Artigo 08 A

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Applied Energy 235 (2019) 1637–1650

Contents lists available at ScienceDirect

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

Impacts of electric vehicles on the electricity generation portfolio – A T


Scandinavian-German case study
M. Taljegard , L. Göransson, M. Odenberger, F. Johnsson

Department of Space, Earth and Environment, Energy Technology, Chalmers University of Technology, 412 96 Gothenburg, Sweden

HIGHLIGHTS GRAPHICAL ABSTRACT

• The hourly impact of electric vehicles


(EVs) on the electricity system is ana-
lysed.
• The models include individual driving
data to optimise EV-electricity gen-
eration.
• The demand for EVs in Scandinavia-
Germany is met by wind power and
thermal power.
• EVs with vehicle-to-grid substantially
reduces the need for peak power ca-
pacity.
• ERS will increase the current net load
profile by up to 18 GW.

ARTICLE INFO ABSTRACT

Keywords: This study investigates how electrification of the Scandinavian and German road transportation sectors under a
Electric vehicle stringent CO2 cap will influence investments in new electricity generation capacity up to Year 2050 and the
Energy system modelling dispatch of the electricity generation portfolio in Year 2030. We apply a cost-minimisation investment model
Electric road systems and an electricity dispatch model of the Scandinavian and German electricity systems, assuming both optimised
Peak power
charging and a vehicle-to-grid (V2G) charging strategy for passenger electric vehicles (EVs). Different EV battery
Vehicle-to-grid
sizes and EV deployment levels are investigated in 11 different scenarios, whereby two of the scenarios include
Smart charging
also electric trucks and buses using electric road systems (ERS). The results of the modelling show that with a cap
on CO2 emissions, the additional electricity demand from an electrified road transport sector is met mainly by
increases in the outputs from wind power and thermal power plants, in the form of coal in combination with
carbon capture and storage. In Year 2030, wind power generation in Scandinavia and Germany increases by
7–30% depending on the EV scenario, as compared to a scenario without EVs, which corresponds to a few more
percentage points than the increased demand from EVs in absolute terms. Furthermore, a V2G charging strategy
for passenger EVs smoothens the net load curve and almost completely reduces the need for peak power capacity
in the Scandinavian-German electricity system. The value of investing in solar power is also reduced in all the EV
scenarios by 22–42%, as compared to a scenario without EVs. This is due to the fact that in Northern Europe
solar power competes with EVs for peak power supply. ERS for mainly trucks and buses will increase the current
load profile by up to 18 GW in the Scandinavian-German electricity system.


Corresponding author.
E-mail address: maria.taljegard@chalmers.se (M. Taljegard).

https://doi.org/10.1016/j.apenergy.2018.10.133
Received 7 March 2018; Received in revised form 27 October 2018; Accepted 30 October 2018
0306-2619/ © 2018 Elsevier Ltd. All rights reserved.
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

1. Introduction avoid increasing the late-afternoon peak in electricity demand, the


vehicle charging can instead be used for ‘valley filling’ of the net
Electrification of road transportation is commonly seen as a way to electricity load, thereby reducing the competition between wind power
reduce the impact on climate of the transport sector (e.g. the European and base load at low wind shares (i.e., up to 20%). The electricity
Commission [1], Fridstrøm and Alfsen [2] and Johansson [3]). Several system of Northern Europe was modelled by Hedegaard et al. [14]
governments are promoting zero-emission vehicles through different using a linear cost optimisation model without possibilities for power
policies, such as tax reductions and other economic benefits. Using an transmission between countries. They showed that controlled charging
optimistic deployment scenario for electric and hybrid vehicles in (i.e., delaying the charging time to avoid increasing the evening peaks
Norway, Fridstrøm and Alfsen [2] have estimated a penetration of 1 or using the V2G strategy) and discharging a small fraction of the EVs
million and 1.5 million electric vehicles (EVs) and plug-in hybrid facilitated a significant increase in wind power investments if the in-
electric vehicles (PHEVs) by Year 2030 and Year 2050, respectively. troduction of EVs was followed by economic support for renewable
This would correspond to vehicle shares of 34% in Year 2030 and 53% energy technologies. However, the results varied significantly from
in Year 2050. In a fossil-free scenario for Sweden, the EV fleet is as- country to country. Lund and Kempton [15] studied a power system
sumed to increase to a share of 22% in Year 2030 and to 60% in Year with large-scale wind power employment using a deterministic model
2050 [3]. Sweden recently established a target of a 70% reduction in that optimises the operation of a given energy system. They concluded
CO2 emissions by Year 2030 relative to Year 2010 [4]. Meeting this that EVs could facilitate the integration of more wind power into the
target, will require substantial deployment of EVs over the coming electricity system, assuming that charging is allocated to hours of low
decades. This will create an additional electric load and necessitate demand or that charging takes place during periods of excess electricity
efficient integration of EVs into the electricity system. Although pre- generation. Jochem et al. [17] analysed the CO2 emissions for con-
vious studies have shown that full deployment of passenger EVs will trolled and uncontrolled EV charging in Year 2030 using a cost-opti-
have only a limited effect on the yearly electricity demand for a given mising model of the German electricity system. They concluded that the
country (typically an increase of 5–6% in electricity consumption vehicle CO2 emissions for Year 2030 could range from zero emissions to
[3,5]), electrification may enhance peaks in the net load. Different about 110 g/km, and they recommended controlled charging as a way
charging alternatives [e.g., static charging or electric road systems to reduce the emissions from an electricity system that included EVs.
(ERS)] and charging strategies [e.g., charging directly upon home- Schill & Gerbaulet [18] modelled also the integration of EVs in the
coming, an optimisation of the charging time to minimise the correla- German electricity system by using a dispatch model with one year time
tion with peak load or vehicle-to-grid; V2G strategy] will result in resolution. They concluded that a controlled EV charging strategy can
different EV load profiles. These will have different effects on the in- smoothen the load curve and not only increases the utilization of re-
vestments in capacity and dispatch of the electricity system. In addition, newable energy, but also strongly increase the utilization of thermal
the expected expansion in the coming decades of variable renewable power plants in Year 2030 [18].
energy sources (vRES), e.g., solar and wind power, will force the elec- All the above-mentioned modelling studies investigated a single
tricity system to handle more variable electricity generation, for ex- country as a closed system, thereby focusing on regional or national
ample, through load management and storage [6]. electricity systems. Hadley and Tsvetkova [19] and Verzijlbergh et al.
Several studies have been carried out in recent years to determine [16] used a multi-regional dispatch model to study the impact of EVs on
the load profiles and impacts of passenger EVs on, for example, the the dispatch of the electricity system, including trading between re-
dispatch of generation technologies, CO2 emissions, and peak power gions. A multi-regional study captures the geographical smoothening
demand [7]. Veneri[7] concludes in a review of several studies that a effects, which have importance for investment in wind power [20,21],
controlled charging of EVs can increase the share of renewable elec- and the distribution of hydro power. Using a minimum-cost unit com-
tricity, when at the same time increase the load factor for base load mitment dispatch model, Verzijlbergh et al. [16] compared the impacts
power plants. This will also reduce the need for peak power. The studies of controlled passenger EV charging on cross-border transmission for
conducted by Grahn [8], Weiller [9], Darabi et al. [10], Vatne et al. different levels of penetration of vRES. They found that transmission
[11], and Stamati & Bauer [12] have used detailed driving patterns and lines and a controlled passenger EV charging strategy were additive in
simulation models to determine the load curve for passenger EVs. For terms of handling a high volume of vRES in the German electricity
example, using a model that includes stochastic individual driving be- system. However, the multi-regional modelling conducted by Hadley
haviours, Grahn [5,8] has concluded that with 50% adoption of pas- and Tsvetkova [19] and Verzijlbergh et al. [16] did not include the
senger PHEVs in Sweden and assuming that 5 kW of charging power impact on new capacity investments (for example investments in peak
from ERS are available for all types of trips, a 2-GW peak load from EVs power) from controlled charging of EVs. The results from the previous
in Sweden would occur at around 12 pm. The analyses made by Vatne studies mentioned above underline the importance of controlled char-
et al. [11] of large-scale adoption of passenger EVs in a Norwegian ging in avoiding any increase in the late-afternoon peak generation
region using a low-voltage distribution network model have shown that through large-scale introduction of passenger EVs. This raises the pos-
a V2G charging strategy for passenger EVs will not exert extra pressure sibility for EVs to handle more variable renewable energy (vRE) in a
on the grid during the peak hours of electricity consumption in Norway. system that employs V2G charging. Surveys of travelling habits are
Another group of studies has looked at the combined effect of in- commonly used in these types of modelling studies to define travelling
tegration of passenger EVs and vRES on the electricity system, applying patterns that can be used to estimate the aggregated impacts on the
electricity investment modelling and/or dispatch modelling to de- electricity grid of EVs. However, such surveys do not provide in-
termine the load curve for passenger EVs [13,14,15,16]. Such model- formation at the level of individual vehicle trips, which makes it diffi-
ling studies take into account other factors that may be important when cult to capture the potentials of optimised charging and V2G charging.
delineating the optimal charging of EVs. In some of these energy system The aim of the present study is to elucidate how massive deploy-
modelling studies (e.g., Göransson et al. [13] and Hedegaard et al. ment of EVs affect the combined Scandinavian (Norway, Sweden and
[14]), passenger EVs have been proposed as a way to facilitate the Denmark) and German electricity system while meeting stringent CO2
accommodation of more intermittent electricity generation in the constraints. The present study adds to the previous modelling studies
electricity system. For example, Göransson et al. [13] have shown by by: (i) investigate the impacts of EVs both on investment decisions re-
modelling western Denmark that PHEVs that are charged at home or/ lated to new power capacity and the dispatch of the power plants,
and in public places may increase the peak load if the charging period is considering an electricity system with the geographical scope of several
mainly at the home arrival time after office hours. However, they have inter-connected countries; (ii) apply the GPS information from 429
also shown that if the charging patterns are delayed by a few hours, to privately driven passenger vehicles to determine the input data on the

