Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

Assessment of future EV charging infrastructure


scenarios for long-distance transport in Sweden
Francisco J. Márquez-Fernández∗‡ Member, IEEE, Joschka Bischoff† , Gabriel Domingues-Olavarría∗ ,
Mats Alaküla∗‡

∗ Lund
University † Technische Universität Berlin ‡ Swedish
Electromobility Centre
Div. of Industrial Electrical Dep. of Transport System Electrical machines, drives and charging
Engineering and Automation Planning and Telematics Chalmers Univ. of Technology
Lund, Sweden Berlin, Germany Göteborg, Sweden

Abstract—Over the last two decades, electrification has gained highlight the importance of the battery as both a limiting
importance as a means to decarbonise the transport sector. As and/or fostering factor for transport electrification. Besides
the number of Electric Vehicles (EVs) increases, it is important being the single most expensive component in an electric
to consider broader system aspects as well, especially when
deciding the type, coverage, size and location of the charging vehicle, the battery also determines the vehicle’s range, the
infrastructure required. charging time, and the lifetime (alternatively the second-hand
This article proposes a new approach using agent-based value) of the vehicle. For heavy vehicles, it may also affect
simulations to assess the impact that different system parameters the payload significantly, due to its weight.
have on the total energy consumption, the charging infrastructure Nonetheless, some of the issues related to the battery may be
needs or the overall system cost for all electromobility related
technologies. To demonstrate the capabilities of this approach, alleviated with a more comprehensive, better designed charg-
five potential future scenarios for charging infrastructure de- ing infrastructure. By allowing for more frequent charging the
ployment are analyzed, assuming that all long-distance transport battery size (and thus its cost, weight and volume) may be
in Sweden is electrified. For each of the scenarios the total reduced, or its lifetime may be extended due to less aggressive
energy consumed and the charging infrastructure needs are cycling. Furthermore, under the last decade a number of
assessed. Finally, the cost associated with the electromobility
related technology in each scenario is estimated. technical solutions have emerged, allowing for continuous
The results show that the lowest system cost corresponds to a supply of energy to the vehicles as they move along the road
scenario with Electric Road Systems (ERS) widely available to (so called dynamic charging) [6]. These solutions receive the
all vehicle types, mostly due to the potential reduction of their generic name of Electric Road Systems (ERS) and are often
battery pack. However, such scenario may incur in a higher classified according to the energy transferring principle into
overall energy consumption, if the drivers decide to alter their
routes to use the ERS, thus avoiding stopping for charging. conductive - when there exists a physical electrical connection
Keywords: Electric vehicles, charging infrastructure, dynamic between the ERS and the vehicle, inductive - when the energy
charging, Electric Road Systems, cost analysis. is transferred wirelessly by means of coupled magnetic fields,
or capacitive - when the energy is also transferred wirelessly
I. I NTRODUCTION using coupled electric fields.
RANSPORT electrification has proven an effective way The relation between the battery size in the vehicles and
T of reducing CO2 emissions in the urge to mitigate the
effects of climate change, as reflected in several national
the charging infrastructure can be illustrated by two extreme
hypothetical scenarios: i) A scenario in which every vehicle
and international regulations and agreements over the world. features a battery large enough to cover its daily driven
Though this may be the primary reason for electrifying trans- distance in all circumstances; ii) A scenario in which ERS
port at present, several studies show that, in a not so distant for dynamic charging (i.e. supplying power to the vehicles as
future, electric vehicles are anticipated to be cheaper and they drive) is available everywhere, on every road. In the first
higher performing than their fossil fuel counterparts, specially case the vehicles would only be charged when at rest, after a
if all additional societal benefits and tax benefits are taken full working day. Charging infrastructure would thus only be
into account, hence economic reasons are expected to promote needed at the places where the vehicles stay overnight: garages
electrification even further [1]–[5]. Most of these studies also or public parking places for passenger cars, depots for many
commercial vehicles and buses, and rest- and service areas for
long-haul trucks and coaches. In the second scenario, just a
Financial support of this work by the Swedish Electromobility Centre very small battery would be needed to get in and out of the
through the project IDEAS is gratefully acknowledged.
E-mail: fran.marquez@iea.lth.se, joschka.bischoff@sbb.ch, parking space, since the vehicles would constantly receive the
gabriel.domingues@iea.lth.se, mats.alakula@iea.lth.se power needed for driving from the road quite in the same way

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

as trains do, and no additional charging infrastructure other board charging equipment, and ERS infrastructure).
than the Electric Road System would be needed. To demonstrate the capabilities of this approach, five poten-
After just a brief reflection over these two extreme scenar- tial future scenarios for charging infrastructure deployment as-
ios, it is clear that they are not likely to happen in reality. In suming that all long-distance transport (> 100 km) in Sweden
the first case some vehicles may need an unrealistically large is electrified. Although relevant and illustrative, these scenarios
battery, while in the second case there may be roads so remote do not intend to cover all possible charging infrastructure
that it might not even be possible to bring an electric power configurations, since that would entail a much more extensive
connection close enough. Moreover, even if both scenarios work than what can be presented in a single article. A
were technically possible, they are most likely not optimal Multi-Agent Transport Simulation (MATSim) [16], [17] model
from a cost and use of resources point of view. An optimal of Sweden has been developed, depicting the long-distance
charging infrastructure scenario will likely be somewhere in travelling nationwide. In this model each vehicle (agent) in
between, with vehicles featuring a battery pack allowing to a reduced-scale fleet is represented, providing very detailed
cover a reasonable distance depending on the vehicle type information of all relevant aspects of each trip, including every
and purpose. A wide-spread "slow" charging infrastructure charging event. Preliminary results from the first simulation
would allow to charge at moderate power levels when the runs have been reported in [18].
vehicles are at rest (e.g. overnight charging or charging at
the workplace), and a supplementary charging infrastructure II. M ETHODOLOGY
consisting of either "fast" (high-power) chargers, Electric Road
Systems (dynamic charging) or a combination thereof, will The first step is to define relevant scenarios to be analysed.
serve those vehicles travelling longer distances than their In this work, five scenarios are proposed. All scenarios share
battery allows with a single charge. the same fleet, road network, trip distribution and population
Finding this optimal combination of different charging data, corresponding to the electrification of long-distance
infrastructure types and battery sizes on-board the vehicles transport (> 100 km) in Sweden. However, they differ on the
is easier said than done. Although extensive research efforts assumed battery sizes, ERS extension and availability of fast
from both industry and academia have been devoted in the charging stations.
last years to the electrification of transport, these have been Each scenario is simulated in MATSim to obtain the energy
mostly directed towards the development of the technology on- flow profiles over time that serve as the basis for the subse-
board the vehicles as well as, lately, the charging equipment quent analysis. These profiles can be aggregated in various
(examples of such efforts in the EU and the US are reported ways to calculate e.g. the energy consumed by a certain
in [7], [8] respectively). As the number of electric vehicles vehicle, vehicle class, or the total fleet, the energy delivered by
increases, it becomes essential to consider broader system a specific charging equipment or within a certain geographical
aspects also. Decisions concerning the deployment of charging area, or the energy consumed per kilometre on any given road.
infrastructure such as the choice of charging mode (static or The size of the battery for each vehicle class and the
dynamic), the power rating of the charging connection, the extension of ERS deployed is predefined for each scenario.
density and spreading of the charging network, or the pricing However, all scenarios are simulated considering an unlimited
policies applied will likely have a significant impact on the rest number of fast charging connections located at the same places
of the transport system, affecting for example the distribution as today’s fuel stations. Once the scenario is simulated, the
of the traffic flow in different roads, the total traffic volumes final number of fast charging stations required to supply all
and the corresponding energy consumption, the choice of vehicles is estimated based on their energy throughput.
transport mode, and the number, location and duration of the Finally, the cost of the electromobility related technology,
charging events [2], [9]–[13]. Additionally, the deployment of battery, electrified drivetrain, on-board and off-board charging
charging infrastructure will also have an impact on, or even equipment, and ERS infrastructure, necessary for the electrifi-
be limited by other related systems such as the electric power cation of the whole fleet, is estimated for each scenario using
grid to name the most obvious one [14], [15]. own-developed cost models for all relevant components.
This article does not intend to find the optimal charging in- This process is illustrated in Figure 1.
frastructure configuration, but rather proposes a new approach
based on the use of microscopic agent-based simulations to III. D ESCRIPTION OF THE S CENARIOS
assess the influence of different system parameters such as the
battery capacity installed in the vehicles, the type, power rating Five potential future scenarios for the deployment of charg-
and density of the charging network, or the pricing policies. ing infrastructure with regard to the electrification of long-
Agent-based simulations are able to represent the decision distance transport (> 100 km) in Sweden are proposed:
processes related to charging for the vehicles in the fleet, thus 1 No ERS installed. All in-travel charging needs are sup-
enabling to assess how the aforementioned parameters affect plied by high power (fast) static charging stations.
e.g. the total energy consumption, the charging infrastructure 2a Overhead conductive ERS in all roads with a speed limit
needs or the overall system cost for all electromobility related of 80 km/h or higher (approx. 11800 km of ERS in
technologies (battery, electrified drivetrain, on-board and off- total), supplemented with fast charging stations.

