indi-
time, storage losses, power related costs ($/kWe) and storage cating a limited suitability for storing large amounts of energy. Very
related costs ($/kWh), electricity storage technologies can be cat- fast response times make flow batteries suitable for providing
egorised according to the time scales they will typically be suited primary reserves [24].
for. The term power related costs refers to costs related to the power
production capacity, e.g. the water turbine for pumped hydro, while 3.3. Electric vehicles
storage related costs cover cost for the storage facilities, e.g. costs of
pumping and storing water in reservoirs for pumped hydro. Storage With implementation of intelligent charging, electric vehicles
losses are defined as losses from the storage depending on the can form a flexible electricity demand supporting integration of
duration while loading/unloading losses refer to losses from loading wind power. If in addition, intelligent vehicle-to-grid power capa-
and unloading in total. bility is implemented, the batteries can form large-scale distributed
electricity storage controlled by the needs of the electric system [1].
3.1. Batteries Thus, charging when electricity prices are low and discharging
when electricity prices are high can be prioritised. In addition, since
Many different battery types exist whereof the leadeacid and electric vehicles can deliver high power within seconds for short
the sodiumesulphur batteries are the most common for large-scale duration of time they are ideally suited for delivering primary
installations. Batteries can be regarded as a mature technology with reserves [28].
many commercial MW-scale installations worldwide [25]. Batteries However, power flow between vehicle and grid can only take
can react instantly to system disturbances and can thus deliver place when the vehicle is grid-connected, i.e. when not driving. A
primary reserve capability. Currently, battery applications covering typical weekday driving pattern can be assumed to express
durations of less than one second to around 5 h seem to be a diurnal cycle of driving in the morning when going for work and
commercially feasible [26]. Batteries have low power related costs in late afternoon when returning from work. Depending on the
and high storage related costs compared to many of the other diffusion of charge spots at employer lots and e.g. mass transit
electricity storage technologies, which supports the assessment of stations, part of the electric vehicles could be grid-connected in
batteries being mainly suitable for electricity storage of shorter periods of the day when not driving. However, the system flexibility
Table 1
Characteristics for a number of large-scale electricity storage technologies
Power related Storage related Response Loading/unloading Storage loss Ref. Ref. Ref.
cost ($/kWe) cost ($/kWh) time loss cost response loss
Batteries 270e530 330e660 <1 s 30e40% 0e20%/monthc [25] [26] [25]
Flow batteries 1100e4500a 110e320a <1 s 30e40% 0% [25] [26] [25]
Electric vehicles 1750 n/a Few sec. 20e30% 1e20%/monthf [11] [26] [29]
CAES 400e500 1-40 9e12 min 30%d Very low [25] [10] [9,30]
Pumped hydro 600e1000 1e20 Few sec. 15e30% in total [25] [10,24] [9,10,24]
Underground pumped hydro 1100e1300 33e60 Few sec. 15e30% in total [25]
Electrolysis, SOFC, hydrogen Electrolysis: <1 Few 55e65%e <0.1%/year [31e33] [27,32] [32, 38]
storage in caverns 290e2000 secehours
SOFC: 1170b
a
Low numbers are future prices assuming large-scale production.
b
Medium term goal for costs.
c
0 % for sodium-sulphur batteries, 2e5%/month for lead-acid batteries and 5e20% for nickelecadmium batteries [26].
d
In addition to the power consumption for compression, around 1.2 kWh natural gas per 1 kW power output is typically needed [30].
e
Electrolysers with 60e73% electricity to fuel efficiency are commercially available and SOFCs have a potential electric efficiency of around 60%. The process heat from the
fuel cell can be utilised for district heating yielding CHP production [32].
f
The batteries used in electric vehicles will be largely lithium ion, (loss of 1%/month), or nickelmetal hydride (loss of 5e20%/month for nickelecadmium) [26].
Table 2
Categorisation of selected large-scale electricity storage technologies with respect to the time scales at which they can support wind power integration.
Batteries X X X
Flow batteries X X X (X)
Electric vehicles X X X
CAES X X X X
Pumped hydro X X X X X
Electrolysis, SOFC (X)*
(X)* X X X
hydrogen storage in caverns
Few to 30 s response 15 min response Intra-day/Day-ahead Several days Seasonal
(primary reserves) (regulating power,
secondary reserves)
Intra-hour
“X” and “(X)” indicate suitable and potentially suitable applications, respectively. *Possible only if the plant is already in operation.
provided by electric vehicles can generally be expected to be 3.6. Electrolysis combined with fuel cells
highest at evening/night and more limited during the day. More-
over, with current battery technology, a typical electric vehicle has Use of hydrogen as a storage of electrical energy requires the
a driving range up to around 150 km per full charge [23] corre- conversion from electrical to chemical energy through electrolysis.
sponding to a few hours of driving. As a result, charging (and dis- Among different hydrogen storage options, pressurised hydrogen
charging) on a daily basis will likely be pursued in most cases. storage in underground caverns is expected to have the lowest
Overall, at least with the current battery technology, electric costs [33]. Several types of fuel cells exist whereof Solid Oxide Fuel
vehicles can mainly be expected to provide balancing intra-hour Cells (SOFC) have a potential for replacing existing technologies in
and intra-day/day-ahead. The main vehicle-to-grid function is distributed central or local CHP/power plants. Compared to other
likely to be fast reserves within a limited period [27]. fuel cells, SOFCs have the strength of potentially higher efficiencies,
rather long life-times at constant operation and are rather fuel-
3.4. CAES flexible. The costs of electrolysers and fuel cells are still high,
SOFCs are still in the developmental stage and none of the
CAES have been studied for many years but only two systems hydrogen storage options are easily applied in practice [24,32].
have been constructed in the world so far, both dominated by Electrolysis plants in operation can be up and down regulated
combustion of natural gas. CAES relies on the possibilities of using within few seconds and are therefore well suited for providing
a cavern or other underground geological formation enabling regulating power and primary and secondary reserves. From cold,
a large storage volume since the technology becomes prohibitively an electrolysis plant however needs hours to start-up [27]. Due to
expensive in the lack of such storage facilities [27]. a high operating temperature (500e1000 C) and the temperature
A CAES plant has a start-up time of around 9 min for an emer- gradients, SOFCs have start-up time of several hours. Hydrogen
gency start and around 12 min under normal conditions. Thus, storage in underground caverns offers potentially low storage
a CAES system is slower in its response than batteries, flow related costs and could therefore be suited for storing large
batteries and vehicle-to-grid applications but approximately the amounts of energy. In addition, preliminary experiences show
same as conventional combustion turbine peak plants [10]. As such, negligible storage losses [38]. As such, hydrogen storage in under-
CAES is suitable for providing regulating power and secondary ground caverns is suited for storage of several days’ production as
reserves intra-hour [2] and for levelling out daily power imbalances well as for seasonal storage.
[34] while its ability to deliver primary reserves requires that the
plant is already in operation. Relatively low costs of storage capacity 3.7. Technology overview
($/kWh) further makes CAES suited for storing large amounts of
energy [25] and very low energy losses makes it possible to store Table 2 summarises the categorisation of the different electricity
energy for more than one year [10]. CAES is suitable to level out storage technologies with respect to the time scales they are suited
daily load fluctuations [34] and according to Cavallo [35], seasonal for in supporting wind power integration.
energy storage with CAES/wind systems can be economically It can be seen that while all the storage technologies are suitable
feasible. for intra-hour and intra-day/day-ahead balancing, only some are
suitable for responding to several days with high or low net load
3.5. Pumped hydro and even fewer for seasonal electricity storage.
Pumped hydro has been at commercial level for a long time in 4. Conclusion
many countries where the topography is suitable. However, new
techniques of utilising underground caverns or subsurface reser- In this study, system impacts of wind power are analysed from
voirs [36,37] are opening up possibilities of using pumped hydro in a time scale perspective for an energy system with 57% wind
areas without mountains, such as Denmark. Like other hydroelec- penetration, using Western Denmark in 2025 as case. The system
tric plants, pumped hydro can respond to load changes within impacts of wind power are categorised into intra-hour impacts,
seconds and can thus provide primary reserve capability [10]. intra-day/day-ahead impacts, several days’ impacts and seasonal
Moreover, like CAES, pumped hydro has relatively low storage impacts. Based on hourly net load variations, the physical potential
related costs ($/kWh) [25] and storage losses and pumped hydro is for power balancing on different time scales is investigated.
currently the most cost-effective means of storing large amounts of Expected lengths of consecutive negative and high net load periods
electrical energy. Pumped hydro is also suitable to level out daily indicate a significant potential for electricity storage or flexible
power imbalances [34]. However, capital costs and the presence of electricity demand being able to charge/activate demand and
suitable geography are critical factors for the feasibility of pumped discharge/inactivate demand in periods of 1 h to one day, i.e.
hydro installations [10]. responding to intra-day/day-ahead impacts of wind power. The
results suggest a lower but also significant potential for flexibility [13] Meibom P, Kiviluoma J, Barth R, Brand H, Weber C, Larsen HV. Value of electric
heat boilers and heat pumps for wind power integration. Wind Energy 2007;
measures capable of charging/activating demand in periods of
10:321e37.
several days. Finally, the estimates indicate a physical potential for [14] Lund H, Münster E. Modelling of energy systems with a high percentage of
seasonal electricity storage. Intra-hour impacts could not be evalu- CHP and wind power. Renewable Energy 2003;28(11):2179e93.
ated since only hourly data where available. [15] Lund H, Hvelpelund F, Østergaard PA, Möller B, Mathiesen BV, Andersen AN,
et al. Danish wind power export and costs. Denmark: Aalborg University; 2010.
Based on technology properties, number of large-scale elec- [16] Energinet.dk. Download of market data, www.energinet.dk/en/menu/
tricity storage technologies are moreover categorised with respect Frontpage.htm; 2009 (accessed 30.12.09).
to the time scales at which they are suited to support wind power [17] Energinet.dk. Miljørapport 2009-baggrundsrapport, www.energinet.dk; 2010
(accessed 18.01.10).
integration. All technologies covered, i.e. batteries, flow batteries, [18] Østergaard PA. Ancillary services and the integration of substantial quantities
electric vehicles, CAES, pumped hydro and electrolysis combined of wind power. Appl Energy 2006;83(5):451e63.
with fuel cells (SOFC, underground hydrogen storage) are found [19] Energinet.dk. Forudsætninger for analyser af regeringens energistrategi - en
visionær dansk energipolitik 2025-til system plan 2007, www.energinet.dk;
suitable for intra-hour and intra-day/day-ahead balancing. Only 2007 (accessed 18.02.10).
CAES, pumped hydro, electrolysis/fuel cells and potentially flow [20] Energinet.dk. Effektiv anvendelse af vindkraftbaseret el i Danmark, www.
batteries are considered suitable for balancing responding to energinet.dk; 2009 (accessed 02.03.10).
[21] The Nordel Market Development Group. Description of balance regulation in
several days with high or low net load. Merely CAES, pumped hydro the Nordic countries. Organisation for the Nordic Transmission System
and electrolysis/fuel cells combined with underground hydrogen Operators, www.energinet.dk; 2008 (accessed 24.10.10).
storage are assessed suitable for seasonal balancing. [22] Energinet.dk. Systemydelser til levering i Danmark - udbudsbetingelser,
www.energinet.dk; 2009 (accessed 03.02.10).
[23] Holttinen H, Meibom P, Orths A, van Hulle F, Lange B, O’Malley M, et al. IEA
wind task 25-design and operation of power systems with large amounts of
Acknowledgements wind power. Finland: VTT Technical Research Centre of Finland; 2009.
[24] Risø DTU National Laboratory for Sustainable Energy. Risø energy report 8:
The study is financed by Risø DTU as part of a PhD study on the intelligent energy system infrastructure for the future, http://www.dtu.
dk/subsites/klima/English.aspx; 2009 (accessed 24.10.10).
flexibility measures facilitating wind power integration. Thanks are [25] Ekman CK, Jensen SH. Prospects for large scale electricity storage in Denmark.
given to the Danish TSO, Energinet.dk, for contributing with data to Energy Convers Manage 2010;51(6):1140e7.
the analysis. [26] Divya KC, Østergaard J. Battery energy storage technology for power sys-
temsdan overview. Electr Power Syst Res 2009;79(4):511e20.
[27] Danish Technological Institute. EcoGrid.dk e steps towards a danish power
system with 50% wind energy, phase 1, WP4: new measures for integration of
References large scale renewable energy, www.energinet.dk; 2007 (accessed 05.03.10).
[28] Tomi c J, Kempton W. Using fleets of electric-drive vehicles for grid support.
[1] Lund H, Kempton W. Integration of renewable energy into the transport and J Power Sources 2007;168(2):459e68.
electricity sectors through V2G. Energy Policy 2008;36:3578e87. [29] Weiss MA, Heywood JB, Drake EM, Schafer A, AuYeung FF. On the road in
[2] Energinet.dk. System plan 2007, www.energinet.dk; 2007 (accessed 20.02.10). 2020: A life-cycle analysis of new automobile technologies. USA: Massachu-
[3] Holttinen H. Hourly wind power variations in the Nordic countries. Wind setts Institute of Technology, http://tris.trb.org/view.aspx?id¼743178; 2000
Energy 2005;8:173e95. (accessed 02.06.10).
[4] Holttinen H. Impact of hourly wind power variations on the system operation [30] Lund H, Salgi G. The role of compressed air energy storage (CAES) in future
in the Nordic countries. Wind Energy 2005;8:197e218. sustainable energy systems. Energy Convers Manage 2009;50(5):1172e9.
[5] Holttinen H. The Impact of Large Scale Wind Power Production on the Nordic [31] Danish Energy Authority, Elkraft system, Eltra. Technology data for electricity
Electricity System, Ph.D. Thesis. Helsinki University of Technology, Finland; and heat generating plants, http://193.88.185.141/Graphics/Publikationer/
2004. Forsyning_UK/Technology_Data_for_Electricity_and_Heat_Generating_Plants/
[6] Jónsson T, Pinson P, Madsen H. On the market impact of wind energy fore- pdf/Technology_Data_March05.pdf; 2005 (accessed 24.10.10).
casts. Energy Econ 2010;32(3):313e20. [32] Mathiesen BV. Fuel Cells and Electrolysers in Future Energy System, Ph.D.
[7] Grohneit PE, Andersen FM. Is more than 25% wind power feasible in a regional Thesis. Aalborg University, Denmark; 2008.
electricity market. 10th IAEE European Conference, Vienna, Austria, 7e10 [33] Mathiesen BV. Analysis of power balancing with fuel cells & hydrogen
September 2009. production plants in Denmark, Energinet.dk, http://vbn.aau.dk/da/
[8] Kazempour SJ, Moghaddam MP, Haghifam MR, Yousefi GR. Electric energy publications/search.html; 2009 (accessed 24.10.10).
storage systems in a market-based economy: comparison of emerging and [34] Kondoh J, Ishii I, Yamaguchi H, Murata A, Otani K, Sakuta K, et al. Electrical
traditional technologies. Renewable Energy 2009;34(12):2630e9. energy storage systems for energy networks. Energy Convers Manage 2000;
[9] Ibrahim H, Ilinca A, Perron J. Energy storage systems e characteristics and 41(17):1863e74.
comparisons. Renew Sustain Energy Rev 2008;12(6):1221e50. [35] Cavallo A. Controllable and affordable utility-scale electricity from intermit-
[10] Hadjipaschalis I, Poullikkas A, Efthimiou V. Overview of current and future tent wind resources and compressed air energy storage (CAES). Energy 2007;
energy storage technologies for electric power applications. Renew Sustain 32(2):120e7.
Energy Rev 2009;13(9):1513e22. [36] Uddin N, Asche M. Preliminary design of an underground reservoir for
[11] Mathiesen BV, Lund H. Comparative analyses of seven technologies to facili- pumped storage. Geotech Geol Eng 2003;21:55e331.
tate the integration of fluctuating renewable energy sources. IET Renew [37] de Boer WW, Verheij FJ, Zwemmer D, Das R. The energy island - an inverse
Power Gen 2009;3:190e204. pump accumulation station. Poster Presented at EWEC, Milan, Italy, 7e10 May
[12] Ummels BC, Pelgrum E, Gibescu M, Kling WL. Comparison of integration 2007.
solutions for wind power in the Netherlands. IET Renew Power Gen 2009;3: [38] Crotogrin, F, Donadai S, Grid scale energy storage in salt caverns. Risø Inter-
279e92. national Energy Conference, Roskilde, Denmark, 10e12 May, 2011.
Paper II
Effects of electric vehicles on power systems in Northern

Europe
Karsten Hedegaard, Hans Ravn, Nina Juul, Peter Meibom.

Energy 48 (2012) 356e368
Contents lists available at SciVerse ScienceDirect
Energy
journal homepage: www.elsevier.com/locate/energy
Effects of electric vehicles on power systems in Northern Europe

Karsten Hedegaard a, *, Hans Ravn b, Nina Juul a, Peter Meibom c
a
Systems Analysis Division, DTU Management Engineering, Frederiksborgvej 399, Build. 130, P.O. 49, DK-4000 Roskilde, Denmark
b
RAM-lose.dk Æblevangen 55, DK-2765 Smørum, Denmark
c
Danish Energy Association, Rosenørns allé 9, DK-1970 Frederiksberg, Denmark
Article history: In this study, it is analysed how a large-scale implementation of plug-in hybrid electric vehicles and
Received 1 November 2011 battery electric vehicles towards 2030 would influence the power systems of five Northern European
Received in revised form countries, Denmark, Finland, Germany, Norway, and Sweden. Increasing shares of electric vehicles (EVs)
20 April 2012
are assumed; comprising 2.5%, 15%, 34%, and 53% of the private passenger vehicle fleet in 2015, 2020,
Accepted 5 June 2012
Available online 6 July 2012
2025, and 2030, respectively. Results show that when charged/discharged intelligently, EVs can facilitate
significantly increased wind power investments already at low vehicle fleet shares. Moreover, due to
vehicle-to-grid capability, EVs can reduce the need for new coal/natural gas power capacities. Wind
Keywords:
Electric vehicles
power can be expected to provide a large share of the electricity for EVs in several of the countries.
Wind power integration However, if EVs are not followed up by economic support for renewable energy technologies, coal based
Investments power will in several cases, particularly in the short term, likely provide a large part of this electricity.
Energy systems analysis The effects of EVs vary significantly from country to country and are sensitive to fuel and CO2 price
Model variations. The EVs bring CO2 reductions of 1e6% in 2025 and 3e28% in 2030 while total costs are
Heat measures generally increased.
1. Introduction Texas power system is used to simulate system operations with

fleets of plug-in hybrid electric vehicles (PHEVs), estimating the
Electric vehicles (EVs) can potentially play an important role in value of vehicle-to-grid (V2G) services. Soares et al. [7] evaluate the
transforming the transport sector towards sustainability. Various possibility of using a fleet of PHEVs to regularise possible energy
fields related to EVs have been studied recently, i.e., building of imbalances for a north-eastern Brazilian case. In [8], it is investi-
infrastructure, how to move towards 100% renewable energy in the gated how it affects the power system when EV charging is dis-
transport system, and potential benefits for vehicle owners as well patched optimally. Kiviluoma and Meibom [9] analyse the influence
as the power system. The concept of vehicle-to-grid (V2G) has been of PHEVs, heat pumps, electric boilers, and heat storages on power
defined and explained by Kempton and Tomi c in [1], where also system investments in Finnish high wind power scenarios. In [10],
potential benefits have been touched upon. In [2], Kempton et al. the same authors analyse the value of smart charging of EVs
have looked more into the services to be provided by EVs and compared to charging immediately when being connected to the
economics of providing these services. Specific focus on peak load grid. Lund and Kempton [11] have developed a rule based model of
shaving in Japan is found in [3] and analyses of regulation and an integrated power and transport system, focussing on the value of
ancillary services are found in [4]. including V2G in different wind penetration scenarios. In [12],
Modelling of the integrated power and transport system has Kristoffersen et al. calculate the optimal charging patterns of EVs
only been the focus of few studies so far. McCarthy, Yang, and when buying and selling electricity on the Nordic day-ahead power
Ogden [5] have developed a simplified dispatch model for Cal- market. Model generated investments in different vehicle types
ifornia’s energy market to investigate the impacts of integrating have been introduced in [13e15].
EVs into the energy system. In [6], a unit commitment model of the The transition path towards large-scale deployment of PHEVs
and battery electric vehicles (BEVs), has e.g. in [16,17] been studied
with focus on how to ensure a smooth transition. However, none of
* Corresponding author. Frederiksborgvej 399, P.O. Box 49, Build. 130, DK-4000
the existing studies identified investigate how transition towards
Roskilde, Denmark. Tel.: þ45 4677 5158; fax: þ45 4677 5199. increased electrification of the transport sector would affect the
E-mail addresses: [email protected], [email protected] (K. Hedegaard). power system. This is the subject of this paper.
http://dx.doi.org/10.1016/j.energy.2012.06.012
K. Hedegaard et al. / Energy 48 (2012) 356e368 357
Barriers are to be overcome before a large-scale deployment of including satisfaction of demands for electricity, heat, and transport
EVs can be realised. One of the obstacles is the current significant in each time period, renewable energy potentials, vehicle restric-
price difference between EVs and conventional vehicles faced by tions, and technical restrictions on units in the power system.
the consumers [18]. In [19], it is found that battery costs must drop Balmorel operates with three geographical entities: countries,
significantly before EVs will find a mass market without subsidies. regions, and areas. Countries are divided into regions connected
With respect to gaining consumer’s acceptance, BEVs furthermore with transmission lines and regions are further divided into areas.
have disadvantages in the form of a relatively short driving range The model balances electricity and road transport supply and
and a long charging time. Electricity consumption for air condi- demand on regional level, whereas district heating is balanced on
tioning or cabin heating can also shorten the driving range in hot/ area level. The optimisation is performed with a yearly time
cold weather [20]. Thus, further technology development and likely horizon. In Balmorel, the year is divided into seasons, which may be
also economic incentives, as established by several national and used to represent weeks, and into time periods, which may
local governments [21], are required before widespread use of EVs represent hours.
can be realised. Finally, large-scale implementation of EVs requires
a set up of sufficient infrastructure in terms of charging spots etc. 2.1. Transport add-on
The focus of this study is not on how to realise a large-scale use of
EVs but on how such a large-scale implementation would affect the Road transport is modelled using the add-on presented by Juul
power system. and Meibom in [14]. Further model development has been made in
A large-scale implementation of electric vehicles would not only order to handle the gradual implementation of different vehicle
affect power system operation but also investments. As power vintages in the vehicle fleet towards 2030. The transport model
system investments are realised continuously, these effects are best includes demand for transport services, vehicle investments and
investigated by analysing a period of several years. In this study, it is operational costs, and electricity balancing in the integrated road
analysed how a gradual large-scale implementation of PHEVs and transport and power system. As such, the model makes it possible
BEVs in the Northern European countries, Denmark, Finland, to analyse interactions between the two systems and to identify
Germany, Norway, and Sweden, would influence power system benefits and optimal investments, and operation. In this study,
investments and operation towards 2030. Inspired by scenarios set vehicle investments are fixed to an assumed development path,
up the Electric Power Research Institute (EPRI) and the Interna- while investments in the power system are generated endoge-
tional Energy Agency (IEA), increasing shares of electric vehicles are nously. Among the vehicle technologies available in the model, the
assumed; comprising 2.5%, 15%, 34%, and 53% of the private following are included in the analysis:
passenger vehicle fleet in 2015, 2020, 2025, and 2030, respectively.
The analyses performed are based on the model of the integrated Internal combustion engine vehicles (ICEs): vehicles driving on
power, district heat, and transport system described in [14,22]. petrol, diesel or the like (for simplicity represented as diesel
Simulations are made with five year intervals where optimal fuelled vehicles in this analysis)
investments identified in previous years are included in the opti- BEVs: battery electric vehicles driving on electricity only
misations of subsequent years. Plug-in patterns based on a national PHEVs: plug-in hybrid electric vehicles driving on electricity as
investigation of transport habits are implemented as in [14] and well as a complementary fuel, i.e. electric vehicles with range
intelligent charging/discharging is identified as part of the energy extenders using an internal combustion engine. All PHEVs are
system optimisation. A socio-economic optimisation is applied in for simplicity assumed to use diesel as liquid fuel.
order to investigate how EVs would affect the power systems in the
absence of taxes, tariffs and subsidies. Under these conditions, the In the model, all EVs are assumed to leave the grid with a fully
results reveal which power generation technologies are likely to charged battery, restricting the charging to meet this load factor.
meet the large electricity demand induced by EVs over the period. The plug-in hybrids are assumed to use the electric storage (the
In a sustainability perspective, the relevant question is to which usable part of the battery) until depletion before using the engine.
extent renewable energy sources would cover this electricity This assumption is considered reasonable due to the high efficiency
demand. Additionally, effects of EVs on CO2 emissions and costs are of the electric motor compared to that of the combustion engine
evaluated. Putting the impacts of EVs into perspective, these are and due to the low price of electricity (average prices in the
compared to the effects of investing in heat storages, and flexible neighbourhood of V50/MWh in the simulations) compared to the
electricity demand options in the form of heat pumps, and electric price of diesel (64e80 V/MWh in 2015e2030, cf. Table 8). More-
boilers in the district heating system, forming alternative ways of over, the batteries have no loss of power before almost depleted,
increasing system flexibility. leaving the motor able to perform as demanded until down to the
Section 2 in the article presents the model, Balmorel, and the minimum state of charge.
transport add-on used for the analyses. In Section 3, the application Integrating the power and transport systems and introducing
of the model is described including scenarios and input data. intelligent charging and discharging requires a number of additions
Section 4 covers presentation and discussion of results. Finally, to the existing system. E.g. communication between vehicles and
a conclusion is given in Section 5. the power system, vehicle aggregators communicating with power
markets, and agreement upon connection standards is needed. In
2. Model the model, all such changes are assumed to be in place. Infra-
structure costs covering charging spots and potential local grid
The integrated power and road transport system is modelled in enhancements are difficult to quantify and are in [23] assessed to be
Balmorel, which is a deterministic partial equilibrium model moderate compared to vehicle costs. Against this background,
assuming perfect competition [9,14,22] .The model optimises infrastructure costs are not included. The model works with
investments in power/heat production, storage, and transmission a capacity credit restriction ensuring enough production capacity to
capacities. This is done minimising total costs in the energy system, meet peak power demand as presented in [9]. Due to V2G capa-
covering annualised investment costs, operation and maintenance bilities of BEVs and PHEVs, they are able to contribute in meeting
costs of existing and new units, as well as fuel and CO2 quota costs. peak power demand. The modelling of this contribution is taken
The optimisation is performed subject to a number of constraints from the PhD thesis by Nina Juul [24].
358 K. Hedegaard et al. / Energy 48 (2012) 356e368
3. Application heat storages, heat pumps, and electric boilers in the district
heating system should be taken into account. This is relevant as
The model includes the power sector, the district heating sector, these represent competing measures of increasing system flexi-
and the part of the road transport sector comprising private bility. Furthermore, it is intended to compare the effects of EVs with
passenger vehicles. With the intent to obtain reasonable compu- the effects of heat measures. Based on these considerations, four
tation times, Norway, Sweden, and Finland are each treated as one main scenarios are created with/without EVs and with/without the
power region. Thus, internal transmission bottlenecks in these possibility to invest in heat measures in the district heating system
countries are not modelled. Germany is aggregated into two (see Table 1).
regions, representing the transmission bottlenecks between When assessing the effects of EVs, the Heat and EVHeat scenario
Northern Germany with its large share of wind power and the large are compared. When comparing the effects of EVs with the effects
consumption centres in Central & Southern Germany. Denmark is of heat measures, differences between the Base scenario and the EV
divided into two regions: Western Denmark being synchronous and Heat scenario, respectively, are used as basis.
with the UCTE power system and Eastern Denmark being In addition, a number of sensitivity scenarios are made, covering
synchronous with the Nordel power system. Also based on e.g. low/high fuel and CO2 price developments. The sensitivity
computation time limitations, Norway, Sweden, and Finland, are scenarios are described and analysed in Section 4.5. Keeping main
each modelled as one district heating area, Germany as two, and focus on the effects of EVs when competing with heat measures,
Denmark as four district heating areas. the sensitivity scenarios comprise variants of the Heat and EVHeat
In order to capture wind power fluctuations and to obtain scenario.
a good representation of power flows between grid and vehicles, an
hourly time resolution is chosen. To ensure reasonable computa-
3.2. Input data
tion times, 7 weeks are simulated and weighted to represent a full
year. Calculation time for a model run with EVs and heat measures
Electricity, district heating and transport demands as well as
covering 2015, 2020, 2025, and 2030 with this time resolution is
annual driving per vehicle are all given as data inputs to the model
approximately 24 h on a 3.4 GHz quad core computer with 8 GB
(see Tables 2 and 3).
RAM.
The expected development of vehicle technologies in terms of
The assumed implementation of EVs is based on scenarios set up
costs, efficiencies, electric storage capacities, and battery ranges is
by EPRI [25] and IEA [26]. In the Medium scenario in [25], a devel-
taken into account. The data applied for the different vehicle
opment in PHEVs new vehicle shares as outlined in Fig. 1a is
technologies and vintages are given in Table 4. As in the study in
assumed. Based on the relative development in sales of PHEVs and
general, all costs are given in V2008.
BEVs towards 2030 in the Blue Map scenario in [26], additional BEV
As in [14], plug-in patterns for BEVs and PHEVs have been
market shares corresponding to half of the PHEV new vehicle
derived from driving patterns obtained from the investigation of
shares have been assumed. Applying an average vehicle lifetime of
transport habits in Denmark [33]. In this regard, it has been
16 years, the resulting development in the vehicle fleet shares
assumed that the EVs are plugged-in at all times when parked.
towards 2030 is illustrated in Fig. 1b. Consequently, EVs are
Furthermore, it is assumed that driving habits are the same for all
assumed to comprise around 2.5%, 15%, 34%, and 53% of the vehicle
the countries in the simulation. An iterative process has been
fleet in 2015, 2020, 2025, and 2030, respectively. This represents an
required in order to make the total transport demands fit with the
ambitious scenario for implementation of EVs and should not be
number of each type of vehicle, the annual driving distances for
interpreted as a forecast. Rather, the scenario is used for analysing
ICEs/PHEVs and BEVs, and the driving patterns. Total transport
how such a large-scale implementation would affect the power
demands have thus been adjusted and are still close to the demands
system. The assumed BEV implementation, comprising 1/3 of the
in the sources used. It is assumed that BEVs of vintage 2015 and
EV fleet shares, can be interpreted to represent use as second/third
2020 can cover trips lasting up to 2 h (corresponding to 115 km)
cars of the households. This is reasonable since statistics show that
yielding an annual driving of 10,230 km/yr and that BEVs of vintage
a large share of all trips, 25% for the case of Denmark, are driven
2025 and 2030 can cover trips of up to 3 h (corresponding to
with second/third cars [15].
205 km) yielding 12,671 km/yr. This is considered reasonable based
on the distances supported by the BEV battery capacities in Table 4;
3.1. Scenarios assuming that people will be reluctant to drive close to emptying
the battery and that spare battery capacity will in some cases be
In order to investigate the effect of EVs, scenarios are set up required for a second trip in the day.
with/without the gradual large-scale EV implementation given in The model includes comprehensive data on capacities, effi-
Fig. 1b. When evaluating the effects of EVs, possible investments in ciencies, operation costs, technical lifetimes etc. for existing units
a b
100% 100%
New vehicle shares
80% 80%
Vehicle fleet
60% ICE
60%
BEV
40% 40%
PHEV
20% 20%
0% 0%
Fig. 1. a) Development in plug-in hybrid electric vehicle (PHEV) new vehicle shares in the Medium scenario in [22] and illustration of the assumed relation between battery electric
vehicle (BEV) and PHEV new vehicle shares based on the Blue Map Scenario in [23]. b) Assumed gradual penetration of PHEVs and BEVs in the vehicle fleet.
Table 1 Table 4
Main scenarios in the analysis. Vehicle technology data.
Base EV Heat EVHeat Veh. Vintage Inv. cost O&M cost Elec. stor. Eff. Bat. range
Gradual implementation of PHEVs and BEVs þ þ type (V/yr)a (V/yr) [30] capb (kWh) (km/kWh) (km)d
towards 2030 [26,30] [26] [14] c [14,31]
Inv. in heat storages, heat pumps, and electric þ þ ICE 2015 1058 1168 e 1.8 e
boilers allowed 2020 1058 1168 e 1.9 e
2025 1058 1168 e 1.9 e
2030 1058 1168 e 2.0 e
BEV 2015 3035 1101 40 5.5 220
for power/heat production, storage, and transmission. As such, 2020 2509 1101 43 6.0 260
gradual decommissioning of existing power/heat production 2025 1962 1101 47 6.5 303
2030 1745 1101 50 7.0 350
capacities towards 2030 is included. The current electricity PHEV 2015 2122 1168 12 5.5 65
production distributed on sources for each of the five countries is 2020 1784 1168 11 6.0 65
illustrated in Fig. 2 (as generated by the model). 2025 1521 1168 10 6.5 65
Based on national policies, nuclear power investments are 2030 1387 1168 9 7.0 65
assumed allowed only in Finland and Sweden. The Swedish a
A discount rate of 5% is applied in fixed prices based on [32].
b
government has decided that investments in new nuclear power The usable storage capacity of the battery.
c
plants can only be made if replacing existing plants [34]. Hence, the 5 km/kWh for BEV/PHEV vintage 2010 and 7 km/kWh for vintage 2030 [14].
d
Battery range of 150 km for BEV vintage 2010 [31] and 350 km for vintage 2030
maximum allowed installed Swedish nuclear power capacity is set [14].
to the current capacity (around 10,200 MW). For Finland, the new
Olkiluoto reactor 3 (1600 MW) being under construction and
planned for operation from 2013 is included in the model. The climate policies. However, if restricting wind power investments in
maximum allowed installed nuclear power capacity in Finland the model, e.g. to meet national wind targets for each year as
towards 2030 is set to 7260 MW, representing current capacity plus a minimum, the effects of EVs on wind power investments and
the capacity of Olkiluoto 3 and of two additional nuclear power generation would not be fully reflected. Therefore, such restrictions
units considered for construction towards 2030 (Olkiluoto 4 and have not been included. Accumulated onshore wind power capac-
Pyhäjoki, around 1500 MW each) [35]. Based on experiences con- ities have, however, been constrained by onshore wind potentials
cerning time scales for construction of nuclear power plants, as given in Table 6. In addition, capacity growth limits on wind
investment in Finnish nuclear power plants is only allowed from power have been applied to ensure reasonable capacity increases
year 2020 and only up to 1500 MW per five years, corresponding to per year.
the implementation of one unit. In response to the Fukushima Hydro power is characterised by costs and implementation
incident, the current German government has decided to phase out barriers that are site specific to a higher degree than many other
nuclear power. Acknowledging this but also considering possible sources of electricity generation [43]. Against this background,
delays in the phase out and shifts in the political opinion, German investments in new hydro power capacity are not identified as part
nuclear power capacities are assumed decommissioned based on of the energy system optimisation. Instead, expected increases in
technical lifetimes. hydro power production are included as fixed generation levels
As coal based power plants are not considered among the (see Table 7).
feasible options for future energy supply in the hydro power Historically, Finland has had a significant net import from
dominated Norwegian energy system, investment in this tech- countries outside the system boundary of this analysis; mainly
nology is in this case excluded. The power system units assumed from Russia; in 2010 around 13.4 TWh corresponding to 15% of
available for investment are given in Table 5. In addition, invest- domestic electricity demand. Therefore, based on hourly data for
ment in transmission capacities between regions is allowed. 2010 [44], this net import is included as a fixed electricity exchange.
Wind power investments are in reality not alone based on
economic rationales but are also influenced by national energy and
100%
Photo voltaic
Table 2 90%
Electricity demand (TWh/yr)/District heating demand (TWh/yr)/Transport demand Wind
(109 person km/yr) given in rounded numbers. 80% Peat
70% Wood waste
2010 2015 2020 2025 2030
Denmark 33/28/57 34/28/60 34/28/63 35/28/66 38/28/69 60% Wood and straw
Finland 89/45/72 95/50/73 99/55/73 101/56/73 104/56/73 50% Municipal waste
Germany 554/94/ 585/96/ 600/100/ 614/101/ 620/102/
1025 1069 1092 1103 1116 40% Hydro
Norway 119/2.7/53 124/2.8/55 127/2.8/57 128/2.8/59 129/2.8/61 30% Natural gas
Sweden 141/45/112 147/46/117 150/47/121 152/47/125 153/46/130
20% Lignite
Sweden, Finland, and Germany: [27], Norway: based on current relation between Coal
Norwegian and Swedish demands/number of cars. Denmark, electricity and district 10%
heating: [28], transport: based on [27,29]. Nuclear
0%
NORWAY
FINLAND
GERMANY
SWEDEN
DENMARK
Table 3
Annual driving for each vehicle type.
km/yr 2015 2020 2025 2030

ICE/PHEV 18,072 18,401 18,676 19,126 Fig. 2. Electricity generation in the present power systems of the five Northern
BEV 10,230 10,230 12,671 12,671 European countries distributed on sources. The distribution is generated by the model
for the Base scenario for 2010 when not allowing investments in new capacities.
Table 5
Technologies available for investment in the optimisation.
Technology Fuel Period available Inv. costa Variable Fixed O&M cost Lifetime Effb CB CV Ref.
(MV/MW) O&M cost (kV/MW/yr) (years)
(V/MWh)
Onshore wind turbine e 2011e2020 1.33 12.50 e 20 1.00 e e [36]
e 2021e2030 1.24 11.75 e 25 1.00 e e [36]
Offshore wind turbine e 2011e2020 2.50 17.00 e 20 1.00 e e [36]
e 2021e2030 2.25 15.50 e 25 1.00 e e [36]
Steam turbine, extraction, CHP Coal 2011e2020 1.43 7.00 e 40 0.46 0.75 0.15 [36]
2021e2030 1.40 7.00 e 40 0.50 0.93 0.15 [36]
Open cycle gas turbine, condensing Natural gas 2011e2030 0.32 2.40 16 20 0.37 e e [37]
Combined cycle gas turbine, Natural gas 2011e2020 0.52 3.20 20 25 0.59 1.55 0.13 [36,37]
extraction, CHP 2021e2030 0.47 3.20 20 25 0.62 1.75 0.13 [36,37]
Nuclear, condensingc Uranium 2011e2030 2.81 7.7 56 40 0.37 e e [37]
Steam turbine, extraction, CHP Wood 2011e2020 1.68 3.20 23 30 0.46 0.53 0.15 [36]
2021e2030 1.60 3.20 23 30 0.48 0.58 0.15 [36]
Steam turbine, back pressure, CHP Wood 2011e2020 4.40 e 154 20 0.25 0.30 e [36]
2021e2030 3.95 e 138 20 0.25 0.30 e [36]
Steam turbine, back pressure, CHP Straw 2011e2020 4.35 e 174 20 0.30 0.49 e [36]
2021e2030 3.90 e 156 20 0.30 0.49 e [36]
Heat boiler Wood 2011e2030 0.50 e 24 20 1.08 e e [36]
Heat boiler Natural gas 2011e2030 0.09 e 3.2 20 1.01 e e [36]
Heat pumpd Electricity 2011e2020 0.65 e 6.9 20 2.8 e e [36,38]
2021e2030 0.65 e 6.9 20 3.0 e e [36,38]
Electric boiler Electricity 2011e2030 0.06 0.5 1 20 0.99 e e [36]
Heat storage e 2011e2030 0.00185 e e 20 0.99 e e [39]
a
Based on [32], investment costs are in the model annualised with a discount rate of 5% given in fixed prices. Investment costs for heat storage are given in MV/MWh
storage.
b
For heat boilers, heat efficiency, for heat pumps, coefficient of performance, and for other units, electric efficiency.
c
Allowed in Finland and Sweden only.
d
Investment costs for heat pumps given in MV/MW-thermal.
Table 6 net import from countries outside the system boundary is assumed
Wind targets for 2030 and assumed onshore wind potentials. diminished from year 2020.
Medium wind High wind Onshore wind potential For Denmark, the system boundary covers all countries with
target [40] target [40] assumed which electricity is exchanged. For Sweden and Norway, electricity
Denmark 7291 8020 4500 (East: 1000 MW, exchange with countries outside the system boundary is of very
West: 3500) [41,42] low magnitude1 and has therefore been excluded. For Germany,
Finland 3200 6000 12,000b
annual import/export in 2008e2010 from/to countries outside the
Germany 54,244 63,587 63,600a
Norway 5980 11,970 12,000a system boundary corresponded to 5e7%/9e11% of domestic elec-
Sweden 10,000 17,000 17,000a tricity demand yielding a net export corresponding to 2e6% of the
a
Due to the large areas of these countries and uncertainties in estimating the
demand [46]. As such, for the case of Germany, this exchange is
onshore wind potential, the maximum onshore capacity is assumed limited to the moderate but not insignificant in magnitude. However, as German
high wind target. electricity exchange with countries outside the system boundary is
b
The Finnish high wind target is considered unrealistically low and therefore, distributed over many countries2 the exchange is complex. Hence,
onshore wind power in Finland is assumed limited to 12,000 MW corresponding to
a satisfactory inclusion of this electricity exchange would require
the Norwegian high wind target.
expanding the model to include all relevant countries with which
Germany trades electricity. This would require comprehensive data
However, part of the background behind installing new nuclear collection and increase model calculation times to a level chal-
power capacities in Finland is supposedly to reduce its dependency lenging completion of model runs and the analysis. This has been
on electricity import from Russia [45]. In addition to the new without scope of this study. Due to its large volumes, the German
nuclear power 1600 MW installed in 2013, model optimisations energy system is highly influential on total CO2 emissions and costs
suggest an additional capacity of 1500 MW to be installed in 2020. for the five Northern European countries as a whole. Thus, when
This leads to a total increase in nuclear power generation of around interpreting the results for Germany and for the five countries as
23 TWh/yr (assuming a capacity factor of 0.85 [35]), thus, exceeding a whole it should be keep in mind that not all German electricity
current Finnish net import from Russia. Based on this, the Finnish exchange options have been modelled.
Based on [47], CO2 prices are assumed to increase from 20 V/
tonne in 2015 to 39 V/tonne in 2030, and the assumed fuel prices
correspond to an oil price of $88/barrel in 2015 and $117/barrel in
Table 7 2030 (see Table 8).
Hydro power generation assumed.
TWh/yr 2010 2015 2020 2025 2030 Ref.