1638
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

Table 1
Total number of electric vehicles for Sweden, Norway, Denmark and Germany combined.
Passenger cars [thousands] Light trucksa [thousands] Busesa [thousands] Heavy trucksa [thousands]
(high/low deployment scenario) (high/low deployment scenario) (high/low deployment scenario) (high/low deployment scenario)

Year 2016 200/200 3/– 0/– 0/–


Year 2030 37,970/12,260 2180/– 75/– 490/–
Year 2050 76,050/44,260 4070/– 125/– 1000/–

a
Trucks and buses are only modelled for a high electric vehicle deployment scenario.

characteristics and distribution of the vehicle fleet; and (iii) incorporate EVs, while at the same time fulfilling a given hourly EV demand with
the electrification of several vehicle types (passenger cars, trucks and the number of EVs, the battery size, and the hourly EV demand being
buses) through two types of charging strategies (static charging and exogenously given. Thus, there is no optimisation of vehicle or battery
ERS). investments in the models. In ELIN, every decade until Year 2050 is
The models used in this study minimises the system cost for elec- modelled with 480 h per year (20 representative days with 1-hour time
tricity system over the investment period investigated, in this case up to resolution). The new load from EVs and the Optimisation + V2G
Year 2050. It also minimises the cost of electricity generation for a strategy (i.e., EV batteries providing storage) may affect investment
particular year (in this work for Year 2030), including cross-border decisions regarding electricity generation capacity in ELIN. In EPOD,
trading of electricity. The modelling results for an electricity system Year 2030 is modelled with 8760 h to determine the optimal dispatch of
with and without the inclusion of EVs are compared for different the power plants, as well as, the hourly charging and discharging of
electrification scenarios, two of which also include electrification of passenger EVs during a full year.
heavy vehicles assuming ERS. In addition, this study uses the GPS data
from 429 privately driven passenger vehicles to determine the different 2.1. Data and assumptions
daily driving patterns. This captures the spread in the individual driving
patterns in a better way than can be achieved using an aggregated fleet, To estimate the impact of EVs on the electricity system, data are
and thus better represents the limitations that apply to the ability of the needed for: (i) the total number of EVs; (ii) the total yearly driving
vehicles to distribute the charging over time. Two different charging distance using electricity; (iii) the fuel consumption per km; and (iv) the
strategies for passenger EVs are analysed: (i) an Optimisation strategy, in travelling patterns of persons driving EVs. The trucks and buses are
which the charging time of the EVs is optimised in order to minimise assumed to use ERS for all their vehicle kilometres in two of the in-
the electricity system cost of meeting the electricity demand; and (ii) an vestigated scenarios. In those two scenarios, it is also assumed that all
optimisation plus V2G strategy that also includes discharging of pas- trips for passenger EVs that are longer than the battery driving range
senger EVs to the grid (Optimisation + V2G). The remaining part of the use ERS as a range extender. The yearly aggregated electricity demand
paper is organized as follows: Section 2 described the methodology and for the EV fleet (EiEV
, y ) in region i and for year y is given by the equation:
modelling used, Section 3 presents the results from the modelling and
Section 4 discusses the results. Section 5 summarizes the conclusions EiEV
, y = Ni, y × vkm × FC i I, y Y (1)
drawn from the work.
where Ni, y is the number of EVs in region i for year y, vkm is the number
of yearly kilometres per vehicle using electricity, and FC is the elec-
2. Methodology tricity consumption per kilometre.

Integration of EVs in the combined Scandinavian and German 2.1.1. Number of electric vehicles and vehicle specifications
electricity system is analysed using an Electricity Systems Investment Table 1 shows the assumptions made regarding the total number of
model (ELIN) and an Electricity System Dispatch model (EPOD), ori- passenger cars, electric trucks and buses in Years 2016, 2030 and 2050.
ginally constructed by Odenberger et al. [22] and Unger et al. [23] and The total numbers of passenger vehicles in Norway, Sweden, Denmark
further refined by Göransson et al. [24] and Nyholm et al. [25]. These and Germany in Year 2016 were 2.6 [26], 4.6 [27], 2.4 [28], and 46
two models have previously been used to study the transformation of [28] million vehicles, respectively. It is further assumed that there will
the European electricity system to meet European policy targets on CO2 be an increase of approximately 13% and 37% in the numbers of pas-
emissions. The ELIN model is a cost-minimisation model that is de- senger vehicles by Years 2030 and 2050, respectively, based on an es-
signed to analyse transformation of the European electricity system by timate made for Sweden by Johansson et al. [3]. In the present study, a
making investment decisions in new electricity generation capacity high EV deployment scenario assumes an EV share of the total fleet of
until Year 2050, while meeting assumptions related to key scenario 60% by Year 2030 and 100% by Year 2050. For all four countries, this
parameters, e.g. a CO2 emissions trajectory. The EPOD model takes the entails 38 million and 76 million passenger EVs by Years 2030 and
description of the power system, fuel and CO2 prices, and transmission 2050, respectively, as shown in Table 1. In a sensitivity analysis, a low
lines as obtained from ELIN for a certain year (in the present work, Year deployment scenario has been investigated with an EV share of 20% by
2030) and carries out optimisation to find the least-cost hourly dispatch Year 2030 and 60% by Year 2050. Trucks are categorised as light or
of the system. The investment model ELIN is run until 2050, since the heavy. The numbers of light and heavy trucks and buses in each country
model has perfect foresight and might impact the results for Year 2030. in Year 2016 are taken from Eurostat [28], and the total numbers of
The geographical scope of the modelling is Sweden, Norway, Denmark light/heavy trucks and buses are assumed to increase in all regions by
and Germany. The results are presented in aggregated form for the Year 2050, according to the scenarios developed for Sweden by Jo-
investigated countries. hansson et al. [3]. Furthermore, the same rates of electrification as for
In the present study, the two models are expanded to include an passenger vehicles are assumed for trucks and buses. The number of
electrified road transport sector in the form of static charging of pas- EVs per country is exogenously given in the model.
senger EVs and/or ERS, where the number of EVs is exogenously given The rates of fuel consumption are assumed to be 0.16, 0.33, 1.19
as new load. Thus, a new demand for electric transportation has been and 2.06 kWh per km for passenger cars, light trucks, buses, and heavy
added to both the investment model and the dispatch model. The two trucks, respectively, and these values exclude the electric motor effi-
models optimise the times of charging and discharging of the passenger ciency and other electrical and mechanical power transmission losses

1639
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

[29]. The yearly driving ranges, based on national statistics for Sweden,
are 14,000, 41,000, and 57,000 km per year for light trucks, buses, and
heavy trucks, respectively, and are assumed to be the same for all the
countries [27]. The power required for static charging of passenger EVs
is assumed to be 3.7 kW. The charging and discharging efficiency of the
battery is assumed to be 95% (i.e., a round-trip efficiency of 90%) for
both static charging and ERS in the models [29].