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

Figure 2: Alternative location of conductive ERS with respect


to the vehicle.

Figure 1: Diagram illustrating the proposed method. wide spread deployment of ERS they can benefit from, while
the passenger car batteries remain unchanged. In Scenario 2b,
since ERS is available for all vehicle classes, the reduction of
2b Roadbound conductive ERS in all roads with a speed the battery size applies to all vehicles. In Scenarios 3a and 3b
limit of 80 km/h or higher (approx. 11800 km of ERS the battery sizes remain the same as in Scenario 1, given the
in total), supplemented with fast charging stations. relatively small coverage of ERS.
3a Overhead conductive ERS in all roads with a speed limit It is worth mentioning that, in this study, conductive ERS
of 110 km/h or higher (approx. 2200 km of ERS in technology is considered in all cases due to it being the
total), supplemented with fast charging stations. most cost effective according to previous work by the au-
3b Roadbound conductive ERS in all roads with a speed thors [2]. Overhead catenary lines are always conductive,
limit of 110 km/h or higher (approx. 2200 km of ERS but commercial solutions for both inductive and conductive
in total), supplemented with fast charging stations. roadbound ERS exist. While there are some differences in
The first obvious difference between the scenarios is the performance, e.g. in the power transfer capabilities per vehicle
extension of the road covered with ERS for dynamic charging. or the efficiency, affecting the total energy consumption or the
In Scenarios 2a and 2b all sections of the road network with amount of supplementary static charging stations needed, the
a speed limit of 80 km/h or higher are equipped with ERS, most significant difference is expected to be in the system cost.
totalling 11800 km, i.e. 89% of all the major roads in the A comparison with scenarios including inductive ERS is out
country, and 12% of all the roads managed by the Swedish of the scope of this article.
Transport Administration. In Scenarios 3a and 3b only those
sections of the road with a speed limit equal to or higher IV. MATS IM T RANSPORT MODEL
than 110 km/h (which is the most common speed limit in Transport models traditionally used in traffic planning, or to
motorways in Sweden) are equipped with ERS. These add up support the development of new road infrastructure, are based
to 2200 km, representing 14% of all the major roads in the on a macroscopic approach in which aggregate data are used
country, and just 2% of all the roads managed by the Swedish to predict traffic flows in the different parts of the transport
Transport Administration. system. Although very computationally effective, macroscopic
The second important difference is the physical location models are too coarse and do not take into account the
of the ERS with respect to ground, which determines the behaviour of individual vehicles, their interactions and their
type of vehicles that can use it. Inductive and capacitive ERS decision-making processes when it comes to e.g. where, when
technologies are limited to be installed under the vehicle, and for how long to charge the batteries. For this reason, they
integrated in the road [19]–[21]. Conductive solutions however are less suitable to analyse the impact of different charging
can be installed as overhead catenary lines about 5 meters infrastructure deployment scenarios.
above ground [22] (corresponding to (e) in Figure 2), as a As mentioned in Section I, activity-based models, in partic-
side mounted rail on the side of the roadway [23], [24] ((d) ular agent-based models in which every agent is simulated
in Figure 2), or as a rail under the vehicle, either embedded independently may be used to have a better understanding
on the road [25] ((a) in Figure 2), flush with the road surface of the impact of a certain charging infrastructure scenario.
[26] ((b) in Figure 2) or mounted on top of the road [27] ((c) An agent in this context may represent any individual in a
in Figure 2). Due to its location high above the ground, the synthetic population such as a private traveller commuting to
overhead catenary can only be used by large vehicles such work, a truck part of a larger fleet, or a scheduled bus service
as trucks and coaches while the other options enable dynamic between two distant cities [28], [29]. Each agent has a number
charging of almost all vehicle types with three or more wheels. of missions or activities to fulfil at specific geographical
Each of the vehicle classes considered in the fleet (described locations and times, and depending on how effectively it
in the next section) has a certain battery capacity installed, moves between missions it will receive a score (utility value).
according to its expected use. In Scenario 2a the size of the Accomplishing a mission earns utility points while the time
batteries installed in trucks is significantly reduced given the spent either in traffic or while charging the battery during

