You are on page 1of 13

Applied Energy 268 (2020) 114961

Contents lists available at ScienceDirect

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

A scenario-based study on the impacts of electric vehicles on energy T


consumption and sustainability in Alberta
Ganesh Doluweeraa, Fabian Hahnb, Joule Bergersonc, Marco Prucknerb,

a
Canadian Energy Research Institute, 3512 33 St NW, Calgary, AB T2L 2A6, Canada
b
Energy Informatics, Friedrich-Alexander University Erlangen-Nürnberg, Martensstraße 3, 91058 Erlangen, Germany
c
Dept. of Chemical and Petroleum Engineering, Schulich School of Engineering, The University of Calgary, 2500 University Dr. NW, Calgary, AB T2N 1N4, Canada

HIGHLIGHTS

• Develop a simulation model to assess energy and emissions impacts of the use of EVs.
• Use the model to assess the impacts of EVs in the Canadian province of Alberta.
• The electricity demand increases due to the use of EVs in Alberta is marginal.
• Cleaner electricity generation and EVs reduce Alberta’s GHG emissions significantly.
• Controlled EV charging can avoid potential increases in peak electricity demand.

ARTICLE INFO ABSTRACT

Keywords: Canada has committed to the Paris Agreement and aims to reduce its emissions to 30 percent below 2005 levels
Transportation electrification by 2030. One of the Canadian provinces with a significant potential to reduce greenhouse gas emissions in
Carbon management Alberta due to its fossil fuel extraction, electricity generation from coal-fired power plants, and the transport
Electric vehicle charging sector. In this paper, we examine the energy and greenhouse emissions impacts of electric vehicles on Alberta’s
Electricity system operations
electricity and transport systems. We utilize a hybrid simulation model and develop a new component to model
electric vehicle fleets. The adapted model is used to investigate the impact of six scenarios with varying as-
sumptions about electric vehicle penetration rates and charging strategies. The analysis shows that the adap-
tation of electric vehicles can contribute to Alberta’s 2030 emissions reduction target of 30% below 2005 levels.
About a third of Alberta’s 2030 greenhouse emissions reduction target can be achieved through the measures to
reduce greenhouse emissions from electricity generation and adaptation of electric vehicles for passenger
transportation.

1. Introduction offers significant emissions reduction potential. Coal-fired electricity


generation is a significant source of GHG emissions in Alberta. In 2016,
Canada is in the group of more than 90 countries that have ratified 50% of Alberta's electricity was produced by coal-fired power plants. In
the Paris agreement and committed to act to reduce greenhouse gas 2015, the government of Alberta announced their Climate Leadership
(GHG) emissions. Canada has set a GHG emission reduction target of Plan, that includes a transition to cleaner electricity generation system,
30% below 2005 levels by 2030 [1]. The federal government of Canada capping oil sands emissions and reducing methane emissions [4]. A key
is currently implementing policies and programs to achieve these component of the plan is a commitment to phasing out coal-fired
emissions reduction targets. Alberta is a province in the western region electricity generation by 2030. Furthermore, a renewable electricity
of Canada. The current population in Alberta is 4.4 million, which is program was initiated to achieve a share of 30% of electricity from
about 12% of the total Canadian population [2]. Of all Canadian pro- renewable energy sources by 2030. These commitments will lead to a
vinces, Alberta has the highest GHG emissions. In 2017, Alberta emitted fundamental shift in how electricity is produced in the province in the
273 million tonnes of carbon dioxide equivalent (MtCO2eq) of GHG, future. However, further GHG emissions reductions will be required in
which was 38% of total Canadian GHG emissions [3]. As such, Alberta the province for Alberta to contribute its proportionate share towards


Corresponding author.
E-mail address: marco.pruckner@fau.de (M. Pruckner).