2.1.2. Travelling patterns


Individual travelling patterns determine the time periods during
which the passenger EVs are parked and available for charging/dis-
charging back to the grid. In the present study, the assumptions made
regarding passenger travelling patterns are based on GPS data from 429
randomly chosen gasoline and diesel vehicles that completed a total of
107,910 trips between Years 2010 and 2012 in the Västra Götaland
region (for more details of the dataset, see Karlsson and Kullingsjö [30]
and Kullingsjö and Karlsson [31]). In total, 12,400 week-days and
5800 days during weekends and holidays were included in the mea-
surements. The average yearly driving distance per vehicle for the
Fig. 2. Load profile for trucks and buses for an average day, where Hour 1
measured fleet was 17,400 km and the average two-way daily com-
correspond to 1 am.
muting distance was 58 km [30]. This is higher than the Swedish na-
tional average of 12,400 km per year, possibly due to a larger share of
the measured vehicle fleet being run on diesel, as compared to the patterns for all days of the year. The drawback of the applied modelling
Swedish national average. The passenger EVs are assumed to be methodology in using representative driving days is that it does not
available for charging at all stops that last at least 1 h. Fig. 1 shows the allow the taking into account of the electricity stored in the vehicle
share of the aggregated passenger fleet that is available for charging batteries from one day to another. Thus even though there is good
(i.e., vehicles that are parked for more than 1 h). In this study, the overall representation of the demand profiles from the representative
driving patterns from 200 vehicle-week-days out of 12,400 vehicle- days, there is no information on the inter-linkages between such re-
week-days were used for optimisation of the model, so that they to- presentative days.
gether match the profile and driving distance of the aggregated fleet Trucks and buses are assumed to be electrified using ERS. The load
shown in Fig. 1. This was performed to capture in a better way the profile for an average day for buses and trucks, as shown in Fig. 2, is
spread of the individual driving patterns than can be achieved using an taken from the analysis of traffic flow data presented by Taljegard et al.
aggregated fleet, and thus better represent the limitations related to the [29]. The demand from ERS is then the yearly load profile, i.e., the
ability of the vehicles to distribute the charging over time. hourly share of the yearly demand, multiplied by the total amount of
The clustering method K-means was used to choose the 200 vehicle- yearly driving for the fleet.
days that are representative in terms of distance and driving profile,
whereby each day is assigned a weighting factor. Running the model 2.2. Investment and dispatch model structure
with more than 200 daily driving profiles was computationally onerous
and it was found that 200 vehicle-days were needed to represent ac- The Scandinavian-German electricity generation system analysed in
curately the aggregated driving profile. These 200 vehicle-week-day the present work is as previously mentioned, derived from two opti-
profiles were used repetitively in the model to represent the driving misation models: the ELIN investment model and the EPOD dispatch
model. The investment model minimises the total investment and
running cost, while the dispatch model minimises the total running cost
of the electricity generation system. Eqs. (2) and (3) describe the ob-
jective functions of the investment model and the dispatch model, re-
spectively. The demand in ELIN and EPOD (i.e. Year 2030 for the EPOD
model) must then be satisfied for each year, time-step and region,
which implies the constraint seen in Eq. (4). A full mathematical de-
scription of the EPOD dispatch model is described in Goop et al. [32].
run
MINCtot = i I p P y Y t T (ci, p, y, t · gi, p, y, t )

inv fix
+ i I p P y Y (Ci, p, y i i, p, y + C i, p, y x i, p, y ) (2)

MINCtot = i I
run
p P t T (ci, p, t · gi, p, t + cicycl
, p , t )+ i I
fix
p P Ci, p x i, p (3)

p P gi, p, y, t + j I , j i qy, t , i, j Di, y, t i I, y Y, t T (4)


where

Ctot is the total system cost


I is the set of all regions
P is the set of all technology aggregates
Y is the set of all years in the investment period
Fig. 1. Share of the aggregated fleet that is parked, i.e., available for charging/ T is the set of all time-steps (differ between the models)
discharging to the grid, and share of the aggregated fleet that is being driven on cirun
, p, y , t is the running cost of region i, with technology aggregate p in year y at time-
an average day, where Hour 1 corresponds to 1 am. step t

1640
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

gi, p,y,t is the electricity generation in region i, technology aggregate p, for year y at resolution for 20 representative days per year until Year 2050 are in-
time-step t
cluded in the investment modelling of this work. The electricity de-
cicycl is the cycling costs (summed start-up costs and part-load costs) in region i
, p, t
with technology aggregate p in year y at time-step t mand and the distribution of the numbers of EVs per country into re-
Ciinv is the investment costs of technology aggregate p in region i and year y gions within each country are given according to GDP [37].
, p, y
is the fixed operational and maintenance costs of technology aggregate p in The present study takes the results from the investment model (i.e.,
Ci,fix
p, y
region i and year y the description of the power system, fuel and CO2 prices, and trans-
ii,p, y is the investment in region i and technology aggregate p in year y mission lines) for Year 2030 and performs optimisation in the dispatch
xi, p,y is the existing capacity in region i and technology aggregate p in year y model (EPOD), to uncover the least-cost hourly dispatch of the system
Di,y,t is the demand for electricity in region i and year y at time-step t for Year 2030. The EPOD model analyses the dispatch of power tech-
qy, t, i, j is the flow of power, positive or negative, from region j to region i in year y
nologies, demand for peak power (i.e., use of gas turbines to cover only
at time-step t
peak demand), and the generation patterns of the power technologies in
Due to computational limitations, the geographical scope of the mod- the system. Furthermore, the model provides the: (i) annual electricity
elling in this work is limited to Scandinavia and Germany (although production by fuel and technology; (ii) total yearly system cost; (iii)
ELIN and EPOD are normally run for the EU-27 countries plus Norway charging and discharging profiles of the EVs; and (iv) hourly marginal
and Switzerland). Scandinavia and Germany are sub-divided into 14 cost for electricity. The EPOD model also takes into consideration the
regions based on the current bottlenecks in transmission capacity. linkages between the operation of a combined heat and power plant
In this work, the ELIN model includes an annual constraint on CO2 and the district heating load, as well as the cycling properties of power
emissions, with a linear reduction in emissions between Year 2010 and plants (minimum load requirements, start and stop decisions). The
Year 2050. The levels of emissions from the modelled electricity system electricity demand, as well as wind power and solar power fluctuations
correspond to reductions of 60% and 93% in Year 2030 and Year 2050, are included on an hourly basis in EPOD. The hourly data are based on
respectively, as compared to Year 1990, which is in line with the data from the European Network of Transmission System Operators for
emissions target set by the European Union [33]. A scenario with a Electricity ENTSO-E (load) [38], and MERRA metrological data (solar
stricter CO2 cap designed to achieve a 99% reduction in emissions by [39,40] and wind [41]). EV transportation demand and the battery
Year 2050 is also investigated in a sensitivity analysis. The version of interaction with the electricity system are all parts of the model version
the investment model applied in this study has a time horizon up to applied in this work.
Year 2050, with investment decisions made every decade and an intra-
annual time resolution of 20 representative days and an hourly time
resolution chosen based on Nahmmacher’s method [34], with some 2.3. Electric vehicle implementation
minor modifications. This approach to represent time was further
evaluated by Reichenberg [35]. The data used for choosing the re- The inclusion of 200 individual daily passenger EV driving profiles
presentative days with Nahmmacher’s method are derived from the and ERS in the ELIN and EPOD models requires modification of the
wind and solar power profiles, as well as the load profile. models used in previous studies. Fig. 3 illustrates the structure of the
The development of the electricity supply system over time is based models and their main input and output parameters. The EVs are im-
on phasing out current power plants based on projected technical life- plemented using the same equations in the investment model (ELIN)
times and then making investments in new generation capacity to meet and the dispatch model (EPOD), although in the investment model
the electricity demand, subject to a number of constraints. Information every tenth years (y) until Year 2050 are modelled with 480 h per year
on the current European electricity supply system in the model is from (20 representative days with 1-hour time resolution). In contrast, the
the Chalmers Energy Infrastructure databases, which have almost full EPOD calculations are limited to Year 2030 (i.e., y = 2030) with 8760 h
coverage of power plants that have a rated net capacity > 10 MW (for (1-hour time resolution). A new demand for passenger EVs and a de-
more information on the database, see Kjärstad and Johansson [36]). mand for other types of vehicles (i.e., trucks and buses) using electricity
Possible generation technologies for the model to invest in are on- and through ERS have been added to the balancing constraint, which on an
off-shore wind power, solar power, heat pumps, and different types of hourly time-scale balances electricity generation and demand (see Eq.
thermal power plants (e.g., condensing, with carbon capture, and
combined heat and power), which can be run on coal, lignite, natural
gas, oil, biomass or waste. For the scenario considered, new investments
in nuclear and hydro power are not allowed, due to both political and
nature preservation reasons. The investment costs, operational and
maintenance (O&M) costs and technical life-times for the different
technological and fuel options are given in Appendix A. Carbon capture
and storage (CCS) in combination with fossil sources has in the model a
capture efficiency of about 90%. The development over time of the
electricity demand, excluding EVs, for different countries in all the in-
vestigated scenarios is based on the Energy Roadmap scenario titled
“High Energy Efficiency” [33] (see Table A2 in Appendix A). A mod-
erate increase (approximately 2.5% in total by Year 2050 compared to
Year 2015) in electricity demand is assumed. The national electricity
demand is divided into regional demand based on the statistics for gross
domestic product (GDP) obtained from Eurostat [37]. The transmission
network between regions is modelled according to the current expan-
sion plans with their specific capacities and limits. There is also the
possibility to invest in additional transmission capacity in the invest-
ment model. The energy demand (excluding EVs), possible generation
technologies for the model to invest in, and the modelling of the
transmissions lines are the same in all the investigated scenarios. An
hourly EV demand and the balancing of the EV batteries for the 200 Fig. 3. A schematic of the main modelling elements applied in this work, in-
individual daily passenger EV driving profiles on an hourly time cluding the input and output parameters used in this study.