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

the mission loses points [30]. Although the objective of this each link is derived from the elevation of the extreme points -
study is to analyse the impact of different types of charging and some intermediate points if the link is longer than 500 m -
infrastructure for EVs (both light and heavy vehicles) the obtained from the Copernicus Land Monitoring Service [37].
model is in principle multi-modal, thus agents can choose
between different modes of transportation if that leads to better C. Static charging infrastructure
results. Since the model focuses on long-distance transport some
MATSim does not perform a time-step integration; instead, degree of planning ahead is expected, hence all vehicles are
for each agent all values are calculated as an average over assumed to be fully charged at the beginning of the simulated
a link in the network (the road section between two intersec- day. This, in turn, implies that a comprehensive slow static
tions), depending on the average speed of the vehicle over that charging infrastructure network is present in all scenarios.
link and the slope of the road. On the other hand, charging When it comes to fast (high-power) static charging, all
is always performed at the highest possible power (depending scenarios are simulated considering an unlimited number of
on the limitations of the battery and the charger as explained fast charging connections, located at the same places as today’s
later), and charging losses are not considered. fuel stations. Once the scenario is simulated, the final number
In order to limit the size of the model and keep it solvable, of fast charging stations required to cover the demands of
detailed information such as driving lanes and acceleration / all vehicles is estimated based on their usage, eliminating
deceleration profiles are excluded. Moreover, only a fraction those stations that deliver less than 10% of the average energy
of the population is simulated (commonly between 10% - throughput (see Section V-B).
20% based on previous experience [29], [31]). The model is
D. Vehicle model
solved iteratively and, for each iteration, slight variations in
the travel plan of the agents are introduced. For the model to The Swedish road vehicle fleet of interest for this study is
give accurate results, several components are needed: represented by four vehicle classes: three sizes of passenger
• Input data in order to create the synthetic population, with
cars (totalling 5 million vehicles), corresponding to the B-
the characteristics corresponding to each individual (e.g. segment - subcompact in the US - 12 % (e.g. Volkswagen
vehicle type, size, weight, battery capacity on board, etc.) Polo, Renault Clio, Ford Fiesta), C-segment - compact in
and the transport missions that should be fulfilled the US - 50% (e.g. Volkswagen Golf, Renault Mégane, Ford
• The transport network in which the agents move, with
Focus), and D-segment - mid-size in the US - 38% (e.g.
an adequate level of discretisation, specifying the traffic Volkswagen Passat, Renault Koleos, Ford Mondeo); and heavy
capacity of the different links, their free flow speed, slope, vehicles represented by an average 21 ton truck (approx. 65000
and the charging infrastructure available (specifying the vehicles). The main characteristics of each vehicle type are
number of charging connections, type, power capabilities, compiled in Table I. Two values are given for the installed
geographical location, etc.) battery capacity, the "reduced" one corresponding to those
• A model of the Electric Vehicles, providing the energy
scenarios in which ERS is widely available for that vehicle
consumption per kilometre as a function of the vehicle’s type (Scenarios 2a and 2b for trucks and Scenario 2b for
speed and the slope of the road, and the battery charging passenger cars).
process as a function of the capacity and state of charge
Table I: Main characteristics of the different vehicle classes.
• Calibration data (e.g. traffic counts) to validate the mod-
elling approach and results B-seg C-seg D-seg Truck
A brief description of each of the previous components is Mass [kg] 1300 1700 2000 19300
given below. A more detailed explanation of the development Mass ERS [kg] 1160 1455 1545 16325
of the MATSim model for long-distance transport in Sweden Power [kW] 80 120 160 200
is provided in [32]. Aux. P. [kW] 1 1.5 2 3.5
Battery [kWh] 40 60 100 500
A. Population input data Batt. ERS [kWh] 20 25 35 75
The synthetic population is based on the following data: the Front Area [m2 ] 2.04 2.19 2.37 7.3
composition of the Swedish vehicle fleet (from the Swedish Airdrag Coef. [-] 0.32 0.27 0.31 0.58
Transport Agency), goods transport flows from the Samgods
model calibration case [33] long-distance commuting trips (> For each vehicle type, energy consumption maps are created
100 km) from the Sampers model [34] and Statistics Sweden as a function of the vehicle average speed and the road slope
[35], and traffic counts at different points of the national road through the simulation of pre-defined drive cycles in Matlab-
network also from the Swedish Transport Agency. Simulink. For passenger cars, the relevant part of the WLTP
(Worldwide harmonised Light-vehicle Test Procedure) [38]
B. Road network for the whole Sweden drive cycle is used, scaling the speed to obtain the desired
The transport network used is derived from OpenStreetMap average speed value. For average speeds between 1 and 43
[36] with enough resolution to be able to capture congestion km/h the first part of the WLTP cycle is used (low speed,
patterns in and around the largest cities. The average slope of with an original average speed of 26 km/h), between 10 and

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

58 km/h the medium speed part (original average speed of 45 charging events were logged, Tesla superchargers were limited
km/h), between 45 and 90 km/h the high speed part (original to 120 kW .
average speed of 61 km/h) and between 83 and 123 km/h
the extra-high one (original average speed of 94 km/h). For V. R ESULTS
heavy vehicles, a much simpler drive cycle consisting on an For each scenario, a full day is simulated with the MATSim
acceleration phase (limited to 0.5 m/s2 ) followed by constant model. Since the model only depicts long-distance transport, it
speed driving is used, since it resembles better the type of is assumed that the drivers will plan ahead and thus all vehicles
driving expected for these vehicles. In the Simulink model, the start the simulation with a fully charged battery. Before the
tractive force needed to follow the drive cycle is obtained from simulation starts, a plan for the day is set up for every vehicle,
the motion equation (1), with mv representing the mass of the specifying the route that should be followed to accomplish
vehicle, a its acceleration, Ftr the required tractive force, Fg the designated transport missions. This plan also includes the
the gravitational force component opposing movement as a charging stops, if needed. In each iteration, small changes are
result of the slope of the road, Fad the air drag resistance and introduced in the daily plan (i.e. choosing a different road
Frol the rolling resistance on the tires. This tractive force is or charging station) aiming to fulfil the missions in the most
then translated into a tractive torque on the vehicle’s wheels, effective way, thus maximising the utility value.
and with the help of loss models for the traction motor, the
power electronics converter, the mechanical transmission as A. Energy consumption
well as assumptions regarding auxiliary loads, the total energy As a result from the MATSim model, the energy consump-
consumption per kilometre for that particular vehicle at the tion profile for every vehicle and the charging profile for
selected speed and road slope values is calculated [39]. These every piece of charging equipment are obtained. These can
values are stored in a look-up table for subsequent use in the be aggregated in different ways, in order to calculate e.g. the
MATSim model. energy consumption per km of road (directly related to the
As an example, the resulting energy consumption map for traffic intensity), the total energy consumed by each vehicle
a C-segment car is shown in Fig. 3, left. class, or the total energy delivered by a certain type of charging
infrastructure. Figure 4 presents the total energy consumption
mv a = Ftr − Fg − Fad − Frol (1) by source (overnight slow static charging, in-travel fast static
charging and ERS) in each scenario for passenger cars (top
left), heavy vehicles (top right) and the whole fleet (bottom).
Battery charging profile
2

106 Energy Consumption - Cars 106 Energy Consumption - Trucks


1.5
Charging rate [-]

8
6
1

Energy [kWh]
Energy [kWh]

6
4
0.5 4

Envelope (based on Tesla Model 3 LR user data) 2


Charging profile in the model Fast-charging 2 Fast-charging
0
0 20 40 60 80 100 ERS ERS
Overnight Overnight
State of charge [%] 0 0
1 2 3 4 5 1 2 3 4 5
Figure 3: Energy consumption map for a C-segment car as a Scenarios
6
Scenarios
10 Energy Consumption - Total
function of road slope and speed (left) and charging profile as 15

a function of the battery state of charge (right).