Germany 19.7 20.8 20.9 21.4 22.0 [27]
Finland 13.8 13.9 14.0 14.2 14.4 [27] 1
In 2008e2010, Swedish import/export to/from countries outside the system
Norway 126.8 131.8 136.8 141.8 146.8 [43]
boundary corresponded to 0.1e0.3%/0.5e1.5% of domestic electricity demand and
Sweden 66.4 66.7 67.0 67.0 67.0 [27,34]a
corresponding numbers for Norway were 0.3e2.1%/1.1e2.7% [46].
a 2
Swedish hydro power production in 2010 set to the average production for the Austria, Switzerland, Czech Republic, France, Lithuania, The Netherlands and
last five years based on [34] and relative increase based on [27]. Poland [46].
Table 8
Fuel and CO2 quota prices assumed (V/GJ) [47].
Fuel oil Diesel Natural gas Coal Lignite Uranium [48] Wood Straw Wood waste [9], CO2 (V/tonne)
municip. wastea
2010 6.7 14.8 6.0 2.9 1.5 0.7 6.0 5.1 0 14
2015 8.3 17.7 8.2 2.9 1.4 0.7 6.6 5.8 0 20
2020 9.4 19.7 9.2 3.2 1.6 0.7 6.9 5.9 0 25
2025 10.2 21.0 10.0 3.4 1.7 0.7 7.2 6.1 0 32
2030 10.9 22.4 10.7 3.4 1.7 0.7 7.5 6.2 0 39
Fuel costs include distribution costs.

a
Municipal waste is assumed to have zero cost applying a socio-economic perspective.
4. Results investments are observed in Denmark and Germany, and particu-

larly in Sweden where wind power investments are increased
Results of the main scenarios are presented in Sections 4.1e4.4 manifold from around 800 MW to 6000 MW. In Finland, the
and results of sensitivity scenarios in Section 4.5. assumed onshore wind potential of 12,000 MW is reached in 2030
in both the Heat and EVHeat scenario. Hence, in this case the EVs
4.1. Effects on power system investments push forward the investments in wind power. The reason for the
increased investments in wind power when implementing EVs is
Socio-economic optimal investments in new power production that their flexible charging/discharging facilitates integration of the
capacities generated in the Base, EV, Heat, and EVHeat scenario are variable production from wind power into the power systems.
illustrated in Fig. 3. As shown, the increases in production capacity As a result of the large-scale EV implementation and the
mainly comprise coal CHP and wind power. Nuclear power resulting increase in electricity demand, one might expect signifi-
investments are also realised for the case of Finland and Sweden, cantly increased investments in dispatchable power production
where this option is allowed. In addition, moderate investments in capacity in terms of e.g. coal or natural gas power plants. However,
wood CHP extraction plants and open cycle gas turbines are the results in Fig. 3 show that when EVs are charged/discharged
observed; the latter used for ensuring sufficient capacity to meet intelligently, increased investments in thermal power production
peak power demands. capacity are only observed in very few cases. As such, the only
Due to increasing fuel and CO2 prices, the economic conditions significant increases are observed in Germany in 2020e2030 where
for wind power generally improve over the period. This is clearly investments in coal CHP are increased with around 1100e1500 MW
illustrated for the cases of Denmark, Germany, and Sweden where in the EVHeat scenario compared to the Heat scenario. In fact,
wind power investments mainly occur in the last part of the period rather than increasing investments in thermal production capacity,
towards 2030. Moreover, existing wind power capacities in for the case of Denmark, Finland, and Sweden, EVs result in reduced
Denmark and Germany are significantly decommissioned from need for new thermal power capacities. Thus, investments in open
2020 to 2025 (from around 3200 MW to 0 MW in Denmark and cycle gas turbines and/or coal CHP capacities are reduced signifi-
from around 23,000 MW to 11,000 MW in Germany), which is also cantly in these countries in 2025e2030 with the implementing of
part of the explanation for the large wind power investments in EVs. This is explained by the V2G capability of EVs which contrib-
these countries in 2025. In Norway, Sweden, and Finland, where utes in covering peak loads resulting in a reduced need for dis-
onshore wind power resources are relatively high in terms of patchable production capacity.
obtainable full load hours,3 wind power investments occur earlier Like EVs, heat pumps, electric boilers, and heat storages in the
than for the cases of Denmark and Germany. The accumulated wind district heating system can also be seen to support wind power
power capacities in 2030 for the EVHeat scenario can be identified investments in several of the countries. While the optimisation
to 5100 MW for Denmark, 12,000 MW for Finland, 56,600 MW for generates heat measure investments in all five countries, the effect
Germany, 10,700 MW for Norway, and 12,000 MW for Sweden. on wind power investments is most significant in Finland and
Comparing these with the national medium wind targets for 2030 Sweden, where e.g. wind power investments in 2025 are signifi-
in Table 6, it can be seen that the target is met or exceeded for cantly higher in the EVHeat scenario compared to in the EV
Finland, Germany, Norway, and Sweden and not far from being met scenario (and in the Heat scenario compared to in the Base
for the case of Denmark. This indicates that the accumulated wind scenario). This is due to the combination of relatively large onshore
power investments generated in the optimisation are reasonable. wind potentials and obtainable full load hours in Finland and
Comparing the EVHeat scenario and the Heat scenario, the Sweden (as opposed to in Germany and Denmark) combined with
implementation of EVs can be seen to facilitate increased wind a significant district heating volume (as opposed to Norway, cf.
power investments in all five countries. In Finland and Norway, this Table 2).
effect is observed already in 2020, where EVs comprise 15% of the The Finnish and Swedish investments in heat measures cover
vehicle fleet. The effect is most significant in Finland where wind heat storages and heat pumps.4 Heat pumps (and electric boilers)
power investments in 2020 are doubled from approximately are able to support wind power investments as they represent
2700 MW to 5400 MW. At higher EV fleet shares of 34% in 2025, EVs a flexible electricity demand that can be activated when electricity
make wind power investments more attractive in Germany and prices are low, corresponding to periods with large amounts of
particularly in Sweden where wind power investments are wind power and low electricity demand. Additionally, the activa-
increased from around 4100 MW to 6000 MW. In 2030, where EVs tion of heat pumps can contribute in relieving CHP plants from heat
comprise 53% of the vehicle fleet, increased wind power bound power production. This also enhances system flexibility and
3 4
Assumed full load hours for onshore wind power: Norway: 3000 [43,50], E.g. for the Heat scenario, Finland: 1.2e6.2 GWh/yr heat storage capacity,
Finland, Sweden: 2600 [43,51], Denmark, West: 2440, Denmark, East: 1960 [42], 600e4600 MW-thermal/yr heat pump capacity, Sweden: 9.6e47 GWh heat storage
Germany: 1750 (based on [52,53]). capacity, 60e950 MW-thermal/yr heat pump capacity.
WI-onshore WI-offshore ST-WOOD-EX OC-GT ST-COAL NUC
DENMARK
6,000
5,000
4,000
MW
3,000
2,000
1,000
0
EVHeat
EVHeat
EVHeat
EVHeat
Heat
Heat
Heat
Heat
EV
EV
EV
EV
Base
Base
Base
Base
2015 2020 2025 2030
DENMARK
FINLAND
12,000
10,000
8,000
MW
6,000
4,000
2,000
0
EVHeat
EVHeat
EVHeat
EVHeat
Heat
Heat
Heat
Heat
EV
EV
EV
EV
Base
Base
Base
Base
2015 2020 2025 2030
FINLAND
GERMANY
40,000
30,000
MW
20,000
10,000
0
EVHeat
EVHeat
EVHeat
EVHeat
Heat
Heat
Heat
Heat
EV
EV
EV
EV
Base
Base
Base
Base
2015 2020 2025 2030
GERMANY
NORWAY
6,000
4,000
MW
2,000
0
EVHeat
EVHeat
EVHeat
EVHeat
Base
Base
Base
Base
Heat
Heat
Heat
Heat
EV
EV
EV
EV
2015 2020 2025 2030

SWEDEN
NORWAY
8,000
6,000
MW
4,000
2,000
0
EVHeat
EVHeat
EVHeat
EVHeat
Heat
Heat
Heat
Heat
EV
EV
EV
EV
Base
Base
Base
Base
2015 2020 2025 2030
Fig. 3. Investments in power production capacities in the Base, EV, Heat, and EVHeat scenario, representing accumulated investments over each five year period. E.g. investments in
2020 represent accumulated investments from 2016 through 2020. WI: Wind power; ST-COAL: Steam turbine, extraction, coal; ST-WOOD-EX: Steam turbine, extraction, wood; OC-
GT: Open cycle gas turbine; NUC: Nuclear power.
the integration of wind power. Heat storages facilitate increased power is low, possibilities for increasing power production at CHP
wind power investments as they increase the flexibility of heat plants are improved since surplus heat production can be stored.
pumps (and electric boilers) and improve the integration of wind Fig. 3 shows that in terms of increasing socio-economically
power into CHP systems. When wind power is high, possibilities for optimal wind power investments, heat measures can have effects
reducing power production from CHP plants are improved since of the same magnitude as EVs. As such, the accumulated wind
heat demand can be satisfied from the heat storage, and when wind power capacities by 2030 in the EV and Heat scenario, respectively,

8 60 4
6 50
3
40
4 2
TWh
TWh
TWh
30
2 20 1
10
0 0
0
2015 2020 2025 2030 2015 2020 2025 2030
-2 -10 2015 2020 2025 2030 -1

15 15 80
10 10 60
TWh
5 40
TWh
TWh
5
0 20
0 2015 2020 2025 2030
-5 0
2015 2020 2025 2030 2015 2020 2025 2030
-5 -10 -20
Fig. 4. Changes in annual electricity generation due to implementation of electric vehicles (EVHeat scenario vs. Heat scenario). Due to import/export and possible changes in
electricity consumption for heat pumps/electric boilers, generated power increases in each year will not necessarily correspond to the electricity demand for electric vehicles.
are identical or close to being identical for the case of Germany, Denmark and Germany in the first part of the period. An opposite
Norway, and Denmark; and while EVs in Sweden generate larger trend can be observed for Finland, where EVs generate increased
accumulated wind power investments than heat measures wind power production until the assumed onshore wind potential
(9200 MW in the EV scenario vs. 6900 MW in the Heat scenario) the is reached.5 After that point, EVs in Finland are mainly driven on
opposite is the case for Finland (10,900 MW in the EV scenario vs. coal based electricity. However, if the Finnish onshore wind
12,000 MW in the Heat scenario). Apart from influencing wind potential was set higher, increased wind power generation would
power investments, heat measures can also be seen to affect most likely also have been observed for 2030.
investments in thermal production capacities. For instance in In some cases, e.g. in Denmark and Finland in the last part of the
Finland and Sweden in 2025, where wind power investments are period, increased production on existing biomass power plants also
increased significantly in the Heat scenario compared to in the Base provides a share of the electricity for EVs. This is a consequence of
scenario, this displaces a significant amount of new coal CHP the increasing CO2 prices and the relatively large increase in fossil
capacity. To compensate for this loss in dispatchable capacity, an fuel prices compared to biomass prices over the period (cf. Table 8).
increased investment in open cycle gas turbines is observed for Norway is a large net exporter and the most optimal response to
covering peak loads. the implementation of EVs is therefore largely to reduce the export.
The results do not show any clear indications of EVs having This explains the gaps between generated electricity production
lower impact on wind power investments when competing with and the electricity demand for EVs in 2025 and 2030. Also in
heat measures. This is reflected in similar wind power investment Sweden, electricity for EVs is in 2020e2025 mainly supplied by
increases when comparing the Base and EV scenario, and the Heat cutting down export. The reduction in Swedish wind power
and EVHeat scenario, respectively. Rather, an example of a synergy production in 2020 is due to the lower wind power investments in
effect from combining EVs and heat measures is evident for the the EVHeat scenario compared to the Heat scenario for this year (cf.
case of Denmark in 2030, where wind power investments are Fig. 3). This is explained by the high Finnish wind power invest-
observed in the EVHeat scenario but not in the EV, Heat or Base ments in the EVHeat scenario in 2020, which significantly reduce
scenario. the economic potential for Swedish electricity export to Finland. As
a consequence, Swedish wind power investments in 2020 are made
4.2. Effects on electricity generation less attractive. In 2025, the difference in Swedish wind power
generation in the two scenarios is diminished and coal based power
By observing electricity generation in the EVHeat scenario production further displaced.
compared to in the Heat scenario, it can be revealed how electricity Focussing on the five Northern European countries in total, the
for EVs is produced in the optimisation (see Fig. 4). average EV is driven on a mix of mainly coal and wind power. Coal
Fig. 4 shows that the EVs generate significantly increased wind dominates in the first part of the period while wind power
power production in all five Northern European countries. This is comprises increasing shares towards 2030.
a direct consequence of the increased wind power investments
identified in Fig. 3. In some countries (Denmark, Finland, and
4.3. CO2 emissions
Sweden), the increases in wind power generation caused by EVs are
even considerably larger than domestic electricity demand for EVs
For the five countries as a whole, CO2 emissions from the power,
in the given year. Consequently, a significant amount of the increase
heat, and transport sector modelled are only slightly reduced in
in wind power generation is exported or is displacing coal based
2015e2020, 0.1e0.8%, while significant reductions are obtained in
power production in the domestic energy system (Denmark and
2025e2030, 4e7% (EVHeat scenario vs. Heat scenario, see Fig. 5a).
Sweden in 2030).
However, for several of the countries, i.e. Denmark, Germany,
and Sweden, wind power is not included in the electricity mix for 5
In addition, in 2025 a wind power capacity increase limitation of 6000 MW per
EVs until the last part of the period. As a result, coal based power five years results in identical wind power investments in the Heat and EVHeat
production provides a large share of the electricity for EVs in scenario, respectively.
a b
600 600
Transp.
CO2 (Mtonnes)
500 fuel,
400 PHEV
CO2 (Mtonnes)
Base
400 Transp.
Heat 200 fuel, ICE
300
EV
200 0 Power&
EVHeat
EVHeat
Heat
Heat
100
0
2015 2020 2025 2030 2030
Fig. 5. a) Total CO2 emissions for the modelled power, heat, and transport system in the Base, Heat, EV and EVHeat scenario b) Distribution of total CO2 emissions in 2030, divided
on sources for the Heat and EVHeat scenario.
The most important factors behind the improvement in the CO2 The cost increase is partly caused by larger investment costs per
balance over the period are 1) the increasing shares of wind power vehicle for BEVs and PHEVs compared to ICEs. Furthermore, due to
in the electricity mix for EVs, 2) the gradual improvement in the the lower annual driving of BEVs compared to ICEs, a larger amount
efficiency of EVs (cf. Table 4), and 3) the increasing shares of EVs in of BEVs is required to provide a given transport demand. This
the fleet. As illustrated for year 2030 (Fig. 5b), while EVs generate increases total investment and O&M costs for the transport sector
increased CO2 emissions from power & heat production and fuel and the cost reduction from displacing fuel use in ICEs is not enough
combustion in PHEVs, a larger CO2 reduction is obtained by dis- to compensate for this. Fig. 7b illustrates this for year 2025. The cost
placing fuel combustion in ICEs. Comparing the EV scenario with effects of EVs are, however, based on an assumed implementation of
the Heat scenario, heat measures and EVs can be seen to provide PHEVs as well as BEVs while PHEVs alone have in [14,15] been
CO2 reductions of the same magnitude. shown to provide system cost reductions. The cost increase is thus
As shown in Fig. 6, also on national level the EVs bring signifi- likely due to the forced implementation of BEVs. This is confirmed
cant CO2 emission reductions in 2025e2030 and in some countries by the fact that BEVs have also in [15] been shown to result in cost
already from 2020. The only exception is the case of Finland in 2025 increases. It should also be noted that the possibility for using EVs
where EVs generate coal power production significantly larger than for providing regulating power and power reserves is not included.
the electricity demand for EVs, leading to a CO2 emission increase. Likewise, a socio-economic valuation of reduced local air and noise
The Norwegian and Swedish CO2 emission reductions in pollution has not been performed. As shown, the cost increase is
2025e2030 and Danish emission reductions in 2030 are high; highest in 2020 (3.9%) and then lower in 2025 (1.2%) and 2030
reaching 17e40%. This is partly explained by the fact that electricity (0.8%); reflecting the influence of expected technical and economic
for EVs in these cases is covered purely by renewable energy improvements of EVs over the period (cf. Table 4).
sources or export reductions. In some cases (Denmark and Relating the cost increases (excluding CO2 quota costs) to the
Sweden), further displacement of coal based power production also CO2 emission reductions, the average CO2 reduction costs for EVs
occurs (cf. Fig. 4). Moreover, due to large shares of hydro power can be estimated. This reveals that CO2 reduction costs are reduced
and/or nuclear in the electricity supply (cf. Fig. 2), CO2 emissions manifold over the period, from very high levels of around 7100 V/
from the Swedish and Norwegian power sector are low. As a result, tonne in 2015 and 1500 V/tonne in 2020, to 140 V/tonne in 2025
CO2 emission reductions obtained in the transport sector lead to and 80 V/tonne in 2030. However, even in 2030, the CO2 reductions
relatively large reductions for the three sectors as a whole. costs for EVs are high compared to the expected CO2 price level of
around 39 V/ton [47]. Comparing the EV scenario with the Heat
4.4. System costs scenario, it can be seen that while EVs result in cost increases
(0.8e3.9%) heat measures reduce total costs (0.004e0.9%).
The analysis shows that the assumed large-scale implementa-
tion of EVs results in an increase in total system costs for the 4.5. Sensitivity analysis
Northern European countries as a whole; around 1.5e7.1 V bill./yr
depending on the year, corresponding to increases of 0.8e3.9% Variants of the Heat and EVHeat scenarios are set up assuming
(for EVHeat vs. Heat scenario, see Fig. 7). low/high fuel prices and low/high CO2 prices, respectively.
Comparing the two scenarios makes it possible to analyse the
10% effects of EVs under different price conditions:
5%
Change in CO2 emissions
0% 2015 Fuel prices: set to low at $80/barrel in 2015 (corresponding to

-5% Denmark Finland Germany Norway Sweden 2020 2010 level) increasing linearly to $90/barrel in 2030 and at high
-10% 2025 increasing linearly from $80/barrel in 2010 to $95/barrel in
-15% 2015 and $140/barrel in 2030. Ratios between prices on
2030
-20% different fuels are kept constant and are based on [47].
-25% CO2 prices: set to low at 15 V/tonne in 2015 increasing linearly
-30%
to 20 V/tonne in 2030 and at high increasing linearly from
-35%
14 V/tonne in 2010 to 26 V/tonne in 2015 and 60 V/tonne in
-40%
-45% 2030.
Fig. 6. Change in national CO2 emissions from the power, heat and transport sector These analyses show that also at low/high fuel and low/high CO2
modelled when implementing electric vehicles (EVHeat compared to Heat scenario). prices, EVs facilitate increased wind power investments and
a b
200 200 Transport,
180 Inv.+ O&M
150
Bill. €/yr
160 Base Transport,
140 CO2 from fuel
Heat 100
Bill. €/yr
120
Transport,
100 EV 50 fuel
80
EVHeat Power&Heat
60 0
EVHeat
Heat
40
20
0
2015 2020 2025 2030 2025
Fig. 7. a) Total annualised costs for the power, heat and transport systems modelled, for the Base, Heat, EV, and EVHeat scenario. b) Distribution of total costs in 2025 divided on
sources for the Heat and EVHeat scenario.
reduced need for new coal/natural gas production capacities in years earlier than at the fuel prices in the main scenario. In Finland,
several countries. However, the changes in power system invest- the higher fuel prices lead to EVs facilitating increased wind power
ments and electricity generation caused by EVs are found to be investments from year 2015, i.e. also earlier than originally. For
sensitive to the development in fuel and CO2 prices. This is illus- Germany, the electricity demand induced by EVs is still largely
trated in Fig. 8 using the Danish case as an example. comprised by coal power and only moderate increases in wind
Under the low fuel price conditions, in Denmark, wind power is power generation are observed in the last part of the period.
not included in the electricity mix for EVs and in Germany not At the high CO2 price conditions, EVs facilitate considerably
before 2030. In Finland, Sweden, and Norway, where wind power increased Danish wind power investments in 2025 and 2030 and
investments are generally more attractive, lower fuel prices have increased Finnish wind power investments from 2015. In addition,
the effect of increasing the relative importance of EVs in facilitating the higher CO2 prices make investment in wood CHP extraction
wind power investments. As an example, EVs facilitate increased plants attractive in several countries. Thus, wind power largely
wind power investments in Norway already at the low vehicle fleet competes with wood based power production rather than coal
shares of 2.5% in 2015, i.e. five years earlier than at the fuel prices in based power production. For the Danish case, this is reflected in
the main scenario. increased wind power generation in 2025 displacing wood based
At the low CO2 price conditions, electricity demand for EVs is in power production (cf. Fig. 8). In Germany, the increase in electricity
Germany met by increased coal based power production. In production for EVs is in 2025 partly and in 2030 solely based on
Denmark, the lower CO2 prices have the effect of reducing onshore renewable energy sources; mainly wood and secondarily wind.
wind power investments in the Heat scenario below the onshore However, coal based power still provides a large share of the
potential for Western Denmark (3500 MW). This facilitates an electricity for EVs in Germany and Denmark, particularly in the first
increase in onshore wind power investments in 2025 when adding part of the period.
EVs to the system. In Finland and Sweden, EVs do not generate large Further sensitivity analyses have been set up investigating the
increases in wind power investments before year 2030. As a result, influence of other boundary conditions. As shown in Table 7,
electricity demand for EVs in Finland and Sweden is to a large a significant expansion of Norwegian hydro power generation from
extent met by coal based power production. In Norway, the lower 2010 to 2030 is included in the model (20 TWh). A sensitivity
CO2 prices have the effect of increasing the impact of EVs on wind scenario has been set up, assuming that only 50% of this increase is
power investments, leading to increased wind power generation realised. In this scenario, due to reduced electricity export from
from 2015. Norway to Denmark, EVs facilitate increased Danish wind power
Under the high fuel price conditions, EVs generate considerable investments from year 2025, i.e. five years earlier than in the main
increases in Danish wind power generation from year 2025, i.e. five scenario. Moreover, Danish wind power investments generated by
Low fuel prices High fuel prices Less Norwegian hydro power
15 15 15
10 10 10
TWh
TWh
TWh
5 5 5
0 0 0
2015 2020 2025 2030 2015 2020 2025 2030 2015 2020 2025 2030
-5 -5 -5
Low CO2 prices High CO2 prices Main scenario
15 15 15
10 10 10
TWh
TWh
TWh
5 5 5
0 0 0
2015 2020 2025 2030 2015 2020 2025 2030 2015 2020 2025 2030
-5 -5 -5
Coal Wind Wood Straw Electricity for EVs
Fig. 8. Changes in Danish electricity generation due to implementation of electric vehicles, illustrated for different scenarios.
30% Main scenario

25% Low fuel prices
CO2 reduction
20% High fuel prices
15% Low CO2 prices
10% High CO2 prices
5% Wind targets, medium
Less Norwegian hydro power
0%
2015 2020 2025 2030 German nuclear phase out
Fig. 9. CO2 emission reductions in the five countries as a whole due to implementation of electric vehicles, illustrated for different scenarios.
EVs in 2030 are increased significantly (from around 1600 MW to is explained by the fact that high fuel prices result in larger cost
2600 MW). For Norway, the impact of EVs on wind power invest- benefits from fuel savings caused by the high efficiency of EVs.
ments is reduced in the short term and increased in the long term. In all sensitivity scenarios, average CO2 reduction costs for EVs
Another sensitivity analysis has been set up assuming a phase are reduced manifold towards 2030. However, the average CO2
out of nuclear power in Germany according to [49] leading to reductions cost in a given year varies significantly among the
a reduction from 17.2 GW in 2010, to 10.7 GW in 2015, 8.1 GW in sensitivity scenarios. Overall, average CO2 reduction costs develop
2020, and 0 MW in 2022. The main effect of this is that German coal from high levels of around 1600e11,900 V/tonne in 2015, to
and wind power investments are increased over the period. 1400e2300 V/tonne in 2020, 115e359 V/tonne in 2025 and ()16
However, wind power investments in Germany are still not to 168 V/tonne in 2030 (negative number for high fuel price
observed before 2025. Furthermore, German wind power invest- scenario due to cost reduction).
ments become highly attractive in the last part of the period even
without the implementation of EVs, leading to only small increases 5. Conclusion
in wind power investments when adding EVs to the system. A
phase out of German nuclear power represents a reduction in In this study, it is analysed how a gradual large-scale imple-
nuclear power generation of 40, 63, and 129 TWh/yr in 2015, 2020, mentation of PHEVs and BEVs towards 2030 would affect the power
and 2025e2030, respectively, i.e. a considerable change in the systems of five Northern European countries; Denmark, Finland,
German power system. Nevertheless, it shows to have no signifi- Germany, Norway, and Sweden. This is done using a model opti-
cant influence on the effect of EVs in the four other countries. This mising power system investments and operation and applying
indicates that the fact that not all German electricity exchange socio-economic costs, i.e. excluding taxes, subsidies, and tariffs.
options are modelled; is also likely to have little influence on the The results reveal that if charged/discharged intelligently EVs
results for the other countries. can due to vehicle-to-grid capability contribute in meeting peak
Finally, a sensitivity analysis has been made where future wind power demand and thus, reduce the need for new coal/natural gas
power capacities in 2015, 2020, 2025, and 2030 are restricted to power production capacities in several of the countries. Further-
meet national wind targets in [40] as a minimum level. As more, EVs facilitate significantly increased wind power invest-
a consequence of this restriction, effects of EVs facilitating addi- ments in all of the countries analysed. For the fuel and CO2 price
tional wind power investments are not observed in Denmark and developments assumed in main scenarios, increases in wind
significant increases in Finnish wind power investments due to EVs power investments are first seen in 2020 (in Norway and Finland)
are not apparent until 2030. This illustrates that when forcing in when EVs are assumed to comprise 15% of the private passenger
future wind power capacities based on political targets it becomes vehicle fleet. In sensitivity scenarios with low/high fuel prices or
difficult to analyse how EVs can influence the investments. low CO2 prices, respectively, EVs facilitate increased wind power
Nevertheless, this sensitivity scenario might give a more realistic investments even at the low vehicle fleet shares of 2.5% assumed
picture on the development in wind power capacities over the in 2015 (in Norway and Finland). As such, depending on the
period and thus, how EVs can influence operation costs of future country and fuel/CO2 prices, EVs can in a transition period even at
power systems. However, this sensitivity analysis shows increases early stages support the integration of wind power into power
in total costs of 0.7e3.9% depending on the year, i.e. very close to systems.
the cost increases identified when wind power investments are The study shows that wind power would likely provide a large
identified without wind target restrictions (0.8e3.9%). The expla- share of the electricity for EVs towards 2030 in several of the
nation is that total costs of the power and district heating systems countries analysed. However, in the optimisation, for several
in the five countries only comprise a moderate part compared to countries (Denmark, Germany, and Sweden), wind power does not
the investment, O&M and fuel costs of the transport system (cf. contribute in providing electricity for EVs before 2025/2030 and in
Fig. 7b). some sensitivity scenarios not at all. As a result, electricity demand
Among all sensitivity scenarios, CO2 emission reductions for EVs is in many cases largely met by an increase in coal based
provided by EVs for the five countries as a whole in 2025 and 2030 power production; particularly for the case of Denmark and
are lowest for the low CO2 price scenario, 1% and 3% respectively, Germany in the first part of the period. As such, this would likely be
and highest for the high CO2 price scenario, namely 6% and 28%, the outcome if an introduction of EVs is not followed up by
respectively (see Fig. 9). As such, CO2 reductions provided by EVs economic support for renewable energy technologies (other than
are particularly sensitive to CO2 price variations. CO2 quotas).
The cost balance for EVs is worst in the low fuel price scenario The effects of EVs on the power system vary significantly from
with increases of 2.4% and 3.1% in 2025 and 2030, respectively. The country to country and are found to be sensitive to variations in fuel
high fuel price scenario represents the best cost balance for EVs and CO2 prices. In addition, the extent with which Norwegian hydro
with a cost increase of 0.5% in 2025 and a cost reduction of 0.9% in power expansions and a phase out of German nuclear power is
2030. That high fuel prices are beneficial for the cost effects for EVs realised have an influence on the results for some countries. It is
therefore difficult to draw more general conclusions regarding how planning model. In: Proceedings of the IEEE Energy 2030 Conference, Atlanta,
USA, 2008.
electricity for EVs would be produced over the period.
[14] Juul N, Meibom P. Optimal configuration of an integrated power and transport
The sensitivity towards CO2 price variations is expressed in CO2 system. Energy 2011;36(5):3523e30.
emission reductions for the five countries as a whole varying from [15] Juul N, Meibom P. Road transport and power system scenarios for Northern
1% to 6% in 2025 and from 3% to 28% in 2030, depending on whether Europe in 2030. Appl Energy 2012;92(4):573e82.
[16] Kempton W, Tomi c J. Vehicle-to-grid power implementation: from stabilizing
low or high CO2 price levels are assumed. Changes in total costs due the grid to supporting large-scale renewable energy. J Power Sources 2005;
to the EV implementation vary from 0.5% to 2.4% in 2025 and from 144:280e94.
()0.9% to 3.1% in 2030, depending on whether low or high fuel [17] Williams BV, Kurani KS. Commercializing light-duty plug-in/plug-out
hydrogen-fuel-cell vehicles: mobile electricity technologies and opportuni-
price development is assumed. Average CO2 reduction costs for the ties. J Power Sources 2007;166:549e66.
assumed implementation of PHEVs and BEVs are reduced manifold [18] Sovacool BK, Hirsh RF. Beyond batteries: an examination of the benefits and
towards 2030. Nevertheless, the CO2 reduction costs are in most barriers to plug-in hybrid electric vehicles (PHEVs) and a vehicle-to-grid
(V2G) transition. Energ Policy 2009;37(3):1095e103.
scenarios high compared to expected CO2 price levels. As such, the [19] Hidrue MK, Parsons GR, Kempton W, Gardner MP. Willingness to pay for
CO2 reductions achieved with the assumed large-scale imple- electric vehicles and their attributes. Resour Energy Econ 2011;33(9):
mentation of PHEVs and BEVs are rather costly, particularly in the 686e705.
[20] Smith WJ. Can EV (electric vehicles) address Ireland’s CO2 emissions from
short term. However, an implementation of only PHEVs might have transport? Energy 2010;35(12):4514e21.
shown better cost effectiveness since PHEVs previously have been [21] Green Car Congress. Growing number of EU countries levying CO2 taxes on
found to bring cost reductions (for 2030) [15]. It should also be cars incentivizing plug-ins. Available at: http://www.greencarcongress.com/
2010/04/acea-tax-20100421.html#more; 2010 [accessed 12.04.12].
mentioned that EVs have the benefit of increasing security of supply
[22] Ravn H, Hindsberger M, Petersen M, Schmidt R, Bøg R, Grohnheit PE, et al.
by reducing the transport sector’s dependency on oil. Balmorel: a model for analyses of the electricity and CHP markets in the Baltic
The effects of EVs have been compared with the effects of Sea region. Ballerup, Denmark. Available at: http://www.balmorel.com/;
allowing investment in heat measures, i.e. heat pumps, electric 2001.
[23] Nielsen LH, Ravn H, Larsen E, Meibom P, Hansen AB, Hedegaard K, et al.
boilers, and heat storages, in the district heating system, repre- Electricity for road transport, flexible power systems and wind power. Tech.
senting alternative ways of increasing system flexibility. The Rep. Risø-R-1804 (EN). Roskilde, Denmark: Risø DTU National Laboratory for
comparison shows that in terms of increasing socio-economically Sustainable Energy; 2011.
[24] Juul N. Modelling and analysis of distributed energy systems with respect to
optimal wind power investments, and reducing CO2 emissions, sustainable energy e focus on electric drive vehicles. PhD thesis. Lyngby,
heat measures can have effects of the same magnitude as EVs. Denmark: DTU Management Engineering, Technical University of Denmark,
However, while the assumed implementation of EVs result in costs 2011.
[25] Electric Power Research Institute. Environmental assessment of plug-in
increases heat measures reduce total costs. The results do not show hybrid electric vehicles, volume 1: nationwide greenhouse gas emissions.
any clear indications of EVs having lower impact on wind power California, USA: Electric Power Research Institute. Available at: http://
investments when competing with heat measures. miastrada.com/yahoo_site_admin/assets/docs/epriVolume1R2.36180810.pdf;
2007.
[26] International Energy Agency. Technology roadmap e electric and plug-in
Acknowledgements hybrid electric vehicles. International Energy Agency. Available at: http://
www.iea.org/papers/2009/EV_PHEV_Roadmap.pdf; 2009.
[27] European Commission, Directorate-General for energy and transport. Euro-
The study is based mainly on the project “Electricity for road pean energy and transport e trends to 2030 d update 2007. European
transport e flexible power systems and wind power” financed by Commission. Available at: http://ec.europa.eu/dgs/energy_transport/figures/
trends_2030_update_2007/energy_transport_trends_2030_update_2007_en.
the Danish Energy Authority (EFP07-II) [23]. Thanks are given to
pdf; 2008.
the reviewers for valuable comments on the article. [28] Danish Energy Authority. Denmark’s energy sector projections, April 2010
[Danmarks energifremskrivning, april 2010]. Copenhagen, Denmark: Danish
Energy Authority. Available at: http://www.ens.dk/Documents/Netboghandel
References %20-%20publikationer/2010/danmarks_energifremskrivning.pdf; 2010 [In
Danish].
[1] Kempton W, Tomi c J. Vehicle-to-grid power fundamentals: calculating [29] Fosgerau M, Brems C, Jensen C, Pilegaard N, Holmblad M, Kveiborg O, et al.
capacity and net revenue. J Power Sources 2005;144:268e79. Long term projection of road traffic: indication of future problem areas,
[2] Kempton W, Tomi c J, Letendre S, Brooks A, Lipman T. Vehicle-to-grid power: background report [Langsigtet fremskrivning af vejtrafik: indikation af
battery, hybrid, and fuel cell vehicles as resources for distributed electric fremtidige problemområder, baggrundsrapport]. Lyngby, Denmark: Danish
power in California. Tech. Rep. UCD-ITS-RR-01e03; 2001. Transport Research Institute. Available at: http://www.transport.dtu.dk/
[3] Kempton W, Kubo T. Electric-drive vehicles for peak power in Japan. Energ Nyheder/Nyhedsarkiv/Arkiv,%20DTF.aspx?guid¼{559A9B5A-6527-4728-
Policy 2000;28:9e18. AB83-FA7A4024B5E3}; 2007 [In Danish].
[4] Tomic J, Kempton W. Using fleets of electric-drive vehicles for grid support. [30] COWI. Technology assessment of alternative fuels for the transportation
J Power Sources 2008;168:459e68. sector [Teknologivurdering af alternative drivmidler til transportsektoren].
[5] McCarthy RW, Yang C, Ogden J. Impacts of electric-drive vehicles on Cal- Copenhagen, Denmark: Danish Energy Authority. Available at: http://www.
ifornia’s energy system. Tech. Rep. UCD-ITS-RP-08e24; 2008. ens.dk/da-dk/klimaogco2/transport/alternativedrivmidler/documents/cowi%
[6] Sioshansi R, Denholm P. The value of plug-in hybrid electric vehicles as grid 20rapport%20teknologivurdering%20af%20alternative%20drivmidler%20i%
resources. Energy J 2010;31:1e23. 20transportsektoren.pdf; 2007 [In Danish].
[7] Soares B, Borba MC, Szklo A, Schaeffer R. Plug-in hybrid electric vehicles as [31] Risø DTU. Risø energy report 8: the intelligent energy system infrastructure
a way to maximize the integration of variable renewable energy in power for the future. Roskilde, Denmark: Risø DTU; 2009.
systems: the case of wind generation in northeastern Brazil. Energy 2012; [32] The Danish Commission on Climate Change Policy. Documentation of the
37(1):469e81. report by the danish commission on climate change policy: green energy, the
[8] Denholm P, Short W. An evaluation of utility system impacts and benefits of road towards a danish energy system without fossil fuels [Dokumenta-
optimally dispatched plug-in hybrid electric vehicles. Tech. Rep. NREL/TP- tionsdelen til klimakommissionens samlede rapport e grøn energi e vejen
620e40293. NREL; 2006. mod et dansk energisystem uden fossile brændsler]. The Danish Commission
[9] Kiviluoma J, Meibom P. Influence of wind power, plug-in electric vehicles, and on Climate Change Policy. Available at: http://www.klimakommissionen.dk/
heat storages on power system investments. Energy 2010;35:1244e55. da-DK/OmKlimakommissionen/Klimakommissionensrapport/Sider/Forside.
[10] Kiviluoma J, Meibom P. Methodology for modelling plug-in electric vehicles in aspx; 2010 [In Danish].
the power system and cost estimates for a system with either smart or dumb [33] Danish Transport Research Institute. Investigation of transport habits
electric vehicles. Energy 2011;36(3):1758e67. [Transportvaneundersøgelsen]. Lyngby, Denmark: Danish Transport Research
[11] Lund H, Kempton W. Integration of renewable energy into the transport and Institute. Available at: http://www.dtu.dk/centre/modelcenter/TU.aspx; 2006
electricity sectors through V2G. Energ Policy 2008;36:3578e87. [In Danish].
[12] Kristoffersen TK, Capion K, Meibom P. Optimal charging of electric drive [34] Swedish Energy Agency. Energy in Sweden: facts and figures 2010. Bromma,
vehicles in a market environment. Appl Energ 2011;88(5):1940e8. Sweden: Swedish Energy Agency. Available at: http://webbshop.cm.se/
[13] Ibáñez E, McCalley J, Aliprantis D, Brown R, Gkritza K, Somani A, et al. National System/TemplateView.aspx?p¼Energimyndigheten&view¼default&cat¼/
energy and transportation systems: interdependencies within a long term Broschyrer&id¼e0a2619a83294099a16519a0b5edd26f; 2010.
[35] World Nuclear Association. Nuclear power in Finland. Available at: http:// [45] The Nuclear N-former. Finland to cut Russia’s energy imports by doubling
world-nuclear.org/info/inf76.html; [accessed 09.10.11]. atomic power output. Available at: http://www.nuclearcounterfeit.com/?
[36] Danish Energy Authority. Technology data for energy plants. Copenhagen, p¼2252; 2011 [accessed 10.10.11].
Denmark: Danish Energy Authority. Available at: http://www.ens.dk/ [46] European Network of Transmission System Operators for Electricity. Data
Documents/Netboghandel%20-%20publikationer/2010/Technology_data_for_ portal, exchange data, detailed electricity exchange. Available at: https://
energy_plants.pdf; 2010. www.entsoe.eu/resources/data-portal/exchange/; [accessed 08.10.11].
[37] International Energy Agency. Tackling investment challenges in power [47] Danish Energy Authority. Preconditions for socio-economic assessments
generation. Paris, France: International Energy Agency. Available at: http:// in the field of energy [Forudsætninger for samfundsøkonomiske analyser
www.iea.org/textbase/nppdf/free/2007/tackling_investment.pdf; 2007. på energiområdet]. Copenhagen, Denmark: Danish Energy Authority.
[38] Morthorst PE. Heat pumps and low temperature district heating. Report from Available at: http://www.ens.dk/da-DK/Info/TalOgKort/Fremskrivninger/
two workshops. Part of the reporting from the project efficient district heating beregningsforudsatninger/Documents/Br%C3%A6ndselspriser%20foruds%C3%
in the future energy system [Varmepumper og lavtemperaturfjernvarme. A6tninger%20notat%20april%202010.pdf; 2010 [In Danish].
Rapportering fra to workshops. Delrapportering fra EFP-projekt Effektiv [48] Münster M, Meibom P. Optimization of use of waste in the future energy
fjernvarme i fremtidens energisystem]. Tech. Rep. Risø-R-1234(DA). Roskilde, system. Energy 2011;36(3):1612e22.
Denmark: Risø DTU; 2008 [In Danish]. [49] German Ministry of Justice. Act on the peaceful uses of nuclear energy and the
[39] Ellehauge K, Pedersen TE. Solar heat storages in district heating networks. protection against its hazards x approval of power plants [Gesetz über die
Fredericia, Denmark: Energinet.dk. Available at: http://www.elle-kilde.dk/ friedliche verwendung der kernenergie und den schutz gegen ihre gefahren x
downloads/Solar_heat_storages_in_district_heating_networks_02-08-2007. 7 genehmigung von anlagen]. Germany Ministry of Justice. Available at:
pdf; 2007. http://www.gesetze-im-internet.de/atg/__7.html; 2011 [In German].
[40] Lemström B, Uski-Joutsenvuo S, Holttinen H, Cutululis NA, Sørensen P, [50] Vogstad K. Utilising the complementary characteristics of wind power and
Warland L, et al. Effects of increasing wind power penetration of the power flows hydropower through coordinated hydro production scheduling using the
in the European grids. Available at: http://www.trade-wind.eu/fileadmin/ EMPS model. Trondheim, Norway: Department of Electrical Engineering;
documents/publications/D5-1_Effects_of_increasing_wind_power.pdf; 2008. 2000.
[41] Mathiesen BV, Lund H, Karlsson K. The IDA Climate Plan 2050: background [51] The Swedish Government. The Swedish national action plan for the promo-
report. Copenhagen, Denmark: The Danish Society of Engineers. Available at: tion of the use of renewable energy in accordance with Directive 2009/28/EC
http://ida.dk/omida/laesesalen/Documents/UK-Future%20Climates% and the commission decision of 30.06.2009. The Swedish Government.
20Background%20Report.pdf; 2009. Available at: http://ec.europa.eu/energy/renewables/transparency_platform/
[42] Energinet.dk. Analysis preconditions 2010e2020, 2025 and 2030 [Analy- action_plan_en.htm; 2010.
seforudsætninger 2010-2020, 2025 og 2030]. Fredericia, Denmark: Ener- [52] Fraunhofer Institute for Wind Energy and Energy System Technology. Full
ginet.dk; 2010 [In Danish]. load hours onshore-offshore. Available at: http://windmonitor.iwes.
[43] Ea Energy Analyses. Sustainable energy scenarios: energy perspectives for the fraunhofer.de/windwebdad/www_reisi_page_new.show_page?page_
Baltic Sea region. Copenhagen, Denmark: Ea Energy Analyses. Available at: nr¼470&lang¼en&owa¼&owa_own_header¼0; [accessed 12.10.11].
http://www.bsssc.com/upload/dokumenty/f_17.pdf; 2009. [53] Federal Republic of Germany. National renewable energy action plan in
[44] Nord Pool Spot. Historical market data, exchange, hourly. Available at: http:// accordance with Directive 2009/28/EC on the promotion of the use of energy
www.nordpoolspot.com/Market-data1/Downloads/Historical-Data- from renewable sources. Available at: http://ec.europa.eu/energy/renewables/
Download1/Data-Download-Page/; [accessed 10.10.11]. transparency_platform/action_plan_en.htm; 2010.
Paper III
Wind power integration using individual heat pumps –