1641
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

Table 2
Charging strategies and electrification scenarios investigated in this work.
Scenario Share of EVs by Transport modes CO2 emissions reduction in Charging strategy Passenger EV Passenger EV battery Share of the road
Year 2030/ included the electricity system by battery size [kWh] capacity by Year 2030/ distance on electricity
2050 Year 2050 2050 [TWh]

S1 0%/0% – 93% – – – –
S2 20%/60% Passenger EV 93% Optimisation 30 0.4/1.3 0.93
S3 20%/60% Passenger EV 93% Optimisation + V2G 30 0.4/1.3 0.93
S4 20%/60% Passenger EV 93% Optimisation + V2G 15 0.2/0.7 0.85
S5 20%/60% Passenger EV 93% Optimisation + V2G 85 1.0/3.8 0.98
S6 60%/100% Passenger EV 93% Optimisation 30 1.1/2.3 0.93
S7 60%/100% Passenger EV 93% Optimisation + V2G 30 1.1/2.3 0.93
S8 60%/100% Passenger EV 93% Optimisation + V2G 15 0.6/1.1 0.85
S9 60%/100% Passenger EV 93% Optimisation + V2G 85 3.2/6.5 0.98
S10 60%/100% Passenger EV 99% Optimisation + V2G 30 1.1/2.3 0.93
S11 60%/100% Passenger EV, 93% Optimisation + V2G 30 1.1/2.3 1.0
trucks and buses
S12 60%/100% Passenger EV, 93% Optimisation + V2G 15 0.6/1.1 1.0
trucks and buses

(5)). Eqs. (6)–(10) are added to the models and are implemented with EiDPEV
, dp, y, t
is the passenger EV discharging to the wheels for driving profile dp in
region i, for year y at time-step t
the aim of optimising the time of charging and discharging of the
BSi, dp, y is the maximum storage capacity of the EV batteries for driving profile dp
passenger EV, while at the same time fulfilling a given hourly passenger in region i and for year y
EV demand assuming 200 individual daily passenger EV driving pro-
files. There are constraints on: (i) the maximum amount of charging at Eqs. (6) and (7) limits the amount of electricity that can be charged and
each time-step (Eq. (6)); (ii) the maximum amount of discharging to the discharged each hour depending on the number of passenger EVs that
grid at each time-step (Eq. (7)); (iii) the amount of EVs connected to the are being parked for more than one hour. The maximum storage ca-
grid available for charging and discharging to the grid (Eq. (8)); (iv) the pacity of the batteries depends on both the number of passenger EVs
balancing of the charging and discharging of the EV battery (Eq. (9a) and the capacity of each passenger EV battery. In both the dispatch
and (9b)); and (v) the maximum EV battery storage capacity (Eq. (10)). model and the investment model, the balancing equations of the battery
All variables in the model can be zero or positive values. (i.e., Eqs. (9a) and (9b)) is executed only on a daily basis and not be-
tween days. This means that the storage level of the battery at hour
p P gi, p, y, t + j I , j i qy, t , i, j Di, y, t + CPEV
dp DP (Ei, dp, y, t EiDgrid ERS
, dp, y , t · n )+Ei, y , t t + 1 depends on the storage level and the charging and discharging of
i I, y Y, t T (5) the battery at hour t, except for the last hour of the day (see Eq. (9b)).

EiCPEV NCi, dp, y, t ·CP i I , dp DP , y Y, t T (6)


, dp, y, t
2.4. Charging strategies and electrification scenarios
EiDgrid BSi, dp, y i I , dp DP , y Y, t T (7)
, dp, y, t
Two charging strategies for passenger EVs in the electricity system
NCi, dp, y, t FAdp, t ·Ni, dp, y i I , dp DP , y Y, t T (8) are investigated, both of which use a form of controlled charging:

SLiPEV
, dp, y, t + 1 SLiPEV CPEV
, dp, y, t + Ei, dp, y, t ·n EiDGrid
, dp, y, t EiDPEV
, dp, y, t i I , dp • Optimisation strategy: Charging of the electric vehicles is part of the
electricity system optimisation and charging is allocated in time
DP , y Y, t T (9a)
such that the total cost of the electricity system is minimised, while
SLiPEV SLiPEV CPEV
EiDGrid EiDPEV at the same time it fulfils a given hourly EV demand. If the charging
, dp, y, tt 23 , dp, y, tt + Ei, dp, y, tt ·n , dp, y, tt , dp, y, tt i I , dp
strategy is Optimisation, the variable for discharging of the EV bat-
DP , y Y , tt TT (9b) tery back to the electricity grid is set to zero.
SLiPEV BSi, dp, y i I , dp DP , y Y, t T (10)
• Optimisation + V2G strategy: Compared to the Optimisation strategy,
, dp, y, t
the Optimisation + VG2 strategy has in addition the possibility to
where discharge the passenger EVs back to the grid based on the most
advantageous case from the electricity system point-of-view, while
at the same time fulfilling a given hourly EV demand. Therefore, in
TT is the set consisting of only the last hour of each day the Optimisation + V2G strategy, the passenger EVs become an ac-
DP is the set of individual daily passenger EV driving profiles
tive player in grid operation by providing capacity- and energy-
EiCPEV is the passenger EV charging for driving profile dp in region i, for year y
, dp, y, t
at time-step t balancing services on an hourly basis. The flexibility of the charging
EiDgrid is the passenger EV discharging to the grid for driving profile dp in region strategy is limited by the number of passenger EVs connected to the
, dp, y, t
i, for year y at time-step t electricity grid at each time-step, the battery storage capacity, and
is the electricity demand for the electric road system in region i, for year y
EiERS
, y, t the maximum charging power (see Eqs. (6)–(10)).
at time-step t
n is the charging and discharging efficiency of the EV battery
NCi, dp, y, t is the number of passenger EV that are connected to the grid for driving Table 2 shows the twelve scenarios (S1–S12) investigated, whereby
profile dp in region i, for year y at time-step t each scenario consists of a combination of battery size, charging
CP is the charging power strategy, EV deployment level, CO2 emission reduction level by Year
FAdp, t is either 1 or 0 depending if the vehicles belonging to driving profile dp
2050, and transport modes. Scenario 1 (S1) is a reference scenario
are connected or not connected to the grid at time-step t
Ni, dp, y is the number of passenger EVs belonging to driving profile dp in region i, without EV. In Scenarios S2–S10, only the electrification of passenger
for year y EVs with static charging is investigated, assuming different battery
SLiPEV
, dp, y, t
is the storage level of the passenger EV battery for driving profile dp in capacities and passenger EV deployment levels (high deployment sce-
region i, for year y at time-step
nario or low deployment scenario, as described in Table 1). Three dif-
ferent battery sizes are investigated for the passenger EVs (15 kWh,

1642
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

30 kWh, and 85 kWh per battery), enabling a usable driving range of

Wind power generation [share of


approximately 90 km, 190 km, and 530 km, respectively. The distance

total production in Year 2030]


that can be travelled solely on electricity depends on the daily driving
profile (in this study, 200 profiles). In the model, the distance that can
be travelled solely on electricity is maximised. The remainder of the
distance (i.e., not covered by electricity) is in Scenarios S2–S10 as-
sumed to be covered by, for example, a conventional engine, assuming
that the vehicle is a PHEV. A shift of technology mode has not been
included in the modelling of the electricity system in these scenarios. In

0.40

0.42
0.43
0.42
0.43

0.42
0.44
0.43
0.44

0.43

0.44
0.44
Scenarios S6–S12, a high deployment of EVs is assumed with the
number of EVs presented in Table 1. In S10, the CO2 emissions level by
Year 2050 is restricted to 99% (instead of 93% as in all the other sce-

period 2020–2050 [GW]


New inter-connectors in
narios). In S11 and S12, full electrification of the road transport sector
is assumed, since in addition to static charging of passenger EVs, trucks
and buses use ERS. ERS is also used in S11 and S12 as a range extender
for passenger EVs for those trips that can’t be completed using only
electricity, due to the battery size and driving pattern. For trucks and
buses, ERS is seen as the main way of using electricity. As shown in

10.7

10.2
9.7

8.8
8.8
9.2
8.8

9.0
9.4
8.9

9.1
9.4
Table 2, the aggregated passenger battery energy capacity for the four
countries investigated ranges from 0.2 TWh to 6.5 TWh, depending on

Wind power curtailed


the scenario and year.

in Year 2030 [%]