Energy [kWh]

10

To preserve the life of the batteries, charging and discharg-


ing power must be limited. This limitation depends mostly 5
Fast-charging
on the capacity of the battery and its temperature for a given ERS
Overnight
battery chemistry. Additionally, the maximum power limitation 0
1 2 3 4 5
Scenarios
imposed by the actual charging equipment must be observed.
In order to have an accurate estimation of the charging times, Figure 4: Total energy consumption in each scenario for
a charging power profile has been implemented (see right plot passenger cars (top left), heavy vehicles (top right) and the
in Fig. 3). Depending on the State of Charge (SOC) of the whole fleet (bottom).
battery when the vehicle arrives to the charging station, the
charging power delivered is limited as follows: 1.75 C between Scenario 1 may be used as a reference case for comparison.
0 and 50% SOC, 1.25 C between 50 and 75% SOC and 0.5 In this scenario, most of the energy consumed by the vehicles
C between 75 and 100% SOC, where the charging rate C (in comes from slow (overnight) charging, supplemented by in-
kW ) is the power needed to charge / discharge the battery travel fast charging for the longer trips. The total energy
in 1 hour. This charging profile is based on publicly available consumption, aggregated and scaled up to the full population
data from Tesla Model 3LR users [40] since this is considered is 148 GW h/day. According to the Swedish Energy Agency,
state-of-the-art battery technology, although at the time these 78 T W h (mostly from fossil fuels) were consumed in road

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

Table II: Energy consumed per scenario (GWh in 10% model) utility points are deduced based on the total travel time, and
Passenger cars Heavy vehicles there is no price difference (or any other difference) between
Slow ERS Fast Total Slow ERS Fast Total charging at a static fast charging station or while driving on
S1 5.00 - 1.38 6.38 6.98 - 1.43 8.41 an ERS. Thus, the vehicles choose to drive on those roads
S2a 5.00 - 1.38 6.38 0.54 7.03 0.02 7.60 equipped with an ERS, even if it results in longer distances,
S2b 0.76 5.96 0.02 6.74 0.54 7.03 0.02 7.59 as long as the total travel time is shorter. While heavy vehicles
S3a 5.00 - 1.38 6.38 2.71 5.74 0.12 8.56 usually drive on the main roads, and hence benefit from the
S3b 2.53 4.33 0.13 6.99 2.72 5.74 0.12 8.56 ERS without major route changes, passenger cars tend to use
smaller roads as well if they do not need to worry about
charging. This result illustrates the influence that charging
transport in Sweden during 2018 [41]. Accounting for the infrastructure may have on traffic flows, which depends not
differences in efficiency between electric and combustion only on the availability of charging infrastructure and its type,
drivetrains, and assuming 300 active days per year, this but also on the length of the trip, the battery size, the State of
corresponds to 80 − 110 GW h/day of electricity used in Charge (SOC) at the beginning of the trip and the desired SOC
transportation. This somewhat lower figure compared to the at the destination [42]. Modelling such complex influence is
simulated one is not unrealistic, considering that not all days a good example of what can be achieved with agent-based
of the year are busy traffic days. models (e.g. MATSim) compared to more traditional transport
Despite being significantly fewer, heavy vehicles consume modelling tools.
about 57% of the total energy used for long distance travelling
during the simulated day. However, due to the characteristics In Scenarios 3a and 3b the ERS coverage is limited to the
of the vehicles and the nature of the trips, both passenger cars motorways (2200 km of roads with a speed limit of 110 km/h
and heavy vehicles require approximately the same amount of or more). In Scenario 3a with overhead catenary lines, only
energy from fast charging. heavy vehicles have access to it. Given that this type of
In Scenario 2a ERS is introduced at large scale for heavy vehicles usually drive on such roads, they take great advantage
vehicles, hence no noticeable differences are observed for from it. Even if the length of ERS installed is less than one
passenger cars - just those due to the randomness in the fifth of that of the previous scenarios, it still supplies 67%
model. For heavy vehicles, a 10% reduction of the total of the total energy consumed (compared to 92% in Scenarios
energy consumption is achieved due to the smaller size (and 2a and 2b). Besides, and given that the vehicles retain the
weight) of the batteries. Since the model does not take into large battery from Scenario 1, the remaining 33% of the
consideration the price of electricity but just the time penalty energy comes mostly from slow static charging (overnight)
when deciding when and where to charge, whenever an ERS is and the contribution of fast charging is almost negligible -
available the vehicles will use it for both traction and charging although necessary to complete some of the trips. The total
as long as the battery is not fully charged already. For this energy consumed by heavy vehicles is very similar, although
reason, and due to the wide availability of ERS in this scenario, somewhat higher than in Scenario 1. This is to be expected
most of the energy consumed by heavy vehicles comes from since the vehicles retain the large batteries from that scenario,
the ERS, the fraction of the consumed energy coming from and not many vehicles are expected to modify their travel plan
night chargers is greatly reduced (down to 7.7% of the value to benefit from the ERS. Passenger cars in this scenario remain
in Scenario 1) and the contribution of fast chargers is almost the same as in Scenarios 1 and 2a.
negligible (22 MWh in the 10% scale model, i.e. 1.6% of In Scenario 3b passenger cars have access to the 2200 km
the corresponding value for Scenario 1). Consequently, the of ERS as well. In this scenario, ERS supply 62% of the total
number of fast charging stations required is also reduced, energy consumption of passenger cars (compared to 89% in
although not in the same proportion as shown later in this Scenario 2b, with 11800 km of ERS), and 36% and 2% come
section. from slow and fast static charging respectively. However, it is
In Scenario 2b the 11800 km of ERS infrastructure from the worth remembering that some vehicles change their route to
previous scenario become available for passenger cars also, as use the ERS thus avoiding to wait at the fast charging stations.
the energy supply is moved from overhead catenary lines down This ERS attraction effect, together with the fact that passenger
to the road level. This implies a substantial reduction of the cars also retain the large batteries from the first scenario given
slow and fast static charging needs for passenger cars, in the the comparatively short length of ERS installed, motivates
same way as for heavy vehicles, down to 15.2% and 1.5% a 9.6% increase in the energy consumed by passenger cars
respectively of the corresponding value in Scenario 1 where compared to Scenario 1.
fast charging is the only charging infrastructure available.
Contrary to heavy vehicles, the total energy consumed by Table III compiles the distribution of the amount of energy
passenger cars in this scenario is 6% higher than in Scenario 1 delivered to the vehicles by both the fast chargers and the
despite the lower weight of the batteries onboard. This again ERS during the simulated time for each scenario. These
is a consequence of the way the model is implemented, in distributions depend mostly on the size of the battery on-board
which the time spent on charging implies a penalty, since and the length of the trips assigned to the different vehicles.