https://doi.org/10.1016/j.apenergy.2020.114961
Received 12 January 2020; Received in revised form 30 March 2020; Accepted 3 April 2020
0306-2619/ © 2020 Elsevier Ltd. All rights reserved.
G. Doluweera, et al. Applied Energy 268 (2020) 114961

the 2030 federal target. 1.1. Related work


Moving people and goods is an essential energy-consuming service
for society to function. Energy demand for transportation has grown Assessments of technical, economic, and environmental implica-
over the past decades in most countries. For example, from 2000 to tions of transportation electrification is currently an active area of re-
2016, the energy demand for transportation in Canada increased by search. Researchers have employed different analytical methods to as-
about 15% [5]. Currently, fuels used for transportation are pre- sess the impacts of transportation electrification in various
dominantly fossil fuels. Consequently, the transportation sector is a jurisdictions. Several studies show that there can be unintended con-
significant contributor to Canada’s GHG emissions. In 2017, the trans- sequences of EV adaptation in terms of GHG emissions and impacts on
portation sector was the second-largest contributor to Canada’s GHG the electricity sector.
emissions and produced 174.3 MtCO2eq [3]. Due to the continued Li et al. [11] study different scenarios for the deployment of EVs in
growth of transportation service demand and fossil fuel dominance of China to assess the energy system and environmental impacts. The
the fuel mix, reducing GHG emissions from transportation is challen- authors utilize an integrated framework to model the transportation
ging. One technology option currently gaining interest is replacing system and power system operations. The transportation model calcu-
fossil fuel fired vehicles with electric vehicles (EV). EVs do not produce lates the electricity demand of EVs and predicts the availability of the
any GHGs at the point of use. Instead, the emissions are moved from the EVs connected with the grid. The power system model uses clustered-
tailpipes of vehicles to the electricity sector. If electricity is produced integer unit commitment for the power dispatch. The main findings are
with low GHG emissive sources (e.g., renewable energy, nuclear that deploying EVs in China moves transportation energy services from
power), overall GHG emissions reductions can be achieved. Further- gasoline to coal-fired power generation, which leads to higher GHG
more, the energy efficiency of EVs is approximately three times higher emissions. The result remains the same even under a smart, coordinated
than fossil fuel fired internal combustion engine (ICE) vehicles, low- charging scenario as it tends to feed EVs with electricity generated by
ering the per-unit distance traveled energy demand. cheap coal-fired power plants. Nevertheless, coordinated charging re-
In Alberta, the transportation sector produced 30.7 MtCO2eq of duces the peak load arising from EV charging and facilitating renewable
GHG in 2017, which was 11% of the province’s total emissions [3]. energy sources generation. Huo et al. [12] also study the impacts of EV
Passenger transportation contributed 40% of Alberta’s transportation adaptation in China and make similar conclusions. Apart from GHG
sector GHG emissions, the remainder coming from freight transporta- emissions, Huo et al. also investigate SO2 and NOx emissions of EVs in
tion. A transition to EVs for passenger transportation along with the China. Results show that the adaptation of EVs does not lead to higher
GHG emissions reduction plans for the Alberta electricity system shows benefits in reducing GHG emissions as long as coal-fired power plants
promise as a potentially viable path to achieving emissions reductions mainly generate electricity.
in Alberta. However, the timing and method of this transition are still De Tena and Pregger [13] study the environmental benefit of EVs in
uncertain and therefore require further investigation. Germany utilizing different scenarios of EV penetration and renewable
A survey among Canadians in 2015 shows that much of the popu- power generation. The scenario-based analysis is conducted using an