Fig. 4 shows the yearly electricity demands exogenously given in the
model for EVs for Years 2030 and 2050, aggregated for Denmark,
Sweden, Norway and Germany for Scenarios S2–S12. As shown in
Fig. 4, the electricity demand varies within the ranges of 30–170 TWh

2.1

1.7
2.0
1.9
1.9

1.7
1.9
1.9
2.1

1.1

1.6
1.9
by Year 2030 and 110–350 TWh by Year 2050. The EVs will in these Summary of the results with respect to main effects of including EVs in the electricity system models for Scandinavia and Germany.
scenarios increase the total electricity demand by 3–19% by Year 2030 Total investments in period
and 12–39% by Year 2050. 2020–2050 [GW]

3. Results

3.1. System impacts of integrating EVs


335

332
319
317
319

344
338
335
341

383

379
383
All the results are presented in aggregated form for Scandinavia and
Germany. Table 3 summarises the main modelling results for Scenarios
2030/2050 [€/tonne]

S1–S12 in relation to the system cost, new investments in capacity and


Cost of CO2 in Year

transmission lines until Year 2050, curtailment, and the generation of


wind power in Year 2030. The total CO2 emissions from the electricity
system are approximately the same for all the investigated scenarios,
22/112
20/31

20/35
19/34
19/34
19/34

21/40
21/40
21/39
21/40

26/46
26/46
since the emissions are capped. Electrification of the transport sector
will, of course, also reduce the tailpipe emissions from the vehicles. To
keep the CO2 emissions in the electricity system below the CO2 cap as
[billion €] and ([€/MWh])

the total electricity demand increases with the charging of EVs, a higher
System cost from EPOD

shadow price for CO2 is reached by Year 2050 (Table 3). An increase in
26 (28.7)

(28.0)
(27.2)
(27.2)
(27.2)

(29.2)
(27.8)
(27.9)
(27.8)

29 (28.7)

32 (29.9)
32 (29.6)
26
25
25
25

29
28
28
28
[billion €] and ([€/MWh])
System cost from ELIN

65 (17.1)

(17.0)
(16.9)
(16.9)
(16.9)

(17.4)
(17.3)
(17.3)
(17.3)

74 (17.7)

79 (17.5)
79 (17.5)
68
67
67
67

73
72
71
73

S10 (Opt+V2G-30 KWh)

S11 (Opt+V2G-30 KWh)


S12 (Opt+V2G-15 KWh)
S3 (Opt+V2G-30 KWh)
S4 (Opt+V2G-15 KWh)
S5 (Opt+V2G-85 KWh)

S7 (Opt+V2G-30 KWh)
S8 (Opt+V2G-15 KWh)
S9 (Opt+V2G-85 KWh)
99% reduction in 2050

ERS for trucks/buses


High EV deployment
Low EV deployment
S2 (Opt.-30 KWh)

S6 (Opt.-30 KWh)
S1 (without EV)

Fig. 4. Estimated annual electricity demands for electric vehicles (EV) in


Table 3

Scandinavia and Germany combined for Year 2030 and Year 2050 for Scenarios
S2 to S12.

1643
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

the CO2 shadow price makes investments in more expensive electricity electricity generation in Year 2030 for the same scenario.
generation technologies profitable. Fig. 6a shows that hydro power, fossil (gas and coal) power and
The cost per generated unit of electricity will in the EV scenarios be wind power dominate the annual electricity generation in Year 2030 in
approximately the same as in the scenario without EV in both models, Scandinavia and Germany. The additional electricity demand from EVs
as shown in Table 3. A small reduction (up to 1%) in investment, fuel is mainly covered by increased production from wind power and flex-
and fixed O&M costs can be noticed in the investment model for the ible complements to wind power, i.e. natural gas and biogas-based
Scenarios S3–S5 compared to the scenario without EVs. The running generation (Fig. 6b). The share of generation of wind power in Year
cost per unit of electricity is reduced with 3–5% in the dispatch model 2030, compared to the total generation in Year 2030, is 40% in S1
EPOD for the Scenarios S2–S9 compared to the scenario without EVs. (without EV), and 42–44% in S2–S12, as shown in Table 3. Therefore,
The main cost reductions in EPOD are in the cycling costs (i.e., start-up EV batteries facilitate only to a minor degree the integration of wind
costs and part-load costs) and the variable and fixed O&M cost. In the power, given that the batteries can only provide diurnal storage in the
Scenarios S11–S12 which includes also ERS, the cost per unit of elec- model, whereas wind power is characterised by fluctuations in gen-
tricity will instead increase with 3–4% compared to the scenario eration on longer time-scales (i.e., several days).
without EVs as seen in Table 3. In the longer time perspective, i.e. to Year 2050, the modelling re-
The total investment in new capacity for the period 2020–2050 in sults give – although not shown in the figures, that the additional
Scandinavia and Germany (combined) is 335 GW in the scenario electricity demand from EVs is mainly covered by increased production
without EVs (S1). The investments in capacity can, when the from wind power and some thermal power plants with CCS. In Fig. 5b,
Optimisation + V2G strategy is employed, be reduced in these scenarios it can be seen that the presence of EVs results in greater investments up
through low deployment of passenger EVs (S3–S5), by approximately to 2050 in wind power capacity with the addition of 7–30 GW (i.e., an
5% compared to the scenario without EV. The main reason for the re- increase of 4–17%) for EV Scenarios S2–S12. Towards the end of the
duction in capacity investments in the Optimisation + V2G scenario, time period investigated, very low CO2 emission levels are allowed,
despite the higher demand for electricity, is the possibility for passenger thus inhibiting natural gas to act as complement to wind power. Lack of
EVs to reduce the need for peak power capacities in the Scandinavian- low cost flexible generation (e.g., the price of biogas has increased by
German electricity system. In addition, the Optimisation charging Year 2050) gives thermal base load generation in the form of coal with
strategy (S2 and S6) reduces the need for peak power demand, albeit to CCS a competitive advantage relative wind power. At the same time,
a lesser extent, since only an optimised charging is taking place and the sites with most favourable wind conditions in Germany have al-
there is no possibility for the discharge of electricity back to the grid. In ready been deployed.
S6–S12, which entail extensive deployment of EVs, the investments in The flexible electricity demand of EVs stimulates a shift from peak-
new capacity increase by up to 14% to meet the new demand from EVs, load generation to base-load generation as seen in Fig. 5b. In the sce-
as compared to the scenario without EVs (Table 3). However, the per- narios with EVs, the thermal plants and nuclear power plants will in-
cent increase in investments is still lower than the percent increase in crease the full load hours and be cycled to a lesser extent. Therefore,
demand when integrating EVs. there will be fewer emissions from the use and cycling of fossil fuel-
An excess of electricity generation under high-wind condition may fired power plants in the scenarios with EV. This will create, within the
lead to the curtailment of wind power certain hours if there is no confines of the CO2 cap, room for more coal with CCS. In S10, in which
possibility for storage or flexible use of electricity. The modelling re- the CO2 emissions are set to almost zero by Year 2050, higher levels of
sults show that flexibility can be provided by passenger EVs in the form biomass (both biomass steam and biogas CCGT plants) together with
of so-called “valley filling” in all the scenarios, with the EVs being CCS lignite co-fired with biomass are used to cover some of the in-
charged when there is an excess of electricity in the system. In all the creased demand from the EVs.
EV scenarios, there is a minor reduction in the share of the wind power In all the EV scenarios that include Optimisation + V2G (i.e., S3–S5
that is curtailed (Table 3), as compared to the case without EV (S1). In and S7–S12), capacity investments in peak power capacity for Years
the investigated region, regulation of generation from reservoir hydro 2020–2050 will in total decrease by 13–15 GW, as compared to the
power and geographical smoothening through trade between regions scenario without EVs (S1) (Fig. 5b), which corresponds to a reduction of
can also help to reduce curtailment in all 12 scenarios. close to 100%. Starting up thermal plants to cover the demand gen-
Table 3 also gives the total investments in new inter-connector ca- erates start-up and part-load costs and emissions, which can be reduced
pacity, of which a major fraction is the connections between Denmark or avoided in all the scenarios that include Optimisation + V2G of pas-
and Germany. With the Optimisation + V2G charging strategy, the in- senger EVs owing to a smoothening of the net load curve. Therefore,
vestment in transmission lines decreases in some of the passenger EV EVs support the integration of more wind power to a small extent, while
electrification scenarios (S2–S5, S7–S9 and S11–S12), as compared to they also support the integration of more thermal power plants by in-
the scenario without EV. The decrease in investment in transmission creasing the full-load hours for these technologies. For the passenger EV
lines is attributed to the possibility to discharge the EVs to the grid so as scenarios with the Optimisation strategy without the possibility to dis-
to smoothen the load curve and cover the peak power demand, rather charge the EV batteries to the grid (S2 and S6), the investments in peak
than using trade during those specific hours. The total amount of power for the period 2020–2050 are reduced by 3 GW (21%) and 9 GW
electricity that is traded between the regions increases by up to 10% in (59%), respectively, compared to the scenario without EVs. Investments
the scenarios with EVs (i.e., S2–S12), thereby increasing the use of the in solar power are taken place after Year 2030 in mainly Germany. The
transmission lines compared to the scenario without EV. value of investing in solar power, as compared to wind power, is re-
duced in all the EV scenarios for Year 2020–2050, by 12–20 GW (i.e.,
3.2. New investments in capacity and electricity generation 22–42%). The reduction in solar power in Scandinavia and Germany
could be due to the fact that in Northern Europe, which has poor
Fig. 5 shows the sum of investments in new capacity for the time conditions for solar power, solar power mainly contributes to reducing
period 2020–2050 in the absence of EVs (Fig. 5a), and the differences in the day-time peak load. Under these conditions and with no cost at-
new investments between an electricity system without EVs (S1) and tached to using EV batteries for discharging to the grid, solar power
Scenarios S2–S12 with EVs (Fig. 5b) for Scandinavia and Germany competes with EVs for peak power supply.
combined. The largest differences between the scenarios with and
without EVs are in the capacity investments in peak power, solar power,
thermal power, and wind power, as shown in Fig. 5. Fig. 6 shows the