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

Table III: Distributions of the amount of energy delivered by fast chargers and ERS (kWh in the 10% reduced scale model).
The red dashed lines marks 90% of the total energy.
Passenger cars Heavy vehicles
Fast chargers ERS Fast chargers ERS
S1 1000 80
Number of vehicles [-]

Number of vehicles [-]


60

500 40

20

0 0
0 50 100 150 200 0 200 400
Energy [kWh] Energy [kWh]
S2a 1000 500 250
Number of vehicles [-]

Number of vehicles [-]

Number of vehicles [-]


400 200

300 150
500
200 100

100 50
0 0 0
0 50 100 150 200 0 10 20 30 40 0 500 1000
Energy [kWh] Energy [kWh] Energy [kWh]
S2b 600 10000 500 250
Number of vehicles [-]

Number of vehicles [-]

Number of vehicles [-]

Number of vehicles [-]


400 200
400
300 150
5000
200 100
200
100 50

0 0 0 0
0 5 10 15 20 0 200 400 0 10 20 30 40 0 500 1000
Energy [kWh] Energy [kWh] Energy [kWh] Energy [kWh]
S3a 100 300
Number of vehicles [-]

Number of vehicles [-]


Number of vehicles [-]

1000

200

500 50
100

0 0 0
0 50 100 150 200 0 200 400 0 500 1000 1500
Energy [kWh] Energy [kWh] Energy [kWh]

S3b 250 4000 100 300


Number of vehicles [-]
Number of vehicles [-]

Number of vehicles [-]

Number of vehicles [-]

200
3000
200
150
2000 50
100 100
50 1000

0 0 0 0
0 50 100 150 200 0 500 1000 1500
0 200 400 0 200 400
Energy [kWh] Energy [kWh]
Energy [kWh] Energy [kWh]

B. Static fast charging infrastructure needs the average energy delivered by all passenger car and heavy
vehicle charging stations is calculated, and those stations that
As described in Section II, all scenarios are initially simu- deliver less than 10% of the average are dismissed.
lated with a virtually unlimited number of charging connec- Once decided which of the 2506 stations are retained, the
tions for both heavy vehicles and passenger cars, at 2506 number of charging connections (plugs) available for each
different locations, corresponding to today’s fuel stations in vehicle type should be estimated, i.e. how many vehicles of
Sweden. Fig. 5 illustrates the use of the fast static charging each type can charge simultaneously at each station. In order
infrastructure in the different scenarios, by means of the total to do that, the total operation time of each station, defined
energy delivered by it aggregated in circles of 50 km diameter. as the time while at least one charging connection is in use,
However, not all of the 2506 stations are used in all is calculated from the simulation results. Then, the average
scenarios. Thus, to be able to compare the different scenarios, power output for each station is computed as the total energy
it is necessary to determine which stations are needed and delivered by that station divided by its operation time.
the number of charging connections they comprise. First, Finally, the number of charging connections is calculated

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

S1 S2a S2b S3a S3b


Figure 5: Total energy delivered by the static fast charging infrastructure aggregated in circles of 50 km diameter

dividing the average power output of each station by a described is estimated. The costs considered in this analysis
representative power value for that scenario and vehicle type, comprise:
estimated by the average charging power for those stations • the battery pack: assumed to 100 e/kW h according to
with only one charging event during the simulation. In this the cost development and forecast reported by Bloomberg
way, the representative power value accounts somehow for NEF [44]
the variability of charging conditions, state of charge at the • the electrified drivetrain, including the electrical machine,
beginning of the charging event and battery capacity among power electronic converter and mechanical transmission,
other factors. using the models presented in [45]–[47]
Table IV summarises the results obtained for each scenarios: • the static charging equipment, including the onboard slow
the number of stations initially used; out of those, the ones AC charger, and both the slow and fast static charging
that deliver more than 10% of the average energy and thus infrastructure, based on the same models as above
are retained; and the number of charging connections they • the dynamic charging equipment, including the onboard
comprise. DC/DC converter (using the same models as above) and
pick-up device (estimated based on actual prototype costs
Table IV: Fast charging infrastructure needs (10% reduced in 16 e/kWERS where kWERS represents the power that
scale model) can be taken from the ERS: 50 kW for B-segment cars,
Passenger cars Heavy vehicles 65 kW for C-segment cars, 90 kW for D-segment cars
Initial Retained Charging Initial Retained Charging and 350 kW for heavy vehicles in order to cover both
Connect. Connect. traction and charging), as well as the ERS infrastructure.
S1 1560 1290 2986 743 743 1262 The cost of the ERS infrastructure accounts for the feed-
S2a 1523 1240 2953 96 83 152 in stations, the power distribution cables along the road
S2b 255 204 409 91 80 149 and the actual power transferring hardware installed in the
S3a 1529 1255 2946 83 77 147 road. However, it does not include the cost corresponding
S3b 529 522 889 83 77 145 to bringing sufficient power capacity to the road, since it
would not vary significantly with the type of charging
In Scenario 1, with no ERS is available approximately infrastructure implemented, static or dynamic [48].
30000 charging points for passenger cars are needed in 1290
All vehicle equipment is depreciated over 5 years, the slow
stations (over 23 points per station on average), and 12500
and fast charging infrastructure over 10 years, and the ERS
for heavy vehicles in 743 stations (17 points per station on
infrastructure over 15 years.
average). As an example, these numbers can be related to the
824 manned service stations - with restrooms and a shop/café Only the four vehicle classes specified in Section IV-D
- in Sweden in 2018 [43], yielding 36 chargers for passenger are considered in the cost comparison. The cost to electrify
cars and 15 for heavy vehicles on average in each charging other vehicles such as urban buses or distribution trucks is
station. not included. The reason is that, even if these vehicles may
regularly drive long distances, they are confined to a limited
geographical area, which makes them less prompt to use and/or
C. Cost comparison
need high-power charging infrastructure on countryside roads.
In this section the cost corresponding to the electromobility Instead, urban buses are likely to have their own charging
related technology required in the five scenarios previously infrastructure, while distribution trucks would benefit more for

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

lower power and more abundant infrastructure for opportunity Of all the costs considered, those corresponding to the
charging while loading and unloading goods. These vehicle ERS related components are the most uncertain. On-board
types are nonetheless highly interesting when modelling the the vehicles these comprise the current collector device (the
effect of electrification in regional and local transport. actual contact to the road supply), a DC-DC converter interface
Fig. 6 shows the estimation of the aforementioned costs for between the road supply and the high-voltage traction system
the five scenarios described for all passenger cars and long- in the vehicle, and the necessary cabling and connections.
haul trucks in Sweden. Off-board the vehicles, the cost of the ERS infrastructure
accounts for the feed-in stations, the power distribution cables
along the road and the actual power transferring hardware
Yearly societal cost of the different scenarios
16 installed in the road [48]. The cost estimations for all these
components, except the DC-DC converter, are based on the
14
experience gathered from existing pilot tests [49], [50]. Since
12 ERS technology is currently under development, the number
10
of pilot tests and hence the data available is scarce. Therefore,
a sensitivity analysis has been conducted, assuming that the
8 on-board, respectively the off-board components’ cost changes
ERS
6 High Power Charger between 50% and 150% the initially assumed values.
Low Power Charger
Pick-Up
4 Sensitivity analysis: cost of ERS related technology on-board and off-board
ERS DC-DC
Drive 16 16
2