lation has an open mind about electric mobility [6]. In the country, a electricity system dispatch model, including the interchange of elec-
growing interest in EVs is observed from recent vehicle sales statistics. tricity between European regions. The study shows that an optimized
Light-duty EV sales, including battery-powered electric vehicles (BEVs) charging strategy can lead to 3.5 to 4.5 GW reduction of the residual
and plug-in hybrid electric vehicles (PHEVs) in Canada, has seen rapid peak load in Germany.
growth in recent years. Throughout 2011 to 2018, sales of light-duty Hedegaard et al. [14] assess the impacts of EV adaptation in Den-
EVs in Canada has grown by 41 times with the passenger vehicle market mark, Finland, Germany, Norway, and Sweden. The integrated power
share of EVs at a mere 0.03% in 2011, rising to 2.1% in 2018 [7]. In and road transport system is modeled in a deterministic partial equili-
Alberta, the total number of EVs is still less than 0.05% of total pas- brium model called Balmorel. Their results show that EVs can reduce the
senger vehicles. However, EV sales in Alberta show signs of scaling by need for new coal and natural gas power plants and can facilitate sig-
more than 100% year to year over three years, from 2015 to 2018 [7]. nificantly increased wind power investments. The impacts of EVs sig-
The two major cities in Alberta, the City of Calgary and the City of nificantly vary by country, and the GHG emissions reduction in 2030
Edmonton, are currently developing EV strategies to support the in- differs considerably by country in the range of 3–28%.
tegration of EVs and to GHG emissions reductions in the transport Bellocchi et al. [15] study the positive interactions between electric
sector [8,9]. Furthermore, the government of Canada, through the vehicles and renewable energy sources in energy scenarios for GHG
Electric Vehicle and Alternative Fuel Infrastructure Initiative, is cur- reductions in Italy. For modeling purposes, the authors use the En-
rently supporting the establishment of a nationwide EV charging net- ergyPLAN integrated energy system analysis tool and define different
work [10]. These developments can facilitate the adoption of EVs for scenarios that are compared to a 2015 reference case where fossil fuels
passenger transportation in Alberta and Canada. are the dominant transportation fuel. The main conclusion is that with
While EVs reduce GHG emissions in the transport sector, they create the lack of large-scale energy storage, the overall GHG emissions re-
additional emissions in the electricity sector because they increase ductions will be limited to about 20% below the reference case.
electricity demand. EVs are more efficient than ICE vehicles in terms of Jochem et al. [16] use four different assessment methods (average
converting onboard energy into motive energy services. However, the annual electricity mix, average time-dependent electricity mix, mar-
overall GHG emissions associated with operating EVs are influenced by ginal electricity mix, and balancing zero emissions) to assess CO2
the fuel mix and efficiency of the generation system that produced the emissions associated with EV adaptation in Germany. The authors use
electricity consumed by the EVs. Furthermore, the electricity system an energy system optimization model (PERSEUS-NET-TS) to implement
may need additional infrastructure such as power plants to satisfy the the four assessment methods. The results show that different assessment
increased electricity demand when EV penetration increases to a large methods lead to conflicting results. Depending on the assessment
portion of the fleet. These factors may add uncertainty into the net GHG method employed, the GHG emissions from EV usage varies from zero
emissions reduction benefits that can be achieved by adopting EVs for to 110 g CO2/km.
passenger transportation. Energy systems-level assessments are re- McCarthy et al. [17] study the impact of electric vehicles on the
quired to gain insights into the net benefits as well as the unintended energy system of California. The authors use a spreadsheet-based dis-
consequences of EV adaptation. patch model and classify generation units in California by technology.