Analysis of different heat storage options
Karsten Hedegaard, Brian Vad Mathiesen, Henrik Lund, Per Heiselberg.

Energy 47 (2012) 284e293
Contents lists available at SciVerse ScienceDirect
Energy
journal homepage: www.elsevier.com/locate/energy
Wind power integration using individual heat pumps e Analysis of different heat
storage options
Karsten Hedegaard a, *, Brian Vad Mathiesen b, Henrik Lund c, Per Heiselberg d
a
Systems Analysis Division, DTU Management Engineering, Frederiksborgvej 399, DK-4000 Roskilde, Denmark
b
Department of Development and Planning, Aalborg University, A.C. Meyers Vænge 15, DK-2450 Copenhagen SV, Denmark
c
Department of Development and Planning, Aalborg University, Vestre Havnepromenade 9, DK-9000 Aalborg, Denmark
d
Department of Civil Engineering, Aalborg University, Sohngaardsholmsvej 57, DK-9000 Aalborg, Denmark
Article history: Significant installations of individual heat pumps are expected in future energy systems due to their
Received 3 March 2012 economic competitiveness. This case study of the Danish energy system in 2020 with 50% wind power
Received in revised form shows that individual heat pumps and heat storages can contribute to the integration of wind power.
6 July 2012
Heat accumulation tanks and passive heat storage in the construction are investigated as two alternative
Accepted 11 September 2012
Available online 13 October 2012
storage options in terms of their ability to increase wind power utilisation and to provide cost-effective
fuel savings. Results show that passive heat storage can enable equivalent to larger reductions in excess
electricity production and fuel consumption than heat accumulation tanks. Moreover, passive heat
Keywords:
Household heat pumps
storage is found to be significantly more cost-effective than heat accumulation tanks. In terms of
Flexible electricity demand reducing fuel consumption of the energy system, the installation of heat pumps is the most important
Passive heat storage step. Adding heat storages only moderately reduces the fuel consumption. Model development has been
Heat accumulation tanks made to facilitate a technical optimisation of individual heat pumps and heat storages in integration with
Model the energy system.
Energy system Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction management can also contribute to the flexible operation of power

systems with high wind power penetrations [7e9]. In this regard,
Wind power is considered as a key renewable energy tech- small (electric) compression heat pumps in households (in this
nology in future energy systems. However, wind power is highly study referred to as individual heat pumps) could also contribute.
variable, which challenges a fuel-efficient and cost-effective inte- As such, ground heat pumps and air/water heat pumps can be
gration of large amounts of wind power. In systems with high wind operated flexibly by storing heat in the central heating system and
penetration, forced electricity export can thus occur when wind in the construction. Energy system analyses show that in relation to
power is high and electricity demand low. In the near future costs, fuel consumption, and CO2 emissions, individual heat pumps
significant expansion of wind power is planned in many European together with district heating form the best heat supply solutions
countries [1]. The Danish energy system is characterised by a large [10]. This is found to be valid for present systems that are mainly
share of wind power, around 20% on annual basis [2], and this share based on fossil fuels, and for potential future systems that are fully
is expected to increase to 50% in 2020. The Danish case thus forms based on renewable energy sources. The cost-competitiveness of
an interesting case for analysing challenges of integrating wind heat pumps is confirmed in Ref. [11], where individual heat pumps,
power that may be faced by other countries in the coming years. in socio-economic cost optimisations outcompete all other indi-
Studies suggest that flexible technologies such as large heat vidual heating technologies. This is found for an energy system in
pumps, electric boilers, and heat storages in (combined heat and 2025 with 30% renewable energy and for a system in 2050 with 90%
power) CHP systems, and electric vehicles can play a significant role CO2 emission reduction. In the future, individual heat pumps will
in facilitating the integration of wind power [3e6]. Demand side thus likely represent a significant electricity demand that could be
made flexible. Investments in heat storage and/or control system
units are however required in order to enable significant flexibility
and different heat storage options exist. From a socio-economic
* Corresponding author. Systems Analysis Division, DTU Management Engi-
perspective, it is therefore relevant to investigate the possible
neering, Frederiksborgvej 399, P.O. Box 49, Build.130, DK-4000 Roskilde, Denmark.
Tel.: þ45 4677 5158; fax: þ45 4677 5199. benefits that different heat storages for individual heat pumps can
E-mail addresses: [email protected], [email protected] (K. Hedegaard). bring to the energy system.
http://dx.doi.org/10.1016/j.energy.2012.09.030
the energy system analyses tool EnergyPLAN, applying an hourly

Nomenclature time resolution and covering a full year. Analyses are made for the
aggregated building stock, for different representative building
Afloor heated floor area (m2)1 categories, and for a range of different reference systems, covering
c active heat capacity depending on the type of realisation of heat savings, different amounts of wind power, and
building (Wh/m2 floor area/K) installation of other flexible technologies. Further development of
CF heat pump capacity factor, representing heat pump the model has been made to make it possible to represent flexible
capacity relative to maximum heat demand heat pump operation using the construction as thermal storage. In
COP coefficient of performance the process of reviewing state-of the-art heat pump systems,
EB(t) heat production from electric boiler in h t (Wh) experts and producers of small heat pumps have been contacted to
HD(t) heat demand in h t (Wh) provide a solid ground for the analyses. The study is aimed at
HDmax highest hourly heat demand over the year (Wh) decision makers and researchers dealing with heat pumps in an
HPel(t) electricity consumption for heat pump operation in energy system context.
h t (Wh) The article is structure in the following way. Section 2 includes
HPheat(t) heat production from heat pump in h t (Wh) the methodology applied, while Section 3 presents the modelling of
HWD hot water demand (Wh) heat pumps and storage options. Section 4 covers the results
Load(t) loading of heat storage in h t (Wh) including sensitivity analyses. Finally, a conclusion is given in
Unload(t)unloading of heat storage in h t (Wh) Section 5 and suggestions for future work in Section 6.
Stor(t) heat storage level in h t (Wh)
Storcap heat storage capacity (Wh)
DTin temperature change of the indoor air allowed when 2. Methodology
utilising passive heat storage ( C, K)
DTpas temperature change of the construction utilised for The potentials of individual heat pumps and different heat
passive heat storage ( C, K) storage options are analysed in the context of a large-scale instal-
U heat transfer coefficient between construction and lation in the system. Heat pumps are assumed installed where they
indoor air (W/m2 floor area/K) are considered particularly relevant in a near term perspective, i.e.
in houses presently heated with oil boilers and/or electric heating.
Only existing buildings are considered as these will for many years
comprise the majority of all buildings, i.e. 85e90% in 2030 for the
Among existing studies, Ref. [12] discusses the possibilities and Danish case [10]. The analyses focus on detached houses as these
challenges of individual heat pumps and thermal storage options in comprise by far the largest heat demand of houses in non-district
terms of increasing system flexibility. A group of studies analyse heating areas in Denmark (close to 90%2) [22].
possibilities for flexible heat pump operation on a single building Only heat pumps capable of storing heat in the central heating
level, e.g. [13e16]. A more large-scale perspective is applied in system and in the construction are considered, i.e. ground heat
Ref. [17], where a simulation of a 2 week period is performed of pumps and air/water heat pumps. For the large majority of existing
a high wind power system with 3000 households, of which 1500 houses with central heating, the heating system will mainly be
are supplied with heat pumps using buffers for space heating and based on radiators. Installation of floor heating in existing houses is
hot water. Only a few studies deal with individual heat pumps in associated with high costs and is therefore typically only carried out
energy system models on national scale, e.g. [10,11,18,19]. However, in connection with large renovation projects. Against this back-
the studies do not analyse the possibilities and effects of flexible ground, only radiator systems are considered. In some houses,
operation of individual heat pumps. Ref. [20] presents a method- radiator substitution and/or renovation of the building envelope
ology for assessing the potential flexibility that a building stock would be required prior to heat pump installation. This aspect is
equipped with heat pumps can offer to power systems with however not important for the focus of this study.
significant penetration of wind power. This methodology is applied Ground heat pumps and air/water heat pumps are today typi-
to a case study of the German energy system in 2020 and 2030 with cally installed in combination with a hot water tank (typically
high renewable energy penetration levels, however, without pre- around 150e200 L) and a small buffer tank (typically around 40e
senting quantitative results. Ref. [21] touches upon the effect of 80 L) connected to the central heating system. The hot water tank
individual heat pumps in an energy system perspective and also ensures that sufficient hot water is available during the day and is
includes scenarios with and without assumed flexibility. However, loaded continuously. The buffer tank ensures that the number of
flexible heat pump operation is not modelled but merely repre- heat pump start-ups can be minimised, thereby enabling better
sented as a modified fixed demand profile based on assuming that operating conditions and improving COP and the technical life-time
heat pump operation can be distributed within the day. [23,24]. The buffer tank is not large enough to enable shifting
No existing studies have been found, which model and analyse operation from one hour to the other. However, such flexibility can
the energy system effects of individual heat pumps and different be obtained through different options, e.g. by investing in a heat
heat storage options. This is therefore the subject of this paper. accumulation tank connected to the central heating circuit [24].
Using a case of the Danish energy system in 2020 with a wind Given space requirements, it is realistically to insert an accumula-
power share of around 50%, the potential of individual heat pumps tion of up to around 1000 L in each house [23]. Alternatively, if
and heat storages is investigated in terms of their ability to increase allowing a given indoor air temperature variation, heat can be
wind power utilisation and provide cost-effective fuel savings. In stored passively in the construction. The analyses cover different
this regard, the potentials of heat accumulation tanks and passive levels of passive heat storage and different sizes of heat accumu-
heat storage in the construction are compared as two alternative lation tanks (300, 500, and 1000 L). Based on a need to limit the
heat storage options. Several energy system analyses are made in analyses and considering that space heating normally comprises
1 2
All floor areas in the study refer to heated floor areas. When including farm houses.
Table 1
Characteristics of the buildings stock in which heat pumps are assumed installed.
Existing detached houses Unit Weighted Construction period Ref.

average
1850e1960d 1961e1972 1973e1978 1979e1998
or total
Heat demand per floor area kWh/m2 yr 172 210 143 117 90 [25]
Hot water share of annual pct. 16% 15% 15% 18% 27% [26]
heat demand
Economically feasible space pct. 34% 39% 20% 27% 28% [26]
heat savings potential
(pct. of space heating)
Passive heat storage cap.a kWh/house 18e72 18e72 18e72 18e72 18e72
Accum. tank storage kWh/house 5e17 5e17 5e17 5e17 5e17
cap., 300e1000 Lb
COPc e 2.7 2.8 2.8 2.7 2.5
Heat demand converted TWh/yr 5.0 3.5 0.78 0.44 0.28 [2,22,31]
to heat pumps
a
Assuming passive heat storage capacity of 60e120 Wh/K/m2 [27,28], an allowed indoor air temperature variation of H1e2 C [30] and an average house size of 151 m2
[22].
b
Assuming 15 C temperature difference in the radiator system corresponding to a typical configuration of heat pump radiator systems [23,24,28] prioritising a reasonable
COP.
c
COP (coefficient of performance) estimated based on Refs. [10,25] depending on hot water share. Based on Ref. [22], ground heat pumps are assumed to cover 25% of the
heat demand converted to heat pumps, and air/water heat pumps the remaining 75%. The COPs apply for use of heat pumps in the Danish climate, which is temperate with
outdoor air temperatures typically ranging from around 5 C to 15 C in the heating season [32].
d
Due to close resemblances, the categories of houses built in year 1850e1930, 1931e1950, and 1951e1960 have been aggregated into one category (1850e1960).
the dominating part of the heat demand,3 possibilities for using hot 2.2. Heat demand variations
water tanks for flexible heat pump operation have not been
investigated. Heat demand variations in the model are based on hourly heat
demand profiles on an aggregated level for a large number of Danish
households [33]. As part of the study, model development has been
2.1. Modelling of the building stock made to ensure that heat stored in the construction or in the heat
accumulation tank can only be used to satisfy space heating
The existing building stock of detached houses is modelled based demand. This is implemented by ensuring that the heat pump as
on data in Refs. [25,26] for different representative building cate- a minimum has to satisfy the hot water demand in each hour. In the
gories. The buildings are categorised depending on the year of model, the hot water demand is identified as the minimum heat
construction, reflecting historical changes in insulation standards and demand over the year. Therefore, the heat demand profiles are
heat demands (see Table 1). The heat demands in Table 1 have been modified in the sense that a constant hot water demand profile is
calibrated with energy statistics of the Danish Energy Agency. applied to represent the average hot water demand (see Fig. 1).
Passive heat storage capacities are estimated in Table 1 Additionally, separate heat demand profiles have been created
assuming utilisation of the thermal storage capacity facing the depending on the hot water share for the given building category.
rooms inside the insulation, which can be utilised actively during
a diurnal temperature variation. The capacity is mainly comprised
by internal constructions in walls, ceiling, and floor while windows, 2.3. The EnergyPLAN model
doors, and furniture have minor influence. The storage capacities
are estimated as c$DTin where c is the active heat capacity (Wh/K/ EnergyPLAN is a deterministic input/output model that optimises
m2) and DTin is the temperature change of the indoor air allowed in the operation of an energy system over a full year, by performing hour-
connection with utilising passive heat storage. by-hour analyses. Inputs are demands and demand distributions,
Based on current data and knowledge, typical existing buildings capacities of technologies included, distributions of fluctuating
in Denmark are expected to have c values of around 60e120 Wh/K/ (renewable energy sources) RES, fuel and CO2 costs etc. The model
m2 floor area, the low end representing buildings of extra-light to outputs comprise energy balances, energy productions, fuel
medium-light materials and the high end representing buildings of consumptions, electricity imports/exports, CO2 emissions, and costs.
mediumeheavy materials [27,28]. The analyses have been made The model covers the whole energy system, i.e. individual heating and
assuming active heat capacities in the low end and high end, district heating, the electricity, transport, and industry sector [34].
respectively. Typically, Danish households keep an indoor EnergyPLAN is further described in Refs. [34,35] and in Ref. [36] where
temperature of around 22 C [29] but if variations around the previous applications and comparison with other models can also be
traditional level are accepted heat pumps can be operated flexibly found. The model makes it possible to use different regulation strat-
utilising passive heat storage. Based on recommendations in egies depending on the purpose of the analysis.
Ref. [30] for an acceptable comfort level in residential buildings, an
allowed indoor temperature variation of H2 C from the initial 2.4. Reference energy system
level can reasonably be assumed (DTin ¼ 4 C). Considering that
thermal comfort preferences are likely to vary among residents, the Apart from increased wind power capacities, the Danish
effect of a lower indoor temperature variation, namely H1 C, is energy system in 2020 is assumed identical to the current system
also investigated (DTin ¼ 2 C). based on Ref. [37]. The effects of individual heat pumps and heat
storage options are then analysed in the context of increased
wind power production. Expected wind power capacities in 2020
3
On average 84% for existing detached houses in Denmark [26]. [38] have been applied, yielding a wind power production of
3.5
Relative to average
3.0 Original heat
demand profile
2.5
2.0
1.5
Modified heat
1.0 demand profile
0.5
0.0
2000
4000
5000
7000
1000
3000
6000
8000
Hour
Fig. 1. Hourly heat demand profile applied for the aggregated building stock of existing detached houses.
16.5 TWh and corresponding to around 50% of annual electricity 2.6. Scenarios
demand.4 Supplemental analyses have been made for wind power
production varying from 0 TWh to 34 TWh, corresponding to 0% Naming conventions used in the main scenarios are presented in
to 100% of annual electricity demand. Compared to other coun- Table 2. Heat demand distributions before and after the installation
tries, the Danish energy system is characterised by a large share of of heat pumps are given in Table 3.
wind power and district heating (46% [10]), and high total energy It can be noted that for countries without large-scale district
efficiency with a large share of CHP (55% of thermal power heating, the market potential for individual heat pumps would be
production and 80% of district heating [2]). Fuel prices corre- significantly larger than for the Danish case considered here (all
sponding to $100/barrel are applied and a CO2 price of 25 V/tonne other things equal).
assumed for 2020 based on the Danish Energy Agency and the
International Energy Agency [39].
2.7. Heat storage and control system costs
2.5. Methodology for the energy system analyses
For the passive heat storage concept, the heat pump and central
The potential of heat pumps and different heat storage options heating system is assumed controlled through an automatic vari-
in supporting wind power integration is investigated in terms of ation of the indoor temperature set point, depending on when it is
their ability to increase wind power utilisation, reducing (excess optimal to load/unload the storage. In order to facilitate this and
electricity production) EEP and fuel consumption5 of the system. still ensure satisfactory regulation of the indoor air temperature,
The cost-effectiveness of the different heat storage options is electronic thermostats (wireless) for the radiators are assumed
identified as the annualised average socio-economic costs per fuel required.6 In a low cost sensitivity scenario (LCOST), it is alterna-
saved (referred to as fuel saving costs). tively assumed that a satisfactory temperature regulation can be
A technical optimisation of a closed system is applied, where handled through a single electronic indoor air thermostat placed at
the electricity and heat supply/demand are balanced in a way that a suitable reference point in the house (wireless, programmable,
minimises the overall fuel consumption and that utilises as much 67 V/house [40]). With this type of control system, the automatic
wind power as possible. In the optimisation, the electricity regulation of the individual heat supply from each radiator would
production from CHP plants and from RES is prioritised hour-by- have to be shut off. As such, a less accurate regulation of the
hour. The remaining electricity demands are met by condensing temperature in each room would be expected in the LCOST scenario
power production, and the remaining district heating demands [24,28].
are met by boilers. By utilising heat storages and extra capacity at When using either passive heat storage or a heat accumula-
the CHP plants, the production at the condensation plants is tion tank for intelligent heat pump operation based on signals/
minimised. To the extent that it can help reduce EEP, CHP prices from the energy system, a central controller would be
generation is replaced with boiler operation in the district heat- required. Such a device does not yet exist on the market and
ing system. As a result, the presence of CHP does not contribute to future prices are therefore difficult to predict. In the main
generating EEP in the optimisation. The remaining EEP is thus scenarios, no additional costs for a central controller are included,
mainly due to wind power fluctuations and required minimum assuming that it will be integrated in intelligent heat pumps in
power supply from central power plants for ensuring grid the future. In a high cost sensitivity scenario (HCOST), the current
stability (450 MW). This makes the results relevant also for price of a related type of controller recently introduced on the
countries without large-scale CHP. market is added7 (348 V/house [40]). Experience shows that new
Heat storages added to individual heat pumps allow the heat innovative electronic products normally drop significantly in
pumps to prioritise their operation for hours with EEP, while price over a period of years. Thus, the control system cost
intending to avoid operation in hours with condensing power included in the HCOST scenario represents a high estimate of
production. In this way, the heat storages contribute to reduce EEP the future cost for a controller including installation. Installation
and fuel consumption of the system. In situations where the costs are expected to be low considering that the control system
measures introduced to ensure balance between electricity can be installed in connection with the heat pump and
demand and supply from CHP and RES are insufficient, EEP will that a wireless system is considered. Assumed heat storage
occur. For an open system, the EEP will represent forced electricity and control system costs in the main scenarios are given in
export. Table 4.
4 6
Excl. electric heating and electricity for individual heat pumps, which vary An electronic air temperature sensor is also required but this device is normally
across the scenarios (1.4e2.1 TWh for main scenarios). delivered together with the heat pump [24].
5 7
Excl. fluctuating RES, i.e. wind power for the Danish case. Used for digital intelligent control of a central heating system (wireless).
Table 2 Table 4
Naming conventions used in main scenarios. Storage and control system costs applied in main scenarios.
Scenario Description Storage concept Inv. cost Technical Ref.

NOHP No heat pumps in households. life-time
V/house V/100 L V/house/yra
HP-NOSTOR Heat pumps installed in detached houses with (yr)
oil boilers and/or electric heating without Heat accum. 300 L 856 285 37 40 [41]
investment in heat accum. tanks or passive tank
heat storage. 500 L 1138 228 49 40 [41]
Tank300/500/1000 Inv. in 300/500/1000 litre heat accum. tanks 1000 L 1485 149 64 40 [41]
in all houses where heat pumps are installed.a Passive heat Electronic 265 e 22 15 [40]
PasLow Inv. in passive heat storage assuming low end storage thermostats
passive heat storage cap. (60 Wh/K/m2)a a
A discount rate of 3% in fixed prices is applied.
PasHigh Inv. in passive heat storage assuming high end
passive heat storage cap. (120 Wh/K/m2)a
1C Allowing an indoor air temp var. of H1 C when
utilising passive heat storage heat demand exceeds the heat pump capacity and when sufficient
2C Allowing an indoor air temp var. of H2 C when heat storage is not available to cover demand either. In hours with
utilising passive heat storage EEP, the heat storage is loaded, thereby maximising heat pump
a
Heat pump installation as in HP-NOSTOR. operation and utilisation of wind power in the system. In hours
with condensing power production, the heat storage is unloaded,
3. Modelling of heat pumps and heat storages intending to minimise heat pump operation. Unloading of the heat
storage is thus prioritised for reducing condensing power produc-
The most important equations and rules, representing the tion in the system and thereby obtaining highest possible fuel
model of the individual heat pumps, are presented in this section. displacement.
Eq. (1) is the heat balance equation and Eq. (2) defines the relation Further model development has been made to implement the
between electricity input and heat output for the heat pump. Eqs. following restrictions on use of the passive heat storage:
(3) and (4) represent heat storage balance and heat storage capacity

restrictions and Eq. (5) the heat pump capacity constraint. Eq. (6) Storðt 1Þ
LoadðtÞ U,DTpas , 1 ,Afloor ct (7)
has been introduced as part of this study, restricting the heat Storcap
pump to meet hot water demand in all hours. In this way, it is
ensured that the heat storages considered can only enable flexible
Storðt 1Þ
heat pump operation in satisfying space heating demand. UnloadðtÞ U,DTpas , ,Afloor ct (8)
Storcap
HDðtÞ ¼ HPheat ðtÞ þ EBðtÞ þ UnloadðtÞ LoadðtÞ ct (1) Due to the low heat capacity of air (0.8 Wh/m2/K [9]) and the
fact that a radiator system is considered, it is assumed that a desired
HPheat ðtÞ ¼ HPel ðtÞ,COP ct (2) indoor air temperate can quickly be reached. The temperature of
the construction is assumed to vary from hour to hour, depending
StorðtÞ ¼ Storðt 1Þ þ LoadðtÞ UnloadðtÞ ct (3) on the amount of heat stored in it (Stor(t)). This reflects the large
thermal mass of the construction. When loading and unloading of
StorðtÞ Storcap ct (4) the passive heat storage occurs over periods of sufficient length,
DTpas will be equal to DTin. Thus, the product U$DTpas represents the
highest obtainable loading/unloading of the passive heat storage.
HPheat ðtÞ CF,HDmax ct (5)
This corresponds to loading at empty storage and unloading at full
storage. The loading/unloading capacity in a given hour depends on
HWD HPheat ðtÞ ct (6) the passive heat storage level (Stor(t)) as expressed in Eqs. (7) and
Ground heat pumps and air/water heat pumps are typically (8).
supplemented by an electric boiler to cover peak loads, in order to The temperature variation of the construction will be parallel-
limit heat pump investment costs. As shown in Eq. (1), this is also shifted downwards compared to the temperature variation of the
represented in the model. A heat pump (capacity factor) CF of 0.80 indoor air (typically with around 1 C for existing buildings). This is
is assumed, corresponding to the typical dimensioning of these due to the fact that transmission losses from the construction to the
types of heat pumps [28,42]. As a result, the heat pump covers more outdoor environment will occur continuously during the heating
than 99% of annual heat demand in the model. Due to the low season. In the model, loading of the passive heat storage corre-
efficiency of the electric boiler, it is only operated in hours where sponds to a situation where the heat transfer from indoor air to
construction is higher than the transmission loss from construction
to outdoor environment. Correspondingly, unloading of the storage
Table 3 represents a situation where the heat transfer from indoor air to
Net heat demand distribution in the scenario without heat pumps (NOHP) and in construction is lower than the transmission loss from construction
scenarios with heat pumps.
to outdoor environment. U is estimated to 22 W/m2 floor area/K
TWh/yr Heat Oil Elec. Biomass Natural District based on [43,44]. The principle of the passive heat storage model is
pumps boilers heating boilers/stoves gas boilers heating illustrated in Fig. 2 for a loading and subsequent unloading period.
(excl.
grid loss)
Fig. 2 shows that the heat pump is operated at maximum capacity
in the first part of the loading period. In the end of the loading period,
NOHPa 0 4.8 1.4 8.6 7.0 28.4
HP scenariosb 5.0 0.9 0.3 8.6 7.0 28.4 limitations on loading (Eq. (7)) imply that the heat pump production
a
must be lowered accordingly. In the start of the unloading period, the
Current heat demands as given in Danish energy statistics [2]. Individual heat
demands represent households. The existing amount of heat pumps is low and has
heat pump production is lowered to the minimum level, thereby
been neglected for simplicity. satisfying only hot water demand. Due to limitations on unloading
b
Heat pumps installed in detached houses with oil boilers and/or electric heating. (Eq. (8)) and the increase in heat demand, the heat pump production
2500
2000 Excess electricity
MW
1500 production
1000
Condensing power
500
production
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64
Hour
2000
Heat demand
1500
MW
1000
Heat pump
500
production
0
-500 Loading(-)
/unloading(+)
-1000
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64
Hour
100%
Storage level
80%
60% Stor(t)/Storcap
40%
20%
0%
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64
Hour
Fig. 2. Optimisation of a large-scale installation of heat pumps using passive heat storage, modelled in integration with the energy system (allowing H1 C indoor air temp. var. and
assuming a passive heat storage cap. of 120 W/m2 floor area/K). A loading and subsequent unloading period is shown, together representing around 2e3 days in mid January. Excess
electricity production and condensing power production is shown after the optimised heat pump operation minimising their occurrence.
is then increased towards the end of the unloading period. Due to the assumed heat pump installations partly displace an amount of electric
influence of Eq. (7), the marginal increase in the storage level, heating in households, which due to its low efficiency represents
Stor(t)/Storcap is reduced during the loading period. Correspondingly, a relatively large electricity demand. This lowers the overall increase
Eq. (8) implies that the marginal decrease in the storage level is in electricity demand resulting from the heat pump installations.
reduced during the unloading period. When heat pumps are equipped with heat storages they are able
In the model of the passive heat storage, full/empty storage to place a larger amount of their operation in hours with EEP. This
corresponds to a situation where the temperature of the construction results in larger EEP reductions. If investing in heat accumulation
is 1 C or 2 C above/below the traditional level. Depending on the tanks, an EEP reduction of 0.15e0.19 TWh (9e12%) is thus obtained.
passive heat storage level, increased transmission losses to the outdoor If enabling passive heat storage (1C/2C-PasLow/High), equivalent to
environment will thus occur in some periods while reduced trans- larger EEP reductions can be achieved, 0.19e0.30 TWh (12e19%).
mission losses will occur in other periods. Overall, transmission losses The patterns illustrated in Fig. 3 have also been identified at
are therefore assumed unchanged. Correspondingly, overall ventila- lower/higher wind penetrations.
tion losses are assumed unchanged. Heat losses from heat accumula- Fig. 4 illustrates the reduction in EEP over the year for the scenario
tion tanks are moderate [45] and have not been modelled. with heat pumps and a high level of passive heat storage (2C-PasHigh).
The modelling methodology developed for representing heat As shown, a large part of the EEP exceeds the total capacity of
accumulation tanks and passive heat storage is compatible with the heat pumps and thus their ability to absorb EEP. In addition, the
use of aggregated heat demand profiles, which are commonly used in heat pumps ability to reduce EEP is very limited in the summer
energy system models. The methodology thereby has the advantage period (around 3700e6000 h) where they only operate to satisfy
of being easy to integrate in the typical structure of energy system hot water demand (46 MW-e).
models. The focus of this paper is not to address the model and
controllers for the operation of individual heat pumps in detail.
HP-NOSTOR
4. Results Tank300
Main results and sensitivity analyses are presented in the Tank1000

following sections for the assumed large-scale installation of heat
1C-PasLow
pumps and heat storages.
1C-PasHigh
4.1. Excess electricity production
2C-PasLow
As illustrated in Fig. 3, heat pumps even without the flexibility 2C-PasHigh

provided by heat storages (HP-NOSTOR scenario), can contribute to
increase wind power utilisation and thus reduce excess electricity -0.35 -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 0.00
production (EEP). As such, EEP is reduced with around 0.12 TWh (from Change in excess electricity production (TWh)
1.58 TWh to 1.46 TWh, i.e. 8%) for the system in 2020 with 16.5 TWh Fig. 3. Change in excess electricity production depending on type and size of heat
wind power. This is alone due to the increased electricity demand storage available for individual heat pumps, compared to a system without heat
resulting from the heat pump installations. It can be noted that the pumps.
4000
EEP (NOHP)
3500
3000 EEP (2C-PasH)
MW-e
2500
2000 EEP (NOHP),
duration curve
1500
EEP (2C-PasH),
1000
duration curve
500
Heat pump cap.
0 1000
3000
6000
2000
4000
5000
7000
8000
Hour
Fig. 4. Excess electricity production (EEP) for a scenario without individual heat pumps (NOHP) and for a scenario with large-scale installation of heat pumps, displacing oil boilers
and electric heating, assuming utilisation of passive heat storage (2C-PasHigh).
4.2. Fuel consumption