Electromobility Tech. Cost

Electromobility Tech. Cost


Aux. DC-DC
Battery
0 14 14
1 2a 2b 3a 3b
12 12
Scenario 1
Figure 6: Yearly cost of the electromobility related technology Scenario 2a
for the five studied scenarios. 10 10 Scenario 2b
Scenario 3a
Scenario 3b
8 8
0.6 0.8 1 1.2 1.4 0.6 0.8 1 1.2 1.4
From the cost results it is clear that the battery packs, in Cost multiplier ERS on-board comp. [-] Cost multiplier ERS off-board comp. [-]
particular the battery packs of the passenger cars, are the most
significant contributor to the analysed electromobility cost (as Figure 7: Yearly cost of the electromobility related technology
anticipated in previous studies [2]). Moving from Scenario 1 assuming that the cost of the on-board (left) and off-board
to 2a, ERS is widely available to all heavy vehicles which (right) ERS related components vary between 50% and 150%
then feature a much smaller battery pack. However, given the of the initially assumed values.
much lower number of heavy vehicles compared to passenger
Looking at the left plot in Figure 7, it can be seen that
cars, the cost reductions due to these smaller batteries are
the cost of the on-board components has a greater influence
scanty, and they are shadowed by the additional cost of the
in those scenarios in which passenger cars can use the ERS
ERS infrastructure.
(Scenarios 2b and 3b). However, compared to those scenarios
Scenario 2b, with extensive ERS available for all vehicle with no ERS for passenger cars, the total cost will be reduced
classes, presents the lowest cost of all. The cost reduction due as long as the added cost of the ERS related components
to the smaller battery packs in all vehicles is larger than the is lower than the savings associated with the smaller battery
additional costs corresponding to the ERS infrastructure and pack. On the other hand, the impact of the ERS infrastructure
the vehicle components needed to interface the ERS. cost is mostly noticeable in those scenarios with many ERS
In Scenarios 3a and 3b the same battery sizes as in Scenario kilometers (Scenarios 2a and 2b). Looking at the right plot
1 are considered, hence no cost reduction is to be expected for in Figure 7, in particular to Scenarios 2a and 3a, it can be
the battery packs. The cost of the limited ERS infrastructure deduced that the cost of the ERS infrastructure to electrify
considered in these scenarios (2200 km) is negligible com- 11800 km is similar to the savings resulting from reducing
pared to the cost of the batteries, but in order to be able to the battery pack of the trucks. Then, the variation of the ERS
use it, the vehicles must be equipped with a current collector infrastructure cost affects whether Scenario 2a is competitive
device (the sliding contact, or pick-up) and an isolated DC- or not compared to Scenario 1. Nevertheless, in all cases
DC converter to interface the ERS (shown in green and purple Scenario 2b remains the most advantageous one from a cost
in Fig. 6). The extra cost of this equipment is particularly perspective, even if the costs of both the on-board and the
noticeable in Scenario 3b, since it is installed in all 5 million off-board components increased to 150%.
passenger cars in Sweden. It is therefore obvious that adapting
all passenger cars in the country to use ERS only results in a VI. C ONCLUSIONS AND F UTURE W ORK
benefit if an equivalent reduction of the battery size (on a fleet In this article a new method using a Multi-Agent Trans-
average) can be achieved, as it is the case in Scenario 2b. port Simulation (MATSim) model to assess the influence of

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

different parameters in the need of charging infrastructure over an ERS (with the corresponding increased energy
is proposed. To illustrate the method, five potential future consumption) if that results on an overall time saving.
scenarios for the electrification of long-distance transport in In addition, vehicles driving on an ERS will always
Sweden are analysed. Each of the scenarios depicts a different retrieve energy from it regardless of whether they have
configuration of the charging infrastructure, with a certain enough energy in the battery to complete the trip. This
combination of static and dynamic charging possibilities. For could be solved by introducing different pricing strategies
each of the scenarios all relevant energy flows are obtained, depending on the type of charging infrastructure used.
and based on these results, the total energy consumed and the • the battery sizes for the different scenarios and vehicle
charging infrastructure needs are assessed. Finally, the cost classes, which have a direct impact on the need for charg-
associated with the electromobility related technology needed ing infrastructure and the potential energy consumption
in each scenario is estimated and all scenarios are compared. savings associated to battery downsizing. However, it is
From the results, heavy vehicles present the lowest energy difficult to foresee how smaller a battery the users are
consumption for Scenarios 2a and 2b, in which ERS is widely prepared to accept given a certain ERS coverage available
available and thus they feature a significantly smaller (and in the country.
lighter) battery pack. However, passenger cars consume more • the assumption that the initial SOC for all vehicles is
energy in those scenarios in which they have access to ERS, 100%, i.e. all vehicles start the day with a fully charged
since due to the model assumptions, they choose to drive battery. Although it is reasonable to expect that, given that
on roads equipped with an ERS even if it results in longer the model focuses on long-distance transport, the users
distances, as long as the total travel time is shorter. will prepare for the trip, once the whole fleet is electrified
When assessing the charging infrastructure needs, it is there may be users without access to slow charging
interesting to realise how heavy vehicles drive mostly on the infrastructure. The initial SOC of the fleet together with
motorways, and the number of supplementary fast charging the battery size have a significant impact on the needs for
station needed is almost the same in Scenarios 2 (a and in-travel charging.
b) with 11800 km of ERS and in Scenarios 3 (a and b) • neither the geographical location of the fast static charg-
with 2200 km ERS - although in the latter with a much ing stations nor the selection of road segments equipped
larger battery pack. For passenger cars this effect is also with ERS are optimized in any way in this study. The first
present, although less pronounced, since their driving routes assumes the same locations as today’s petrol stations and
are much more geographically spread. Furthermore, it is worth the latter is based on the speed limitation on the road. A
highlighting the huge number of high power charging points different placement of the charging infrastructure, taking
required to fulfil the long-distance transport needs when no into account aspects such as the availability of the electric
ERS is available (Scenario 1): approximately 30000 charging power grid or the expected traffic volumes may produce
points for passenger cars and 12500 for heavy vehicles. These different results.
numbers can be related to the 824 manned service stations - • the lack of logged data to validate the results of the
with toilets and a shop/café - in Sweden in 2018 [43], yielding utilization of charging stations. Vehicle usage patterns
36 and 15 chargers for passenger cars and heavy vehicles and charging infrastructure usage patterns are to a certain
respectively on average in each, if those stations were to be extent affected by the charging infrastructure availability.
the only places used for fast charging. For that reason, logged driving data from combustion
The cost comparison shows that the battery pack (especially vehicles cannot be directly used to assess EV charging
that of passenger cars) is the largest single contributor to the infrastructure needs. However, logged data from EVs
total electrification cost. Thus, Scenario 2b - in which ERS (when available in the future) will reflect the influence
is widely available for passenger cars so they can benefit of the existing charging infrastructure at the time it
from a substantial reduction of the battery size - is the most was logged, and thus can be used in Machine Learning
favourable one from a cost perspective, despite the extra cost schemes (e.g. to train neural networks) to predict the
for the ERS infrastructure itself and the additional vehicle impact of future charging infrastructure deployment.
components needed to interface the ERS.
When developing both the transport (MATSim) and the Likewise, for the cost models:
cost models it is necessary to make certain assumptions that, • the assumption of 100 e/kW h for the cost of the battery
although reasonable, may have a strong influence in the final and 5 year depreciation time, perceived as conservative
results. Revising such assumptions and investigating their when the project started, seems now almost pessimistic
impact by means of e.g. sensitivity analyses is paramount to given the rapid evolution of battery technology and may
ensure the validity of the model’s outcome. In that respect, need to be revised in the future, given the importance of
the following aspects of the transport model (MATSim) have the battery packs in the cost estimation
been identified for future research: • for the same reason, the battery sizes assumed for the
• the zero marginal cost (in utility points) associated to different scenarios and vehicle classes should be reconsid-
the use of the ERS infrastructure compared to fast ered, since they have a direct impact on the cost savings
static charging leads to vehicles driving longer distances related to battery downsizing in those scenarios with ERS