1644
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

Fig. 5. New capacity investments in Scandinavia and Germany for the period 2020–2050 S1 without EV (a) and the differences in GW capacity investments between
an electricity system without EV (S1) and EV Scenarios S2–S12 (b). CCS = carbon capture and storage; BW-CSS = lignite co-fired with biomass carbon captured and
storage; Opt = Optimisation charging strategy; Opt + V2G = optimisation and vehicle-to-grid charging strategy; ERS = electric road systems.

Fig. 6. Total annual electricity generation in Scandinavia and Germany in Year 2030 for the scenarios without EV (S1) and the differences in electricity generation
between an electricity system without EV (S1) and EV Scenarios S2–S12 (b). Opt = Optimisation charging strategy; Opt + V2G = optimisation and vehicle-to-grid
charging strategy; ERS = electric road systems.

3.3. Power plant dispatch during 1 week. However, in the Optimisation + V2G case, the peak
power capacity can be reduced by discharging the EV batteries to the
Fig. 7 compares the hourly dispatch of the power plants in Scenario grid (Fig. 7b). Fig. 7 also shows that the solar PV installed by Year 2030
S1 without EVs (Fig. 7a) with the S7-Optimisation + V2G scenario mainly acts as peak power in the Scandinavian-German electricity
(Fig. 7b) during a 1 week period in October of Year 2030. Fig. 7b in- system.
cludes also the hourly load for the charging and discharging to the grid
of the EVs in the same scenario (S7). In the Optimisation + V2G sce- 3.4. Charging and discharging of electric vehicles
nario, the usage of thermal power is increased compared to the case
without EVs, and it is used to a greater extent as traditional base-load Fig. 8 displays the charging and discharging to the grid profiles from
generation in the electricity system. With EVs, nuclear-powered, as well the EV fleet for an average day, as obtained from the modelling in
as fossil fuel- and bio-powered generation plants are cycled less fre- Scenario S7 with Optimisation + V2G, as well as the annual maximum
quently with EVs than without EVs as seen in Fig. 7. In both of the spread of the levels of charging and discharging to the grid. It is clear
scenarios depicted in Fig. 7, the peak power is used for several hours that the major share of the charging in Year 2030 occurs during night-

1645
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

Fig. 7. Model results for the generation of electricity and for charging and discharging from the EVs back to the electricity grid during a period of 1 week in October
in Year 2030 for S1 without EVs (a) and for S7 with Optimisation + V2G (b). CCS = carbon capture and storage.

time when the load from other sectors is low. In contrast, discharging to sector, both if burning fossil fuels in combustion vehicle engines or CO2
the electricity grid occurs mainly during the peak hours in the morning emissions from the electricity system when charging EVs. The annual
and afternoon (Fig. 8). The times of charging and discharging of pas- level of discharging to the grid is in the range of 17–45 TWh in Year
senger EVs in Year 2030 are heavily influenced by the load curve from 2030. This level is low compared to the total generation in Scandinavia
other sectors. and Germany of approximately 900 TWh per year, although it provides
Table 4 shows, for Year 2030, the total amount of discharging from flexibility to the system, which is important for reducing the levels of
the vehicle to the electricity grid (i.e., Optimisation + V2G operation) peak power, as shown in Fig. 7. This service can to a large extent be
and the total amount of CO2 emissions in Year 2030 from the transport provided by the EV:s already in the scenarios with small batteries

Fig. 8. Charging (a) and discharging back to the electricity grid (b) from the EV fleet for a yearly average day, and the spread between the maximum and minimum
values for EV Scenario S7. The minimum values are zero for all hours.

1646
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

Table 4
Levels of power charging and discharging from vehicles to the electricity grid and the total CO2 emissions in Year 2030 from the road transport sector in Scandinavia
and Germany if using fossil fuels or electricity.
S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12
Battery capacity in Year 2030 [TWh] 0.4 0.4 0.2 1.0 1.1 1.1 0.6 3.2 1.1 1.1 0.6

Discharge in Year 2030 [TWh] – 17 17 18 – 20 20 18 17 45 30


Max charging/discharging power in Year 2030 [GWh/h] 28/0 43/29 41/21 44/29 56/0 62/35 60/37 63/34 60/35 72/59 71/40
Total emissions from the transport sector if using electricity in Year 2030 [MtCO2]a 4 3 3 3 10 11 10 11 11 17 15
Total emissions from the transport sector if using fossil fuels in Year 2030 [MtCO2] 29 29 27 31 91 91 83 95 91 119 136
Emissions per unit of electricity to the electric vehicles from the electricity system Year 112 85 98 86 94 96 98 95 95 101 94
2030 [gCO2/kWh]

a
Calculated as the CO2 emissions per GWhel multiplied by the amount of EV charging minus the discharging (GWh) each hour followed by summing for all hours
of Year 2030.

(15 kWh). However, larger batteries (85 kWh), as in S5 and S9, could 4. Discussion
contribute to less severe cycling of the batteries than in the scenarios
with smaller batteries (15 kWh and 30 kWh), as in S3, S4, S7 and S8, in Although the modelling results presented in this work provide in-
which the battery is discharged more frequently from 90% down to sights into the influences that EV introduction would have on invest-
20% of battery capacity. ments in the electricity generation portfolio, several parameters that
Passenger EVs will smoothen the net load curve so that the hour could influence the outcomes remain uncertain. These include: (i) fu-
with maximum net load is reduced with 7 GW (from 120 GW to ture driving patterns and the market diffusion of autonomous vehicles;
113 GW) if the Optimisation + V2G strategy is applied. With the (ii) the dimensioning of battery size in relation to individual driving
Optimisation strategy (and no V2G), the peak net load will be reduced by patterns and preferences; and (iii) the willingness of individual house-
less than 1 GW. In contrast, ERS of mainly trucks and buses will increase holds to participate in an optimisation and V2G solution. On the one
the current net load profile by a maximum of 18 GW, assuming the hand, cheaper batteries might lead to over-dimensioning of the battery
current travelling patterns. However, despite ERS for trucks and buses, in relation to the daily travelling demand. The excess capacity could be
EVs as a whole decrease the need for peak power compared to the used for balancing the electricity system. On the other hand, the con-
scenario without EVs, since the Optimisation + V2G charging strategy is tinuous development of the charging infrastructure, including ERS and
applied for the passenger vehicles in those scenarios with ERS. charging stations in public places, may lead to smaller batteries being
The present study used 200 daily driving profiles to represent the installed in passenger EVs. However, the present study shows that the
spread in the individual driving patterns in a more detailed way than effects on the electricity system are marginal when the battery fleet for
can be achieved using one profile for the aggregated fleet, and to enable Scandinavia and Germany increases from, for example, 0.2 TWh to
a more realistic representation of battery availability in the electric 6.5 TWh. This is, however, probably dependent upon the fact that the
grid. In Fig. 7, charging and discharging to the grid are shown for investment model can only balance EV battery storage within days. A
1 week in October for the sum of all the driving profiles. Fig. 9 shows, region with a lot of good resources for wind power, as is the case for the
for the same week in October, the charging (9a), EV load (9b) and investigated region, needs more long-term storage (i.e., of at least
discharging back to the electricity grid (9c) for 3 of the 200 individual several days) to increase further the wind power share of the total
daily passenger EV driving profiles. The profiles chosen are those with generation. EVs neither substantially increases or decreases the share of
the maximum, minimum and median daily driving distances. In Fig. 9, wind power in the system provided the assumptions in the present
the results from Scenario S7, which assumes a 30-kWh vehicle battery study. The share of wind power linked to the integration of EVs might
and the Optimisation-V2G charging strategy are presented. As shown in be under-estimated in the present study, since no between-days storage
Fig. 9, the different driving profiles are charged and discharged in very of electricity in the EV batteries is allowed.
different ways due to differences in the daily driving distance and Naturally, a smart charging strategy that includes V2G might in-
profile. The profile with the maximum daily driving distance (802 km crease the degradation rate of batteries, owing to the higher frequency
per day) is almost never discharged back to the grid, since the charging of cycling. For battery owners to participate in a V2G system, they
that takes place, mostly during the night-time, needs to be used for would need to be compensated for such degradation. This may be
driving during the day-time. The profile with the minimum daily economically feasible, as EV batteries would thereby, to a large extent,
driving distance (5 km per day) can to a large extent allow discharging reduce the need for peak power during hours when the electricity price
to the grid, since a shorter driving distance means more time connected is high. The EV batteries could potentially also be used as regulating
to the grid, as well as, fewer hours charging the battery to be used for power (i.e., capacity available to meet demand within a short interval
driving. of time) as investigated by Andersson et al. [42]. One decade after
significant EV employment, there is the potential to obtain substantial
battery capacity from batteries that are too degraded for EV purposes,
3.5. CO2 emissions yet, have a potential role to play in the stationary electricity system.
However, the life-span and remaining capacity of such batteries are
The total CO2 emissions in Year 2030, calculated based on the EPOD highly uncertain. A V2G charging solution would most likely require
results for electricity used by all road transport in Scandinavia and reinforcement of the local grid and the installation of additional char-
Germany are in the range 3–17 MtCO2 (Table 4), while the emissions ging stations. It would be of interest to compare the value for the
from burning fossil fuels are at maximum 29–136 MtCO2. Therefore, the electricity system of being able to charge and discharge EVs only at the
saving in CO2 emissions by switching from fossil fuels to electricity in home location and in all parking lots. There may also exist other var-
the road transport sector is about 85% for all the EV scenarios in Year iation management strategies to reduce peak power, such as stationary
2030. The level of CO2 emissions from the electricity system for pas- batteries/secondary-life EV batteries, household temperature manage-
senger EVs is approximately 15 gCO2/km. This can be compared with ment, and hydrogen storage. For the scenarios investigated, the options
an average of 180 gCO2/km for the current vehicle fleet and the goal of for handling variability in net load output are restricted to trade, EV
95 gCO2/km in Year 2020 set by the European Union. batteries, and the ramping of thermal power plants. An interesting