HP-NOSTOR
Changes in the systems fuel consumption due to the installation Tank300

of heat pumps and heat storages are illustrated in Fig. 5 at the wind
Tank1000 Oil
power level expected for 2020.
As shown, the installation of heat pumps (HP-NOSTOR scenario) 1C-PasLow Coal
alone results in a large reduction in oil consumption due to the
1C-PasHigh
displacement of oil boilers. Due to the higher efficiency of heat pumps Natural
compared to the displaced heating technologies (oil boilers and gas
2C-PasLow
electric heating) only moderate increases in coal and natural gas
consumption for electricity production are observed. As a result, 2C-PasHigh
significant net fuel savings are obtained, around 3.5 TWh (2%8). When -5 -4 -3 -2 -1 0 1 2
adding heat storages to the heat pumps, a larger share of the wind
Change in fuel consumption excl. wind (TWh)
power production is utilised and condensing power production
reduced. This results in a lower consumption of coal and natural gas. In Fig. 5. Changes in fuel consumption excl. wind when installing individual heat pumps
line with the observed effects on EEP, passive heat storage provides and heat storages compared to a system without heat pumps (NOHP).
equivalent to larger fuel savings compared to heat accumulation
tanks, 3.7e3.9 TWh vs. 3.6e3.7 TWh (2%). However, compared to the
the fuels displaced, i.e. 20 V/MWh for coal, and 33 V/MWh for
fuel savings provided by heat pumps alone (HP-NOSTOR scenario),
natural gas (when including CO2 costs and fuel handling costs).
the additional fuel savings provided by heat storages (Tank300...2C-
For the case of heat accumulation tanks, the economic benefits
PasHigh) are moderate. The high efficiency of heat pumps compared
are far from being sufficient to cover the investment costs. As such,
to the displaced heating technologies is of course part of the expla-
the annualised investment costs are around 4 times higher than the
nation for this. Moreover, the occurrence of EEP is moderate (occur-
cost savings for the 1000 L tank and around 6 times higher for the
ring around 20% of the year in the NOHP scenario), compared to the
300 L tank. As a result, fuel saving costs are very high, around 86e
many operating hours needed for the heat pumps to satisfy heat
140 V/MWh. The results in Fig. 7 indicate that heat accumulation
demand throughout the year. The heat pumps limited ability to
tanks are far from cost-effective as a measure of increasing wind
absorb EEP also plays a role (see Section 4.1). The above patterns have
power utilisation for fuel displacement in the system. In contrast,
also been identified at lower/higher wind penetrations.
passive heat storage provides significantly lower fuel saving costs
and demonstrates a reasonable cost-effectiveness. The robustness
of this assessment is tested in a number of sensitivity analyses in
4.3. Costs
the following section. Among the three tank sizes analysed, fuel
saving costs can be seen to be lowest for 1000 L tanks. Therefore,
In Fig. 6, changes in socio-economic costs from adding heat
sensitivity analyses of the tanks are limited to cover this size only.
storages to the heat pumps are presented for the system in 2020. As
a measure of cost-effectiveness, average costs per fuel saved
(V/MWh) are presented in Fig. 7; where costs represent annualised 15
Investment
investment costs and variable O&M costs.
Fig. 6 shows that for the passive heat storage scenarios, the 10
Costs (M€/yr)
savings on fuel, CO2, and variable O&M costs are around the same Fuel
5
level or larger than the annualised investment costs required for
enabling the storage. Thus, passive heat storage provides an CO2
0
approximate break-even or a net reduction in costs depending on
the scenario. This is also reflected in fuel saving costs of 10e29 V/ Variable
-5
MWh, i.e. below or close to the expected price levels in 2020 of O&M
-10
8
Relative to fuel consumption of the whole energy system covering individual Fig. 6. Changes in socio-economic cost when enabling different types/sizes of heat
heating, district heating, the electricity, transport, and industry sector. storages for individual heat pumps.
Fuel saving costs (€/MWh) 160 different building categories in Table 1. Deviations of 15% to þ33%
140 from the results in the main scenario have been found for the case
120 of passive heat storage. Correspondingly, deviations of 9% to þ4%
100 have been identified for the case of 1000 L heat accumulation tanks.
Regardless of the building category, passive heat storage has been
80
found to be significantly more cost-effective than heat accumula-
60
tion tanks.
40
Naming conventions used in further sensitivity scenarios are
20
given Table 5. Fuel saving costs for the different sensitivity
0 scenarios are presented in Fig. 8.
Fig. 8 shows that for the aggregated building stock, heat savings
have very low impacts on the results (HS scenarios). Costs of control
system components are however highly influential. As such, in the
LCOST scenarios for passive heat storage, fuel saving costs are very
Fig. 7. Average annualised costs per fuel saved for different heat pump storage options.
low. Whether the control system in the main scenarios or the
For comparison, the expected coal price, 20 V/MWh (dashed line) and natural gas
price, 33 V/MWh (black line) in 2020, including CO2 costs and fuel and handling costs, control system in the LCOST scenarios is most likely to be installed
is indicated. in practice depends on how the individual household weighs costs
vs. comfort. When including costs of a central controller (HCOST
Table 5
scenarios) based on the current price of a related product newly
Naming conventions used in sensitivity analyses of the system in 2020.
introduced on the market, fuel saving costs are increased consid-
Name Description erably. However, the results of the HCOST scenarios represent
HS Economically feasible heat savings (cf. Table 1) assumed a worst-case-end as the price of new innovative electronic prod-
realised prior to heat pump and heat storage installation. ucts, will normally be reduced considerably over a period of years. If
LCOST Low control system costs for passive heat storage assuming
that a single electronic indoor air thermostat is sufficient
operating with 25 C temperature difference in the radiator system
to ensure satisfactory regulation of the indoor air (25DT scenario) fuel saving costs for heat accumulation tanks are
temperature (cf. Section 2.7). reduced moderately. When assuming installation of electric boilers
HCOST Inclusion of a high cost estimate of a central controller and large heat pumps in the district heating system (EBHP
applying current price of a related product recently
scenarios), fuel saving costs for the heat storages in households are
introduced on the market (cf. Section 2.7).
EBHP Electric boilers and large heat pumps assumed installed increased. This shows that competition with other flexible tech-
in the district heating system (250 MW-e electric boilers nologies can have a significant influence on the cost-effectiveness
[38] and 400 MW-e heat pumps based on Ref. [21]). of adding further flexibility to the system. If also assuming large-
EV Electric vehicles assumed implemented corresponding to scale implementation of electric vehicles (EBHPEV scenarios)
10% of the vehicle fleet in 2020.
25DT Operation with 25 C temp. difference in the radiator system
further increases in fuel savings costs are observed. The use of
on average, yielding larger utilisation of the heat accum. a discount rate of 5% (R5% scenarios) on heat storage and control
tank but reducing COP of the heat pump [24], corresponding system investments, results in significantly increased fuel saving
to 40% reduction when loading the tank.a costs for the heat accumulation tank scenario. For the passive heat
R5% Discount rate of 5% applied in fixed prices.
storage scenarios, the discount rate has minor influence due the
a
COP reduction occurs due to a needed increase in the forward temperature from lower investment cost and technical life-time for the type of
typically around 40 C on average to 55 C and the fact that COP is typically reduced devices in question (cf. Table 4). Overall, the sensitivity analyses
with 2e3% per degree increase in forward temperature [46].
confirm that heat accumulation tanks are far from being cost-
effective while passive heat storage can provide significantly
4.4. Sensitivity analyses lower fuel saving costs, potentially at a reasonable level.
It can be mentioned that in the 25DT scenario, the EEP reduction
Sensitivity analyses on fuel saving costs for the heat storage obtained with a heat accumulation tank of 1000 L is found to be
options have been made considering installation in each of the four 0.24 TWh, i.e. on level with the EEP reduction for the 1C-PasHigh
250
2C-PasHigh 2C-PasLow 1C-PasHigh 1C-PasLow Heat accum. tank, 1000 litres
200
Fuel saving costs (€/MWh)
150
100
50
0
Main HS LCOST HCOST 25DT EBHP EBHPEV R5%
scenarios
Fig. 8. Average annualised costs per fuel saved for sensitivity scenarios of different heat pump storage options. For comparison, the expected coal price, 20 V/MWh (dashed line)
and natural gas price, 33 V/MWh (black line) in 2020, including CO2 costs and fuel and handling costs, is indicated.
scenario, some-what larger than for the 1C-PasLow scenario, but References
still lower than for the 2C-PasHigh scenario. Due to the COP
reduction in the 25DT scenario, the fuel saving obtained is never- [1] Beurskens WM, Hekkenberg M, Vethman P. Renewable energy projections as
published in the national energy action plans of the European member states.
theless still below or at level with the fuel savings obtained in all Report no. ECN-Ee10-069. ECN and European Environment Agency. Available
passive heat storage scenarios. at: http://www.ecn.nl/docs/library/report/2010/e10069.pdf; 2011.
[2] Danish Energy Agency. Energy statistics 2008 [Energistatistik 2008] [in
Danish]. Copenhagen, Denmark: Danish Energy Agency. Available at: http://
5. Conclusion www.ens.dk; 2009.
[3] Mathiesen BV, Lund H. Comparative analyses of seven technologies to facili-
Compression heat pumps in households (individual heat tate the integration of fluctuating renewable energy sources. IET Renewable
Power Generation 2009;3:190e204.
pumps), using heat accumulation tanks or passive heat storage in
[4] Kiviluoma J, Meibom P. Influence of wind power, plug-in electric vehicles, and
the construction, have been analysed in terms of their potentials in heat storages on power system investments. Energy 2010;35(3):1244e55.
increasing wind power utilisation and providing cost-effective fuel [5] Lund H. Large-scale integration of wind power into different energy systems.
savings. The Danish energy system in 2020 has been used as case, Energy 2005;30(10):2402e12.
[6] Hedegaard K, Ravn H, Juul N, Meibom P. Effects of electric vehicles on power
representing a system with a large share of wind power (around systems in Northern Europe. Energy, in press.
50%), district heating, and CHP. Installation of heat pumps in [7] Keane A, Tuohy A, Meibom P, Denny E, Flynn D, Mullane A, et al. Demand side
existing detached houses is assumed, displacing individual oil resource operation on the Irish power system with high wind power pene-
tration. Energy Policy 2011;39(5):2925e34.
boilers and electric heating. For the analyses, a modelling meth-
[8] Papagiannis G, Dagoumas A, Lettas N, Dokopoulos P. Economic and environ-
odology has been developed to enable representation of flexible mental impacts from the implementation of an intelligent demand side
heat pump operation using the heat storages in question. The management system at the European level. Energy Policy 2008;36(1):163e80.
methodology has the advantage of being compatible with the [9] Nyeng P. System integration of distributed energy resources e ICT, ancillary
services and markets. PhD thesis. Kgs. Lyngby, Denmark: Technical University
typical representation of heat demands in energy system models.
of Denmark, Department of Electrical Engineering; 2010.
Results show that by displacing less efficient heating technolo- [10] Lund H, Möller B, Mathiesen BV, Dyrelund A. The role of district heating in
gies and increasing electricity demand, the installation of heat future renewable energy systems. Energy 2010;35(3):1381e90.
[11] Energy Analyses Ea, Risø DTU, RAM-løse edb. Efficient district heating in the
pumps alone can contribute to the integration of wind power,
future energy system [Effektiv fjernvarme i fremtidens energisystem] [in
providing significant reductions in excess electricity production and Danish]. Copenhagen, Denmark: Ea Energy Analyses; 2009.
fuel consumption. If additionally investing in heat accumulation [12] Hewitt NJ. Heat pumps and energy storage e the challenges of implementa-
tanks, moderately increased reductions are obtained. However, if tion. Applied Energy 2012;89(1):37e44.
[13] Halvgaard R, Kjølstad NK, Madsen H, Jørgensen JB. Economic model predictive
instead investing in passive heat storage, equivalent to larger control for climate control of buildings in a smart energy system. In. 50th IEEE
reductions can be achieved. Moreover, heat accumulation tanks are conference on decision and control and European control conference,
found to be far from cost-effective while passive heat storage proves Orlando, USA; 2011.
[14] Okamoto S. A heat pump system with a latent heat storage utilizing seawater
to have a significantly higher, potentially reasonable, cost- installed in an aquarium. Energy and Buildings 2006;38(2):121e8.
effectiveness. [15] Oldewurtel F, Ulbig A, Parisio A, Andersson G, Morari M. Reducing peak
Overall, in terms of enabling flexible operation of individual heat electricity demand in building climate control using real-time pricing and
model predictive control. In. 49th IEEE conference on decision and control,
pumps for a fuel-efficient and cost-effective integration of wind Atlanta, USA; 2010.
power, promising potentials have been identified for passive heat [16] Tahersima F, Stoustrup J, Soroush MA, Rasmussen H. Contribution of domestic
storage but not for heat accumulation tanks. Nevertheless, in terms heating systems to smart grid control. In. 6th Dubrovnik conference on
of reducing fuel consumption of the system, the installation of heat sustainable development of energy water and environment systems,
Dubrovnik, Croatia; 2011.
pumps is the most important step. The additional fuel savings [17] MacDougall P, Warmer C, Kok K. Raising the accommodation ceiling for wind
provided by adding heat storages are moderate in comparison. power by intelligent response of demand and distributed generation. In. 10th
international workshop on large-scale integration of wind power into power
systems as well as on transmission networks for offshore wind power plants,
6. Future work Aarhus, Denmark; 2011.
[18] Mathiesen BV, Lund H, Karlsson K. 100% renewable energy systems, climate
It would be interesting to analyse the effect of the heat storages mitigation and economic growth. Applied Energy 2008;88(2):488e501.
[19] Karlsson K, Balyk O, Zvingilaite E, Hedegaard K. District heating versus indi-
in a model that minimises total system costs, including operation vidual heating in a 100% renewable energy system by 2050. In. 6th Dubrovnik
costs and annualised investment costs, also taking external elec- conference on sustainable development of energy water and environment
tricity trade into account. In this regard, it would be relevant to systems, Dubrovnik, Croatia; 2011.
[20] Papaefthymiou G, Hasche B, Nabe C. Potential of the flexibility of heat pump
investigate to which extent investment in passive heat storage is
operation for wind power integration. In. 10th International workshop on large-
likely to be economically attractive, considering possibilities to scale integration of wind power into power systems as well as on transmission
invest in competing flexible technologies. As in many other energy networks for offshore wind power plants, Aarhus, Denmark; 2011.
system models, the heat pumps were represented by a yearly [21] Energinet.dk. Efficient utilisation of wind power in Denmark [Effektiv
anvendelse af vindkraftbaseret el i Danmark] [in Danish]. Fredericia,
average COP in this study. It could be interesting to analyse the Denmark: Energinet.dk. Available at:, www.energinet.dk; 2009.
effect of heat storages when taking hourly COP variations (caused [22] Olsen L, Poulsen CS, Vorre MH. Description of potentials e individual heat
by variations in outdoor and ground temperatures etc.) into pumps [Potentialebeskrivelse e individuelle varmepumper] [in Danish].
Taastrup, Denmark: Danish Technological Institute; 2010.
account. An inclusion of hourly COP variations is however not ex- [23] Jensen JB, Hvenegaard CM, Pedersen SV. The little blue about heat pumps
pected to change the overall conclusion on the comparison of heat [Den lille blå om varmepumper] [in Danish]. Copenhagen, Denmark: Danish
pumps and the different heat storages. Energy Association; 2011.
[24] Andersen JJ. Vølund Varmeteknik. Personal communication. (August 2010,
January 2012).
Acknowledgements [25] Dyrelund A, Lund H, Möller B, Mathiesen BV, Fafner K, Knudsen S, et al. Heat
plan Denmark, appendix [Varmeplan Danmark, bilagsrapport] [in Danish].
Kolding, Denmark: Danish District Heating Association; 2008.
The authors would like to thank Lars Olsen, Danish Technical [26] Wittchen KB. Potential energy savings in existing buildings [Potentielle
Institute, Brian Elmegaard, Technical University of Denmark, and energibesparelser i det eksisterende byggeri] [in Danish]. Hørsholm,
Jacob J. Andersen, Vølund Varmeteknik, for providing technical Denmark: Danish Building Research Institute, Aalborg University; 2009.
[27] Aggerholm S, Grau K. SBi instruction 213, buildings’ energy demands, PC
insight into heat pump and storage configurations. Thanks are also
programme and calculation instruction [SBi-Anvisning 213, Bygningers
given to Poul Erik Morthorst, Marie Münster, Technical University of Energibehov, Pc-Program Og Beregningsvejledning] [in Danish]. Hørsholm,
Denmark, and the reviewers for valuable comments on the article. Denmark: Danish Building Research Institute; 2008.
[28] Olsen L. Danish Technological Institute. Personal communication. (June, [38] Energinet.dk. Analysis preconditions 2010e2020, 2025, and 2030 [Analy-
September, 2010, April 2011, January 2012). seforudsætninger 2010e2020, 2025 og 2030] [in Danish]. Fredericia,
[29] Wittchen K. Danish Building Research Institute. Personal communication. Denmark: Energinet.dk; 2010.
(June 2010). [39] Danish Energy Agency. Preconditions for socio-economic analyses in the field
[30] Standards European. EN 15251-indoor environmental input parameters for of energy [Forudsætninger for samfundsøkonomiske analyser på ener-
design and assessment of energy performance of buildings addressing indoor giområdet] [in Danish]. Copenhagen, Denmark: Danish Energy Agency; 2010.
air quality, thermal environment, lighting and acoustics. Pilsen, Czech Republic: [40] Danfoss. Danfoss living connect, product prices. Hasselager, Denmark: Danfoss.
European Standards. Available at: http://www.en-standard.eu; 2006. Available at: http://www.billigvvs.dk/Varmesystemer-Radiatorer-Radiator-
[31] Danish Enterprise and Construction Authority. The Danish building register tilbehoer-Danfoss-Living-Connect.html [accessed December 14, 2011].
[Bygnings- og boligregistret] [in Danish]. Copenhagen, Denmark: Danish [41] Varmeteknik Vølund. Pricelist e recommended sale prices, April 2010 [Pri-
Enterprise and Construction Authority; 2010. sliste e vejledende udsalgspriser april 2010] [in Danish]. Videbæk, Denmark:
[32] EMD International. Data for a Danish test reference year contained in ener- Vølund Varmeteknik; 2010.
gyPRO 4.1.2, software for projecting and planning of heat, cooling, and power [42] Jensen KA. Dimensioning of heat pumps most important [Dimensionering af
plants. Aalborg, Denmark: EMD International. Available at: http://www.emd. varmepumper vigtigst] [in Danish]. Glostrup, Denmark: TEKNIQ The Electri-
dk [accessed January, 2012]. cians’ Organisation. Available at: http://www.tekniq.dk/PresseOgNyheder/
[33] Lund H, Mathiesen BV. Note for the Danish society of engineers’ energy year [Notat Nyheder/2010/Marts/Dimensionering-af-varmepumper-er-vigtigst.aspx;
til IDA energiår] [in Danish]. Aalborg, Denmark: Aalborg University; 2006. 2010.
[34] Lund H. EnergyPLAN e advanced energy systems analysis computer model, [43] Artmann N, Jensen RL, Manz H, Heiselberg P. Experimental investigation of
February 2010, documentation version 8.0. Aalborg, Denmark: Aalborg heat transfer during night-time ventilation. Energy and Buildings 2010;42(3):
University; 2010. 366e74.
[35] Lund L, Münster E. Modelling of energy systems with a high percentage of [44] Akbari H, Samano D, Mertol A, Bauman F, Kammerud R. The effect of variations in
CHP and wind power. Renewable Energy 2003;28(11):2179e93. convection coefficients on thermal energy storage in buildings part I e interior
[36] Connolly D, Lund H, Mathiesen BV, Leahy M. A review of computer tools for partition walls. Energy and Buildings 1986;9(8):195e211.
analysing the integration of renewable energy into various energy systems. [45] Vaillant. Ground heat systems 6e17 kW [Jordvarmesystemer 6e17 kW] [in
Applied Energy 2010;87(4):1059e82. Danish]. Karlslunde, Denmark: Vaillant; 2007.
[37] Danish Energy Agency. Denmarks’ energy projection towards 2030, April [46] Danish Energy Agency. The Danish building regulation [in Danish]
2009 [Danmarks energifremskrivning frem til 2030, april 2009] [in Danish]. [Bygningsreglementet]. Available at: http://www.ebst.dk/bygningsregleme
Copenhagen, Denmark: Danish Energy Agency; 2009. ntet.dk/br10_02_id123/0/42 [accessed February 27, 2011].
Paper IV
Energy system investment model incorporating heat pumps

with thermal storage in buildings and buffer tanks
Karsten Hedegaard & Olexandr Balyk.
Accepted for publication in Energy. September 26, 2013.

Paper IV
Energy system investment model incorporating heat pumps with thermal

storage in buildings and buffer tanks
Karsten Hedegaard a, *, Olexandr Balyk a

a
Systems Analysis Division, DTU Management Engineering, Frederiksborgvej 399, P.O. Box 49, Build. 130, DK-
4000 Roskilde, Denmark.
*Corresponding author. Tel.: +45 4677 5158; fax: +45 4677 5199. E-mail addresses: [email protected],
[email protected] Frederiksborgvej 399, P.O. Box 49, Build.130, DK-4000 Roskilde (K.Hedegaard).
ABSTRACT
Individual electric heat pumps constitute a potentially valuable resource in supporting wind power
integration due to their economic competitiveness and possibilities for flexible operation. When
analysing the system benefits of flexible heat pump operation, effects on energy system investments
should be taken into account. In this study, we present a model that facilitates analysing individual heat
pumps and complementing heat storages in integration with the energy system, when optimising both
investments and operation. The model incorporates thermal building dynamics and covers various heat
storage options: passive heat storage in the building structure via radiator heating, active heat storage
in concrete floors via floor heating, and use of thermal storage tanks for space heating and hot water.
It is shown that the model is well qualified for analysing possibilities and system benefits of operating
individual heat pumps flexibly. This includes prioritising heat pump operation for hours with low
marginal electricity production costs, and peak load shaving resulting in a reduced need for peak and
reserve capacity investments.
Key words: Residential heat pumps, Flexible electricity demand, Peak load shaving, Wind power
integration, Thermal building model, Demand side management.
1. Introduction
Wind power is considered a key technology in transforming our energy systems towards sustainability.
However, the variable and partly unpredictable nature of wind challenges an effective integration of
large amounts of wind power and requires dispatchable power capacities to back it up. Flexible
technologies such as large heat pumps and heat storages in combined heat and power (CHP) systems
and electric vehicles can play a significant role in facilitating the integration of wind power [1-6].
Demand side management (DSM) is also expected to play a central role in increasing system flexibility
for incorporating larger shares of wind power [7-9]. Small electric compression heat pumps on
consumer level (individual heat pumps) can develop into an important DSM resource. Analyses show
that individual heat pumps together with district heating form the best heat supply solutions in present
and future energy systems [10, 11], and the use of heat pumps for individual heating in Denmark is
expected to grow considerably [12]. Individual heat pumps will thus represent a significant electricity
demand in future energy systems. This electricity demand becomes flexible if e.g. investments are
1
Paper IV
made in control equipment, enabling intelligent heat storage in the building structure and/or intelligent
use of existing hot water tanks, and/or if investing in heat accumulation tanks for space heating.
A group of recent studies present methods and control concepts for how individual heat pumps (or
electric heating) can be operated flexibly responding to the needs of the power system. Many of these
studies operate on building level, e.g. [13-15]. A larger scale perspective is applied in [16], which
models the operation of a high wind system with 3000 households, of which 1500 are supplied with
heat pumps using buffers for space heating and hot water. Parkinson et al. [17] introduce a demand
response control strategy aimed at thermostatically controlled electric heating and cooling of buildings.
The study simulates management of the aggregated heat pump load of 1000-2000 buildings in
regulating power fluctuations of two wind turbines. In [18], a method is presented to model and control
a population of thermostatically controlled loads applying a simulation, where the loads follow the
output of 138 wind turbines. A distribution system platform is presented in [19] for simulating control
of load flow fluctuations in the distribution grid via self-regulating heat pump cycling. Wang et al. [20]
present a strategy for optimal demand response control based on managing comfort-constrained
individual heat pumps to provide regulation and spinning reserves. A new market approach is presented
in [21] for integrating large distributed populations of heat pumps, electric vehicles, and electrolysers
as demand-side virtual power plants providing spinning reserves. In [22], a thermal building model is
presented for investigating the use of price-responsive electric space heating.
Only few studies integrate flexible operation of individual heat pumps in energy system models on
national scale. Papaefthymiou et al. [23] present a methodology for coupling a model of buildings
thermal behaviour with an electricity market model using the German energy system as a case study.
In [24], model development is presented that facilitates technical optimisation of individual heat pumps
and thermal storages in integration with the energy system minimising fuel consumption of the system
and utilising as much wind power as possible. Finally, [25] presents a simulation approach for covering
day-ahead and intra-day operation of the power system in analysing a broad range of demand response
options including cooling and electric heating in the residential sector.
The above mentioned models mainly focus on operational aspects, i.e. how individual heat pumps can
contribute to peak load shaving, load shifting, provision of ancillary services, and increasing wind
power utilisation. However, investments in heat storage options, required to enable flexible operation
of the heat pumps, could suffer from low economies of scale and will in some aspects compete with
other flexible energy technologies in the system. Moreover, the heat pumps could have important
effects on energy system investments, e.g. on the need for peak and reserve capacities. The objective
of this study has therefore been to develop a model that integrates individual heat pumps and their
flexibility options in an energy system model that optimises both operation and investments of the
whole system. Such model capability has not been identified in previous studies. The model is made
as an add-on to the energy system investment model Balmorel. The add-on covers the following heat
storage options: 1) intelligent passive1 heat storage in the building structure for houses with radiator
1
I.e. heat transfer through radiation, thermal conductance, and convection without the use of a heat transferring
media going in and out of the storage [53].
2
Paper IV
heating (or floor heating), 2) intelligent active heat storage in concrete floor heating systems, and 3)
thermal storage tanks for space heating and hot water, respectively. The model is illustrated using a
case study of a possible future Danish energy system in 2030 with around 60% wind power.
The article first describes the Balmorel model and the add-on. Subsequently, the model functionality
is illustrated by using the case study described and by comparing with measured data. In the end, a
discussion and conclusion is given including suggestions for future work. Nomenclature can be found
in Appendix at the end of the paper.
2. Balmorel
Balmorel is a deterministic partial equilibrium model assuming perfect competition [3, 26]. It is
formulated as a linear optimisation problem2 in GAMS, a high-level modelling system for
mathematical programming and optimisation. The model optimises investments in power/heat
production, storage, and transmission capacities, and the operation of the system. This is done by
minimising total costs in the energy system over a given year covering annualised investment costs,
operation and maintenance costs of existing and new units, and fuel and CO2 quota costs. The
optimisation is performed subject to a number of constraints including satisfaction of demands for
electricity and heat in each time period, renewable energy potentials, and technical restrictions on units
in the system. The CPLEX solver3 is normally used to find the optimal solution.
Balmorel operates with three geographical entities: countries, regions, and areas. Countries are divided
into regions connected with transmission lines and regions are further divided into areas. Electricity
supply and demand is balanced on regional level, whereas heating is balanced on area level. In this
study, the existing model structure in Balmorel is used for representing the integrated power and district
heating system, while individual heating of houses is modelled using the new add-on. In Balmorel, the
optimisation is performed with a yearly time horizon where the year is divided into seasons, that are
often used to represent weeks, and into time periods that are often used to represent hours.
3. Thermal building model add-on

The thermal building model renders it possible to represent houses with radiator heating as well as
houses with floor heating as the central heating system. This division is relevant since it influences the
possibilities for utilising the heat storage capacity of the building. Houses with radiator heating can
store heat passively in the building structure, while houses with concrete floor heating can foremost
store heat actively in the large thermal mass of the concrete floor (in addition heat can be stored
passively in the remaining building structures). The model also includes thermal storage tanks for space
heating and hot water, respectively, representing other options of enabling flexible heat pump
operation.
2
A mixed integer model version is also available.
3
CPLEX is a high-performance solver for linear, mixed integer, and quadratic programming.
3
Paper IV
Fig. 1 and Fig. 2 illustrate how the two different central heating systems are modelled. For simplicity,
only the space heating circuit is shown, i.e. excluding the hot water circuit.
CI, TI QIB QBO

QSH
QTLos CB
TO
HP QPel TB
ACT
QSH QVen
LDACT
CF, TF
Fig. 1. Thermal building model of houses with radiator heating (only the space heating circuit is illustrated). HP:
Heat pump (or other heating installation). ACT: Heat accumulation tank. Q SH = QSHe + QSHn.
CI, TI QIL QLO
QTLos CL
TO
HP QPel TL
ACT QVen
LDACT QFI
F F
QSH QSH C , T
Fig. 2. Thermal building model of houses with floor heating (only the space heating circuit is illustrated). HP: Heat
pump (or other heating installation). ACT: Heat accumulation tank. QSH = QSHe + QSHn.
For houses with radiator heating, the indoor air and the building structure are modelled as two separate
thermal masses. The model of floor heating systems covers three thermal masses, namely the indoor
air, floor, and walls/ceiling. The modelling of both types of central heating systems thus comprises at
least two dominating heat accumulating masses as required in order to sufficiently capture the heat
dynamics of buildings [27]. The model is based on physical definitions and relations of heat transfers
and heat capacities and can be characterised as a linear state space model. It is inspired by the modelling
approach in [22].
Through integration of the add-on into Balmorel, existing constraints on e.g. fuel consumption,
electricity balance, and relation between heat output and electricity input for heat pumps are utilised4.
In the following Sections 3.1-3.6, the new equations introduced in the model add-on are presented.
Cost elements added to the objective function are presented in Section 3.1 and the heat balance
4
Heat output = COP · electricity input, where COP is the coefficient of performance.
4
Paper IV
equations in Sections 3.2-3.4. Section 3.5 covers investment restrictions set up for individual heating
installations, while Section 3.6 deals with methodology implemented to represent peak load shaving
through flexible heat pump operation. The Sections 3.7-3.8 touch upon the influence of solar
transmission and cooling. Finally, model output is compared with measured data in Section 3.9.
3.1 Costs
The use of heat storage in the building structure for flexible heat pump operation can be facilitated by
investing in digital thermostats, enabling intelligent control of the central heating system, and in a
central controller enabling communication with the power system. In addition, frequent measurements
of indoor and ambient temperatures are required (using electronic thermometers, which are typically
included in modern individual heat pumps). The temperature of the building structure can be estimated
based on the calculated heat transfers to/from the building, as given by the temperature levels
measured. The central controller then optimises the operation of the heat pump, based on various data,
such as temperature measurements/estimations, weather forecasts, and various measurements on the
central heating and hot water circuit [24, 28].
The use of thermal storage tanks (in addition to the small hot water tank of 150-200 litres, typically
installed in combination with individual heat pumps [29, 30]) requires investment in the tank in
addition to a central controller [24, 31]. The cost components described above have been added to the
objective function in Balmorel, which covers total system costs. The added costs are shown in Eq. (1).
∑ ∈ ∑ ∈ , ∙ , +∑ ∈ ∑ ∈ , ∙ , +∑ ∈ ∙ ,′ ′
(1)
The investment cost for thermal storage tanks is a function of the cost per storage capacity ( , ) and
the invested new capacity (KN). The cost of intelligent heat storage in the building structure is
determined by the investment cost per house ( ,′ ′ ) and the number of houses investing in this
Inv
option (HS ). The following Eqs. (2)-(27) all represent constraints on the cost minimisation performed
in the optimisation.
3.2 Hot water demand

Eq. (2) balances the hot water supply (from existing, QHWe, and new heating installations, QHWn) and
demand (DHW) in each hour, also taking loading of hot water tanks (LDHWT) into account.
∑ ∈ , , , + , , , − , , = , , ∀ ∈ ; ∈ ;ℎ ∈ (2)
5
Paper IV
3.3 Space heating

General conditions
Space heating is assumed only in the heating season, in Denmark defined as September 15 to May 15
(WHSea) [32]. The need for space heating is modelled by restricting the indoor temperature (TI) to be
within a certain defined minimum (TIMin) and maximum level (TIMax), as shown in Eqs. (3)-(4). When
fixing the indoor air temperature in each hour (TIMin = TIMax = TIRef ), the heating installation is forced
to supply exactly the amount of space heating needed to cover the net heat demand in each hour, i.e.
total heat loss of the building minus heat contributions. This does not allow for intelligent heat storage
in the building structure, and is referred to as the reference situation. If allowing a given indoor
temperature span, i.e. TIMax > TIRef and TIMin < TIRef, the heating installation can be operated flexibly by
utilising intelligent passive/active heat storage in the building structure. In Eqs. (3)-(4) it is
implemented that the larger the share of houses that invest in this option (HSInv/HSTot), the larger the
possibilities for flexible heat pump operation.
, , ≥ , + , − , ∙ ∀ ∈ ; ∈ ;ℎ ∈ (3)
, , ≤ , + , − , ∙ ∀ ∈ ; ∈ ;ℎ ∈ (4)
Eq. (5) ensures that the number of houses investing in this heat storage option (HSInv) can maximum
be equal to the number of houses in the given area (HSTot).
≤ ∀ ∈ (5)
Eq. (6) ensures same average indoor temperature in the reference case and in the case with an allowed
indoor temperature span. This results in the same demand for space heating in the two cases, which
makes them comparable. The difference between the two cases thereby reflects the effect of operating
the pumps flexibly, as intended. If leaving out Eq. (6), the difference between the cases would mainly
show the effect of accepting a generally lower indoor temperature level, i.e. lower thermal comfort,
implying a reduced space heating demand.
∑ ∈ , ,
∑ ∈
≥ , ∀ ∈ ; ∈ (6)
Heat transfers occur each hour between the thermal masses in the model depending on the heat transfer
coefficients and the temperature difference between the masses. The ventilation loss through the
building envelope (QVen) depends on the heat transfer coefficient UVen and the temperature difference
between the indoor air (TI) and the ambient air (TO) (see Eq. (7)).
−6
, , = ∙ , , − , ∙ ∙ 10 ∀ ∈ ; ∈ ;ℎ ∈ (7)
The remaining heat transfer equations represent heat transfer by transmission. These equations vary
depending on whether houses with radiator heating or floor heating are considered.
6
Paper IV
Radiator heating systems

For houses with radiator heating, heat transfer between indoor air and building structure (QIB) is
covered by Eq. (8), and heat transfer between building structure and ambient air (QBO) by Eq. (9).
, , = ∙ , , − , , ∙ ∙ 10 ∀ ∈ ; ∈ ;ℎ ∈ (8)
, , = ∙ , , − , ∙ ∙ 10 ∀ ∈ ; ∈ ;ℎ ∈ (9)
The temperature of a thermal mass will increase if the total amount of heat transferred to the mass
exceeds the amount leaving it; and vice versa. This is expressed in Eq. (10) for the case of the indoor
air. The heat transferred to the indoor air covers heat supply from the heating installations (existing,
QSHe, and new QSHn), heat contribution from persons and electrical appliances (QPel), and the heat loss
from thermal storage tanks contributing to space heating (QTLos). The heat flows leaving the indoor air
cover heat losses through the building envelope (QVen, QIB) and to little extent possible cooling
(QCool,see Section 3.8). Loading of heat accumulation tanks is also accounted for (LDACT). The lower
the heat capacity of the thermal mass (CI in this case), the faster will the temperature change in response
to a net heat transfer to/from it.
, , , , , , , , , , , , , , , , , ,
∈
, , = , , + (10)
∙ ∙
∀ ∈ ; ∈ ;ℎ ∈
where , , = ∙ ∙ , , + ∙ ∙ , ,
Eq. (11) correspondingly represents the development in the temperature of the building structure (TB).
, , , ,
, , = , , + ∀ ∈ ; ∈ ;ℎ ∈ (11)
∙ ∙
where = −
Floor heating systems

For floor heating systems, the heat transfer between floor and indoor air (QFI) is covered by Eq. (12),
heat transfer between indoor air and walls/ceiling (QIL) by Eq. (13), and heat transfer between
walls/ceiling and ambient air (QLO) by Eq. (14).
, , = ∙ , , − , , ∙ ∙ 10 ∀ ∈ ; ∈ ;ℎ ∈ (12)
, , = ∙ , , − , , ∙ ∙ 10 ∀ ∈ ; ∈ ;ℎ ∈ (13)
, , = ∙ , , − , , ∙ ∙ 10 ∀ ∈ ; ∈ ;ℎ ∈ (14)
The temperature development of the floor (TF), indoor air (TI), and walls/ceiling (TL) is represented by
Eqs. (15), (16), and (17), respectively.
7
Paper IV
∑ ∈ , , , , , , , , , ,
, , = , , + ∙ ∙
∀ ∈ ; ∈ ;ℎ ∈ (15)
, , , , , , , , , , , ,
, , = , , + ∀ ∈ ; ∈ ;ℎ ∈ (16)
∙ ∙
, , , ,
, , = , , + ∀ ∈ ; ∈ ;ℎ ∈ (17)
∙ ∙
3.4 Thermal storage tanks

The equations for heat accumulation tanks, which operate on the space heating circuit only [30], are
shown in this section. Eq. (18) determines the development in the heat storage level (SACT) over time
depending on the loading (LDACT), unloading (QSHe, QSHn), and the fractional storage loss (SLACT).
, , = , , ∙ (1 − )+ , , − , , , + , , , (18)
∈
∀ ∈ ; ∈ ;ℎ ∈
Eq. (19) ensures that the storage level is kept within the available storage capacity.
, , ≤∑ ∈ , + , ∀ ∈ ; ∈ ;ℎ ∈ (19)
Eqs. (20)-(22) represent storage loading/unloading constraints.
, , , ≤ , ∀ ∈ ; ∈ ; ∈ ;ℎ ∈ (20)
, , , ≤ , ∀ ∈ ; ∈ ; ∈ ;ℎ ∈ (21)
, , ≤∑ ∈ , + , ∀ ∈ ; ∈ ;ℎ ∈ (22)
Corresponding constraints have been implemented for hot water tanks. In this case, the constraints
apply for the full year, since hot water is not only needed in the heating season.
3.5 Investment constraints for individual heating installations