2
G. Doluweera, et al. Applied Energy 268 (2020) 114961

For the evaluation, different future pathways in energy system devel- investigate the impact of EVs on the electricity sector, we further de-
opment and EV penetration are taken into account. Results show that velop an existing hybrid simulation model [25] developed for the
GHG emissions in the transport sector can be drastically reduced in electricity system of Germany. The model was also applied for several
comparison to a baseline (gasoline) scenario. Weis et al. [18] assessed case studies concerning the impact of EVs [26] and the grid integration
the emissions and cost implications of EV charging in the U.S. PJM of stationary battery energy storage systems [27]. This model is a de-
Interconnection and concluded that the emissions reduction potential is cision support tool and has been used to inform decisions related to the
driven by the nature of the power plant fleet in the electricity system at future development of the federal state of Bavaria during the phase-out
which the EVs are charged. Several recent studies showed that de- of nuclear energy and the increasing electricity generation from re-
pending on the nature of the EVs are charged, the electrical systems newable energy sources in Germany. We integrate a newly developed
may experience exacerbated peak demand conditions, leading to chal- EV electricity demand simulation model into the existing modeling
lenges in maintaining electric power system reliability [19,20]. At the framework. We create scenarios to represent different EV penetration
same time, some assessments provide evidence of EV charging miti- rates and charging methods. We estimate Provincial GHG emissions,
gating peak demand conditions [21]. annual electricity generation supply/demand balance, and electricity
All studies that we reviewed above focus on the impacts in the use demand profile under each scenario.
phase of EVs. Several life cycle assessment (LCA) studies assess the The remainder of the paper is organized as follows. Section 2 pre-
environmental impacts of EVs and ICE vehicles from vehicle manu- sents an overview of the simulation model framework, including the
facturing to their end of life. For example, Tagliaferri et al. [22] conduct newly developed vehicle fleet model and the scenarios developed in this
an LCA study covering several EV, hybrid vehicles, and ICE vehicles. paper. In addition to the main article, we also provide a supplementary
The assessment results show that due to the higher amount of material information section that explains the structure, the logic of our simu-
processing, the EVs have higher environmental impacts in the produc- lation framework, and the validation for Alberta's electricity system
tion phase than an ICE vehicle. The highest amount of environmental model in detail. Section 3 discusses the simulation results. Finally,
impacts are associated with the battery manufacturing phase of EVs. Section 4 concludes the paper and provides an outlook on future work.
However, in terms of GHG emissions, the higher environmental benefits
in the use phase of EVs outweigh the adverse impacts in the production 2. Materials and methods
phase resulting in lower life cycle GHG emissions than ICE vehicles. The
life cycle GHG impacts of EVs could vary due to local conditions. Onat The simulation model developed in this paper extends a hybrid si-
et al. [23] develop a state-based comparative carbon and energy foot- mulation model previously developed by Pruckner [25] for the elec-
print analysis in the United States. The study finds EV to be the least tricity system of Germany. We calibrate the hybrid simulation model to
carbon-intensive vehicle option only in 24 out of the 50 states. Driving simulate the operations of the electricity system of Alberta. The model,
patterns and electricity generation mix in other states lead to the lower which was originally developed to simulate electricity system opera-
GHG emissions benefits of EVs in other states on a life cycle basis. tions, is further enhanced to add the energy demands of the transpor-
As we can see from these cited publications, the impacts of adap- tation system. The simulation model is suitable to represent the elec-
tation of EVs depend on the nature of the electric power system and tricity system of an entire jurisdiction over a long time-horizons (over a
user-specific characteristics such as the charging method employed. decade) while maintaining the system dynamics at a temporal resolu-
Energy system modeling can be utilized to capture the systems level tion of one hour. The simulation model is implemented using the
complexities and to provide reasonably accurate estimates of the en- commercial simulation software AnyLogic 7® and uses a hybrid simu-
vironmental impacts of EV adaptation. Furthermore, a system-level lation approach by combining two simulation paradigms [28]:
modeling study can provide insights into unintended consequences of
changes to the energy systems. In the literature, researchers have de- • System Dynamics: Modeling temporally continuous power flows
veloped energy system models of varying levels of complexity to assess (e.g., the electricity generation of power plants and the charging of
the impacts of EV adaptation. The work by Jochem et al. [16] shows EVs)
that the assessment methods used can also lead to conflicting results. • Discrete Event Simulation: Modeling of events which change para-
Decisions such as the adaptation of mitigation measures to reduce GHG meters at discrete points in time (e.g., discrete modeling of control
emissions from the transportation sector have far-reaching societal level decisions, operating state transitions of conventional power plants
consequences. Therefore, to gain insights into the impacts of mitigation and stochastically changing parameters such as starting of EV
measures, it is important to use assessment methods that are robust charging)
enough to capture the systems-level dynamics but also transparent and
auditable. The model is used to simulate six scenarios that represent different
The impact of EVs on Alberta in terms of technical performance, levels of EV penetration and charging strategies. Simulating different
economics, and environmental impacts will be unique among jurisdic- scenarios enables the identification of potential outcomes, risks, and
tions around the world. In a recent commentary, Kennedy [24] suggests miscalculations at an early stage. The simulation model we developed
that the electricity sector must meet a threshold GHG intensity of 600 for this analysis is depicted in Fig. 1. The model has several sub-models
gCO2eq/kWh for an EV to make lower GHG emissions than a fossil fuel- that represent different system components. The complete model details
fired ICE engine. As of 2017, the GHG intensity of the electricity system and development of the transportation system model are presented in
of Alberta was 750 gCO2eq/kWh [3]. However, the programs and po- the supplementary information section. The remainder of this section
licies that have been put in place could lead to overall GHG emissions presents a brief description of the simulation model.
reductions in Alberta. A study to assess these impacts must consider the
changing conditions in the electricity sector and the transportation 2.1. Electricity system model
sector. The analysis must be conducted transparently to facilitate the
interpretations of the results. The electricity system model reduces the whole electricity system to
a single node to which all generators and demands are connected.
1.2. Scope and objectives Existing and future power plants are simulated by taking into account
the fuel consumed by the units and conversion efficiencies (where ap-
This paper develops a scenario-based simulation study to evaluate plicable). The future development of the electricity generation system is
the energy and environmental impact of EVs on the electricity sector modeled by using a forecast made by the Alberta Electricity System
and the transport sector of the Canadian province of Alberta. To Operator (AESO) [29] and government policies such as the coal phase-

3
G. Doluweera, et al. Applied Energy 268 (2020) 114961

Fig. 1. Overview of the simulation model.