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

Finally, it is worth mentioning that the presented approach, [20] M. Debbou and F. Colet, “Inductive wireless power transfer for electric
used in this article to analyse five potential future scenarios in vehicle dynamic charging,” in 2016 IEEE PELS Workshop on Emerging
Technologies: Wireless Power Transfer (WoW), 2016, pp. 118–122.
Sweden, can be applied anywhere else provided that there are [21] B. Regensburger, S. Sinha, A. Kumar, J. Vance, Z. Popovic, and K. K.
enough input data available, reliable and sufficiently detailed Afridi, “Kilowatt-scale large air-gap multi-modular capacitive wireless
to create a representative synthetic population. power transfer system for electric vehicle charging,” in 2018 IEEE
Applied Power Electronics Conference and Exposition (APEC), 2018,
pp. 666–671.
R EFERENCES [22] Siemens, “eHighway - Electrification of road freight transport,” accessed
2020-07-13. [Online]. Available: https://www.mobility.siemens.com/
[1] G. Crabtree, “The coming electric vehicle transformation,” Science, global/en/portfolio/road/ehighway.html
vol. 366, no. 6464, pp. 422–424, 2019. [Online]. Available: [23] T. Tajima, W. Noguchi, and T. Aruga, “Study of a dynamic charging
https://science.sciencemag.org/content/366/6464/422 system for achievement of unlimited cruising range in ev,” in SAE
[2] P. Fyhr, G. Domingues, M. Andersson, F. J. Márquez-Fernández, Technical Paper. SAE International, 04 2015.
H. Bängtsson, and M. Alaküla, “Electric roads: Reducing the societal [24] T. Tajima and H. Tanaka, “Study of 450-kw ultra power dynamic
cost of automotive electrification,” in 2017 IEEE Transportation Elec- charging system,” in SAE Technical Paper. SAE International, 04 2018.
trification Conference and Expo (ITEC), 2017, pp. 773–778. [25] Elways, “Electric roads for cars, trucks and buses - unlimited
[3] B. J. Limb, Z. D. Asher, T. H. Bradley, E. Sproul, D. A. Trinko, driving with small batteries,” accessed 2020-07-13. [Online]. Available:
B. Crabb, R. Zane, and J. C. Quinn, “Economic viability and environ- https://elways.se/
mental impact of in-motion wireless power transfer,” IEEE Transactions [26] P. Veyrunes, P. Duprat, and J. Hourtane, “Ground-level feeding systems
on Transportation Electrification, vol. 5, no. 1, pp. 135–146, 2019. from rail to road,” in 2017 IEEE Transportation Electrification Confer-
[4] I. Malmgren, “Quantifying the societal benefits of electric vehicles,” ence and Expo, Asia-Pacific (ITEC Asia-Pacific), 2017, pp. 1–4.
World Electric Vehicle Journal, vol. 8, no. 4, pp. 996–1007, 2016. [27] Elonroad, “Make it change - Make it charge,” accessed 2020-07-13.
[Online]. Available: https://www.mdpi.com/2032-6653/8/4/996 [Online]. Available: https://elonroad.com/
[5] S. Wappelhorst, P. Mock, and Z. Yang, “Using vehicle taxation policy to [28] M. Khalesian and M. R. Delavar, “A multi-agent based traffic net-
lower transport emissions: An overview for passenger cars in europe,” work microsimulation using spatio-temporal gis,” in The International
The International Council on Clean Transportation (ICCT), 2018. Archives of the Photogrammetry, Remote Sensing and Spatial Informa-
[6] M. Alaküla and F. J. Márquez-Fernández, “Conductive electric road tion Sciences, vol. XXXVII, 2008.
systems (ers) in sweden -an overview of development and testing [29] J. Bischoff and M. Maciejewski, “Agent-based simulation of electric
activities,” in 2019 JSAE Annual Congress (Spring), 2019, pp. 1–8. taxicab fleets,” Transportation Research Procedia, vol. 4, pp. 191 –
[7] A. Zubaryeva, P. Dilara, and L. Maineri, “Publicly funded research, 198, 2014, sustainable Mobility in Metropolitan Regions. mobil.TUM
development and demonstration projects on electric and plug-in vehicles 2014. International Scientific Conference on Mobility and Transport.
in europe - update,” Publications Office of the European Union, EUR - Conference Proceedings.
Scientific and Technical Research Reports, 2015. [30] D. Charypar, A. Horni, B. Kickhöfer, and K. Nagel, “A closer look
[8] S. Boyd and D. Howell, “An Overview of the Hybrid and Electric at scoring,” in Multi-Agent Transport Simulation MATSim, A. Horni,
Systems R&D at the U.S.–DOE (fiscal year 2015–2016),” World K. Nagel, and K. Axhausen, Eds. Ubiquity Press, 2016.
Electric Vehicle Journal, vol. 8, no. 2, pp. 461–472, 2016. [Online]. [31] K. Nagel and A. Horni, “More about configuring matsim,” in Multi-
Available: https://www.mdpi.com/2032-6653/8/2/461 Agent Transport Simulation MATSim, A. Horni, K. Nagel, and K. Ax-
[9] F. J. Márquez-Fernández, G. Domingues-Olavarría, L. Lindgren, and hausen, Eds. Ubiquity Press, 2016.
M. Alaküla, “Electric roads: The importance of sharing the infrastructure [32] J. Bischoff, F. J. Márquez-Fernández, G. Domingues-Olavarría, M. Ma-
among different vehicle types,” in 2017 IEEE Transportation Electrifi- ciejewski, and K. Nagel, “Impacts of vehicle fleet electrification in
cation Conference and Expo, Asia-Pacific (ITEC Asia-Pacific), 2017, pp. sweden – a simulation-based assessment of long-distance trips,” in 2019
1–6. 6th International Conference on Models and Technologies for Intelligent
[10] M. Taljegård, “Electrification of road transportation - implications for Transportation Systems (MT-ITS), 2019, pp. 1–7.
the electricity system,” Ph.D. dissertation, Chalmers University of Tech- [33] M. Bergquist, V. Bernhardsson, and E. Rosklint, “Representation of the
nology, 8 2019. Swedish transport and logistics system in SAMGODS v. 1.1,” Swedish
[11] E. Figenbaum and S. Nordbakke, “Battery electric vehicle user experi- Transport Administration, Tech. Rep., 2016. [Online]. Available:
ences in norway’s maturing market,” Institute of Transport Economics, http://vti.diva-portal.org/smash/get/diva2:1034226/FULLTEXT01.pdf
Norwegian Centre for Transport Research, 2019. [34] C. Persson and S. Jiang, “Framtagande av kalibreringsmål för SAM-
[12] E. Lorentzen, P. Haugneland, C. Bu, and E.Hauge, “Charging infras- PERS regionala modeller (in Swedish),” Swedish Transport Adminis-
tructure experiences in norway - the worlds most advanced ev market,” tration, Tech. Rep., 2015. [Online]. Available: https://www.trafikverket.
in 2017 Electric Vehicle Symposium EVS30, 2017, pp. 1–11. se/contentassets/c700bc932efd44a4b104b2cbb2a0e79e/kalibreringsmal_
[13] D. Hall and N. Lutsey, “Emerging best practices for electric vehicle 160314.pdf
charging infrastructure,” The International Council on Clean Transporta- [35] “Statistics Sweden,” accessed 2020-07-13. [Online]. Available: https:
tion (ICCT), White paper, 2017. //www.scb.se/en/
[14] J. Xiong, K. Zhang, Y. Guo, and W. Su, “Investigate the impacts of [36] “OpenStreetMap,” accessed 2020-07-13. [Online]. Available: https:
pev charging facilities on integrated electric distribution system and //www.openstreetmap.org/
electrified transportation system,” IEEE Transactions on Transportation [37] “Copernicus - Land Monitoring Service,” accessed 2020-07-13.
Electrification, vol. 1, no. 2, pp. 178–187, 2015. [Online]. Available: https://land.copernicus.eu/
[15] L. Calearo, A. Thingvad, K. Suzuki, and M. Marinelli, “Grid loading [38] Working Party on Pollution and Energy, “Proposal for a new global
due to ev charging profiles based on pseudo-real driving pattern and user technical regulation on the Worldwide harmonized Light vehicles Test
behavior,” IEEE Transactions on Transportation Electrification, vol. 5, Procedure (WLTP),” United Nations Economic Comision for Eurpe
no. 3, pp. 683–694, 2019. - UNECE, 2014. [Online]. Available: http://www.unece.org/fileadmin/
[16] “Multi-Agent Transport Simulation (MATSim),” accessed 2020-07-13. DAM/trans/doc/2014/wp29/ECE-TRANS-WP29-2014-027e.pdf
[Online]. Available: http://matsim.org [39] F. J. Márquez-Fernández, “Electric traction machine design for an e-rwd
[17] A. Horni, K. Nagel, and K. Axhausen, Eds., Multi-Agent Transport unit,” Ph.D. dissertation, Lund University, 4 2014.
Simulation MATSim. London: Ubiquity Press, Aug 2016. [40] “A Better RoutePlanner - Tesla battery charging data from
[18] F. J. Márquez-Fernández, J. Bischoff, G. Domingues-Olavarría, and 801 cars,” accessed 2020-07-13. [Online]. Available: https:
M. Alaküla, “Using multi-agent transport simulations to assess the //forum.abetterrouteplanner.com/blogs/entry/6- tesla- battery- charging-
impact of ev charging infrastructure deployment,” in 2019 IEEE Trans- data-from-801-cars
portation Electrification Conference and Expo (ITEC), 2019, pp. 1–6. [41] Swedish Energy Agency (Energimyndigheten), “Transportsektorn
[19] A. Ahmad, M. S. Alam, and R. Chabaan, “A comprehensive review of energianvändning (in Swedish),” accessed 2020-07-13. [Online].
wireless charging technologies for electric vehicles,” IEEE Transactions Available: http://www.energimyndigheten.se/statistik/den- officiella-
on Transportation Electrification, vol. 4, no. 1, pp. 38–63, 2018. statistiken/statistikprodukter/transportsektorns-energianvandning/