Eq. (23) ensures that the total heat generation capacity is dimensioned to comply with Danish building
regulations in terms of being able to cover the dimensional heat loss at an ambient temperature of -12
°C (TODim) and an indoor temperature of 20 °C (TIDim); and satisfy the average hot water demand at the
same time (DHWAv) [33, 34].
∑ ∈ , + , ≥( + )∙ − ∙ ∙ 10 + (23)
∀ ∈
8
Paper IV
Due to Eq. (23), the required capacity of individual heating installations cannot be reduced through
flexible operation of these, as in reality. The heat transfer coefficient between indoor air and ambient
air (UIO) is identified by using principles of series and parallel connections of thermal resistances:
∙ ∙
= ∀ ∈ , = ∀ ∈
+ +
Ground source heat pumps and air/water heat pumps are typically supplemented by an electric boiler
to cover peak loads in order to reduce heat pump investment costs. In the model, this is reflected in Eq.
(24), which ensures that the heat pump is dimensioned to constitute a given share (CS) of the total
capacity, while an electric boiler covers the rest. Heat pumps are typically dimensioned to cover 72%-
82 % of the dimensional heat loss defined in Eq. (23) [35]. This corresponds to that the heat pump
covers around 99 % of the annual heat demand5.
∑ ∈ , = ∙∑ ∈ , ∀ ∈ (24)
Residents in individually heated areas will normally invest in only one type of primary heating
technology (in Denmark typically a fuel boiler or heat pump). In order to reflect this in the model, it is
therefore avoided that different types of invested primary heating technologies are used to supplement
each other within a given area; e.g. using one technology for base load operation and another for peak
load operation. Eq. (25) ensures this by setting a minimum number of full load hours (FLHMin) for new
primary individual heating technologies.
∑ ∈ , , , + , , , ∙ℎ , ≥ , ∙ ∀ ∈ ; ∈ (25)
∈
Given space requirements, it is realistic to insert a heat accumulation tank of up to around 1000 litres
in each house (VMaxACT) [29]. This is taken into account in Eq. (26).
∑ ∈ , + , ≤ ∙ ∙ ∀ ∈ (26)
3.6 Capacity balance restriction

In Balmorel, a reasonable amount of dispatchable power capacity in each power region can to some
extent be ensured by including the period of the year with highest net load, i.e. gross load minus power
from wind (and other fluctuating renewable energy sources), in the optimisation. However, this will
likely underestimate the need for peak and reserve capacity; firstly due to the assumed perfect foresight
on wind power and load. Secondly, electricity import can in the optimisation contribute to cover high
net loads, while in reality risk of failures on transmission cables should be taken into account.
Inspired by [3], a capacity balance restriction is therefore added to ensure adequate production capacity
and reserve margin in each power region (Eq. (27)). The term capacity credit (cc) is here applied,
which determines the share of total installed capacity that is available for electricity generation at a
5
For a full year simulation in the thermal building model using ambient temperatures in [51].
9
Paper IV
certain level of confidence. Based on [3], the capacity credit of conventional units is set to 0.99, while
the capacity credit for wind power is set to 0.14. Eq. (27) states that sufficient electricity generation
capacity must be available in each power region to cover the peak load; covering both conventional
electricity demand (DElConv) and flexible electricity demand of heat pumps/electric boilers (and
electricity storages) (DElFlex).
∑ ∈ ( )∑ ∈ , + , ∙ ≥ , , +∑ ∈ ( )∑ ∈ , , , (27)
∀ ∈ ; ∈ ;ℎ ∈
The equation reflects that the need for production capacity and reserves can be reduced through peak
load shaving (by reducing DElFlex). As the constraint is applied for each hour and week it allows for
lowering the heat pump operation in peak load hours taking the hourly pattern of conventional and
flexible electricity demand into account. This is a new development compared to the capacity balance
restriction in [3], where the right hand side of the equation is a constant peak load set for each country.
3.7 Solar transmission

Heat contribution from solar transmission through windows has been excluded due to lack of data.
This delimitation can also be found in other thermal building model simulations, e.g. [15, 22, 36]. The
heat contribution from solar transmission will vary greatly from house to house depending on in
particular the total window area, the orientation of the windows, the shading effects from surroundings,
and transmittance of the window type (depending on glazing an number of glass panes) [37, 38].
According to estimates in [39], the largest share of the total free heat contribution6 over the heating
season in existing one-family houses however stem from electrical appliances and persons i.e. around
2/3, which is included in this model. The relative influence of solar transmission will generally be
highest in the most well insulated houses. New houses constructed towards 2030, having very low heat
demands (10 kWh space heating annually per m2), have therefore not been included in this study. This
delimitation is reasonable since new houses only constitute about 6 % of the individual heat demand
expected by 2030.
3.8 Cooling
The possibility of cooling has been introduced to ensure that the indoor temperature can be kept below
TIMax in all periods of the year. In the optimisation, the possibility of cooling is only used in houses
with (concrete) floor heating and mainly when requiring a fixed indoor temperature. Since the floor
heated houses constitute only 5 % of the modelled individual heat demand the cooling has no
importance for the model illustration.
6
I.e. contribution from other sources than the heat installation.
10
Paper IV
3.9 Comparison with measured data

Measurements of 28 Danish heat pump installations in one-family houses indicate that a constant
indoor temperature setting over the day/night is typical in the reference situation (21.5 °C on average)
[40]. Model generated diurnal space heating profiles for this indoor temperature setting (21.5 °C) have
been compared with corresponding measured profiles for Danish heat pump installations (see Fig. 3).
Nov 1 - Dec 31 Mar 15 - May 15, Sep 15 - Dec 31

0.07 0.07

Measured av.
Measured av.
0.06 profile
0.06 profile
0.05 0.05
0.04 Modelled
0.04 profile, Modelled profile,
estimation A estimation A
0.03 0.03
0.02 0.02
Modelled profile, Modelled profile,
0.01 0.01
estimation B estimation B
0.00 0.00
1 3 5 7 9 11 13 15 17 19 21 23 1 3 5 7 9 11 13 15 17 19 21 23
Hour Hour
Fig. 3. Model generated average diurnal space heating profile compared to a corresponding measured average
profile for heat pumps in Danish one-family houses [40]. Radiator heating systems are illustrated. Estimation A: UBO
for monitored houses estimated based on construction year. Estimation B: UBO for monitored houses estimated based
on measured space heating over the given period. a) Comparison covering November 1 to December 31 (data
available for 24 houses). b) Comparison covering March 15 to May 15 and September 15 to December 31 (data
available for 5 houses).
Fig. 3 indicates that the thermal building model gives a reasonable representation of actual space
heating profiles. Furthermore, it confirms that the assumption of constant indoor temperature setting
in the reference situation is realistic. The larger diurnal variation observed in the right part of Fig. 3 is
mainly explained by the fact that this period also covers autumn/spring periods, where there is a larger
relative influence of diurnal ambient temperature variations and heat contributions from solar
transmission, persons, and appliances.
The measured space heating profile has a valley around mid-day, when solar transmission is highest.
The valley of the model generated space heating profile occurs a few hours later, due to the exclusion
of solar transmission (see Section 3.7), and since the heat contribution from electrical appliances (and
persons) is highest in the early evening. Fig. 3 shows that the modelled space heating profile is in some
situations potentially7 a bit lower than the measured profile, in the hours where conventional electricity
demand typically peaks, i.e. around 8.00-9.00 and 18.00 o’clock (in Denmark). This means that the
model could slightly underestimate the peak load shaving achieved when operating the heat pumps
flexibly. However, the deviation between the profiles is acceptable as it is believed to be within the
general uncertainties related to analyses of individual heat pumps on energy system level: the
uncertainty of representing a large-scale building stock on aggregated level (covering around 716 000
7
If assuming that the estimation procedure B is the best approximation of the building stock data (see Fig. 3).
11
Paper IV
houses for the Danish case) and the uncertainties on other input data (fuel and CO2 prices, investment
costs for technologies etc.).
4. Application
For the illustrative case study of the Danish energy system by 2030, the most important preconditions
are outlined in the following. Energy demands and fuel potentials are based on the Future scenario in
the report of The Danish Commission on Climate Change Policy [41]. Technology data are mainly
based on technology catalogues developed by the Danish Energy Agency [42, 43] and secondarily on
[10, 44]. Electricity trade with neighbouring countries is taken into account by including Germany,
Norway, Sweden, and Finland in the optimisation.
The thermal building model add-on is applied for the stock of existing individually heated one-family
houses in 2030. Houses with radiator heating are assumed to constitute 95 % of this heat demand. This
is based on the tradition for installing radiator-based central heating systems in new houses until the
1990'ies [45, 46] and the high costs of installing floor heating in existing houses [24, 47]. The one-
family houses are divided into 3 different categories, depending on the year of construction, reflecting
changes in insulation standards. The heat capacity of the building that can be utilised for passive heat
storage diurnally (effective heat capacity) is based on typical values for Danish buildings as given in
[48] and on the following general distribution 60 Wh/°C/m2 floor area: 25 % of the houses, 100
Wh/°C/m2 floor area: 50 % of the houses, and 140 Wh/°C/m2 floor area: 25 % of the houses [49]. The
effective heat capacity of the building is mainly comprised by internal construction in walls, ceiling,
and floor, while windows, doors, and furniture have minor influence [24]. For houses with floor
heating, a typical concrete layer of 10 cm [46, 50] has been assumed representing a thermal mass of
67 Wh/°C/m2 floor area.
Overall, the stock of individually heated one-family houses is aggregated into 10 different areas both
in Eastern and Western Denmark (20 areas in total). Based on a need to limit computation time, five
weeks are simulated to represent a full year; hereof four weeks in the heating season and one summer
week. Weighting factors are applied to account for the period length that each of the simulated weeks
represent. The heat transfer coefficients for the houses, UBO and ULO, have then been calibrated to give
the annual space heating demand for each building category. The calibrated heat transfer coefficients
deviate less than 2 % from the heat transfer coefficients corresponding to a full year simulation. This
shows that the simulated weeks give a very good approximation of the space heating demand over a
full year.
Typical diurnal and seasonal variations in ambient temperatures (TO) are represented using hourly
average temperature data for the Danish design reference year [51]. The hourly hot water demand
variation is based on a demand profile on aggregated level for a large number of Danish households
[52]. The average heat contribution from persons and electrical appliances is set to 5 W/m2 floor area;
hereof 1.5 W/m2 from persons and 3.5 W/m2 from electrical appliances [48]. The variation in the heat
contribution from electrical appliances is based on the aggregated electricity demand variation for
Danish households.
12
Paper IV
5. Results illustrating model functionality

Fig. 4 shows the development in space heating from the heat pumps in a situation without allowing
heat storage investments, and where the required indoor temperature is fixed to 21.5 °C. In this
situation, the heat pumps are restricted to provide the space heating needed to cover the net heat
demand in each hour, i.e. transmission and ventilation loss minus heat contribution from other sources
(persons and electrical appliances in this case). It can be seen that when the ambient temperature is
low, space heating is high due to high heat losses.
4500 20 Space heating

4000 15
3500 10
Ambient temp. ( C)
3000 Transmission loss
5 from indoor air to
MW-th
2500
0 construction
2000
-5 Heat from persons
1500 and appliances
1000 -10
500 -15
Ventilation loss
0 -20
T113
T117
T001
T021
T041
T061
T081
T101
T121
T141
T161
T013
T033
T053
T073
T093
T133
T153
T005
T025
T045
T065
T085
T105
T125
T145
T165
T017
T037
T057
T077
T097
T137
T157
Ambient
S02 S12 S40 S51 temperature ( C)
Week, hour
Fig. 4. Space heating from individual heat pumps, other heat contributions, heat losses, and ambient temperatures
for situation without heat storage investments and with a fixed indoor temperature requirement of 21.5 °C (the
simulated heating season is illustrated focusing on houses with radiator heating).
When allowing investments in thermal storages complementing the heat pumps, investments in
intelligent heat storage in the building structure are observed. In houses investing in this option, the
indoor temperature is allowed to vary within an interval of 20-23 °C during the day and 19-22 °C at
night, ensuring thermal comfort [14, 15, 40]. The optimisation suggests that it is socio-economically
feasible to invest in this heat storage option in around 34 % of heat pump installations in one-family
houses. The optimisation reveals no feasible investments in heat accumulation tanks (or hot water
tanks). The feasibility of investing in control equipment, enabling intelligent use of the existing small
hot water tanks, has not been tested in this model illustration. The investments in intelligent heat
storage in the building structure are observed across all the defined categories of insulation levels,
central heating systems, and heat capacities of the building structure (see Fig. 5).
13
Paper IV
60
Share of houses inv. in
50
HP flexibility (% )
Effective heat capacity

40 of the building :
30 60 Wh/°C/m2
20 100 Wh/°C/m2
10 140 Wh/°C/m2
0
Low Med High High
insul. insul. Insul. Insul.
Radiator h. Floor h.
Fig. 5. Shares of existing individually heated one-family houses investing in intelligent heat storage in the building
structure for flexible heat pump operation. Houses with (concrete) floor heating will foremost utilise the thermal
mass of the concrete floor of 67 Wh/°C/m2 floor area.
However, the investments in intelligent heat storage in the building structure are generally larger the
larger the heat capacity of the building. This is expected, since the possibilities for flexible heat pump
operation utilising heat storage in the building structure will be higher the larger the heat capacity of
the building. Fig. 5 furthermore indicates a higher feasibility of investing in intelligent heat storage in
the building structure in houses with lower insulation level. This might seem counterintuitive since low
insulation level results in higher heat losses in the process of storing heat in the building structure. The
explanation is that a low insulation level corresponds to a higher space heating demand per house. This
means higher electricity consumption for heat pump operation and thus a higher potential benefit of
enabling the storage in absolute terms.
Fig. 6 illustrates the modelled flexible operation of the heat pumps. The electricity price on the figure
represents the marginal cost of generating electricity in energy system optimisation. It can be seen that
the electricity consumption of the heat pumps is low, when the electricity price is high and vice versa.
This shows that the model works as it should, i.e. that investments in heat storage capability render it
possible to optimise the heat pump operation in integration with the energy system.
1600 140
1400 120
1200 Electricity
100 consumption for
€/MWh-e
1000 individual heat

MW-e
80
800 pumps (MW-e)
60
600
400 40
Electricity price
200 20
(€/MWh-e)
0 0
T001
T021
T041
T061
T081
T101
T121
T141
T161
T013
T033
T053
T073
T093
T113
T133
T153
T005
T025
T045
T065
T085
T105
T125
T145
T165
T017
T037
T057
T077
T097
T117
T137
T157
S02 S12 S40 S51

Week, hour
Fig. 6. Modelled flexible operation of individual heat pumps over the heating season shown for Western Denmark
by 2030.
14
Paper IV
The use of an aggregated system model means that a heat storage capacity in a given area can be shared
among all heat pumps within the area. This is evident from Fig. 6, where the heat storage investments
realised in 34 % of the houses render it possible to operate a larger share of the heat pumps flexibly to
some extent. The flexibility offered by heat storages can thus be distributed among the houses, which
implies best-case conditions. Ideally, each of the many thousands of individually heated houses should
be modelled separately (716,000 in this case). However, it would hardly be realistic to run such a
model.
As illustrated in Fig. 7, the flexible heat pump operation contributes to peak load shaving due to the
capacity balance restriction implemented in Eq. (27). The week with the highest peak loads of the year
is illustrated representing the week determining for the need for peak and reserve capacities in the
optimisation.
Non-flexible operation of indiv. heat pumps Flexible operation of indiv. heat pumps
7000 7000
Electricity consumption Electricity consumption
6000 for6000
individual heat pumps for individual heat pumps
(MW-e) (MW-e)*
5000 5000
MW-e
MW-e
4000 4000 consumption

Electricity Electricity consumption
for large heat pumps and for large heat pumps and
3000 3000 boilers (MW-e)
electric electric boilers (MW-e)
2000 2000
Conventional electricity Conventional electricity
1000 1000 (MW-e)
demand demand (MW-e)
0 0
T001
T011
T021
T031
T041
T051
T061
T071
T081
T091
T101
T111
T121
T131
T141
T151
T161
T001
T011
T021
T031
T041
T051
T061
T071
T081
T091
T101
T111
T121
T131
T141
T151
T161
Hour Hour
Fig. 7. Peak load shaving achieved due to flexible operation of individual heat pumps. The week with the highest
peak loads of the year is illustrated (week no. 51). *Incl. supplemental electric boilers.
The peak load shaving is facilitated by using the heat storage capacity of the building for shifting the
load of the individual heat pumps a few hours within the day: from the peak load hours in the morning
and evening to the load valleys in the afternoon; and to some extent also to the night. As a result, the
investment in peak/reserve capacity is in the optimisation reduced by 440 MW corresponding to the
size of a large power plant.
Fig. 7 illustrates that if peak loads were to be shaved down further it would require lowering the
individual heat pump load further in longer consecutive periods lasting up to 15 hours (hour T056 to
T070). This is difficult since ambient temperatures are very low in this week (see Fig. 8). As a result,
the individual heat pumps are forced to operate at their capacity limit in longer periods, and the electric
boilers needed for covering peak loads. This means that the capacity margin available for load shifting
is limited regardless of the heat storage possibilities.
15
Paper IV
2000 4
Electricity
1800 2 consumption for indiv.
1600 0 electric boilers
Ambient temp. ( C)
1400 -2 (MW-e)
MW-e
1200 -4
1000 -6 Electricity
800 -8 consumption for indiv.
600 -10 heat pumps
(MW-e)
400 -12
200 -14
0 -16 Ambient temperature
( C)
T111
T001
T012
T023
T034
T045
T056
T067
T078
T089
T100
T122
T133
T144
T155
T166
Hour
Fig. 8. Electricity consumption for individual heat pumps and supplemental electric boilers shown for the week with
the highest peak loads of the year (week no. 51). The situation without flexible heat pump operation is illustrated.
The two cases exhibit different computational times. The case without flexible heat pump operation,
i.e. fixed indoor temperature (and no allowed investments in thermal storage tanks), takes only about
3 hours to complete. The case with flexible operation of the heat pumps is more complex and takes
around 16 hours. These computational times are considered reasonable and acceptable given the size
of the problem.
Overall, the results presented illustrate that the model is well qualified for analysing the possibilities
and system effects of flexible heat pump operation.
6. Discussion and Conclusion

We have presented a model that renders it possible to analyse individual heat pumps and
complementing heat storages in integration with the energy system when optimising both investments
and operation. The model covers various heat storage options: i.e. passive heat storage in the building
structure via radiator heating, active heat storage in concrete floors via floor heating, and use of thermal
storage tanks for space heating and hot water, respectively. Through comparison with measured data,
it is indicated that the model gives a reasonable representation of actual space heating profiles.
The capability of the model is illustrated in a case study of the Danish energy system by 2030 with
optimised wind penetrations of around 60 %. The case study shows that the model is well qualified for
analysing the possibilities and system benefits of operating individual heat pumps intelligently. This
includes prioritising heat pump operation for hours with low marginal electricity production costs and
reducing required investments in peak and reserve capacity through peak load shaving.
As the model is linear, it does not include start-up costs, minimum load requirements, or part load
efficiencies. Moreover, the model assumes perfect foresight and therefore does not include power
balancing responding to errors in forecasts of wind and load; only requirements for reserve capacities
are included. As illustrated in [23], start-up costs and minimum load requirements for thermal power
plants can have a significant influence on the operation cost savings obtained when using individual
heat pumps for demand side management. This means that potentials for reducing operation costs
16
Paper IV
through flexible heat pump operation are conservatively represented in the model. However, we
incorporate effects on energy system investments in contrast to previous models focusing on individual
heat pumps.
As in many other energy system models, the heat pumps in this study were represented by a yearly
average COP. It could be interesting to include hourly COP variations in the model caused by variations
in the heat source temperature and in the output temperature of the heat pump. Another subject for
further research would be to model how flexible operation of the heat pumps influences required
enhancements in the distribution grids. The inclusion of heat contribution from solar transmission in
the model also forms a relevant subject for future work.
Nevertheless, the model development is considered a big step forward, in providing the methodology
needed for analysing individual heat pumps and heat storages in an energy system context. It would be
interesting to use the model for analysing the impact that heat pumps can have in terms of integrating
wind power at a national level, including impacts on investments, system costs, fuel consumption, and
CO2 emissions.
Acknowledgments
The authors would like to thank Marie Münster for giving constructive comments on the article and
Per Heiselberg, Aalborg University, and Helge Christensen, Danfoss, for inputs to preconditions for
the thermal building model. We also owe thanks to the reviewers for valuable comments on the article.
APPENDIX
Nomenclature
Indices
a, A heating area, set of heating areas
AInd set of individually heated areas
ARad , AFlo set of individually heated areas with radiator heating and floor heating, respectively
g, G generation technology/storage, set of generation technologies/storages
G ACT, GHWT set of individual heat accumulation tanks and hot water tanks, respectively
GH set of individual heating technologies (heat pumps, boilers, solar thermal and heat storages)
G El set of electricity generation technologies
G PH set of primary individual heating technologies (heat pumps, fuel boilers)
G HPi, GEBi set of individual heat pumps and electric boilers, respectively
h, H, hL time period (hour), set of time periods simulated, length of time period (hours)
h++1 Gams notation ensuring circular connection, i.e. that the value in the last hour of the week is linked to the
value in the first hour of the week
r, R region, set of regions
w, W, WHsea week, set of weeks simulated, set of weeks in the heating season
Variables
D ElFlex electricity consumption for heat pumps/electric boilers and loading of electricity storages (MW)
LDACT, LD HWT loading of hot water tank/heat accumulation tank (MW)
HSInv number of houses investing in intelligent heat storage in the building structure (houses)
17
Paper IV
KN capacity of new generation technologies (MW) and storages (MWh)

SHWT , S ACT storage level in hot water tank, storage level in heat accumulation tank (MWh)
TB temperature of building structure (°C)
TF temperature of floor (°C)
TI temperature of indoor air (°C)
TL temperature of walls/ceiling (°C)
Q Cool cooling (MW)
Q HWe , Q HWn hot water generation from existing and new technologies, respectively (MW)
Q SHe , QSHn space heating from existing and new technologies, respectively (MW)
Q Ven ventilation loss through building envelope (MW)
Q IB heat transfer from indoor air to building structure (MW)
Q BO heat transfer from building structure to ambient air (MW)
Q FI heat transfer from floor to indoor air (MW)
Q IL heat transfer from indoor air to walls/ceiling (MW)
Q LO heat transfer from walls/ceiling to ambient air (MW)
Q TLos heat loss from thermal storage tanks contributing to space heating (MW)
Parameters
AR heated floor area (m2)
CB effective heat capacity of building structure (Wh/°C/m2 floor area)
CI heat capacity of indoor air, furniture and for radiator heating systems, also water in radiators (Wh/°C/m2
floor area)
CFur heat capacity of furniture (Wh/°C/m2 floor area)
CF heat capacity of concrete floor heating system incl. concrete and water in tubes (Wh/°C/m2 floor area)
CPas effective heat capacity of building incl. furniture (Wh/°C/m2 floor area)
CL effective heat capacity of walls/ceiling (Wh/°C/m2 floor area)
CS heat pump capacity share ( )
cc capacity credit of electricity generation unit ( )
coInv annualised investment costs for given heat storage option (€/MWh/yr or €/house/yr)
D ElConv conventional electricity demand, i.e. excl. heat pumps, electric boilers, and electricity storage (MW)
D HW hot water demand (MW)
D HWAv average hot water demand (MW)
FLHmin minimum full load hours required (hours)
HSTot total number of houses (houses)
KE capacity of existing generation technologies (MW) and storages (MWh)
QPel heat contribution from persons and electrical appliances (MW)
SLHWT, SLACT stationary heat loss from hot water tank/heat accumulation tank in share of heat stored per hour ( )
SCHWT, SCACT share of heat loss from hot water tank/heat accumulation tank contributing to space heating ( )
SVACT heat content per volume for heat accumulation tank (MWh/litre)
TO ambient (outdoor) air temperature (°C)
T IDim dimensioning indoor temperature (°C)
T ODim dimensioning ambient temperature (°C)
T IMin minimum indoor temperature required (°C)
T IMax maximum indoor temperature required (°C)
T IRef indoor temperature setting in reference situation without flexible heat pump operation (°C)
U BO heat transfer coefficient between building structure and ambient air (W/m2 floor area/°C)
U IB heat transfer coefficient between indoor air and building structure (W/m2 floor area/°C)
U IL heat transfer coefficient between indoor air and walls/ceiling (W/m2 floor area/°C)
18
Paper IV
U LO heat transfer coefficient between walls/ceiling and ambient air (W/m2 floor area/°C)
U Vent heat transfer coefficient for ventilation loss through building envelope (W/m2 floor area/°C)
U IO total heat transfer coefficient for transmission loss through building envelope (W/m2 floor area/°C)
VMaxAct maximum heat accumulation tank size per house due to space requirements (litre/house)
References
[1] Liu w, Lund H, Mathiesen BV. Large-scale integration of wind power into the existing Chinese
energy system. Energy 2011;36(8):4753-4760.
[2] Østergaard PA. Wind power integration in Aalborg Municipality using compression heat pumps
and geothermal absorption heat pumps. Energy 2013;49(1):502-508.
[3] Kiviluoma J, Meibom P. Influence of wind power, plug-in electric vehicles, and heat storages on
power system investments. Energy 2010;35(3):1244-1255.
[4] Hedegaard K, Ravn H, Juul N, Meibom P. Effects of electric vehicles on power systems in
Northern Europe. Energy 2012;48(12):356-368.
[5] Lund H. Large-scale integration of wind power into different energy systems. Energy
2005;30(10):2402-2412.
[6] Bellekom S, Benders R, Pelgröm S, Moll H. Electric cars and wind energy: Two problems, one
solution? A study to combine wind energy and electric cars in 2020 in The Netherlands. Energy
2012;45(9):859-866.
[7] Strbac G. Demand side management: Benefits and challenges. Energy Policy 2008; 36(12):4419-
4426.
[8] Keane A, Tuohy A, Meibom P, Denny E, Flynn D, Mullane A, O’Malley M. Demand side
resource operation on the Irish power system with high wind power penetration. Energy Policy
2011;39(5):2925-2934.
[9] Fitzgerald N, Foley AM, McKeogh E. Integrating wind power using intelligent electric water
heating. Energy 2012;48(12):135-143.
[10] Lund H, Möller B, Mathiesen BV, Dyrelund A. The role of district heating in future renewable
energy systems. Energy 2010;35(3):1381-1390.
[11] Münster M, Morthorst PE, Larsen HV, Bregnbæk L, Werling J, Lindboe HH, et al. The role of
district heating in the future Danish energy system. Energy 2012;48(12):47-55.
[12] Wittrup S. A large amount of uncontrollable heat pumps requires more power plants [in Danish]
[Et væld af "ustyrlige" varmepumper stiller krav om flere kraftværker]. Ingeniøren/Energi&Miljø,
March 2, 2010. Available at: http://ing.dk/artikel/106844-et-vaeld-af-ustyrlige-varmepumper-stiller-
krav-om-flere-kraftvaerker. Accessed on February 3, 2013.
19
Paper IV
[13] Hong J, Johnstone C, Torriti J, Leach M. Discrete demand side control performance under
dynamic building simulation: A heat pump application. Renewable Energy 2012;39(3):85-95.
[14] Pedersen TS, Andersen P, Nielsen KM, Staermose, HL, Pedersen PD. Using Heat Pump Energy
Storages in the Power Grid. In: 2011 IEEE International Conference on Control Applications,
Denver, CO, USA; 2011.
[15] Halvgaard R, Kjølstad NK, Madsen H, Jørgensen JB. Economic model predictive control for
climate control of buildings in a smart energy system. In: 50th IEEE Conference on Decision and
Control and European Control Conference, Orlando, USA; 2011.
[16] MacDougall P, Warmer C, Kok K. Raising the accommodation ceiling for wind power by
intelligent response of demand and distributed generation. In: 10th International Workshop on Large-
Scale Integration of Wind Power into Power Systems as Well as on Transmission Networks for
Offshore Wind Power Plants, Aarhus, Denmark; 2011.
[17] Parkinson S, Wang D, Crawford C, Djilali N. Comfort-Constrained Distributed Heat Pump

Management. Energy Procedia 2011;12:849-855.
[18] Callaway DS. Tapping the energy storage potential in electric loads to deliver load following
and regulation, with application to wind energy. Energy Conversion and Management
2009;59(5):1389-1400.
[19] Williams T, Wang D, Crawford C, Djilali N. Integrating renewable energy using a smart
distribution system: Potential of self-regulating demand response. Renewable Energy 2013;52(4):46-
56.
[20] Wang D, Parkinson S, Miao W, Jia H, Crawford C, Djilali N. Online voltage security
assessment considering comfort-constrained demand response control of distributed heat pump
systems. Appl. Energy 2012;96(8):104-114.
[21] Wang D, Parkinson S, Miao W, Jia H, Crawford C, Djilali N. Hierarchical market integration of
responsive loads as spinning reserve. Appl. Energy 2013;104(4):229-238.
[22] Nyeng P. System integration of distributed energy resources - ICT, ancillary services and
markets. PhD thesis. Kgs. Lyngby, Denmark: Technical University of Denmark, Department of
Electrical Engineering; 2010.
Engineering, Electric Energy Systems, Kgs. Lyngby, Denmark, 2010.
[23] Papaefthymiou G, Hasche B, Nabe C. Potential of Heat Pumps for Demand Side Management
and Wind Power Integration in the German Electricity Market. IEEE Transactions on Sustainable
Energy 2012;3:636-642.
[24] Hedegaard K, Mathiesen BV, Lund H, Heiselberg P. Wind power integration using individual
heat pumps - Analysis of different heat storage options. Energy 2012;47:284-293.
20
Paper IV
[25] Klobasa M. Analysis of demand response and wind integration in Germany's electricity market.
IET Renewable Power Generation 2010;4:55-63.
[26] Ravn H, Hindsberger M, Petersen M, Schmidt R, Bøg R, Grohnheit PE, et al. Balmorel: a model
for analyses of the electricity and CHP markets in the Baltic Sea region. Ballerup, Denmark.
Available at: http://www.balmorel.com/; 2001.
[27] Madsen H, Holst J. Estimation of continuous-time models for the heat dynamics of a building.
Energy Build. 1995;22(3):67-79.
[28] Braun JE. Load control using building thermal mass. J. Solar Energy Eng. 2003;125:292-301.
[29] Jensen JB, Hvenegaard CM, Pedersen SV. The little blue about heat pumps [in Danish] [Den
lille blå om varmepumper]. Copenhagen, Denmark: Danish Energy Association; 2011.
[30] Andersen, JJ. Vølund Varmeteknik. Personal communication (August 2010, January 2012).
[31] Danish Energy Association. Heat pumps as part of the flexible electricity demand, background
note for the report: The little blue about heat pumps [in Danish] [Varmepumper som en del af det
fleksible elforbrug, baggrundsnotat til rapporten Den lille blå om varmepumper]. Copenhagen,
Denmark: Danish Energy Association; 2010.
[32] Mylin Energi Consult ApS. Good energy advices [in Danish] [Gode energiråd]. Available at:
http://mylliin.s-11.dk/default.asp?pid=72. Accessed on January 5, 2013.
[33] Danish Energy Agency. The Danish building regulation [In Danish] [Bygningsreglementet].
Available at: http://www.bygningsreglementet.dk/br10_02_id123/0/42. Accessed on December 7, 2012.
[34] Pedersen, SV. Danish Technological Institute. Personal communication (January, 2013).
[35] Jensen KA. Dimensioning of heat pumps most important [in Danish] [Dimensionering af
varmepumper vigtigst]. Glostrup, Denmark: TEKNIQ The Electricians' Organisation. Available at:
http://www.tekniq.dk/PresseOgNyheder/Nyheder/2010/Marts/Dimensionering-af-varmepumper-er-
vigtigst.aspx ; 2010.
[36] Tahersima F, Stoustrup J, Soroush MA, Rasmussen H. Contribution of domestic heating systems
to smart grid control. In: 6th Dubrovnik Conference on Sustainable Development of Energy Water
and Environment Systems, Dubrovnik, Croatia; 2011.
[37] Duer K, Svendsen S, Mogensen MM, Laustsen JB. Energy labelling of glazings and windows in
Denmark: calculated and measured values. Solar Energy 2002;73(7):23-31.
[38] Wittchen, KB. Danish Building Research Institute. Personal communication (March 2012).
[39] Wittchen KB. Potential energy savings in existing buildings [Potentielle energibesparelser i det
eksisterende byggeri] [in Danish]. Hørsholm, Denmark: Danish Building Research Institute, Aalborg
University; 2009.
21
Paper IV
[40] Energinet.dk. Measurements of heat pump installations in Danish households, from the
demonstration project Control your heat pump. Fredericia, Denmark: Energinet.dk. Available at:
www.styrdinvarmepumpe.dk. Accessed on March 2, 2012.
[41] Danish Commission on Climate Change Policy. Documentation for the report of the Danish
Commission on Climate Change Policy [in Danish] [Dokumentationsdelen til klimakommissionens
samlede rapport: Grøn energy - vejen mod et dansk energisystem uden fossile brændsler.
Copenhagen, Denmark: Danish Commission on Climate Change Policy; 2010.
[42] Danish Energy Agency. Technology data for energy plants. Copenhagen, Denmark: Danish
Energy Agency; 2010.
[43] COWI. Danish Technological Institute and Danish Gas Technology Centre. Technology data for
energy plants - individual heating plants and energy transport. Copenhagen, Denmark: Danish
Energy Agency and Fredericia, Denmark: Energinet.dk; 2012.
[44] International Energy Agency. Tackling investment challenges in power generation. Paris,
France: International Energy Agency. Available at:
http://www.iea.org/textbase/nppdf/free/2007/tackling_investment.pdf; 2007.
[45] Christian HC. Danish Technological Institute. Personal communication (March, 2012).
[46] Mortensen K. Saint Gobain Weber (producer of construction materials). Personal

communication (November, 2012).
[47] Bolius knowledge center for house owners. Floor heating in older houses [in Danish]
[Gulvvarme i ældre huse]. Bolius Knowledge center for house owners, November 5, 2006.
Available at: http://www.bolius.dk/alt-om/energi/artikel/gulvvarme-i-aeldre-huse/. Accessed on