out, the renewable energy target, and carbon pricing system. Electricity observed data accurately.
production potential of variable renewable power plants (i.e., wind,
solar, and run of the river hydropower plants) is constrained by re-
2.2. Vehicle fleet model
source availability. Renewable energy availability is modeled using a
combination of time series data of historical production levels of ex-
A primary enhancement made in this paper to the original simula-
isting units and estimates made by CERI [30].
tion model developed by Pruckner [25] is the inclusion of a vehicle fleet
At every simulation time step (in this analysis, we set a simulation
module that estimates transportation energy demand and EV charging
time step of 1 h), the available generating units are dispatched to satisfy
demand curves. This vehicle fleet model is based on vehicle aggrega-
the electricity demand. The dispatch of generating units is based on an
tion, where the behavior of individual EVs are tracked and then ag-
economic merit order. The merit order is established for every time step
gregated to construct EV fleet-level energy demand profiles. In the lit-
by ordering the generating units by the ascending marginal costs. The
erature, many different modeling approaches are used to model the
marginal costs of each power plant depend on fuel price, carbon price,
charging behavior of an EV fleet based on the vehicle aggregation ap-
power plant efficiency, fuel GHG emissions intensity, and the variable
proach. Some examples of these approaches include use of empirical
operating and maintenance costs. Transmission and distribution losses
data [31], deterministic approach [32,33], stochastic modeling ap-
are estimates at every time step and added to the demand. When esti-
proach [19,34–37], cloud models [38], and physical modeling methods
mating the GHG emissions, fuel specific emissions intensities are used.
based on diffusion theory [39]. We develop and integrate a stochastic
Note that emissions associated with imported electricity are excluded
vehicle aggregation sub-model into the hybrid simulation model (full
from Alberta’s power sector emissions. Emissions associated with im-
details of the model are presented in the supplementary information
ported electricity remains in the emissions accounts of the exporting
section). Combining the agent-based modeling approach developed by
jurisdictions. Following the same argument, GHG emissions associated
Pouladi et al. [35] with a discrete event simulation, we generate a
with exported electricity is included in Alberta’s GHG accounts.
comprehensive EV fleet charging model suitable for the simulation of a
The power plants are dispatched based on the estimated net de-
large fleet for a long simulation duration such as 15 years. We simulate
mand1 and the generating unit merit order. Must-run power plants in
a growing EV fleet with the potential to reach 1 million EVs. To reduce
Alberta consist mainly of natural gas-fired cogeneration plants. Pro-
the computational effort, we implement a simplification. Only 10,000
duction levels of those cogeneration units are kept at historic levels.
EVs are modeled as single agents.
Power plants are dispatched power balance of the electricity system2.
The vehicle fleet model consists of five sub-models (Fig. 1). The
The imports and exports are used to balance the system through the
vehicle stock sub-model determines the size of the passenger vehicle
interties with neighboring jurisdictions. The model is validated by si-
fleet in each year of the simulation period. We develop the vehicle stock
mulating the electricity supply in two past years. As shown in the
model by conducting an EV market analysis. EV adoption in Alberta has
supplementary information section, the model was able to simulate past
just recently started to emerge with little data to be used to derive a
forecast. Therefore, we use an EV market analysis for Canada in the
1
net demand = provincial electricity demand – variable renewable elec- period 2010–2017 and choose the three most common battery EVs to
tricity generation – production by must-run power plants. ensure representative buying behavior and a realistic EV fleet char-
2
Power balance: residual demand + exports = electricity generation + acteristics. The three vehicles are the Tesla Model S, the Tesla
imports. Model X, and the Nissan Leaf. Since we produced the results presented

4
G. Doluweera, et al. Applied Energy 268 (2020) 114961

Table 1
EV fleet characteristics.
Vehicle model Battery capacity (kWh) Range on a full charge (km) Fuel economy (kWh/km) EV fleet share (%)