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TTE.2021.3065144, IEEE
Transactions on Transportation Electrification

[42] P. Ashkrof, G. Homem de Almeida Correia, and B. van Arem, power electronics converters,” IEEE Transactions on Power Electronics,
“Analysis of the effect of charging needs on battery electric vehicle vol. 32, no. 11, pp. 8681–8692, 2017.
drivers’ route choice behaviour: A case study in the netherlands,” [46] G. Domingues-Olavarríía, F. J. Márquez-Fernández, P. Fyhr, A. Reinap,
Transportation Research Part D: Transport and Environment, vol. 78, M. Andersson, and M. Alaküla, “Optimization of electric powertrains
p. 102206, 2020. [Online]. Available: http://www.sciencedirect.com/ based on scalable cost and performance models,” IEEE Transactions on
science/article/pii/S1361920919309757 Industry Applications, vol. 55, no. 1, pp. 751–764, 2019.
[43] Svenska Petroleum & Biodrivmedel Institutet, “Försäljningställen [47] P. Fyhr, “Electromobility - materials and manufacturing economics,”
(in Swedish),” accessed 2020-07-13. [Online]. Available: https: Ph.D. dissertation, Lund University, 05 2018.
//spbi.se/statistik/forsaljningsstallen/ [48] G. Domingues-Olavarría, “Modeling, optimization and analysis of elec-
[44] Bloomberg New Energy Finance, “Battery pack prices fall as market tromobility systems,” Ph.D. dissertation, Lund University, 11 2018.
ramps up with market average at $156/kwh in 2019,” accessed 2020- [49] EVolution Road project consortium, “Evolution road demonstration
07-13. [Online]. Available: https://about.bnef.com/blog/battery-pack- project,” accessed 2021-01-19. [Online]. Available: https://www.
prices- fall- as- market- ramps- up- with- market- average- at- 156- kwh- in- evolutionroad.se/en/
2019/ [50] D. Bateman, D. Leal, S. Reeves, M. Emre, L. Stark, F. Ognissanto,
[45] G. Domingues-Olavarría, P. Fyhr, A. Reinap, M. Andersson, and R. Myers, and M.Lamb, “Electric road systems: a solution for the
M. Alaküla, “From chip to converter: A complete cost model for future?” World Road Association (PIARC), 2018.

2332-7782 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF NEW MEXICO. Downloaded on May 16,2021 at 16:45:55 UTC from IEEE Xplore. Restrictions apply.

You might also like