November 7, 2012.
[48] Aggerholm S, Grau K. SBi Instruction 213: Buildings' Energy Demands, PC Programme and
Calculation Instruction [in Danish] [SBi-Anvisning 213: Bygningers Energibehov, Pc-Program Og
Beregningsvejledning]. Hørsholm, Denmark: Danish Building Research Institute; 2008.
[49] Olsen, L. Danish Technological Institute. Personal communication (June, September, 2010,
April 2011, January 2012).
[50] Chrisensen H. Danfoss, floor heating department. Personal communication (November, 2012).
[51] Wang PR, Scharling M, Nielsen KP. 2001 - 2010 Design reference year for Denmark.
Copenhagen, Denmark: Danish Meteorological Institute; 2012.
[52] Lund H, Mathiesen BV. Note for the Danish society of engineers' energy year [in Danish] [Notat
til IDA energiår]. Aalborg, Denmark: Aalborg University, Denmark; 2006.
22
Paper IV
[53] Duer K, Dalsgaard T, Ellehauge K, Ladekarl P. Status for active heat storage in buildings [in
Danish] [Status for aktiv varmelagring i bygninger]. Lyngby, Denmark: DTU Civil Engineering,
Technical University of Denmark; 2001.
23
Paper V
Influence of individual heat pumps on wind power

integration – Energy system investments and operation
Karsten Hedegaard & Marie Münster.
Published in Energy Conversion and Management. vol. 75, pp. 673-684. November, 2013
(published online September 11, 2013).
Energy Conversion and Management 75 (2013) 673–684
Contents lists available at ScienceDirect
Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman
Influence of individual heat pumps on wind power integration – Energy

system investments and operation
Karsten Hedegaard ⇑, Marie Münster
Systems Analysis Division, DTU Management Engineering, Frederiksborgvej 399, P.O. Box 49, Build. 130, DK-4000 Roskilde, Denmark
Article history: Individual heat pumps are expected to constitute a significant electricity demand in future energy sys-
Received 11 April 2013 tems. This demand becomes flexible if investing in complementing heat storage capabilities. In this study,
Accepted 11 August 2013 we analyse how the heat pumps can influence the integration of wind power by applying an energy sys-
tem model that optimises both investments and operation, and covers various heat storage options. The
Danish energy system by 2030 with around 50–60% wind power is used as a case study. Results show that
Keywords: the heat pumps, even without flexible operation, can contribute significantly to facilitating larger wind
Residential heat pumps
power investments and reducing system costs, fuel consumption, and CO2 emissions. Investments in heat
Flexible electricity demand
Demand side management
storages can provide only moderate system benefits in these respects. The main benefit of the flexible
Peak load shaving heat pump operation is a reduced need for peak/reserve capacity, which is also crucial for the feasibility
Thermal building model of the heat storages. Socio-economic feasibility is identified for control equipment enabling intelligent
Optimisation heat storage in the building structure and in existing hot water tanks. In contrast, investments in new
heat accumulation tanks are not found competitive.
1. Introduction structure and/or in existing hot water tanks, and/or if investing

in heat accumulation tanks for space heating. This can potentially
Wind power has great potential in mitigating CO2 emissions provide system benefits in terms of e.g. rendering wind power
and improving security of supply through displacing use of fossil investments more attractive and reducing the need for peak and
fuels in the energy system. The share of wind power is high in reserve capacities. However, investments in heat storages in the
the Danish energy system, around 30% of annual electricity de- households could suffer from low economies of scale and will in
mand [1], and the Government plans to increase this share to by some aspects compete with other flexible energy technologies in
2020 [2]. This makes the Danish case interesting for analysing chal- the system. This makes it relevant to analyse individual heat
lenges of integrating wind power that may be faced by other coun- pumps and supplementing heat storages in integration with the
tries in the future. energy system, optimising both investments and operation.
Previous studies have shown that large heat pumps, electric Only few studies evaluate the effects that flexible operation of
boilers, and heat storages in combined heat and power (CHP) sys- individual heat pumps can have on the system by applying a
tems, as well as electric vehicles, can play a significant role in facil- large-scale energy system model. Papaefthymiou et al. [17] assess
itating the integration of wind power [3–8]. Similarly, demand side the system benefits of operating a large amount of individual heat
management can contribute significantly to increasing system pumps flexibly, using the building stock as thermal storage. Apply-
flexibility [9–13], and individual heat pumps in the residential sec- ing a case study of the German power system in 2030 with a
tor could also contribute in this regard. Analyses show that individ- renewable electricity share of 50–70% (hereof, two thirds wind
ual heat pumps together with district heating form the best heat power), it is found that flexible operation of the heat pumps can re-
supply solutions in present and future energy systems [14,15]. Fur- duce system operation costs with 0.45–0.95% corresponding to 25–
thermore, the use of individual heat pumps is expected to grow 40 €/house/year [17]. Hedegaard et al. [18] compare the potentials
considerably in Denmark [16], thereby developing into a signifi- of individual heat pumps and different heat storage options in
cant part of the total electricity demand. The electricity demand terms of facilitating a fuel-efficient and cost-effective integration
of the heat pumps becomes flexible if e.g. investing in control of wind power. In a case study of the Danish energy system in
equipment, enabling intelligent heat storage in the building 2020 with 50% wind power, it is shown that the installation of heat
pumps significantly reduces fuel consumption, while flexible oper-
ation provides only moderate fuel savings. Klobasa [19] analyses a
⇑ Corresponding author. Tel.: +45 4677 5158; fax: +45 4677 5199. broad range of demand response potentials and costs, including
E-mail address: [email protected] (K. Hedegaard).
0196-8904/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.enconman.2013.08.015
674 K. Hedegaard, M. Münster / Energy Conversion and Management 75 (2013) 673–684
cooling and electric heating in the residential sector, and compares the system. The model is deterministic assuming perfect foresight
them to costs of conventional balancing technologies. and the linear version applied does not include start-up costs, min-
The above mentioned studies mainly focus on operational as- imum load requirements, or part load efficiencies [3,24].
pects without investigating how flexible heat pump operation The model operates with three geographical entities, namely
can influence energy system investments. In reports by the Danish countries, regions, and areas. Countries are divided into regions
TSO and the Danish Energy Association [20,21] it is estimated that connected with transmission lines, while regions are further di-
flexible heat pump operation can significantly reduce investments vided into areas. Electricity supply and demand is balanced on
in peak capacity and reduce system costs by 80–130 €/house/year regional level, and heating is balanced on area level. In this study,
on average (socio-economic costs) [22]. However, the applied the existing model structure in Balmorel is used for handling the
model covers only operation of the system, while investments integrated power and district heating system. Individually heated
are estimated separately. Moreover, flexible operation is not mod- one-family houses are represented using the new thermal building
elled but e.g. in [20] merely represented as a modified fixed de- model add-on. In Balmorel, the optimisation is performed with a
mand profile, assuming that operation can be optimised within yearly time horizon, where the year is divided into seasons, which
the day. are further divided into time periods. When applying an hourly
No existing studies have been identified, which analyse the sys- time resolution, seasons represent weeks, while time periods rep-
tem effects of individual heat pumps, applying a national scale en- resent hours. A full description of the Balmorel model can be found
ergy system model that optimises both operation and investments. in [24,26].
This is therefore the subject of this study. In a case study of the Heat pumps (compression heat pumps) are in Balmorel repre-
Danish energy system in 2030 with optimised wind power shares sented as commonly in energy system models, where the relation
of around 50–60%, we analyse how individual heat pumps and between heat output (Qt) and electricity input in a given time per-
complementing heat storages affect investments in the energy sys- iod (Pt) is expressed as Qt = COPPt, where COP is the average an-
tem; particularly to which extent they can facilitate increased nual coefficient of performance for the heat pump. In the thermal
wind power investments and reduce the need for peak and reserve building model add-on, the individual heat pumps (ground source
capacity. Moreover, we investigate to which extent investments in heat pumps and air/water heat pumps) are supplemented by an
individual heat storages are socio-economically attractive. In this electric boiler to cover peak loads in order to reduce heat pump
regard, we include competition with other flexibility measures in investment costs, as in reality. The capacity shares of heat pump
the system – here represented as large heat pumps, electric boilers, and electric boiler, respectively, are defined in the input data.
and heat storages in the CHP system. Effects on system costs, fuel When the heat pumps are connected to heat storage, they can be
consumption, and CO2 emissions are also quantified. The results operated flexibly. The large heat pumps are in Balmorel connected
are intended as input to policy makers in prioritising efforts and to the district heating storages, while the individual heat pumps
incentives within the field of individual heat pumps and wind can utilise the different heat storage options covered in the ther-
power integration. The main research question investigated is: to mal building model add-on, if the required investments are made.
which extent can individual heat pumps and complementing heat The thermal building model applied represents houses with
storages support an effective wind power integration? radiator heating as well as houses with floor heating, as the central
We apply an add-on recently developed for the investment heating system. In the model, houses with radiator heating can uti-
model Balmorel (see Section 2), which incorporates thermal build- lise the option of storing heat passively in the building structure
ing dynamics and various heat storage options for individual heat through radiation, thermal conductance, and convection. The term
pumps. The model has been described in detail in [23]. Socio-eco- passively refers to the fact that heat is stored without the use of a
nomic cost minimisation is applied in order to identify what is heat transferring media going in and out of the storage. The storage
optimal for the system as a whole. The article is structure in the concept mainly utilises the heat capacity of internal constructions
following way; Section 2 describes the Balmorel model and the in walls, ceiling, and floor, while windows, doors, and furniture
add-on developed. Section 3 presents how the model has been ap- have minor influence [18]. Houses with (concrete) floor heating
plied, covering scenarios and input data. In Section 4, results of the can via the water tubes in the floor foremost store heat actively
energy system analysis are presented, and a discussion and conclu- in the large thermal mass of the concrete floor (in addition, heat
sion is given in Sections 5 and 6. Finally, Section 7 outlines sugges- can be stored passively in the rest of the building structure). Ther-
tions for future work. mal storage tanks for space heating and hot water are also covered,
representing other options of enabling flexible heat pump opera-
tion. The add-on is a linear state-space model and is based on phys-
2. Model: Balmorel with thermal building model add-on ical definitions and relations of heat transfers and heat capacities. A
full description of the model add-on is given in [23].
Balmorel is a partial equilibrium model assuming perfect com-
petition [3,24]. The Balmorel model has in previous studies been
used for analysing e.g. wind power development, security of sup- 3. Application
ply, the role of district heating, development of international
electricity markets, the role of demand response, compressed air 3.1. Scenarios
energy storage, unit commitment, international markets for green
certificates and emission trading as well as evaluation of environ- Three main scenarios are compared in the analysis: NOiHP, iHP,
mental policies cf. e.g. [15,25]. and iHP-Flex (see Table 1). By comparing the scenarios NOiHP and
The model minimises total costs of the system, covering annu- iHP, it is analysed how the individual heat pumps affect the energy
alised investment costs, operation and maintenance costs of exist- system, when not operated flexibly. The difference between the
ing and new units, and fuel and CO2 quota costs. In this process, the scenarios iHP and iHP-Flex reveals the effect of investing in heat
model optimises investments in power/heat production, storage, storages complementing the heat pumps, thereby facilitating flex-
and transmission capacities, and the operation of the system. The ible operation. Investment in heat storages will only occur to the
optimisation is done subject to a number of constraints including extent that it is socio-economically attractive, in the optimisation.
satisfaction of demands for electricity and heat in each time period, Focusing on the largest storage potentials, the following heat stor-
renewable energy potentials, and technical restrictions on units in age options are optimised in the main scenarios:
K. Hedegaard, M. Münster / Energy Conversion and Management 75 (2013) 673–684 675
Table 1 3.2. Period simulated

Main scenarios analysed in the optimisation.
NOiHP iHP iHP- Based on a need to limit computation time, five weeks are sim-
Flex ulated to represent a full year, hereof four weeks in the heating
Investment in individual heat pumps allowed + + season and one summer week. In order to ensure good representa-
Investment in intelligent passive/active heat storage + tion of variations in wind power, electricity, and heat demands, an
in the building structure alloweda hourly time resolution within each week is applied. When scaling
Investment in heat accumulation tanks allowed +
up to a full year, weighting factors are applied to account for the
a
Indoor temperature variation allowed in this regard: day: 20–23 °C, night: 19– period length that each of the simulated weeks represent. The
22 °C. weeks have been carefully selected based on the following criteria:
(a) The weeks give a good representation of the varying heat

demands over the seasons.
Intelligent heat storage in the building structure: investment in (b) The highest net load (electricity demand minus wind power)
digital thermostats1, enabling intelligent control of the central over the year for Denmark is covered. This captures the need
heating system, and in a central controller communication with for peak capacity for periods, where electricity demand is
the power system [18,27]. For houses with radiator heating, this high and wind power low.
renders it possible to utilise passive heat storage in the building (c) The highest total load (conventional plus flexible electricity
structure, while for houses with (concrete) floor heating, it also demand) over the year for Denmark is covered. This captures
facilitates using active heat storage in the concrete floor. the need for adequate production and reserve capacity, via a
Heat accumulation tanks: investment in a heat accumulation capacity balance restriction applied.
tank for the space heating circuit of a size up to 1000 l per house (d) Peaks in individual heat demand over the year are covered.
and in a central controller communicating with the power sys- This ensures good representation of the electric boiler,
tem [18]. which supplements the heat pump in covering peak loads
[32]. Due to its low efficiency (around 1 compared to an
A small hot water tank of around 180 l [28] is typically in- applied COP of 3.1–3.8 for the heat pump), the electric boiler
stalled in connection with individual heat pump installations. is highly influential on peaks in electricity consumption for
The hot water tank can to some extent also support flexible heat the total heat pump unit.
pump operation if investing in control equipment (a central con- (e) The weeks cover significant wind power variations.
troller). The feasibility of investing in this option is analysed in a
supplementary analysis (see Section 4.4). In a sensitivity analy- The heat transfer coefficients for the houses have been cali-
sis, the feasibility of the space heating storages listed above is brated to give the annual space heating demand for each building
moreover investigated in a situation where the investment in category (cf. Table 3). The calibrated heat transfer coefficients devi-
intelligent use of the hot water tank has already been made ate less than 2% from values corresponding to a full year simula-
(see Section 4.4). This analysis approach is applied in order to tion. This shows that the simulated weeks give a very good
avoid double counting the cost of the common control equip- approximation of the space heating demand over a full year. Based
ment (the central controller) and to ensure a reasonable compu- on this and the selection criteria a-e, the modelled period is thus
tation time. representative for a full year.
The system model covers the power and district heating sectors
of not only Denmark, but also Norway, Sweden, Finland, and Ger- 3.3. Energy demands applied
many, in order to cover the influence of electricity import/export.
For Denmark, individually heated one-family houses are addition- Electricity and heat demands in 2030 are based on the Future
ally modelled, representing the bulk, app. 80% [29], of individual scenario set up by The Danish Commission on Climate Change Pol-
heating. Existing individual heating installations are assumed icy [31] (see Table 2). In this scenario, the district heating share for
decommissioned by 2030 based on a typical lifetime of around heating of buildings is increased from 46% in 2008 to 55% in 2030
15–20 years [14,30]. [33].
Denmark is divided into a Western power region, being syn- The stock of existing individually heated one-family houses in
chronous with the former UCTE power system, and an Eastern Denmark (app. 716,000 houses) is modelled, representing the large
power region being synchronous with the former Nordel power majority of the individually heated one-family houses by 2030 (94
system (both power systems are today represented by ENTSO- % of the heat demand).
E). In order to limit computation time, the Danish district heat-
ing system is aggregated into 4 areas. This aggregation has no 3.4. Modelling of individually heated one-family houses
real influence on the results since (1) A linear model is applied,
i.e. investments can be made in continuous capacity sizes. (2) The development in the Danish building stock and in the heat
The total district heating demand for each actual heating area demands towards 2030 is based on the Future scenario in [31].
covers several thousands of consumers and is thus already very As a result, the heat demand for existing individually heated
smoothened from aggregation. (3) The district heating expansion one-family houses is assumed reduced by about 25% towards
is in the analysis fixed based on [31]. (4) A large share of exist- 2030, mainly due to heat savings and secondarily due to demoli-
ing district heat generation capacities are decommissioned to- tion of old buildings. The houses are divided into 3 different cate-
wards 2030. Due to the focus on individual heating, 20 gories, depending on the year of construction; reflecting changes in
different individual heating areas with different characteristics insulation standards (see Table 3). The assumed heat savings have
are modelled (see Section 3.4). been distributed on the different construction periods based on
heat saving potentials given in [34].
Radiator systems have for a long period of time been the typical
type of central heating system installed in connection with con-
1
On each radiator for radiator heating and in each room for floor heating. struction of new houses in Denmark. Use of floor heating as central
Table 2
Electricity and heat demands modelled in 2030 ( TWh).
Denmarka Finland Germany Norway Sweden

Electricity 30.1 98.1 569.6 145.3 160.5
District heatingb 32.5 34.1 86.8 1.9 46.4
Indiv. heating, one-family houses 13.1 – – – –
a
Demands of the future scenario in [31] are used.
b
Incl. district heating to industry.
Table 3
Modelled stock of existing individually heated one-family houses in Denmark by 2030.
Unitc Average or total Construction period

1850–1960 1961–1978 1979–2005 Ref.
Av. net heat demanda kWh /m2 122 157 111 68 [31,34–36]
Net heat demand TWh 13.1 7.4 4.1 1.6 [31,34–36]
Heated floor area 106 m2 108 47 37 24 [31,34–36]
House size, av. m2/house 151 142 151 171 [37]
Floor heating share % 5%/9%b 0% 0% 40% [38,39]
Hot water share pct. 17% 13% 16% 28% [37]
a
After heat savings based on [31] and [34], representing improvement of the building envelope. Net heat demand: demand for
heating from installation considering heat losses and heat contributions.
b
5% of net heat demand and 9% of heated floor area.
c
All areas given in m2 refer to heated floor areas.
heating system (i.e. not only in the bathroom) was introduced dur- (measured in living room/kitchen), an interval of 20–23 °C is as-
ing the 1990s and has from around 1995–2000 been the common sumed allowed in the iHP-Flex scenario during the day. The al-
type of central heating system installed in new houses [38,39]. lowed temperature interval at night is set to 19–22 °C based on
Against this background, houses constructed from 1995 are as- [46,47]. Allowing the indoor temperature to vary within such mod-
sumed to have a central floor heating system and houses con- erate intervals facilitates utilising the heat storage capacity of the
structed before 1995, a radiator-based central heating system. building structure for flexible heat pump operation. Typical diurnal
This results in radiator heating covering 95% of the individual heat and seasonal variations in ambient air temperatures are repre-
demand of existing one-family houses in 2030. Installation of floor sented using hourly data for the Danish design reference year [48].
heating as central heating system in existing buildings is much Individual heat pumps can mainly be operated flexibly within
more expensive than when installed in the construction phase the day, given the heat storage options presented [49]. In compar-
[40]. Therefore, installation of floor heating in existing houses is ison, large heat pumps can be connected to the larger district heat-
typically only carried out in connection with full renovation of ing storages, which give a higher level of flexibility. As such, large
the floor [18]. Floor renovations are expensive, as also reflected heat pumps can be activated flexibly in periods of up to several
by the fact that this is not included among the feasible heat saving days, e.g. responding to longer periods with high wind power
measures in [31]. Installation of floor heating in existing houses and low electricity prices. Furthermore, the large heat pumps can
has therefore been neglected. be supplemented by a fuel based unit (CHP unit or a fuel boiler).
Houses with a radiator-based central heating system have been This means that large heat pumps can be turned off in longer peri-
divided into three groups depending on the heat capacity of the ods when wind power is low and electricity prices are high [49].
building structure that can be utilised for passive heat storage
diurnally. This heat capacity is based on typical values for Danish 3.5. Fuel and CO2 prices
buildings as given in [41] and on the following general distribu-
tion: 60 Wh/°C/m2 floor area: 25% of the houses, 100 Wh/°C/m2 The fuel and CO2 prices used in the main scenarios are based on
floor area: 50% of the houses, and 140 Wh/°C/m2 floor area: 25% [31], which is based on IEAs Reference scenario [50] (see Table 4).
of the houses [42]. Overall, the stock of individually heated one- In a sensitivity analysis, fuel and CO2 prices corresponding to a very
family houses is represented as 10 different heating areas in East- ambitious international climate mitigation are applied [31,50].
ern and Western Denmark, respectively. The 10 areas in each re- Fossil fuel prices in this scenario are lower, mainly due to lower de-
gion constitute 1 floor heated area and 9 areas with radiator mand resulting from CO2 reduction requirements, while CO2 prices
heating, expressing combinations of insulation standards and heat are significantly higher.
capacities of the building structure. Concrete floor heating systems
have been applied as this is the dominant type of water-based floor 3.6. Energy policies implemented in model
heating system installed in houses in Denmark, due to its low costs
[39,43]. A typical concrete layer of 10 cm has been assumed The following Danish energy policies towards 2030 have been
(67 Wh/°C/m2 floor area) [39,43]. implemented in the model:
Reflecting the average diurnal indoor temperature profile for 28
Danish households with heat pump installations [44], a constant Wind power must in Denmark by 2020 correspond to at least
indoor temperature requirement of 21.5 °C is applied in the iHP 50% of the national electricity consumption [2].
scenario. [45] confirms that a constant indoor temperature setting The use of coal at Danish heat and power plants is phased out by
is typical for current heat pump installations in one-family houses. 2030 [2]
This setting has moreover been found to give the best correlation Individual oil boilers are phased out by 2030.
with diurnal space heating profiles for the measured houses. Based 1.2 Mtonnes straw and 0.2 Mtonnes wood chips must be incin-
on typical indoor temperature levels for the households [44] erated at Danish heat and power plants annually [52].
Table 4
Fuel and CO2 prices assumed for 2030 (€/GJ).
€/GJ Main scenario [31] Sensitivity analysis [31] Transport cost [51]
Power plant Individual consumer
Crude oil ($/barrel) 120 94 – –
Coal 3.1 1.8 0 –
Natural gas 9.3 7.4 0.4 3.3
Straw 5.3 5.3 1.8 –
Wood pellets 11.5 11.6 0 3.4
Wood chips 6.4 6.4 1.6 –
Biogas 14.1 14.1 0 –
Munic. waste 3 3 – –
CO2 (€/tonne) 38 77 – –
Table 5
The Danish energy policy also includes a target of electricity and
Danish biomass and waste potentials in 2030 [31].
heat generation to be 100% based on renewable energy by 2035
[53]. The policy does not specify whether this applies to district Fuel Potential (PJ)
heating only or all heating including individual heating. In the light Straw 40
of this uncertainty, it is chosen to include the possibility to invest Wood chips/wood wastea 101
Biogas 32
in individual natural gas boilers in the model by 2030. Historically,
Munic. waste 43
the Danish natural gas grid expansion for supplying heat to one-
a
family houses has declined since 1999 [29]. The use of individual Wood residues and energy crop potential available for heat
and power generation.
natural boilers is therefore constrained to the amount supported
by the current natural gas grid, and further reduced due to the heat
savings assumed towards 2030. This corresponds to allowing new
natural gas boilers only in the form of reinvestments. open cycle gas turbines and natural gas boilers are allowed also in
Binding EU energy targets apply for the member states in 2020, the Danish system, since these are suited in scale and technology
ensuring a renewable energy share of 20% for EU as a whole for the for a later shift to using renewable fuels (biogas or gasified bio-
gross final energy consumption for electricity, heating, cooling, and mass) [31]. Based on [56], investment costs are in the model annu-
transport [54]. Each of the EU member states have set up national alised with a discount rate of 3% given in fixed prices. All costs in
renewable energy action plans that will ensure compliance with the study are socio-economic and are given in €-2011.
these targets. Planned renewable energy capacities, as given in In individually heated areas, the model includes possibilities to
these plans, have been implemented in the model for the other invest in the technologies presented in Table 7. Individual heat
EU countries modelled (Germany, Sweden, and Finland). Where pumps are in the analysis dimensioned to 77% of the total heat out-
relevant, planned renewable electricity generation levels have fur- put capacity, reflecting the typical situation in Danish heat pump
thermore been implemented as minimum generation levels. Other installations (72–82% [32]). Thereby, the heat pumps cover around
energy policies implemented in the model include planned nuclear 99% of the annual heat demand, while the supplemental electric
and hydro power capacities and the German policy of phasing out boilers cover the rest.
nuclear power towards 2022 [55]. Planned expansions of electrical transmission capacities to-
The Danish energy policy, targeting a 100% renewable heat and wards 2030 have been included. The possibility to invest in further
power sector by 2035, is more ambitious than the actions needed transmission capacities has however been excluded in the main
in the other countries to comply with EUs renewable energy tar- scenarios. This is most reasonable considering that most of the
gets for 2020. Particularly the phasing out of the coal in Denmark Northern European countries other than Denmark are here mod-
by 2030 could lead to significant net import of electricity from elled without representation of internal bottlenecks in the power
other countries. A large permanent net import of electricity cannot system. Nevertheless, the influence of allowing further transmis-
be considered reasonable from a security of supply perspective, as sion capacity investments is covered in a sensitivity analysis.
neighbouring countries will face similar challenges of converting
their energy systems towards sustainability [31]. Therefore, as
done in [31], Danish electricity generation is constrained to be
minimum equal to the electricity demand on annual level. 4. Results
When not allowing investments in individual heat pumps (NOi-

3.7. Biomass and waste potentials HP scenario), the investments in individual heating installations
constitute natural gas boilers, to the extent supported by the nat-
Use of straw, wood chips/wood waste, firewood, biogas, and ural gas grid (37% of total individual heat demand), and then wood
municipal waste is in the analysis limited to the national resources. pellet boilers to cover the rest of the heat demand. When invest-
Wood pellets import is allowed since wood pellets are to a large ments in individual heat pumps are allowed (iHP and iHP-Flex sce-
extent traded internationally. The Danish biomass and waste nario), air–water heat pumps are installed in all individually
potentials applied are presented in Table 5. heated areas. This represents a significant electricity demand of
4.3 TWh-e.
3.8. Investment options modelled In the iHP scenario, the heat pumps are restricted to cover the
net heat demand in each hour, i.e. heat transmission and ventila-
In the optimisation, it is in all five modelled countries made tion losses minus heat contributions. As shown in Fig. 1, space
possible to invest in the technologies presented in Table 6, for sat- heating will in this situation (non-flexible operation) be highest,
isfying the electricity and district heating demand. Investments in when ambient temperatures are lowest.
Table 6
Power and district heating technologies assumed available for investment in 2030.
Technology Fuel Inv. cost (M€/ Var O&M (€/ Fix O&M (k€/MW/ Lifetime Eff.b CB CV Ref.
MW)a MWh) yr) (years)
Open cycle gas turbine, condensing Natural gas 0.34 2.5 16.8 25 0.46 – – [57,58]
Combined cycle gas turbine, extraction, Natural gas 0.58 3.4 21.0 25 0.61 1.75 0.13 [57,58]
CHPc
Steam turbine, extraction, CHPc Coal 1.47 7.3 – 40 0.52 1.01 0.15 [58]
Steam turbine, extraction, CHP Wood 1.47 7.3 – 40 0.52 1.01 0.15 [58]
pellets
Steam turbine, extraction, CHP Wood chips 1.68 3.4 24.1 30 0.49 0.61 0.15 [58]
Steam turbine, back pressure, CHP Straw 2.31 6.4 39.8 25 1.02 0.4 – [58]
Centralised biogas plant, back pressure, Biogas 5.45 34.6 – 20 0.92 1.07 – [58]
CHP
Steam turbine, back pressure, CHP Munic. 8.91 23.1 162.4 20 0.97 0.37 – [58]
waste
Offshore wind turbine – 2.31 15.7 – 25 1.00 – – [58]
Onshore wind turbine – 1.28 12.0 – 25 1.00 – – [58]
Photo voltaic Sun 1.83 18.9 – 30 1.00 – – [58]
Heat pump, waste-water Electricity 0.70 – 7.0 20 3.0 – – [20,59,60]
Electric boiler Electricity 0.08 0.5 1.0 20 0.99 – – [58]
Heat boiler Wood chips 0.09 – 3.3 20 1.01 – – [58]
Heat boiler Natural gas 0.52 – 24.6 20 1.08 – – [58]
Heat boiler Munic.waste 1.15 5.6 52.4 20 0.98 – – [58]
Solar thermal Sun 0.22 0.5 – 20 1.00 – – [58]
Thermal storage tanks – 0.0030 – – 40 0.99 – – [61]
a
Investment costs for thermal storage tanks are given in M€/ MWh storage and for heat pumps in M€/MW-thermal.
b
For heat boilers, heat efficiency, for heat pumps, coefficient of performance, and for other units, electric efficiency (for CHP extraction plants, electric efficiency in
condensing operation).
c
Excluded for investment in Denmark in the light of the Danish political phase out of coal by 2030, and the goal of 100% renewable energy use for electricity and heat
generation by 2035.
Table 7
Individual heating technologies assumed available for investment in 2030.
Technology Fuel Inv. cost (M€/ Inv. cost (€/ Fix O&M (k€/MW/ Var O&M (€/ Lifetime Eff.c Ref.
MW) house) yr) MWh) (years)
Heat pump, air/waterb Electricity 0.78 – 5.1 – 15 3.1 [14,30]
Heat pump, groundb Electricity 1.56 – 5.1 – 15/40 3.8 [14,30]
Natural gas boiler Natural gas 0.50 – 4.0 7.2 22 1.02 [30]
Biomass boiler, automatic stoking Wood 0.98 – 3.0 – 20 0.91 [14,30]
pellets
Solar thermal Sun 1.10 – 14.8 0 30 1.00 [30]
Heat accum. tank, space heating – 0.073 1275 – – 40 0.99 [62]
(1000 l)a
Digital thermostats (wireless) – – 312 – – 15 [63]
Central controller – – 134 – – 15 – [62]
a
Investment costs for heat storages are given as M€/ MWh storage capacity.
b
Data in [14] for today’s heat pumps in radiator systems (after heat savings) are projected to 2030 based on [30]. Costs include an electric boiler supplementing in covering
peak loads. For ground source heat pumps, a life time of 15 years for the heat pump unit and 40 years for pipes in the ground [14].
c
For heat boilers, heat efficiency, for heat pumps, annual COP including electric boiler, and for heat storages, accounting for 1% heat loss per hour. For simplicity and due to
the low share of floor heating, same COP for floor heating as for radiator heating is applied.
When allowing investments in heat storages for the heat pumps that a heat storage capacity in a given area is shared among all heat
(iHP-Flex scenario), investment in intelligent heat storage in the pumps within the area. This is evident from Fig. 1, where the heat
building structure is observed in 34% of the houses. The heat stor- storage investments occurring in 34% of the houses render it pos-
age investments occur within all 20 defined categories of individ- sible to operate a larger share of the heat pumps flexibly to some
ually heated houses; however the investments are highest in extent, e.g. reflected by the significant load reductions in hour
buildings with high heat capacities (typically corresponding to T109 and T163. As such, the model implies best case conditions
buildings with a high share of concrete) and in buildings with high for the heat storages, since the flexibility can be distributed among
heat demand (low insulation level) (cf. [23]). No investments in the houses in this way. Ideally, each of the many thousands of indi-
heat accumulation tanks are observed, due to their higher invest- vidually heated houses should be modelled separately (716,000 in
ment costs. As shown in Fig. 1, the heat storage capability of the this case); however it would hardly be possible to run such a
buildings renders it possible to prioritise the operation of the heat model.
pumps for hours with low electricity prices. The electricity prices
in the model represent the marginal electricity generation cost in 4.1. Energy system investments
the given power region and time period. Fig. 1 thus illustrates that
when connected to heat storage, the heat pump operation is opti- In all three scenarios for 2030, the bulk of the annual Danish
mised in integration with the energy system. electricity generation is based on wind power (57–61%) and wood
The interaction between heat pumps and heat storages in indi- chips (28–31%), while straw (4%), municipal waste (6–7%), wood
vidually heated areas is modelled on aggregated level. This means pellets (1%), and natural gas (1%) contribute with smaller shares.
Non-flexible heat pump operation (iHP) Space heating (MW-th)

2500 6
Ambient temp. ( C)
4
2000 Transmission loss from indoor
MW-th
2 air to building structure (MW-th)
1500 0 Heat from persons and
1000 -2 appliances (MW-th)
-4 Ventilation loss (MW-th)
500
-6
0 -8 Ambient temperature ( C)
Flexible heat pump operation (iHP-Flex)
800 140
120
€/MWh-e
600 100 Electricity consumption (MW-e)
MW-e
80
400 Electricity price (€/MWh-e)
60
200 40
20
0 0
T118
T001
T010
T019
T028
T037
T046
T055
T064
T073
T082
T091
T100
T109
T127
T136
T145
T154
T163
Hour
Fig. 1. Non-flexible heat pump operation (iHP scenario) and flexible heat pump operation (iHP-Flex scenario) illustrated for a week in the heating season. Total heat output
and electricity consumption is shown for heat pump installations in houses with radiator heating in Western Denmark.
6000 3500 18
5000 3000 16
14
4000 2500 NOiHP
12
MW-th
MW-e
GWh
2000 10 iHP
3000
1500 8 iHP-Flex
2000 6
1000
1000 4
500
2
0 0 0
(a) (b) (c)

Fig. 2. Investments in the Danish system by 2030 for different scenarios (a) investments in electricity generation capacities, (b) investments in district heat generation
capacities and (c) Investment in thermal storages in the district heating system. On: onshore, Off: offshore. WoodC: wood chips/waste, OC-GasT: open cycle gas turbines, DH:
district heating, HO: heat only, MN: municipal waste. SOL: solar thermal.
The diminishing share of fossil-based electricity generation is pumps is distributed over many hours in the day, even without
mainly explained by the implemented political goal of phasing flexible operation (see Fig. 1) and both heat demand and wind
out coal at Danish power plants. power is typically high in cold periods. This creates good possibil-
The investments in new electricity and heat generation capaci- ities for utilising wind power in covering the electricity demand for
ties in Denmark for the three different scenarios are illustrated in the heat pumps while using dispatchable power capacities to back
Fig. 2. As shown, the new electricity generation capacities cover it up. In contrast, e.g. dumb charging of electric vehicles would typ-
onshore and offshore wind power, biomass CHP steam turbines, ically be concentrated in a few hours of the day when people re-
and open cycle gas turbines. The wood chip and straw CHP plants turn from work, making wind power utilisation difficult in this
are mainly used for intermediate load operation, supplementing case.
the large amounts of wind power. Open cycle gas turbines are used The increased investments in wood chip CHP and open cycle gas
as peak/reserve capacities (cf. Fig. 3). turbines are driven by a need to supplement the fluctuating wind
Fig. 2a shows that the installation of individual heat pumps power. In this regard, these dispatchable technologies also cover
facilitates increased investments in wind power, wood chip CHP, an increased need for peak/reserve capacities imposed by the heat
and open cycle gas turbines (iHP vs. NOiHP scenario). The increase pumps. The annual wood chips-based electricity generation is
in wind power capacities occurs in order to meet the increase in however identical in all scenarios since the wood chips consump-
electricity demand from the installation of the heat pumps. As tion is limited by the national biomass resources.
such, wind power covers 97% of the increase in annual electricity As shown in Fig. 2a, the investments in open cycle gas turbine
demand, while natural gas peak power plants cover the rest. Part and wood chip CHP capacities are reduced with 440 MW when
of the explanation for this is that the competitor coal is politically the heat pumps are operated flexibly (iHP-Flex scenario compared
phased out by 2030 in the Danish system and that the other low to iHP scenario). The reason is the heat pumps contribute in peak
cost fuels, wood chips and municipal waste, are constrained by load shaving, when operated flexibly, which reduces the need for
the available national resources. This leaves good conditions for peak/reserve capacity. The flexible heat pump operation moreover
wind power. Moreover, the electricity consumption of the heat facilitates reduced wind power investments onshore and increased
12,000
10,000 Natural gas
Wood pellets
MW-e
8,000
Straw
6,000
Wood chips
4,000
Wind
2,000 Munic. waste
- T115
T118
T001
T020
T039
T058
T077
T096
T134
T153
T004
T023
T042
T061
T080
T099
T137
T156
T007
T026
T045
T064
T083
T102
T121
T140
T159
T010
T029
T048
T067
T086
T105
T124
T143
T162
T013
T032
T051
T070
T089
T108
T127
T146
T165
S02 S12 S31 S40 S51
Week, hour
Fig. 3. Modelled hourly electricity generation in the Danish energy system by 2030 (iHP scenario).
30 1800
1600
25
1400
NOiHP
20 NOiHP 1200
M€/yr
iHP
TWh
iHP 1000
15 800 iHP-Flex
iHP-Flex
10 600
400
5 200
0
0
Wind Wood Munic. Straw Natural Wood
chips waste gas pellets
Fig. 4. Primary energy consumption in the Danish energy system by 2030 for
different scenarios.
Fig. 5. Annualised system costs for the modelled Danish heat and power system by
2030 for different scenarios.
investments offshore. This is due to the fact that flexibility added
to a system generally improves the conditions for technologies
with relatively high investment costs and low variable costs. In this of straw for heat and power generation only occurs to the extent
regard, offshore wind power has higher investment costs, but on enforced by the Danish biomass agreement and is therefore un-
the other hand higher obtainable full load hours (and a different changed across scenarios. The individual heat pumps displace a
variation profile) compared to onshore wind power. significant amount of wood pellets and natural gas used in individ-
The investments in large heat pumps in the district heating sys- ual boilers. This reduces the pressure on biomass resources and
tem are clearly reduced with the installation of individual heat also increases national security of supply in this case since wood
pumps (see Fig. 2b). This is explained by the increase in electricity pellets are in Denmark typically imported.
demand resulting in generally increased electricity prices, making Flexible operation of the individual heat pumps merely results
large heat pumps less attractive. The lower investment in large in a slight increase in wind power generation and a slight decrease
heat pumps is the explanation for the increase in solar thermal in natural gas consumption, due to an avoided use of open cycle
in the iHP scenario. gas turbines for peak load operation (iHP-Flex vs. iHP scenario).
When adding flexibility to the individual heat pumps (iHP-Flex The total fuel consumption excl. wind is reduced significantly as
scenario), the investment in large heat pumps is reduced further. a result of the individual heat pump installations (23%). This is ex-
The reason is that the individual heat pumps compete with the plained by the higher efficiency of the heat pumps, and that the in-
large heat pumps in placing operation in low electricity price peri- crease in electricity demand imposed by the heat pumps is largely
ods. The reduced investments in large heat pumps result in in- met by wind power. When activating flexible operation of the heat
creased investments in natural gas heat only boilers or solar pumps, an additional minor fuel saving is achieved (0.3%). The CO2
thermal and in thermal storages in the district heating system. It emissions in 2030 are reduced considerably with the installation of
can be noted that the solar thermal capacities are relatively large individual heat pumps (37%) (iHP scenario) and further reduced by
due to their low number of full load hours (app. 730). only 2% with flexible operation of the heat pumps (iHP-Flex sce-
nario). It can be noted that the percentagewise CO2 reductions
4.2. Fuel consumption and CO2 emissions are enlarged by the Danish no coal policy, which yields generally
low absolute CO2 emissions.
The primary energy consumption of the Danish heat and power
system in the three scenarios is shown in Fig. 4. It can be seen that 4.3. System costs
the installation of individual heat pumps facilitates a significant in-
crease in wind power generation (iHP vs. NOiHP scenario). This is The annualised socio-economic system costs for the three sce-
due to the increased wind power investments presented in Sec- narios are presented in Fig. 5. As shown, fuel costs are reduced con-
tion 4.1. The consumption of wood chips and municipal wastes is siderably with the installation of individual heat pumps. This is
the same in all scenarios due to the resource constraints. The use explained by the high efficiency of the individual heat pumps,
and that the increase in electricity demand is largely covered by in peak/reserve capacity, while the average system benefit per
wind generation. Furthermore, the displaced technologies, natural house decreases substantially (to 8 €/house/yr compared to
gas boilers and wood pellet boilers, have relatively high fuel costs. 120 €/house/yr in the main scenario). This illustrates that the opti-
Investment costs and fixed O&M costs are on the other hand in- mised heat storage investments in the main scenario have largely
creased, due to the larger investment costs (annualised) of individ- been made to the extent that peak/reserve capacities can be cost-
ual heat pumps compared to individual gas boilers, and the effectively reduced.
increased investments in wind power and dispatchable power If forcing through investments in both 1000 l heat accumulation
capacities needed along with the heat pump installations. In total, tanks and in the enabling of intelligent heat storage in the building
system costs for the Danish heat and power system are reduced structure in all houses, only moderate increases in wind power
significantly with the installation of individual heat pumps (12%). generation (0.1 TWh) and reductions in fuel consumption (0.5%)
The cost reduction achieved when operating the heat pumps flex- are still observed; while system costs are increased moderately
ibly is moderate (0.9%). (1%). Peak/reserve capacity investments are reduced only slightly
The flexible operation of the heat pumps provides a socio-eco- more than when optimising the heat storage investments
nomic cost reduction of around 120 € per year per house investing (460 MW compared to 440 MW).
in intelligent heat storage in the building structure. This cost When assuming a heat pump capacity share in the low end,
reduction is mainly caused by savings on energy system invest- 72%, the impact of the electric boiler in peak load hours becomes
ments (incl. fixed O&M, mainly peak/reserve capacities), as illus- more critical in the situation without flexible operation. As a result,
trated in Fig. 6. a larger reduction in peak/reserve capacity is achieved when oper-
Due to the self sufficiency constraint implemented, electricity ating the heat pumps flexibly (560 MW, with optimised heat stor-
import equals export on annual level. However, significant electric- age investments). Correspondingly, if assuming a heat pump
ity trade is utilised hour-by-hour as expected. For the other coun- capacity share in the high end, 82%, a lower reduction in peak/re-
tries modelled, energy system investment and operation is serve capacity is achieved (290 MW). When assuming higher
practically unchanged by the installation of individual heat pumps investment costs and COP (3.7–4.0) for the heat pumps as given
and heat storages in Denmark. in [30], the heat pumps still outcompete all other individual heat-
Overall, the results indicate that individual heat pumps can con- ing installations. However, the higher COP results in lower impact
tribute significantly to facilitating larger wind power investments on peak/reserve capacities, which are therefore reduced by only
and reducing fuel consumption, CO2 emissions, and system costs. 320 MW with the flexible heat pump operation.
Heat storages for the heat pumps, enabling flexible operation, have If assuming 50% higher/lower investment cost for control
shown to provide only moderate system benefits in these respects. equipment (digital thermostats and central controller), investment
Investment in intelligent heat storage in the building structure is in intelligent heat storage in the building structure occurs in 20%
identified as socio-economically feasible in a given share of the and 75% of the houses, respectively. Peak/reserve capacities are re-
houses, while heat accumulation tanks are not identified as so- duced by 426–430 MW and the average system benefit per house
cio-economically competitive. is 190 €/house/yr and 60 €/house/yr, respectively. If allowing an in-
door temperature of 20–23 °C at night, only moderate changes are
observed since the peak loads occur in the day time. When apply-
4.4. Sensitivity analyses ing a discount rate of 5%, generally lower wind penetration and
higher investments in thermal power plants is observed. The
The general results stated above have been confirmed in all sen- reduction in peak/reserve capacity obtained with flexible heat
sitivity analyses presented in this section. However, the share of pump operation is higher (640 MW).
houses in which intelligent heat storage in the building structure When assuming an ambitious international climate mitigation
is feasible, the achieved reduction in peak/reserve capacity, and (high CO2 prices, lower fossil fuel prices, cf. Table 4), the reduced
the average system benefit per house, is found sensitive to certain investment in peak/reserve capacity mainly comprises wood pellet
factors. These sensitivities are presented in the following, focusing CHP, and the system benefit is increased to 180 €/house/yr. If
on the most important deviations from the main scenario. Other assuming 50% higher biomass prices, wind power investments
exploratory sensitivity analyses are also presented. are increased significantly, wood-based CHP investments reduced
As a first test, investment in intelligent heat storage in the considerably, and open cycle gas turbine investments increased
building structure is forced through in all houses with heat pump considerably. However, the overall outcome of the analysis is not
installations, instead of optimising the heat storage investments as changed. If not restricting balance in electricity import and export
in the main scenario. This provides practically the same reduction on annual level, the net electricity import is increased from 0 to
2–3 TWh/yr without altering the overall outcome of the analysis.
The overall picture is not changed either, if allowing investment
Energy system inv. in transmission capacities further than planned.
and fixed O&M
In another sensitivity analysis, we have assumed intelligent
Fuel utilisation of the existing hot water tanks (180 l [28]) by forcing
through investment in a central controller in all houses. This leads
CO2 emission to a reduction of the required investments in peak/reserve capacity
by about 310 MW. In this situation, the possibility for achieving
further reductions in peak/reserve capacity, by investing in intelli-
Variable O&M
gent heat storage in the building structure, is limited to 120 MW.
However, since investment in the central controller in this case
-200 -150 -100 -50 0 50 100
Inv. in intelligent has already been made, the additional investment cost of enabling
Change in socio-economic costs heat storage in intelligent heat storage in the building structure is reduced (to cov-
(€/house/yr) building structure er only the digital thermostats). As a result, investment in further
Fig. 6. Change in annualised system costs due to flexible operation of individual
flexibility still occurs in a significant share of the houses (48%).
heat pumps; given as average costs per house investing in intelligent heat storage in When optimising the investments in the intelligent use of existing
the building structure. hot water tanks, investments occur in about 20–70% of the houses,
depending on the cost of the controller (applying ± 50% of the cost installation of heat pumps in individually heated one-family
given in Table 7). houses (716,000 houses, 4.3 TWh-e).
We find that individual heat pumps can contribute significantly
in facilitating larger wind power investments and in reducing sys-
5. Discussion tem costs, fuel consumption, and CO2 emissions. In addition, the
heat pumps can reduce the pressure on the limited biomass re-
The results confirm the findings of previous studies [14,15] in sources, due to displacement of individual wood pellet boilers.
identifying individual heat pumps as highly socio-economically The system benefits of the heat pumps are first explained by their
competitive in non-district heating areas. Our results indicate that high efficiency and socio-economic competitiveness. Moreover, the
in terms of making wind power investments more attractive and aggregated electricity demand profile of the heat pumps is well
reducing system costs, fuel consumption, and CO2 emissions, the suited for integrating wind power, even without flexible operation
installation of the heat pumps is the most important step, while of the heat pumps. As a result, the increase in electricity demand
the system benefits of adding flexibility to the heat pumps are imposed by the heat pumps can largely be covered by increased
moderate. This is in line with the findings in [18], which identified wind generation, while supplementing with dispatchable power
a similar pattern concerning effects of heat pumps and heat storag- generation. In the case analysed, the political phase out of coal in
es on fuel consumption of the system. We find that utilising the Denmark by 2030 and the constrained biomass resource availabil-
heat storage capacity of the building is socio-economically feasible, ity creates particularly good conditions for wind power.
while heat accumulation tanks are not competitive. This is also in When investing in individual heat storages, the heat pumps
line with the findings in [18]. contribute in peak load shaving, and their operation is prioritised
The possibility for reducing investment in peak/reserve capacity for hours with low electricity prices. However, the flexible opera-
is identified as crucial for the feasibility of heat storages for the tion provides only moderate system benefits. Hereof, the main
heat pumps. This is illustrated by the fact that savings on energy benefit is a reduced need for peak/reserve capacity investments
system investment generally constitute the largest part of the cost of about 300–600 MW for the case analysed, corresponding to
reduction achieved with flexible heat pump operation. The sub- the size of a large power plant. The possibility for reducing peak/
stantial decrease in the system benefit per house, with an increas- reserve capacity investments is identified as crucial for the feasibil-
ing number of houses with heat storage capability, is another ity of the heat storages.
illustration of this. This result is in agreement with patterns previ- It is found that investments in intelligent heat storage in the
ously identified for the case of EVs in [3], which found that system building structure can be feasible to some extent, i.e. in around
benefits per smart vehicle decrease substantially with an increas- 20–75% of the houses with heat pump installations, depending
ing number of EVs. on the cost of control equipment in particular. The flexible opera-
The achievable reduction in peak/reserve capacity identified, tion of the heat pumps provides a socio-economic cost reduction of
300–600 MW, is on level with the identified peak load shaving po- about 60–200 € per year per house investing in this storage option.
tential in [64] for smart charging of EVs in the Finnish system The result intervals have been identified through various sensitiv-
(540 MW). Furthermore, the identified system costs reduction, ity analyses covering e.g. variation in investment cost of control
due to flexible operation of the heat pumps, generally around equipment, COP, investment costs, and capacity share of the heat
0.6–1.1%, is on level with the system cost reduction of 0.45– pumps, discount rate, as well as fuel and CO2 prices. A reduction
0.95% found in [17]. If forcing through investment in intelligent in the peak/reserve capacity requirement can also be obtained if
heat storage in the building structure and in 1000 l heat accumula- investing in control equipment that enables intelligent use of the
tion tanks in all houses with heat pump installations, and exclud- hot water tank typically installed in combination with heat pumps.
ing investment costs for the storages, the system benefits become Investment in this option is found feasible in around 20–70% of the
comparable to the results in [20,21,17]. For this case, a system ben- houses. In contrast, investments in heat accumulation tanks are
efit of about 50 €/house/year is identified, i.e. close to the result in not found competitive due to their larger investment costs.
[17] (25–40 €/house/year) and significantly lower than the esti- The main benefit of the flexible heat pump operation, namely
mate in [20,21] (80–130 €/house/year). the reduced peak/reserve capacity requirement, will not be visible
As the applied model is deterministic and does not include on the electricity bills paid by the individual heat pump owners.
start-up costs, minimum load requirements, or part load efficien- Some of this benefit should therefore be transferred to the heat
cies, the potentials for reducing operation costs through flexible pump owners when developing incentives for them to invest in
heat pump operation are conservatively represented. However, heat storage capability. The relevance of this is reflected by the fact
we include effects on investments, in contrast to previous studies that under current regulation it is, according to the Danish TSO, not
modelling the impacts of individual heat pumps and supplement- attractive for heat pump owners to invest in flexible operation,
ing heat storages on the energy system. Our use of an aggregated even if gaining access to both the spot market and the regulating
energy system model means that heat storages are represented power market [22].
optimistically in the sense that the flexibility provided by an opti- Overall, large-scale installation of individual heat pumps is
mised heat storage capacity can be distributed among the heat identified as an important step in supporting the integration of
pumps within the given area. Even under these conditions, only wind power. Heat storages, facilitating flexible operation of the
moderate system benefits have been identified for the heat heat pumps, can contribute only moderately in this regard. When
storages. policy makers prioritise efforts and public funds for creating incen-
tives within this field, first priority should thus be given to ensure a
large-scale deployment of the heat pumps. The enabling of flexible
6. Conclusion operation should be given second priority.
The influence of individual heat pumps on wind power integra-