1647
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

variations of solar power, pointing to the need for investigations in such


regions.
An alternative to EVs is fuel cell vehicles that are powered by hy-
drogen. While hydrogen has a lower efficiency from electricity power
source to wheels compared to EVs, it offers better possibilities for en-
ergy storage and is produced during periods of excess electricity gen-
eration, as described in Taljegard et al. [29]. Using a fuel that could be
stored for several weeks and distributed at constant flow could provide
additional flexibility for the electricity system compared to batteries,
and could enhance the integration of more wind power [43]. However,
there are currently few hydrogen vehicles on the market, mainly due to
the lack of hydrogen infrastructure and difficulties associated with
hydrogen storage. ERS have gained a new interest during the past few
years mainly due to the fact that current battery technology are too
heavy for long range vehicle categories such as buses and heavy trucks.
ERS can potentially reduce the fuel costs per kilometre, but it is less
clear what the vehicle cost will be since it depends on the extent of ERS
and the possibility to reduce the on-board battery for trucks and buses,
as well as cars. It is not obvious which option/options that are the best
for the transport sector when moving away from fossil fuels, since both
hydrogen, EVs with static charging and ERS has its own advantages and
disadvantages, and these alternatives need to be investigated further
considering different economical and technical parameters. However,
such an analysis is considered outside the scope of this work. The
present work deals with the relation between future road electrification
and investments in new electricity generation. There are of course many
other aspects which will influence the possibility to electrify road
transportation as well as the conditions for electrification may differ
between different regions. Thus, there is a need for more work to
provide a more comprehensive assessment of electrification of the
transport sector.

5. Conclusions

A model-based investigation of the impacts of EVs on the


Scandinavian and German electricity system is presented. We compare
an Optimisation strategy, in which the charging time of the EVs is op-
timised to minimise the electricity system cost of meeting the electricity
demand, and an Optimisation + V2G strategy that also includes dis-
charging of passenger EVs to the grid. The methodology is superior to
methods found in literature in terms representing individual daily
passenger EV driving profiles. This enables an improved description of
the availability of passenger vehicle batteries connected to the elec-
tricity grid. The modelling results shows that the different driving
profiles are charged and discharged in very different ways due to dif-
ferences in the daily driving distance and profile. This study also in-
vestigates the impacts of EVs on both investment decisions related to
new power capacity and the dispatch of the power plants.
Fig. 9. Modelling results for the charging (a), EV transport load (b) and dis- With respect to the modelling of the variations in electricity supply
charging back to the electricity grid (c) during 1 week in October for three of induced by the EV load, the additional demand from an electrified
the individual daily passenger EV driving profiles, i.e., the maximum, minimum transport sector by Year 2030 is mainly met by increased generation
and median daily driving distances. The results are taken from EV Scenario S7. from wind power complemented with natural gas and biogas genera-
tion, provided that there is a cap on CO2 emissions. The modelling gives
that the CO2 emissions from the electricity system Year 2030 supplying
future study would be to include the cycling cost of batteries and sto- passenger EVs are only 15 gCO2/km. Towards Year 2050, i.e. the end of
rage options other than EV batteries in similar models. the time period investigated, very low CO2 emission levels are allowed,
The results presented in this study are system-specific, i.e., fully resulting in investments in both wind power and thermal base load
valid only for the investigated region, which is characterised by access generation in the form of coal with CCS (or CCS lignite co-fired with
to reservoir hydro power, good conditions for wind power, but less- biomass) to cover the additional EV load. Lack of low cost flexible
good conditions for solar power. For other countries with similar con- generation and moving to sites with less favourable wind conditions
ditions for wind, solar and hydro power, the same type of electricity gives thermal base load generation in the form of coal with CCS a
system modelling might show similar results. A different region, such as competitive advantage relative wind power. The electricity generation
Spain, with more favourable conditions for solar power but little hydro from wind power increases by 7–30% in the EV scenarios. This corre-
power, might instead utilise the EV batteries for handling daily sponds to a few percentage points higher than the increase in transport

1648
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

electricity demand from EVs. Thereby, EVs neither substantially in- in the modelling in this work only allows for intra-day storage. A sce-
creases or decreases the share of wind power in the system provided the nario with full electrification of the road transport sector, including ERS
assumptions in the present study. for trucks and buses increases the net load by 18 GW. However, a sce-
Furthermore, the present study shows that optimised charging, and nario that includes ERS still decreases the need for peak power, as
especially that allowing discharging to the grid (V2G), reduces (by compared to a scenario without EVs, provided that the
20–100%) the need for new investments in peak power capacity, as Optimisation + V2G charging strategy is applied to the passenger EVs.
compared to the scenario without EVs. The value of investing in solar
power is also reduced in all the EV scenarios by 22–42%, as compared Acknowledgements
to a scenario without EVs, due to the fact that in Northern Europe solar
power competes with EVs for peak power supply. We gratefully acknowledge the Norwegian Public Road
In the present work, battery sizes greater than15 kWh per vehicle do Administration for financial support, and Dr. Sten Karlsson and Dr. Lars
not facilitate the integration of more wind power or reduce the cur- Henrik Björnsson at Chalmers University of Technology for valuable
tailment of wind power, bearing in mind that the battery representation data input and fruitful discussions.

Appendix A. Input data

See Tables A1 and A2.

Table A1
Technology investment, operational and maintenances costs (O&M) and life-times of some of the key technologies available in the ELIN model [44–48].
Life-time Investment Fixed O&M Variable O&Ma
[Years] costa [€/kWel] costa cost
[€/kWel/ [€/MWhel]
year]

Hard coal
Condense 40 1560 27.36 2.1
CHP/BP 40 1560 27.36 2.1
CCS 40 3003 90.5 2.1
CCS + bio-cofiring 40 3463 107.6 2.1

Lignite
Condense 40 1560 31.68 2.1
CHP/BP 40 1560 31.68 2.1
CCS 40 3003 90.5 1.36
CCS + bio-cofiring 40 3463 107.6 2.1

Natural gas
GT 30 390 7.92 0.4
CCGT 30 780 12.96 0.8
CHP/BP 30 1014 16.56 0.7
CCS 30 1800 35.1 2.1

Bio & waste


Condense 40 1860
GT 30 380 7.92 0.7
CCGT 30 755 12.96 0.8
Waste 40 6630 443 2.1
CHP/BP 40 3150 57.6 2.1

Intermittent
Wind (onshore) 25 1320 30 1.1
Wind (offshore) 25 2190 100 1.1
Solar PV 25 1280 20 1.1
Small hydro 75 3745 73 1

a
The values shown for investment costs and the fixed/variable O&M costs are based on the World Energy Outlook assumptions of the IEA from the 2011–2014
editions [44–47] and have been extrapolated for Year 2035 to Year 2050. Investment costs for CCS technologies are obtained from the Zero Emission Platform [48].