tion has been analysed using a model that optimises both invest- 7. Future work
ments and operation of the energy system. The Danish energy
system by 2030 with an optimised wind power share of around After having assessed the socio-economically feasibility of
50–60% is used as case. The optimisations result in a large scale investments in individual heat pumps and heat storage options,
the next relevant step would be to analyse the feasibility of the [18] Hedegaard K, Mathiesen BV, Lund H, Heiselberg P. Wind power integration
using individual heat pumps – analysis of different heat storage options.
technologies seen from a private economic perspective. Whether
Energy 2012;47:284–93.
individual heat pumps and supplemental control equipment and [19] Klobasa M. Analysis of demand response and wind integration in Germany’s
heat storages will in fact be installed, depends heavily on the reg- electricity market. IET Renew Power Gener 2010;4:55–63.
ulation and incentives. It would therefore be interesting to analyse [20] Energinet.dk. Efficient utilisation of wind power in Denmark. <http://
www.windpower.org/download/857/
to which extent the existing regulation and incentives promote the Effektiv_anvendelse_af_vindkraftbaseret_el_i_Danmark.pdf>; 2009 [accessed
socio-economically optimal solutions and whether further incen- 08.02.13] [in Danish].
tives are needed. Furthermore, it would be relevant to analyse [21] Danish Energy Association, Energinet.dk. Smart grid in Denmark.
<www.danskenergi.dk>; 2010 [accessed 08.01.13] [in Danish].
how flexible operation of the heat pumps affect investments in dis- [22] Energinet.dk. Potential and possibilities for flexible electricity demand with
tribution grid enhancements. Supplementary analyses, where particular focus on individual heat pumps. <http://www.energinet.dk>; 2011
investments in individual natural gas boilers are excluded, could [accessed 08.01.13] [in Danish].
[23] Hedegaard K, Balyk O. Energy system investment model incorporating heat
also be relevant in the light of the long term goal of moving to- pumps with thermal storage in buildings and buffer tanks. Energy, accepted
wards a renewable energy society. Finally, it would be interesting for resubmission.
to include additional competing flexible demand technologies, [24] Ravn H, Hindsberger M, Petersen M, Schmidt R, Bøg R, Grohnheit PE. Balmorel:
A model for analyses of the electricity and CHP markets in the Baltic Sea
such as electric vehicles, in analysing the impacts on wind power region. <http://www.balmorel.dk>; 2001 [accessed 05.11.12].
integration. [25] 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(4):1059–82.
[26] Ravn H. The Balmorel Model Structure. <http://www.balmorel.com>; 2011
Acknowledgments [accessed 03.07.13].
[27] Braun JE. Load control using building thermal mass. J Solar Energy Eng
The authors would like to thank Bernd Müller and Erika Zving- 2003;125:292–301.
[28] Hammer O. Danfoss, Section for heat pumps, personal, communication,
ilate for providing heat demand data for the Danish building stock October; 2012.
from the heat atlas. Thanks are also given to Kenneth Karlsson and [29] Danish Energy Agency. Energy statistics 2010. <http://www.ens.dk/da-dk/
Olexandr Balyk for discussing preconditions for the analysis and to info/talogkort/statistik_og_noegletal/aarsstatistik/Sider/Forside.aspx>; 2011
[accessed 03.10.12] [in Danish].
the reviewers for constructive comments on the article.
[30] COWI, Danish Technological Institute, Danish Gas Technology Centre.
Technology data for energy plants – individual heating plants and energy
transport. Developed for the Danish Energy Agency. <http://www.ens.dk/
References Documents/Netboghandel%20-%20publikationer/2012/
Technology_data_for_individual_heating_plants_and%
20energy_transport.pdf>; 2012 [accessed 05.10.12].
[1] Danish Energy Agency. Energy statistics 2011. <http://www.ens.dk/da-dk/info/
[31] Danish Commission on Climate Change Policy. Documentation for the report
talogkort/statistik_og_noegletal/aarsstatistik/Sider/Forside.aspx>; 2012
of the Danish commission on climate change policy. <http://www.ens.dk/da-
[accessed 05.01.13] [in Danish].
dk/politik/dansk-klima-og-energi-politik/klimakommissionen/
[2] Danish Ministry of Climate, Energy and Buildings. The Governments’ energy
klimakommissionensrapport/sider/forside.aspx>; 2010 [accessed 15.09.12] [in
and climate political goals and the results of the energy agreement in 2020.
Danish].
<http://www.kemin.dk/Documents/Presse/2012/Energiaftale/
[32] Jensen KA. Dimensioning of heat pumps most important. <http://
Faktaark%202%20energi%20og%20klimapolitiske%20mal.pdf>; 2012 [accessed
www.tekniq.dk/PresseOgNyheder/Nyheder/2010/Marts/Dimensionering-af-
08.12.12] [in Danish].
varmepumper-er-vigtigst.aspx>; 2010 [accessed 25.09.12] [in Danish].
[3] Kiviluoma J, Meibom P. Influence of wind power, plug-in electric vehicles, and
[33] Risø DTU, Ea Energy Analyses. Background note on the reference scenario U
heat storages on power system investments. Energy 2010;35(3):1244–55.
and future scenario U. <http://www.ens.dk/da-DK/Politik/Dansk-klima-og-
[4] Hedegaard K, Ravn H, Juul N, Meibom P. Effects of electric vehicles on power
energi-politik/klimakommissionen/klimakommissionensrapport/
systems in Northern Europe. Energy 2012;48(12):356–68.
baggrundsrapporter/Sider/Forside.aspx>; 2010 [accessed 15.09.12] [in
[5] Mathiesen BV, Lund H. Comparative analyses of seven technologies to facilitate
Danish].
the integration of fluctuating renewable energy sources. IET Renew Power
[34] Wittchen KB. Potential energy savings in existing buildings. Danish Building
Gener 2009;3:190–204.
Research Institute. <http://www.sbi.dk/miljo-og-energi/energibesparelser/
[6] Ekman CK, Jensen SH. Prospects for large scale electricity storage in Denmark.
potentielle-energibesparelser-i-det-eksisterende-byggeri/potentielle-
Energy Convers Manage 2010;51(6):1140–7.
energibesparelser-i-det-eksisterende-byggeri>; 2009 [accessed 14.09.12] [in
[7] Østergaard PA. Wind power integration in Aalborg Municipality using
Danish]
compression heat pumps and geothermal absorption heat pumps. Energy
[35] Rambøll Denmark A/S, Aalborg University. Heat plan Denmark, Appendix.
2013;49:502–8.
Danish District Heating Association. <http://vbn.aau.dk/files/16591726/
[8] Blarke MB. Towards an intermittency-friendly energy system: Comparing
Varmeplan_bilagsrapport.pdf>; 2008 [accessed 12.09.12] [in Danish].
electric boilers and heat pumps in distributed cogeneration. Appl Energy
[36] Müller B. Heat atlas for the stock of existing Danish buildings – data received
2012;91(3):349–65.
in access format from the author. Aalborg, Denmark: Aalborg University; 2008.
[9] Zehir MA, Bagriyanik M. Demand side management by controlling
[37] Wittchen KB. Assessment of the heat saving potential in existing buildings.
refrigerators and its effects on consumers. Energy Convers Manage
Danish Building Research Institute. <http://www.sbi.dk/miljo-og-energi/
2012;64(12):238–44.
energibesparelser/energibesparelser-i-eksisterende-bygninger/vurdering-af-
[10] Perfumo C, Kofman E, Braslavsky JH, Ward JK. Load management: model-based
potentialet-for-varmebesparelser-i-eksisterende-boliger>; 2004 [accessed
control of aggregate power for populations of thermostatically controlled
14.09.12] [in Danish].
loads. Energy Convers Manage 2012;55(3):36–48.
[38] Christiansen CH. Danish Technological Institute, personal, communication;
[11] Callaway DS. Tapping the energy storage potential in electric loads to deliver
March 2012.
load following and regulation, with application to wind energy. Energy
[39] Mortensen K. Saint Gobain Weber, personal, communication; November 2012.
Convers Manage 2009;50(5):1389–400.
[40] Bolius Knowledge center for house owners. Floor heating in older houses.
[12] Qureshi WA, Nair NC, Farid MM. Impact of energy storage in buildings on
<http://www.bolius.dk/alt-om/energi/artikel/gulvvarme-i-aeldre-huse/>;
electricity demand side management. Energy Convers Manage
2006 [accessed 07.11.12] [in Danish].
2011;52(5):2110–20.
[41] Aggerholm S, Grau K. SBi Instruction 213 – buildings’ energy demands, PC
[13] Rankin R, Rousseau PG. Demand side management in South Africa at industrial
programme and calculation instruction, Danish Building Research Institute,
residence water heating systems using in line water heating methodology.
Hørsholm, Denmark; 2008 [in Danish].
Energy Convers Manage 2008;49(1):62–74.
[42] Olsen L, Danish Technological Institute, personal, communication; June 2010,
[14] Lund H, Möller B, Mathiesen BV, Dyrelund A. The role of district heating in
November 2012.
future renewable energy systems. Energy 2010;35(3):1381–90.
[43] Chrisensen H, Danfoss, Floor heating department, personal, communication;
[15] Münster M, Morthorst PE, Larsen HV, Bregnbæk L, Werling J, Lindboe HH, et al.
November, 2012.
The role of district heating in the future Danish energy system. Energy
[44] Energinet.dk. Measurements of heat pump installations in Danish households
2012;48(12):47–55.
– from the project Control your heat pump. <www.styrdinvarmepumpe.dk>;
[16] Wittrup S. A large amount of uncontrollable heat pumps requires more power
2012 [accessed 02.03.12].
plants. <http://ing.dk/artikel/106844-et-vaeld-af-ustyrlige-varmepumper-
[45] Andersen JJ. Vølund Varmeteknik, personal, communication, August, 2010;
stiller-krav-om-flere-kraftvaerker>; 2010 [accessed 22.02.13] [in Danish].
March 2012.
[17] Papaefthymiou G, Hasche B, Nabe C. Potential of heat pumps for demand side
[46] Halvgaard R, Kjølstad NK, Madsen H, Jørgensen JB. Economic model predictive
management and wind power integration in the German electricity market.
control for climate control of buildings in a smart energy system. In: 50th IEEE
IEEE Trans Sust Energy 2012;3:636–42.
conference on decision and control and European control conference (CDC- www.gesetze-im-internet.de/atg/__7.html;>; 2011 [accessed 01.10.11] [in
ECC), Orlando, USA; December 2011. German].
[47] Pedersen TS, Andersen P, Nielsen KM, Staermose HL, Pedersen PD. Using heat [56] Mathiesen B, Lund H, Karlsson K. The IDA climate plan 2050 – background
pump energy storages in the power grid. In: IEEE international conference on report. Danish Society of Engineers. <http://ida.dk/sites/prod.ida.dk/files/UK-
control applications (CCA), Denver, CO., USA; September 2011. Future%20Climates%20Background%20Report.pdf>; 2009 [accessed 02.02.13].
[48] Wang PR, Scharling M, Nielsen KP. 2001–2010 Design reference year for [57] International Energy Agency. Tackling investment challenges in power
Denmark. Danish Meteorological Institute <http://www.dmi.dk/dmi/tr12- generation. Paris, France: International Energy Agency; 2007.
17.pdf>; 2012 [accessed 03.02.13] [in Danish]. [58] Danish Energy Agency. Technology data for energy plants. <http://
[49] Hedegaard K. Wind power integration with heat pumps, heat storages, and www.ens.dk/Documents/Netboghandel%20-%20publikationer/2010/
electric vehicles – energy systems analysis and modelling. PhD thesis. DTU Technology_data_for_energy_plants.pdf>; 2010 [accessed 03.09.12].
Management Engineering, Technical University of Denmark, Kgs. Lyngby, [59] Morthorst PE. Heat pumps and low temperature district heating – reporting
Denmark; 2013. from two workshops. Tech. Rep. Risø-R-1234(DA), Risø DTU National
[50] International Energy Agency. World energy outlook 2009, International Energy Laboratory for Sustainable Energy, Roskilde, Denmark; 2008 [in Danish].
Agency, Paris, France; 2009. [60] Pedersen TE. COWI, Energy department, personal communication on
[51] Danish Energy Agency. Preconditions for socio-economic analyses in the field experiences from the large heat pump installed in Frederikshavn in
of energy. <http://www.ens.dk/da-DK/Info/TalOgKort/Fremskrivninger/ Denmark; April 2011.
beregningsforudsatninger/Sider/Forside.aspx>; 2011 [accessed 01.09.12] [in [61] Danish Energy Agency, Energinet.dk. Technology data for energy plants –
Danish]. generation of electricity and district heating, energy storage and carrier
[52] The Danish Government. Det Konservative Folkeparti, Venstre, Socialistisk generation and conversion. <http://www.ens.dk/Documents/
Folkeparti. Aggreement on increased use of biomass in the energy supply and Netboghandel%20-%20publikationer/2012/Teknologikatalog_2012.pdf>; 2012
for industrial purposes. <http://www.ens.dk/da-dk/politik/dansk-klima-og- [accessed 02.10.12].
energi-politik/politiskeaftaler/Sider/1993og1997Biomasseaftale.aspx>; 1993 [62] Danish Energy Association. Heat pumps as part of the flexible electricity
[accessed 15.10.12] [in Danish]. demand, background note for the report: The little blue about heat pumps,
[53] The Danish Government. Our energy. <http://www.ens.dk/DA-DK/politik/ Copenhagen, Denmark; 2010 [in Danish].
dansk-klima-og-energi-politik/voresenergi/Sider/Forside.aspx>; 2011 [63] Danfoss. Danfoss living connect, product prices. <http://www.billigvvs.dk/
[accessed 08.10.12] [in Danish]. Varmesystemer-Radiatorer-Radiator-tilbehoer-Danfoss-Living-
[54] Beurskens LWM, Hekkenberg M, Vethman P. Renewable energy projections as Connect.html>; 2011 [accessed 14.11.11].
published in the national renewable energy action plans of the European [64] Kiviluoma J, Meibom P. Methodology for modelling plug-in electric vehicles in
member states. Petten, The Netherlands: ECN and European Environment the power system and cost estimates for a system with either smart or dumb
Agency; 2011. electric vehicles. Energy 2011;36(3):1758–67.
[55] German Ministry of Justice. Act on the peaceful uses of nuclear energy and the
protection against its hazards § approval of power plants. <http://
Paper VI
District heating versus individual heating in a 100 %

renewable energy system by 2050
Kenneth Karlsson, Olexandr Balyk, Erika Zvingilaite, Karsten Hedegaard.
Modified version of peer reviewed conference proceeding published at the 6th Dubrovnik
Conference on Sustainable Development of Energy Water and Environment Systems, September
25-29, 2011.
Paper VI
District heating versus individual heating in a 100% renewable energy

system by 2050
Kenneth Karlsson*
DTU Climate Centre
Risø National Laboratory for Sustainable Energy,
Technical University of Denmark, Roskilde, Denmark
email: [email protected]
Olexandr Balyk, Erika Zvingilaite, Karsten Hedegaard

Risø National Laboratory for Sustainable Energy,
Technical University of Denmark, Roskilde, Denmark
ABSTRACT
In a future 100% renewable based energy system is district heating (DH) then socio-economically
attractive? Analyses of 100% renewable energy systems are often made either without or with
existing DH infrastructure as a premise, as most of the energy system models do not have the
possibility to invest in DH grids endogenously. In this paper, we compare total socio-economic
costs for achieving a future renewable energy system for the two following cases: 1) A case where
DH infrastructure is built and operated and 2) A case with no DH infrastructure, i.e. only
individual heating. Denmark is used as “test area”, and the linear optimisation model Balmorel is
used to configure a socio-economic optimal energy system in 2050 for each of the two cases. The
model has a higher level of detail for Denmark and includes the possibility to invest in heat
savings in buildings, as an alternative to heat generation. Additionally, human health externalities
related to local air pollution are internalised in the socio-economic cost optimisation. The results
do not show a clear socio-economic benefit of building or not building a DH system in a future
100% renewable energy system.
INTRODUCTION
What should be the role of district heating (DH) in a future 100 % renewable energy system?
The presence of DH in an energy system can have an impact on the balancing of the power system
via large heat pumps and electric boilers combined with thermal storages. This possibility is more
limited if all heating is based on individual installations. Furthermore, individual boilers based on
biomass have higher negative health impacts compared to biomass in large central combined heat
and power plants. On the other hand, a future building stock with significant heat savings and a
power system with large amounts of electricity generation without surplus heat, i.e. wind, sun and
wave energy, could make DH less favourable. So considering these aspects, is large scale DH then
socio-economically attractive? This question forms the background for this study.
We use Denmark as a test area in trying to come up with an answer to this question, applying a
long term perspective and focusing on a 100 % renewable energy system. To do this we need to
include all costs related to the energy system, also the full cost of a district heating network.
1
Paper VI
The analysis is carried out using the energy system model Balmorel. Balmorel is a linear
optimisation model of heat and power supply in Denmark and power and district heating supply in
Norway, Finland, Sweden and Germany [1]. The model has a higher level of detail for Denmark
and here includes a possibility to invest in heat saving measures in buildings as an alternative to
heat generation technologies. Additionally, local air pollution related to human health externalities
are internalised in Balmorel, in order to design a renewable heat and power system with the least
negative impact on human health. The valuation of effects and cost of impact on human health
from air pollution is based on the work in the Centre for Energy, Environment and Health
(www.ceeh.dk).
The analysis is based on running the model for the year 2050 with no existing power and heat
production capacity, leaving it up to the model to create an energy system. This is done for a case
where a district heating network is present and for a case with no district heating network. Then
two different fuel price scenarios are applied adding up to the four scenarios presented in the
paper.
The paper is built up in sections describing inputs to the model, outputs and conclusions. Section
2 describes projections of future energy prices and demand for electricity and heat in the model
area. Section 3 describes technology data and assumptions implemented in the model database.
Section 4 is about new model developments needed for making the analysis and the paper is
concluded in section 5.
ENERGY DEMAND IN 2050

For all the surrounding countries the total electricity demand is projected from an energy service
growth rate and an efficiency improvement rate. Subtracting the efficiency improvement rate from
the energy service growth rate gives the growth rate for the energy demand. For Denmark the
electricity demand consists of an exogenously projected part which relates to increase in energy
service level and an endogenously derived part comprising electricity demand from heat pumps,
electrolysers etc.
Table 1. Energy demand growth and improved efficiency.
Country Growth rate energy Growth rate energy

demand 2008-2020 demand 2020-2050
(% p.a.) (% p.a.)
Service Efficiency Service Efficiency
Norway 2.4 1.5 1.9 1.5
Sweden 2.3 1.5 1.6 1.5
Finland 1.9 1.5 1.4 1.5
Germany 1.7 1.5 1.0 1.5
Denmark 2.0 2.2 2.0 2.2
Using the assumptions in Table 1 leads to the energy demands for Norway, Sweden, Finland, and
Germany shown in Table 2.
2
Paper VI
Table 2. Electricity and district heating demand in Norway, Sweden, Finland and Germany.
TWh/year 2008 2050

Norway electricity 131.7 165.3
Norway district heat 1.7 2.2
Sweden electricity 151.7 171.9
Sweden district heat 43.9 49.7
Finland electricity 99.2 100.8
Finland district heat 34.4 35.0
Germany electricity 614.2 541.1
Germany district heat 93.6 82.5
Table 3 below shows the assumed demand for heat and electricity used as exogenous inputs to
Balmorel in the two modelled cases (with and without DH). The countries around Denmark are
assumed to have the same demands in both cases, while in Denmark the demand for heat in
district heating areas is transferred to individual heating in the case without district heating.
Process energy is higher in the case without DH, because a part of the DH demand is used for
process in the other case.
Table 3. Demand for heat and electricity in Denmark in 2050 in the two cases.
TWh/year 2008 Case incl. DH 2050 Case without DH 2050

Electrical appliances 26.1 58.6 58.6
Electricity for transport 0.3 29.2 29.2
DH demand 28.6 51.1 0
Indiv. room heat excl. DH 32.8 29.4 62.5
Process energy 30.2 33.3 42.8
TECHNOLOGY DATA AND RESOURCES

When looking at year 2050, there is great uncertainty on the costs and efficiencies of energy
technologies. Some technologies being expensive today are expected to improve through the
period, and a technology as photovoltaic is expected to be competitive by then. The technology
assumptions are based on the Danish Technology Catalogue and IEA data and the used data is
presented in tables below.
The costs of district heating network

In Balmorel district heating (DH) network costs are added per unit (kWh) of heat supplied by heat
and/or cogeneration plants. The length of the Danish district heating network is 29 thousand
kilometres, consisting of transmission and distribution pipelines. One average cost is used in the
model for the whole district heating network in Denmark. Clearly, the actual costs of the network
differ for transmission and distribution pipes due to differences in pipe diameters and network
utilisation time. Distribution pipes are used 2000 hours during a year while transmission network
is exploited for 5000 hours annually [2]. The cost of heat supply by DH network in Table 4 is
calculated based on data available in Heat Plan for Denmark [2].
3
Paper VI
Table 4. Cost of district heating network.
Existing Expanded DH for new Cost data for

DH network DH networka buildingsb Balmorelc
Total value of DH net (M€) 14,500 24,970 1,210 26,200
Total DH production (GWh) 35,800 55,000 2,400 57,600
Annuitized value of DH netd (app. €/year) 563,000,000 970,000,000 46,900,000 1,020,000,000
DH system cost per produced heat (€/kWh) 0.016 0.018 0.019 0.018
a
Expansion of the existing network to cover 70 % of heat market in Denmark.
b
District heating infrastructure for new buildings, build before 2030.
c
Weighted average cost of ‘Expanded DH network’ and of ‘DH for new buildings’
d
3 % discount rate and lifetime of 50 years are assumed.
Heat saving costs

Costs of heat saving measures, included in the model, are based on the costs of building element
improvements (Table 5) and current state of 175 types of buildings by purpose and construction
year.
Table 5. Average costs for energy efficiency improvements of building envelopea.
Improvement costb (€/m2element)

Insulating walls 47.3-82.8
Insulating floor 47.0
Insulating roof 26.9
Replacing windows 40.3
a Based on [3].
b Incremental costs for energy efficiency improvements only.
Calculation of heat savings potential is described in [4] and modelling of heat savings in Balmorel
is documented in [5] and [3].
The resulting heat saving costs included in the model span from around 2 €cent/kWh to around 60
€cent/kWh depending on building type and heat saving measure.
Technologies for individual heating

In a 100% renewable energy scenario without district heating, heat supply will be based on
individual biomass boilers, electric boilers, solar thermal and different types of heat pumps. In
addition, investment in individual heat storage is possible in the model (see Table 6). Heat
storages in the model will only be relevant for solar thermal due to the variability of solar
radiation and heat pumps due to electricity price variations.
4
Paper VI
Table 6. Technologies available for investment in individually heated areas [6].
Life Inv. costa Inv. costb O&M Fixed O&M Fuel eff./
Technology time (€/unit) (M€/MW-th) (% of inv.) (k€/MW-th) COP
Heat pump
ground-water 15 13400 1.89 0.6 11.3 3.2
Heat pump air-water 15 6700 0.88 0.6 5.3 2.6
Biomass boiler 15 6700 0.88 2.8 22.6 1
Electric boiler 20 1100 0.13 0.9 1.2 1
c
Heat storage 40 860-1490 0.089 - - 0.98
a
Costs for a typical house with a heat demand of 15 MWh/year.
b
Based on assuming a heat pump dimensioning covering 80 % of heat demand at an outdoor temperature of -
12°C and an indoor temperature of 20°C [7] and [8].
c
Estimated average investment costs for heat storage tanks (M€/MWh) in the size interval of 300-1000 litres
based on [9].
Power and heat supply technologies

The relevant renewable power and heat producing technologies included in the model database for
both process energy in industries and for power and district heat production are listed in Appendix
1. The model database includes several generations of each technology, but here only the best
technologies available in 2050 are included, from which the model chooses when only running it
for 2050.
For power production, the main technologies are wind power, photovoltaic, wave power and
biomass (also municipal waste) based CHP, but also fuel cell technologies are available. For
district heating, the main technologies are heat pumps, biomass based boilers and solar heating.
Industrial process energy can be covered by electric boilers, biomass based boilers or solar
heating (only low temperature).
RESOURCES AND FUEL PRICES

Some resources are limited, and therefore restrictions are added on some of these resources in the
model. As we only include renewable energy in the scenarios for Denmark biomass can be a
limiting resource, but also wind has restrictions on how much can be installed onshore and
offshore. The surrounding countries are still allowed to invest in fossil fuel based energy
production and hydro power in Norway, Sweden and Finland is kept at the same level as today.
Table 7. Resource restrictions for Denmark in the scenarios.
Max usable resource

Offshore wind 20,000 MW
Onshore wind 5,000 MW
Biomass straw 15 TWh
Biomass wood 22 TWh
Biomass wood waste 5 TWh
Municipal waste 13 TWh
5
Paper VI
The transport sector is not included in this Balmorel model, therefore 33 TWh biomass, not
included in Table 7, are reserved for the transport sector to be transformed into bio-fuels and
assumed not to be available to the rest of the energy system.
Ground heat potential for individual heat pumps

The ground heat resource available for individual ground heat pumps is implemented based on
estimates in [10]. In this report, data on heat demand, ground area and built area have been
collected for existing detached houses in Denmark and it is assumed that heat demand in houses
with heat demand below 57 kWh per m2 available ground area can cover their heat demand with
ground heat pumps. The available ground area, i.e. area available for installing ground heat pipes,
is here defined as half of the difference between ground area and built area. Taking also the
ground heat potential for new buildings into account, the ground heat potential for 2050 is
estimated for the different scenarios (see Table 8).
Table 8. Ground heat potentials (GWh).
Ground heat Ground heat potential 2050

potential 2008a DH-scenariob NODH-scenarioc
Eastern Denmark 1361 2438 3242
Western Denmark 2383 3812 6939
a
Source: [10].
b
All new houses built from 2008 to 2050 are assumed to have sufficiently low heat demands to make possible
that heat supply can be based on ground heat pumps. Total heat consumption for residential houses in
individually heated areas is estimated to increase 17 % from 2008 to 2050.
c
Based on data extraction from heat atlas by Bernd Möller, Aalborg University.
Fuel prices
The used price scenarios are based on the fuel price scenarios from the Danish Commission on
Climate Change Policy [11] where the case with high fossil fuel prices, low CO2-price and high
biomass prices represents a world without a global climate agreement. This leads to increased
demand for fossil fuels and thereby higher prices, on the other hand the global demand for
biomass is low and the CO2-market is not working very well. The case with low fossil fuel prices,
high CO2-price and high biomass prices resembles a situation where an ambitious global climate
agreement are in place, forcing the CO2-price up and increasing biomass prices due to high
demand for biomass in the energy sector. The demand for fossil fuel decrease as result of the
global agreement and therefore market prices on fossil fuels are low.
Table 9. Fuel prices and CO2 price used in the scenarios.
Low fossil fuel price scenario High fossil fuel price scenario
(€/GJ) (€/GJ)
CO2-price (€/ton) 320 106
Coal 1,9 3,3
Natural gas 8,7 11,7
Fuel oil 15,0 21,0
Biomass wood 17,7 14,3
Biomass straw 15,6 9,0
Municipal waste -2,3 -2,3
6
Paper VI
RESULTS
The combination of two set of fuel price assumptions and a system with and without district
heating give four scenarios. The naming convention for the scenarios used in result tables and
figures can be seen in Table 10.
Table 10. Name convention for scenarios.
Low fossil fuel price scenario High fossil fuel price scenario
With district heating WithDH_LowFP WithDH_HighFP
Without district heating NoDH_LowFP NoDH_HighFP
Total Danish system costs

Comparing the total costs of a Danish energy system with district heating with a system without,
is not straight forward when the Danish power system is linked up to a common Nordic power
market. Changing the situation in Denmark will impact the investments in the other countries and
thereby will affect trade of power between the countries. So to compare the system costs we count
all investments, operating and maintenance costs, cost of infrastructure, fuel costs and health cost
from air pollution in Denmark, and then add the net income from power trade in each of the
scenarios. The total yearly cost of running the Danish energy system is around 10 bill. € and the
scenarios with district heating have around 2% higher costs than the scenarios without district
heating.
A problem by not having a district heating system is that the municipal waste is not utilised for
power and heat production. Then the waste has to be treated in incineration plants not utilising the
energy or stored at dump sites. If the negative fuel price on municipal waste is used as the cost of
alternative treatment, then the extra costs for the scenarios with district heating drop to 1.5%.
Energy system configuration and fuel consumption