Table A2
National electricity demand in TWh [38].
Year 2020 Year 2030 Year 2040 Year 2050

Norway 134 125 127 131


Sweden 150 151 148 150
Denmark 35 35 36 38
Germany 563 561 556 576

1649
M. Taljegard et al. Applied Energy 235 (2019) 1637–1650

References statistik-efter-amne/transporter-och-kommunikationer/vagtrafik/fordonsstatistik/
[accessed 7 March 2018].
[28] Eurostat. Road transport equipment - stock of vehicles. European Commission;
[1] European Commission. Electrification of the transport system - studies and reports; 2017.
2017. [29] Taljegard M, Göransson L, Odenberger M, Johnsson F. Spacial and dynamic energy
[2] Fridstrøm L, Alfsen KH. Vegen mot klimavennlig transport; 2014. demand of the E39 highway – implications on electrification options. Appl Energy
[3] Johansson T. Fossilfrihet på väg. Stockholm, Sweden: Ministry of Enterprise, SOU; 2017;195:681–92.
2013. p. 84. [30] Karlsson S, Kullingsjö L-H. GPS measurement of Swedish car movements for as-
[4] Swedish Government. Det klimatpolitiska ramverket. Regeringskansliet 2017. sessment of possible electrification. Electric vehicle symposium and exhibition
[5] Grahn P. Electric vehicle charging impact on load profile [Licentiate Thesis]. (EVS27). 2013. p. 1–14.
Stockholm; 2013. [31] Kullingsjö L-H, Karlsson S. The Swedish car movement data project. In: Proceedings
[6] Robert Meyer MK, Themelis Nickolas, Bronicki Lucien, Soder Lennart, Vega Luis. to EEVC Brussels, Belgium, November 19–22, 2012; 2012.
Renewable energy systems. Springer; 2013. [32] Goop J, Odenberger M, Johnsson F. The effect of high levels of solar generation on
[7] Veneri O. Technologies and applications for smart charging of electric and plug-in congestion in the European electricity transmission grid. Appl Energy
hybrid vehicles. Springer; 2017. 2017;205:1128–40.
[8] Grahn P. Electric vehicle charging modeling [Doctoral Thesis]. Stockholm; 2014. [33] European Commission. Energy Roadmap 2050. Brussels; 2011. https://ec.europa.
[9] Weiller C. Plug-in hybrid electric vehicle impacts on hourly electricity demand in eu/energy/sites/ener/files/documents/2012_energy_roadmap_2050_en_0.pdf [ac-
the United States. Energy Policy 2011;39:3766–78. cessed 7 March 2018].
[10] Darabi Z, Ferdowsi M. Aggregated impact of plug-in hybrid electric vehicles on [34] Nahmmacher P, Schmid E, Hirth L, Knopf B. Carpe diem: a novel approach to select
electricity demand profile. IEEE Trans Sustain Energy 2011;2:501–8. representative days for long-term power system modeling. Energy
[11] Vatne Å, Molinas M, Foosnas JA. Analysis of a scenario of large scale adoption of 2016;112:430–42.
electrical vehicles in nord-trøndelag. Energy Procedia 2012;20:291–300. [35] Reichenberg L. Variability and variation management in a renewable electricity
[12] Stamati T-E, Bauer P. On-road charging of electric vehicles. IEEE transportation system [Doctoral Thesis]; 2017.
electrification conference and expo (ITEC). 2013. p. 1–8. [36] Kjärstad J, Johnsson F. The European power plant infrastructure—presentation of
[13] Göransson L, Karlsson S, Johnsson F. Integration of plug-in hybrid electric vehicles the Chalmers energy infrastructure database with applications. Energy Policy
in a regional wind-thermal power system. Energy Policy 2010;38:5482–92. 2007;35:3643–64.
[14] Hedegaard K, Ravn H, Juul N, Meibom P. Effects of electric vehicles on power [37] Eurostat. Gross domestic product (GDP) at current market prices by NUTS 2 regions;
systems in Northern Europe. Energy 2012;48:356–68. 2012. http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=nama_10r_2gdp
[15] Lund H, Kempton W. Integration of renewable energy into the transport and elec- &lang=en [accessed 7 March 2018].
tricity sectors through V2G. Energy Policy 2008;36:3578–87. [38] European Network of Transmission Operators for Electricity. Hourly load values for
[16] Verzijlbergh R, Martínez-Anido CB, Lukszo Z, de Vries L. Does controlled electric all countries for a specific month (in MW); 2017. http://www.entsoe.eu/db-query/
vehicle charging substitute cross-border transmission capacity? Appl Energy consumption/mhlv-a-specific-country-for-a-specific-month [accessed 7 March
2014;120:169–80. 2018].
[17] Jochem P, Babrowski S, Fichtner W. Assessing CO 2 emissions of electric vehicles in [39] Global Modeling and Assimilation Office (GMAO). MERRA-2 tavg1_2d_rad_Nx:
Germany in 2030. Transport Res Part A: Policy Pract 2015;78:68–83. 2d,1-hourly,time-averaged,single-level,assimilation,radiation diagnostics V5.12.4.
[18] Schill W-P, Gerbaulet C. Power system impacts of electric vehicles in Germany: Greenbelt, MD, USA: Goddard Earth Sciences Data and Information Services Center
charging with coal or renewables? Appl Energy 2015;156:185–96. (GES DISC); 2015 [accessed 7 March 2018].
[19] Hadley SW, Tsvetkova AA. Potential impacts of plug-in hybrid electric vehicles on [40] Norwood Z, Nyholm E, Otanicar T, Johnsson F. A geospatial comparison of dis-
regional power generation. Electric J 2009;22:56–68. tributed solar heat and power in Europe and the US. PLoS ONE 2014.
[20] Reichenberg L, Hedenus F, Odenberger M, Johnsson F. Tailoring large-scale elec- [41] Global Modeling and Assimilation Office (GMAO). MERRA-2 tavg1_2d_slv_Nx: 2d,1-
tricity production from variable renewable energy sources to accommodate base- hourly,time-averaged,single-level,assimilation,single-level diagnostics V5.12.4.
load generation in Europe. Renew Energy 2018. Greenbelt (MD, USA): Goddard Earth Sciences Data and Information Services
[21] Reichenberg L, Wojciechowski A, Hedenus F, Johnsson F. Geographic aggregation Center (GES DISC); 2015 [accessed 7 March 2018].
of wind power—an optimization methodology for avoiding low outputs. Wind [42] Andersson S-L, Elofsson A, Galus MD, Göransson L, Karlsson S, Johnsson F, et al.
Energy 2017;20:19–32. Plug-in hybrid electric vehicles as regulating power providers: case studies of
[22] Odenberger M, Unger T, Johnsson F. Pathways for the North European electricity Sweden and Germany. Energy Policy 2010;38:2751–62.
supply. Energy Policy 2009;37:1660–77. [43] Göransson L, Johnsson F. A comparison of variation management strategies for
[23] Unger T, Odenberger M. Dispatch modelling of the European electricity supply: the wind power integration in different electricity system contexts. Wind Energy
EPOD model. Methods and models used in the project pathways to sustainable 2018;21:837–54.
european energy systems. Mölndal 2011:97–101. [44] International Energy Agency. World Energy Outlook; 2011.
[24] Göransson L, Goop J, Unger T, Odenberger M, Johnsson F. Linkages between de- [45] International Energy Agency. World Energy Outlook; 2012.
mand-side management and congestion in the European electricity transmission [46] International Energy Agency. World Energy Outlook; 2013.
system. Energy 2014;69:860–72. [47] International Energy Agency. World Energy Outlook; 2014.
[25] Nyholm E, Goop J, Odenberger M, Johnsson F. Solar photovoltaic-battery systems [48] Zero Platform. The costs of CO2 capture - post-demonstration CCS in the EU.
in Swedish households – self-consumption and self-sufficiency. Appl Energy Brussels, Belgium; 2011. http://hub.globalccsinstitute.com/sites/default/files/
2016;183:148–59. publications/119801/costs-co2-capture-post-demonstration-ccs-eu.pdf [accessed 7
[26] Statistics Norway. Registered Vehicles. https://www.ssb.no/en/transport-og- March 2018].
reiseliv/statistikker/bilreg [accessed 7 March 2018].
[27] Statistics Sweden. Fordonsstatistik; 2017. https://www.scb.se/hitta-statistik/

1650

You might also like