Balmorel optimises the whole Nordic energy system including Germany and finds the optimal
investments in Denmark.
7
Paper VI
MW
30000
25000
20000
WithDH_HighFP
15000
NoDH_HighFP
10000 WithDH_LowFP
5000 NoDH_LowFP
Figure 1. Investment in power producing capacity in Denmark in the four scenarios.
The wind potential is fully developed in all scenarios; wave and biomass are also almost the same
between the scenarios. The difference in the power sector is mainly in photovoltaic capacity. In
the scenarios without district heating industry is relying more on electricity and therefore has a
higher demand for electricity causing increasing investments in photovoltaic.
When it comes to heat production, heat supply in all scenarios is based on heat pumps and
biomass together with heat storage. Only one of the scenarios utilises waste for heat production.
The district heating scenarios have higher production on the central heat pumps because a part of
the district heat is used by industry and therefore counted as heat production. Process energy is
not included in the results showed in Figure 2.
TWh
90000
80000
70000
60000
WithDH_HighFP
50000
NoDH_HighFP
40000
WithDH_LowFP
30000
NoDH_LowFP
20000
10000
0
Heat Heat Biomass Municipal Hydrogen
storage pumps waste
Figure 2. Heat production divided on fuels and scenarios.
8
Paper VI
Looking at the total fuel consumption for electricity and heat production in Denmark (Figure 3), it
is clear that more biomass is used in the cases with lower biomass price and only one scenario
(WithDH_LowFP) utilises municipal waste.
TWh
90000
80000
70000
60000
WithDH_HighFP
50000
NoDH_HighFP
40000
WithDH_LowFP
30000
NoDH_LowFP
20000
10000
0
Wave Wind PV Municipal Biomass
waste
Figure 3. Total fuel consumption for the Danish energy system divided on fuels and scenarios.
Heat savings are not so profitable in areas with district heating, as district heating is a cheap
efficient heat source. This can also be seen from our results. In Figure 4 the heat saving potential
in Danish buildings today is compared with the heat savings implemented in the different
scenarios.
TWh
30000
25000
20000
Floor
15000
Roof
10000 Wall
Window
5000
Figure 4. Heat saving potential in Denmark divided on building components and implemented savings in the
scenarios.
9
Paper VI
In the scenarios with district heating around 30% of the potential savings are implemented while it
is around 45% in the cases without district heating. We also see slightly more investments in heat
savings when the biomass prices are higher (WithDH_LowFP and NoDH_LowFP).
CONCLUSIONS
Based on the modelling results, there is no clear indication whether it is cheaper to run a future
renewable based energy system with or without a district heating system. Every model has
limitations, and it is important to be aware of these before concluding on the results. In the present
analysis, we included all system costs (except from electricity distribution net) and also the costs
of building and operating a district heating network, in the cases with district heating. But the
district heating piping technologies are not optimised to serve low energy buildings; therefore
more optimised district heating technologies would improve the results for district heating. As
seen from the results, not all scenarios utilise municipal waste for energy production, which
means that the waste has to be removed in another way and this is not included in the modelling.
The results are also very sensitive to assumed development in costs and efficiencies for the
different supply technologies. Taking all these precautions into account our analysis show that
there is no socio-economic argument for building or not building a district heating system in a
future 100% renewable energy system.
ACKNOWLEDGEMENTS
The presented study is a part of the research of the Centre for Energy, Environment and Health,
financed by The Danish Strategic Research Program on Sustainable Energy under contract no
2104-06-0027.
REFERENCES
1. Ravn H et al. Balmorel: A Model for Analyses of the Electricity and CHP Markets in the
Baltic Sea Region. www.Balmorel.com, 2001.
2. Rambøll (2008): Rambøll Danmark, 2008, “Varmeplan Danmark” (Heat plan for Denmark)
Bilagsrapport, 2008.
3. Zvingilaite E. (2010): Zvingilaite E., 2010, “Modelling energy savings in the Danish building
sector combined with internalisation of health related externalities in a heat and power system
optimisation model”, submitted to Energy Policy.
4. Risø DTU (2010): Risø DTU, “Energiforbrug og besparelser i bygninger. Sektoranalyse.
Bygninger”, Klimakommissionen, September, 2010.
5. Zvingilaite E. (2009): Zvingilaite E., 2009, “Modelling of heating sector in Denmark with
focus on local externalities”, Paper presented at the International Energy Workshop 2009, 17-
19 June, 2009, Venice, Italy.
6. Lund et al (2010): H. Lund, B. Möller, B. V. Mathiesen and A. Dyrelund, "The role of district
heating in future renewable energy systems," Energy, vol. 35, pp. 1381-1390, 3, 2010.
7. Lars Olsen, Danish Technological Institute: Lars Olsen, Danish Technological Institute,
Personal communication, June 30 and September 8, 2010.
8. Jensen (2010):, Jensen, K. A. Dimensionering af varmepumper er vigtigst, 2010. Available:
http://www.tekniq.dk/PresseOgNyheder/Nyheder/2010/Marts/Dimensionering-af-
varmepumper-er-vigtigst.aspx
10
Paper VI
9. Vølund A/S (2010): Vølund A/S. Prices on heat pumps. Prisliste - vejledende udsalgspriser
april 2010, 2010.
10.Danish Technological Institute (2010): Danish Technological Institute. potentialebeskrivelse -
individuelle varmepumper, 2010. Available: www.danskenergi.dk
11.Danish Commission on Climate Change Policy (2010): Danish Commission on Climate
Change Policy. Available: http://www.klimakommissionen.dk/da-
DK/OmKlimakommissionen/Klimakommissionensrapport/Sider/Forside.aspx.
11
Paper VI
APPENDIX
Variable Fixed
Fuel Invest., Life,
Technology Description O&M, O&M,
eff. M€/MW year
€/MWh k€/MW
WI-On-ENS50 Wind turbine size 3.5 MW 1.00 1.16 11.00 30

WI-Of-ENS50 Wind turbine size 8-10 MW 1.00 2.00 14.00 30
Heat boiler (biomass),
HO-BP-WO industry 1.00 0.40 12.00 20
SH-PR Solar heating, industry 1.00 0.60 0.62 20
EH-BP Electric boiler, industry 1.00 0.24 2.00 20
Steam turbine, wood pellets,
advanced steam process, ultra-
ST-WO-ENS50 supercritical, 250-400 MW 0.55 1.40 7.00 40
ST-MW-ENS20 Waste to energy CHP 0.97 8.50 22.00 155.00 20
Steam turbine, medium,
ST-WW-ENS30 woodchips, 10-100MW 0.49 1.60 3.20 23.00 30
Steam turbine, medium, straw,
ST-ST-ENS10 10-100MW 1.02 2.20 6.10 38.00 25
Steam turbine, small, straw, 8-
ST-STs-ENS20 10MW 0.90 3.90 156.00 20
Steam turbine, small,
ST-WWs-ENS20 woodchips, 0.6-4.3MW 1.03 3.50 122.50 20
SE-PV-ENS50 Phtovoltaic cells 1.00 0.95 12.00 30
Wave-ENS50 Wave power 1.00 2.55 7.00 20
Large heat pump 1-10MJ/s
(heat output), heat source:
EH-PA-ENS50 ambient temperature 3.20 0.50 4.00 20
Large heat pump 1-10MJ/s
EH-PW-ENS50 (heat output), heat source: 35C 3.80 0.50 4.00 20
EH-BL-ENS10 Electric boiler, 20 MW, 10kV 0.99 0.06 0.50 1.00 20
EH-BM-ENS10 Electric boiler, 10 MW, 10kV 0.99 0.08 0.50 1.00 20
EH-BS-ENS10 Electric boiler, 1-3MW, 400V 0.99 0.24 0.50 1.00 20
Waste to energy, district
HO-MW-ENS20 heating 0.98 1.10 5.30 50.00 20
District heating boiler, wood-
HO-WW-ENS10 chips, 1-50MW 1.08 0.50 23.50 20
Solar district heating,
investment cost are calculated
using full load hours for
SH-DH-ENS30 Denmark 1.00 0.21 0.50 20
Alkaline electrolysis (AEC),
EH2-AL-ENS10 <3.4MW 0.73 0.23 7.00 15
Solid oxide electrolysis
EH2-SO-ENS30 (SOEC), 5MW 0.83 0.57 14.00 20
Solid polymer electrolysis
EH2-SP-ENS20 (PEMEC), 0.045MW 0.88 0.16 6.00 20
FC-SO-ENS30 SOFC CHP 2-3MW 0.90 0.50 25.00 15
FC-PE-ENS20 PEMFC CHP 20MW 0.80 0.50 0.60 7
H2-STO-30 Hydrogen storage 1.00 0.01 25
H-STO Heat storage 0.99 0.00 20
12
Report chapter
Chapter 9: Balmorel model results – EVs and power system

investments.
Karsten Hedegaard, Hans Ravn, Nina Juul, Peter Meibom.
Risø-Report, Risø-R-1804 (EN): Electricity for Road Transport, Flexible Power Systems, and
Wind power. December 2011. Systems Analysis Department, Risø DTU National Laboratory for
Sustainable Energy.
9 Balmorel model results – EVs and power system
investments
The overall goal of this analysis is to investigate how a gradual large-scale implementation of
plug-in hybrid electric vehicles (PHEVs) and battery electric vehicles (BEVs) in the private
passenger vehicle fleet will influence power system investments and operation in the years
towards 2030. The analysis covers not only the Danish power system but also the power
systems of Germany, Norway, Sweden and Finland. This is due to the importance of electrical
interconnections and in order to reveal possible differences in the effects of EVs on different
power systems. However, main focus is put on the effects on the Danish energy system.
The idea of the analysis is to investigate how the power system will be affected by the increase
in electricity demand due to introduction of EVs and by the flexibility of this demand when
assuming intelligent charging of the vehicles. Moreover, the effect of activating vehicle-to-
grid capabilities is investigated by assuming that EVs can deliver power back to the system
when needed.
In the following, the scope, preconditions and results of the analysis are presented. Input data
and further modelling preconditions for the analysis are given in Appendix12.3.
9.1 Scope and preconditions
The assumed implementation of private passenger electric vehicles (EVs) is based on

scenarios set up by the Electric Power Research Institute (EPRI) [91] and IEA [92]. In the
Medium scenario, EPRI assumes a development in PHEVs new vehicle shares as outlined in
Figure 66a. Based on the relative development in sales of PHEVs and BEVs towards 2030
presented by IEA in the Blue Map scenario, we assume additional BEV market shares
corresponding to half of the PHEV new vehicle shares. As a result, we consider a development
in the vehicle fleet shares towards 2030 as illustrated in Figure 66b. Consequently, EVs are
assumed to comprise around 2.5 %, 15 %, 34 % and 53 % of the private passenger vehicle
fleet in 2015, 2020, 2025 and 2030, respectively. This development in the vehicle fleet shares
is assumed for all the Northern European countries.
100% 100%
New vehicle shares
80% 80%
Vehicle fleet
60% 60%
ICE
40% 40%
BEV
20% 20% PHEV
0% 0%
Figure 66 a) Development in plug-in hybrid electric vehicles (PHEV) new vehicle shares in
the Medium scenario in [91] and illustration of the assumed relation between battery electric
vehicle (BEV) and PHEV new vehicle shares based on the Blue Map Scenario in [92]. b)
Assumed gradual penetration of PHEVs and BEVs in the vehicle fleet (ICE: Internal
Combustion Engine vehicles).
Risø-R-1804 (EN) 101

To analyse the impacts of EVs over the period, two scenarios are set up:
o Base: Only internal combustion engine (ICE) private passenger vehicles towards 2030
o EV: Gradual implementation of PHEVs and BEVs as private passenger vehicles
towards 2030 (as outlined in Figure 66b)
The power, district heat and transport system is modelled in integration using the model
Balmorel including the transport-addon developed by Juul and Meibom in [93]. Model
development has been made in order to handle the gradual implementation of different vehicle
vintages towards 2030. Balmorel is a deterministic partial equilibrium model assuming perfect
competition optimising investments in power/heat production, storage and transmission
capacities and minimises total costs in the energy system - covering annualised investment
costs, operation and maintenance costs of existing and new units, as well as fuel and CO2
quota costs (the model is further described in the Chapter 12.2). The transport-addon includes
demand for transport services, vehicle investment and operation costs and electricity balancing
in the integrated road transport and power system. As the gradual implementation of PHEVs
and BEVs is fixed, investments in vehicles are in this study not performed as part of the
optimisation.
Simulations are made with five year intervals, for 2015, 2020, 2025 and 2030, where optimal
investments identified in previous years are included in the optimisations of subsequent years.
Plug-in patterns for BEVs and PHEVs have as in [93] been derived from driving patterns
obtained from the investigation of transport habits in Denmark [94]. It has been assumed that
the EVs are plugged-in at all times when parked and that driving habits are the same for all the
countries in the simulation. Optimal vehicle-to-grid and grid-to-vehicle power flows are
identified as part of the optimisation.
All EVs are assumed to leave the grid with a fully charged battery, restricting the loading to
meet this load factor. The PHEVs are assumed to use the electric storage (the usable part of
the battery) until depletion before using the engine. This is considered a reasonable
assumption due to the high efficiency of the electric motor compared to that of the combustion
engine as well as the low price of electricity (average prices in the neighbourhood of
€50/MWh in the simulations) compared to the price of diesel (64-80 €/MWh in 2015-2030
[95]). The model works with a capacity credit restriction ensuring enough production capacity
to meet peak power demand as presented in [96]. BEVs and PHEVs are due to V2G capability
able to contribute in meeting peak power demand. Modelling of this contribution is taken from
the PhD thesis by Nina Juul [97].
Integrating the power and transport systems and introducing intelligent charging/discharging
requires a number of additions to the existing system, e.g. communication between vehicles
and the power system, vehicle aggregators communicating with power markets, and agreement
upon connection standards. All such changes are in the model assumed to be in place and
infrastructure costs, e.g. charging stations and hardware, are not included.
In the analysis, investment in the following unit types is allowed:
o Wind turbines (onshore, offshore)

o Coal CHP, steam turbine, extraction
o Natural gas CHP, combined cycle gas turbine, extraction
o Natural gas open cycle gas turbine, condensing
o Nuclear power, condensing (in Finland and Sweden only)
o Biomass CHP, medium, extraction (wood chips), Biomass CHP, small, backpressure
(wood chips), Biomass CHP, small, backpressure (straw)
o Natural gas heat boiler, Biomass heat boiler (wood chips)
o Heat pumps
o Electric boilers
o Heat storages
o Transmission capacities between power regions
102 Risø-R-1804 (EN)

Data on capacities, efficiencies, operation costs, and technical lifetimes etc. for existing units
for power/heat production, storage and transmission are included in the model. Gradual
decommissioning of existing power/heat production capacities towards 2030 is thus taken into
account. The optimisation is based on socio-economic costs in order to investigate how EVs
would affect the power system in the absence of taxes, tariffs and subsidies. The idea behind
this approach is that if the outcome does not correspond to what is desired for society, taxes,
tariffs and subsidies can then be designed in order to reach the situation wanted.
9.2 Results
In the following, the results are shown covering the effect of EVs on power system
investments, electricity generation, CO2 emissions and costs.
Effects on power system investments
Socio-economically optimal investments in new power production capacities generated by the

model in the Base and EV scenario are illustrated in Figure 67 for each of the five countries.
As shown, the investments cover on-shore and off-shore wind power, coal based CHP, nuclear
power where this option is allowed, and open cycle gas turbines (OC-GT); the latter for
ensuring sufficient capacity to cover peak loads.
As a result of increasing fuel and CO2 prices, the economic conditions for wind power
generally improve over the period. This is clearly illustrated for the cases of Denmark,
Germany, and Sweden where wind power investments only or mainly occur in the last part of
the period towards 2030. Furthermore, existing wind power capacities in Denmark and
Germany are significantly decommissioned from 2020 to 2025 (from around 3,200 MW to 0
MW in Denmark and from around 23,000 MW to 11,000 MW in Germany). This is likely part
of the explanation for the large wind power investments occurring in these countries in 2025.
Norway, Sweden, and Finland, have relatively high onshore wind power resources in terms of
obtainable full load hours 1. Therefore, wind power investments occur earlier in these countries
than for the cases of Denmark and Germany.
1 Assumed full load hours for onshore wind power: Norway: 3000 [99.1, 99.2], Finland, Sweden: 2600 [99.1] and
based on [99.3], Denmark, West: 2440, Denmark, East: 1960[99], Germany: 1750 (based on [99.4, 99.5] ).
Risø-R-1804 (EN) 103

DENMARK
6,000
5,000 WI-onshore
4,000 WI-offshore
MW
3,000
OC-GT
2,000
1,000 ST-COAL
0
Base EV Base EV Base EV Base EV
2015 2020 2025 2030
DENMARK
GERMANY
40,000
30,000 WI-onshore
WI-offshore
MW
20,000
10,000 ST-COAL
0
2015 2020 2025 2030
GERMANY
SWEDEN
10,000
8,000 WI-onshore
6,000 OC-GT
MW
4,000 ST-COAL
2,000 NUC
0
2015 2020 2025 2030
SWEDEN
NORWAY
6,000
4,000
MW
WI-onshore
2,000
0
2015 2020 2025 2030
NORWAY
FINLAND
12,000
WI-onshore
10,000
8,000 WI-offshore
MW
6,000 OC-GT
4,000 ST-COAL
2,000
NUC
0
2015 2020 2025 2030
Figure 67. Investments in power production capacities in the Base and EV scenario,
representing accumulated investments over each five year period. E.g. investments in 2020
represent accumulated investments from 2016 through 2020. WI: Wind power; ST-COAL:
Steam turbine, extraction, coal; OC-GT: Open cycle gas turbine; NUC: Nuclear power.
Comparing the Base scenario with the EV scenario, it can be seen that the gradual
implementation of EVs facilitates increased wind power investments in all five countries. The
reason is that the flexible charging/discharging of EVs supports the integration of the variable
production from wind power into the power systems. For Norway and Finland, this effect is
observed from 2020, where EVs comprise 15 % of the vehicle fleet. As such, e.g. Finnish
wind power investments are doubled in 2020 with the implementation of EVs. At higher EV
104 Risø-R-1804 (EN)

fleet shares of 34 % in 2025, EVs generate increased wind power investments in Germany and
Sweden and in 2030, where EVs comprise 53 % of the vehicle fleet, increased wind power
investments are observed in Denmark and Germany, and particularly in Sweden where wind
power investments are increased manifold. In Finland, the assumed onshore wind potential is
reached in 2030 in both scenarios. Hence, in this case the EVs push forward the investments in
wind power. The Danish wind power investments in 2025 are made in Western Denmark
where wind resources are highest. The wind power investments generated in the EV as well as
in the Base scenario are constrained by the onshore wind potential for this area (set to 3,500
MW). As a result, identical wind power investments are observed for the two scenarios. In
2030, the effect of EVs on Danish wind power investments is, however, significant, increasing
offshore wind power investments from 0 MW to around 1,600 MW. In the EV scenario, the
accumulated Danish wind power capacity in 2030 is around 5,100 MW, i.e. significantly lower
than the national medium/high wind target for 2030 of 7,300/8,000 MW [98]. As such, the
results suggest that a large scale implementation of EVs is not sufficient to facilitate reaching
the Danish wind target for 2030 by socio-economic optimality.
As a consequence of the large-scale EV implementation and resulting increase in electricity

demand, one might expect a significantly increased need for dispatchable power production
capacity. However, the results show that when EVs are charged/discharged intelligently,
increased investment in dispatchable power production capacity is only observed in a few
cases (Germany in 2020-2030). In fact, rather than increasing investments in thermal
production capacity, EVs result in a reduced need for new thermal power capacities for the
case of Denmark, Finland, and Sweden. As such, investments in open cycle gas turbines
and/or coal CHP capacities are reduced significantly in these countries in 2025-2030 with the
implementing of EVs. This is explained by the V2G capability of EVs contributing in
covering peak loads. In Denmark, the effect is most significant in 2030 where investments in
open cycle gas turbines are reduced from around 2,000 MW to 0 MW.
Effects on electricity generation
By observing electricity generation in the EV scenario relative to generation in the Base

scenario, it can be seen how electricity for EVs is produced in the optimisation (see Figure
68). As shown, the electricity demand for EVs is in Denmark largely covered with coal based
electricity production in 2015 and 2020. The production increase occurs almost exclusively on
existing plants, since the increase in Danish coal power investments caused by EVs is
diminishing (cf. Figure 67). From 2025, Danish electricity demand for EVs is partly met by
biomass based power production. This is a consequence of the increasing CO2 prices and the
relatively large increase in fossil fuel prices compared to biomass prices over the period. In
2025, electricity demand for EVs in Denmark is partly met by reducing electricity export to
Germany and Sweden. This explains the gap in 2025 between the increase in power generation
and the electricity demand for EVs. In 2030, EVs is in Denmark generate a significant increase
in wind power generation. This is a direct consequence of the increased wind power
investment in the EV scenario compared to the Base scenario; i.e. only by increasing wind
power capacities, EVs can result in increased wind power generation. The generated increase
in Danish wind power production in 2030 is much higher than domestic electricity demand for
EVs. As a result, significant displacement of coal based power production occurs and net
electricity import from Norway and Sweden is reduced.
Risø-R-1804 (EN) 105

8 60
WOOD 4
WOOD
6 50
WIND 3
40 WIND
4 2
TWh
TWh
TWh
30
STRAW WATER
2 20 1
NAT_GAS
10 STRAW
0 0
0
MUNI_WAST NUCLEAR
2015 2020 2025 2030 2015 2020 2025 2030
-2 -10
E 2015 2020 2025 2030 -1
15 15 80 Coal
WOOD_WASTE
WOOD
10 60 Wind
10 WIND
WATER
Lignite
TWh
5 40
TWh
TWh
5 STRAW Wood
PEAT
0
NUCLEAR
20 Natural gas
0 2015 2020
NAT_GAS 2025 2030 0 Straw
-5
LIGHTOIL
2015 2020 2025 2030 FUELOIL Electricity
2015 2020for 2025
EDVs 2030
-5 -10
COAL -20
Figure 68. Changes in annual electricity generation due to implementation of electric

vehicles. Due to import/export, possible changes in electricity consumption for heat
pumps/electric boilers, and in the use of pumped hydro electricity storage, generated power
increases in each year will not necessarily correspond to the electricity demand for electric
vehicles.
The German case shows similarities with the Danish in the sense that electricity demand for
EVs is in 2015-2020 largely met by increased coal based power production while wind power
does not contribute in providing electricity for the EVs until in the last part of the period. Also
in Sweden, EVs do not facilitate increased wind power generation until 2030. The Finnish
case stands in contradiction the trend in Denmark, Germany and Sweden. As such, EVs in
Finland generate increased wind power production from 2020 until reaching the assumed
onshore wind potential. After that point, the electricity demand for EVs in Finland is largely
met by coal fired electricity production. However, estimating the Finnish onshore potential is
connected with large uncertainty and the potential might be higher than assumed. If setting the
onshore wind potential higher, electricity for Finnish EVs would in the optimisation, also in
the last part of the period, most likely be met by wind power. Norway is a large net electricity
exporter and to a large extent, the cheapest way of providing electricity for EVs is therefore to
reduce the export. Similarly, in 2020-2025 Sweden largely provides electricity for EVs by
cutting down export. For this part, the implementation of EVs in Norway and Sweden, thus
contributes to the generated increases in power production observed in the other countries.
CO2 emissions
As result of the EV implementation, Danish CO2 emissions from the power, heat and transport
systems modelled, are more or less unchanged in 2015-2020 while significant emission
reductions are obtained in 2025, 7 %, and in 2030, 17 % (see Figure 69 a). The most important
factors behind the significant improvement in the CO2 balance over the period are 1) the
increasing share of renewable energy in the electricity mix for EVs, 2) the gradual
improvement in the efficiency of the EVs and 3) the increasing shares of EVs in the fleet. As
illustrated in Figure 69b, the emission reductions in 2025 and 2030 are mainly caused by
displaced fuel consumption for ICEs. In addition, an emission reduction from power&heat
production is observed in 2030. This is an effect of the significant displacement of coal based
power production that year (cf. Figure 68).
106 Risø-R-1804 (EN)

35 30
30 25 Transport
CO2 (Mtonnes)
CO2 (Mtonnes)
fuel, PHEV
25 Base 20
20 15 Transport
EV
fuel, ICE
15 10
10 5 Power&Heat
5 0
0 Base EV Base EV
2015 2020 2025 2030 2025 2030
Figure 69 a). Danish CO2 emissions for the simulated power, heat and transport system in the
Base and EV scenario. b) Danish CO2 emissions in 2025 and 2030, divided on sources for the two
scenarios.
For the five Northern European countries as a whole CO2 emissions are also more or less
unchanged in 2015-2020 while reductions of 3 % and 7 % are obtained in 2025 and 2030.
Costs
Figure 70 shows that the implementation of EVs results in an increase in total costs for the
simulated power, heat and transport sector of the Northern European countries; around 1.5-7.1
€ Billion/yr depending on the year, corresponding to increases of 0.8-3.9 %.
The cost increase is partly due to larger investment costs per vehicle for BEVs and PHEVs
compared to ICEs. Moreover, due to the assumed lower annual driving of BEVs compared to
ICEs, a larger amount of BEVs are required to provide the same transport demand. Overall,
this increases total investment and O&M costs for the transport sector. As illustrated in Figure
70b, the cost reduction from displacing fuel consumption in ICEs is not enough to compensate
for this. The cost increase is highest in 2020 (3.9 %) and then lower in 2025 (1.2 %) and 2030
(0.8%); reflecting the influence of expected technical and economic improvements of EVs
over the period.
200 200
180 Transport,
Inv.+ O&M
Bill. €/yr
160 150
Bill. €/yr
140 Transport,
120 Base CO2 from fuel
100
100 EV Transport,
80
50 fuel
60
40 Power&Heat
20 0
0 Base EV Base EV
2015 2020 2025 2030 2020 2025
Figure 70 a). Total costs for the simulated power, heat and transport system for the Northern
European countries in the Base and EV scenario. b) Total costs in 2020 and 2025, divided on
sources.
These cost effects are based on an assumed implementation of PHEVs as well as BEVs while
it should be mentioned that PHEVs alone have, in [93], shown to provide system cost
reductions. Furthermore, potential benefits from using EVs for providing regulating power and
power reserves is not included in the cost estimates. Finally, the socio-economic benefit of
reducing the transports dependency on oil, increasing security of supply, is not valuated.
Risø-R-1804 (EN) 107

When relating the cost increases (excluding CO2 quota costs) to the CO2 emission reductions
provided by EVs, average CO2 reduction costs for EVs can be estimated for the five countries
as a whole. This shows that CO2 reduction costs are reduced manifold over the period; from
very high levels of around 7100 €/ton in 2015 and 1500 €/ton in 2020, to 110 €/ton in 2025
and 80 €/ton in 2030. However, even in 2030, the CO2 reductions costs for EVs are rather high
compared to the expected CO2 price level of around 39 €/ton [95]. As such, when comparing
with assumed CO2 price levels, the analysis suggests a low cost efficiency of EVs in
providing CO2 reductions, particularly in the short term.
Sensitivity analysis
In the analysis above, based on [95], CO2 prices are assumed to increase from 20€/ton CO2 in
2015 to 39 €/ton CO2 in 2030, and the assumed fuel prices correspond to an oil price of
$88/barrel in 2015 and $117/barrel in 2030. In a sensitivity analysis, the following low/high
fuel price and low/high CO2 price developments are assumed:
• Fuel prices: set to low at $80/barrel in 2015 increasing linearly to $90/barrel in 2030
and at high increasing linearly from $80/barrel in 2010 to $95/barrel in 2015 and
$140/barrel in 2030. Ratios between prices on different fuels are kept constant and are
based on [95].
• CO2 prices: set to low at 15 €/ton in 2015 increasing linearly to 20 €/ton in 2030 and at
high increasing linearly from 14 €/ton in 2010 to 26 €/ton in 2015 and 60 €/ton in
2030.
These analyses show that also at low/high fuel and low/high CO2 prices, EVs facilitate a
reduced need for new coal/natural gas production capacities in several of the countries,
including Denmark. However, changes in investments and electricity production caused by the
EVs over the period are generally found to be sensitive to the development in fuel and CO2
prices. As such, e.g. for Denmark, at the low fuel price conditions, wind power is not included
in the electricity mix for EVs towards 2030. At the low CO2 price conditions, onshore wind
power investments in the Base scenario are reduced below the onshore potential for Western
Denmark. As a result, EVs facilitate an increase in onshore wind power investments in 2025.
However, no offshore wind power investments are observed and in 2030, wind power only
contributes with a small part of the electricity for EVs. Overall, at the low fuel/CO2 price
conditions, electricity demand for EVs in Denmark is in most of the period towards 2030
largely covered by coal based power or through electricity exchange. When assuming high
fuel or CO2 prices EVs facilitate considerable increases in wind power investments from 2025,
i.e. five earlier than at the original price conditions. The resulting reductions in Danish CO2
emissions depending on price conditions are presented in Table 1.
Table 1. Change in Danish CO2 emissions reductions due to implementation of EVs depending
on fuel and CO2 price conditions
2015 2020 2025 2030

Fuel and CO2 prices based on [95] 0.04% -0.1% -7% -17%
Low fuel prices -0.5% -0.4% -6% -15%
High fuel prices -0.2% -2.0% -14% -22%
Low CO2 prices -0.4% -1.1% -9% -10%
High CO2 prices* 0.1% -0.8% 2% -20%
* The slight CO2 emission increase in 2025 in the high CO2 price scenario is due to reduced net
import, resulting in increased coal based power production in Denmark. Cf. Figure 61 which
illustrates the diverse reactions in different countries. For the five considered countries as a whole,
CO2 emissions are in 2025 reduced with 6 % in this scenario.
108 Risø-R-1804 (EN)

9.3 Conclusions
o When charged/discharged intelligently electric vehicles (EVs) can facilitate increased

wind power investments and can due to vehicle-to-grid capability reduce the need for
new coal/natural gas power capacities
o Wind power will likely provide a large share of the electricity for EVs towards 2030
in several of the Northern European countries
o However, if not followed up by economic support for renewable energy technologies
other than CO2 quotas, wind power will, for the case of Denmark (and Germany and
Sweden) not contribute in providing electricity for EVs until the last part of the period
o As a result, electricity demand for EVs will in Denmark (and Germany) in the short
term likely be met by coal based power
o Large scale implementation of EVs is not sufficient to facilitate reaching the Danish
wind target for 2030 by socio-economic optimality
o Effects of EVs on the power system vary significantly from country to country and are
sensitive to variations in fuel and CO2 prices
o In the last part of the period towards 2030, EVs can provide significant CO2 emission
reductions for the Danish energy system as well as for the Northern European
countries as a whole.
Risø-R-1804 (EN) 109

Indoor temperature measurements in Danish houses

Measured heat supply in Danish houses
Handling of the summer period
Modelled heat supply over the year
Karsten Hedegaard.
Indoor temperature measurements in Danish houses

Indoor temperature measurements for 28 Danish houses, with heat pumps (radiator heating) are
shown in Figure A and Figure B [1].
25
Indoor air temp. ( C)
24
23
22
21
20
19
18
1 3 5 7 9 11 13 15 17 19 21 23
Hour
Figure A. Diurnal indoor air temperature profile for an average day in the period November 1 to December 31,
2011, measured in 28 Danish houses with heat pump installations and radiator heating [1] (measured in living
room or kitchen). The thick black line indicates the average diurnal profile.
27
Indoor air temp. ( C)
26
25
24
23
22
21
20
19
18
1
115
172
229
286
343
400
457
514
571
628
685
742
799
856
913
970
58
1027
1084
1141
1198
1255
1312
1369
1426
Hour
Figure B. Duration curve for indoor air temperatures in the period November 1 to December 31, measured in 28
Danish houses with heat pump installations and radiator heating [1] (measured in living room or kitchen). The
thick black line indicates the average duration curve. The dotted black lines indicate the lower/upper indoor air
temperature levels applied in the analysis.
The temperature measurements above (measured in living room/kitchen) form part of the
background for the temperature settings applied in the analysis of heat pumps and heat storages:
1) the fixed indoor temperature requirement of 21.5 °C in the reference situation without flexible
heat pump operation and 2) the allowed indoor temperature interval of 20-23 °C during the day,
in the situation where heat pumps are operated flexibly, utilising the heat storage capacity of the
building structure (an allowed temperature interval of 19-22 °C at night is assumed when
supplementing with other sources, as mentioned in Section 2.2).
A winter period, November 1 to December 31, has been chosen to reduce the influence of solar
transmission. Thereby, the measured indoor temperatures best reflect the thermal comfort
preferences of the residents. The heat pumps have been operated conventionally, i.e. not flexibly,
in the given period [2]. Houses with radiator heating are used as basis, since the bulk of the heat
1
demand for individually heated Danish one-family houses by 2030 are expected to represent
radiator heating systems (see Section 2.2). Moreover, radiator heating systems enable a
significantly faster regulation in indoor temperatures than (concrete) floor heating systems.
Therefore, indoor temperatures measured in houses with radiator heating best reflect the
temperature preferences of the residents. The same general patterns in indoor temperatures have
however been observed for houses with floor heating systems.
Measured heat supply in Danish houses

Average diurnal space heating profiles for Danish houses with heat pumps and radiator heating
are shown in Figure C and Figure D for two different periods. The resulting average profiles
(indicated with black bold) are used in the comparison with corresponding model generated space
heating profiles in Paper IV.
0.07
House 1
0.06
House 2
0.05
0.04 House 3
0.03 House 4
0.02 House 5
0.01 Av. house

0.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure C. Average diurnal space heating profile measured for 5 Danish houses with heat pump installations and
radiator heating. The data cover the period March 15 – May 15 and Sep 15 - December 31, 2011.
0.07 House 1 House 2


House 5 House 6
House 11 House 12
House 17 House 18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 House 23 House 24
Hour Av. house
Figure D. Average diurnal space heating profile measured for 24 Danish houses with heat pump installations
and radiator heating. The data cover the period 1 November – 31 December, 2011.
Handling of the summer period

Depending on individual comfort preferences and the degree of insulation of the house, Danish
households with heat pump installations typically turn off space heating manually in the summer
or let the space heating be turned on/off automatically, on a day to day basis, through weather
compensation. Weather compensation typically means that when the average ambient
temperature for the past day drops below a certain level (15-20 °C depending on the degree of
2
insulation), space heating is turned off [3]. When the average ambient temperature increases
above this level, space heating is then turned on again. This mechanism is however not possible
to model in a linear optimisation model. Regardless of the type of control, use of heat pumps for
space heating in the summer period is however typically very low [3] [1]. Based on the above, it
is generally assumed that space heating is only needed in the heating season, which in practice
typically covers the period September 15 to May 15 [4]. For computation efficiency reasons, this
is implemented in the model by activating the space heating equations, for the heating season
only.
Modelled heat supply over the year

Figure E illustrates how generation of space heating and hot water over the year is represented in
the thermal building model for an average existing detached house (151 m2).
8
7
6 Space heating
5
kW-th
4 Hot water
3
2
1
0
Apr
Nov
Dec
Mar
Aug
Sep
Jul
Oct
May
Jan
Feb
Jun
Figure E. Space heating generation and hot water demand profile over the year for an average existing
individually heated detached house as generated by the model.
As shown, no space heating occurs in the non-heating season May 15 to September 15. The peaks
in space heating represent the few hours of the year where outdoor temperatures are very low (-11
°C to -13 °C). As can be seen, the total heat pump capacity needed for an average existing
detached house is around 8 kW-th (for space heating and hot water). This fits well with the
capacity of 8.3 kW-th, used in [5] as typical value for individual heating installations.
References
[1] Energinet.dk, "Measured data of heat pump installations in Danish households, from the
project control your heat pump," Energinet.dk, Fredericia, Denmark, Available at:
www.styrdinvarmepumpe.dk (accessed March 2, 2012), 2012.
[2] Steen Kramer. Energinet.dk, "Personal communication," March 2011.
[3] Jacob J. Andersen. Vølund Varmeteknik, "Personal communication," August 2010, January,
March 2012.
[4] Mylin Energi Consult ApS, "Good energy advices [in Danish] [Gode energiråd],"
Available at: http://mylliin.s-11.dk/default.asp?pid=72. Accessed on January 5, 2013.
[5] Rambøll Denmark A/S and Aalborg University, "Heat plan Denmark, appendix [Varmeplan
Danmark, bilagsrapport] [in Danish]," Danish District Heating Association, Kolding, Denmark,
2008.
3
The fluctuating and only partly predictable nature of wind challenges an effective integration of
large wind power penetrations. This PhD thesis investigates to which extent heat pumps, heat stor-
ages, and electric vehicles can support the integration of wind power. Considering the gaps in exist-
ing research, main focus is put on individual heat pumps in the residential sector and the possibilities
for flexible operation, using the heat storage options available.
Extensive model development is performed that significantly improves the possibilities for analys-
ing individual heat pumps and heat storages in an energy system context. Energy systems analyses
reveal that the heat pumps can even without flexible operation contribute significantly to facilitat-
ing larger wind power investments and reducing system costs, fuel consumption, and CO2 emissions.
When equipping the heat pumps with heat storages, only moderate additional benefits are achieved.
Hereof, the main benefit is that the need for investing in peak/reserve capacities can be reduced
through peak load shaving. It is more important to ensure flexible operation of electric vehicles than
of individual heat pumps, due to differences in the load profile.
ISBN 978-87-92706-32-4

Department of Management Engineering
Technical University of Denmark
Risø Campus
Frederiksborgvej 399 / P.O. Box 49
Building 130
DK- 4000 Roskilde
Denmark
Tel. +45 46 77 51 00
Fax +45 46 77 51 99
www.man.dtu.dk

Wind Power Integration With Heat Pumps, Heat Storages, and Electric Vehicles - Energy Systems Analysis and Modelling

Uploaded by

Copyright:

Available Formats

Wind Power Integration With Heat Pumps, Heat Storages, and Electric Vehicles - Energy Systems Analysis and Modelling

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Wind Power Integration With Heat Pumps, Heat Storages, and Electric Vehicles - Energy Systems Analysis and Modelling

Uploaded by

Copyright:

Available Formats

Downloaded from orbit.dtu.

dk on: Nov 09, 2017

Hedegaard, Karsten; Morthorst, Poul Erik; Münster, Marie; Detlefsen, Nina

Link back to DTU Orbit

PhD thesis 7.2013

DTU Management Engineering

Wind power integration with heat pumps, heat storages,

In partial fulfillment of the requirements for the PhD degree

Author: Karsten Hedegaard

Main supervisor: Professor Poul Erik Morthorst

DTU Management Engineering

Print: Schultz Grafisk A/S

Published September, 2013

5.3 Perspectives .................................................................................................................... 53

Begge modeller er blevet videreudviklet til at give en forbedret repræsentation af individuelle

Energisystemanalyserne viser, at med hensyn til at understøtte integration af vindkraft, er

Intelligent varmelagring i bygningskonstruktionen er identificeret samfundsøkonomisk rentabelt i

20-70 % af husene. Derimod er varmeakkumuleringstanke pga. deres højere

Yderligere analyser undersøger systemeffekterne af en gradvis storskala-implementering af

Konkurrerende fleksible energiteknologier, så som store varmepumper, elkedler og varmelagre i

Samlet set kan det konkluderes, at individuelle varmepumper, fleksible teknologier i

I would like to thank... for...

As illustrated in Figure 2, the Danish Transmission System Operator (TSO), Energinet.dk,

Danish electricity demand, 2030

1.3 Goals and research question

1.4 Contents of papers

Table 1. Overview of primary publications in the thesis.

Model Energy systems

1.5 Previous studies

1.5.1 Individual heat pumps and heat storages

Recently, a significant amount of studies focusing on intelligent operation of individual heat

1.5.2 Electric vehicles

1.5.3 Other fields

2 Technologies and power balancing

2.1 Demands for power balancing

Table 2. Current Danish electricity markets.

2.2 Individual heat pumps and heat storages

Heat pump Electric boiler

It is hardly realistic to expect households to engage in individual trading on the electricity

Temp. interval utilised in

Floor heating, Floor heating, new

new houses (constructed

Efficiency Investment cost

Indiv. heat Large steel Large water

2.4 Electric vehicles

A very ambitious large-scale implementation of electric vehicles corresponding to 53 % of the

2.5 Overview of the technologies' power balancing potentials

Intra-hour Intra-day Intra-week Seasonal

Large HPs, DH, High estimate

Large EBs, DH, High estimate

0 50 100 150 200 250 300 350 400 450

3.1 Choice of models

3.3 Model development in EnergyPLAN

Passive heat storage Storage in heat accumulation tank

3.5 Model development in Balmorel

Figure 12. Active heat storage in concrete floor heating systems.

CI, T I QIL QLO

QSH QSH CF, TF

Nov 1 - Dec 31 Mar 15 - May 15, Sep 15 - Dec 31