Tesla Model S 75 416 0.180 42


Tesla Model X 75 381 0.197 42
Nissan Leaf 40 241 0.166 16

in this paper, new vehicle models such as Tesla Model 3 have been We focus mainly on charging at home and charging at public lo-
available in the market and dominate the new EV sales in Canada and cations. Public charging is employed if the vehicles are not able to
elsewhere. However, the important technical attributes such as the fuel complete their daily trip due to a deficit of energy in their battery. The
economy of the vehicles are close enough to the vehicle models we impact of the charging method, which varies by the scenario, is eval-
selected for this analysis. For example, the 2019 Tesla Model 3 (Mid- uated by simulating the following three charging methods.
Range) has an average fuel economy of 0.17 kWh/km and a driving
range of 425 km. Table 1 presents the technical attributes and market • Uncoordinated charging: EVs start charging immediately after
4
share of each vehicle. they reach home, competing their last trip
The second sub-model simulates travel behavior and estimates the • Delayed charging: Delaying the start of the charging at least four
transportation energy demand. This sub-model takes a stochastic ap- hours after the last home arrival as long as the next home departure
proach and uses a probability density function of parameters that re- will have a fully charged battery
present the travel behavior. Values such as the distance driven, the • Valley filling: Here we consider an adaptive charging method that
home arrival time, and the home leaving time are sampled for each day would communicate with the electricity system and move EV
and each vehicle. We use the U.S. 2009 National Household Travel charging to periods where the electricity system sees with relatively
Survey (NHTS) to obtain the probability density functions of different low electricity demands
parameters [40]. A comprehensive travel behavior dataset such as the
U.S. NHTS is not available for Alberta. By using the NHTS data set, we The charging method and charging efficiency determine the calcu-
assume that the driving behavior in Alberta and the U.S. does not differ lation of the EV fleet charging curve. In this analysis, the charging curve
much. Through an examination of high-level travel data from Calgary, of a given vehicle is the charging power demand at each simulation
which is the most populous city in Alberta, we found the use of U.S. time step from the beginning to the end of the charging process. Fig. 2
NHTS data to be acceptable to model the travel behavior in Alberta provides an overview of the calculations procedure that was followed to
[41]. In our travel demand model, the average distance traveled by a determine the charging curve of an individual EV and the aggregated
vehicle is about 50 km/day. The distributions of driving distances are charging curve of the EV fleet. Charging curve estimation procedure
given in the supplementary information section. considers influencing factors such as the distance driven, parked
The Energy requirement for transportation per vehicle is estimated duration, ambient temperature, and charging methods. The charging
by the third sub-model. The energy demand is estimated by multiplying curve of each EV is calculated for every day. Then, the individual
the distance traveled in the period of interest by the fuel economy of the charging curves are aggregated. For each EV, it is checked whether or
vehicle. In the case of EVs, we also estimate the state of charge (SOC) of not it drives on the current day according to the driving schedule,
the battery. The SOC is the remaining energy in the vehicle battery in which is determined by the Travel demand and behavior sub-model. If
percentage. The home arriving SOC (SOCHA) can be derived by com- the vehicle does not drive, every entry of the daily charging curve is set
bining home leaving SOC (SOCHD ), the EV battery capacity (Cbat ) , fuel to zero. If the vehicle drives, the charging curve is calculated.
economy3 (ε), and the driven distance (d ) in Eq. (1). The EV fuel economy values and driving range values listed in
d· Table 1 are obtained from the manufacturer specifications. However,
SOCHA = SOCHD the practical driving range achievable is affected by some use-phase
Cbat (1)
factors, such as driving conditions and ambient temperatures [42,43].
Knowing SOCHA , the energy needed to charge the battery fully The performance (e.g., efficiency, discharge capability, and available
(Echarge ) can be calculated (Eq. (2)) using the efficiency of the charging energy) of the EV battery decreases under low and high temperatures,
equipment ( ), which is assumed to be 90%. leading to a lower driving range than manufacturer specifications.
Cbat Furthermore, auxiliary energy consumption to provide passenger
Echarge = ·(1 SOCHA) comfort under varying weather conditions (e.g., cabin heating and air-
(2)
conditioning) will reduce the driving range. Our simulation model
The parking duration limits the charging duration. The home captures the relative range reduction due to ambient temperature using
leaving SOC (SOCHD ) of an EV is equal to the SOC at the end of the last an empirical data set. The data set provides the observed relative
charging. driving range reduction of a Nissan Leaf under different ambient tem-
The fourth and fifth sub-models determines the charging location, peratures [44]. We adjust the fuel economy of EVs for each driving day
charging rate, charging method, and simulate the charging process. We in our simulations by considering the typical ambient temperatures in
model two charging rates (Level 1 and Level 2) for EV charging in the two major cities in Alberta. We assumed the same range reduction
households and two charging rates for public charging stations (Level 2 values for all EVs listed in Table 1 (See the supplementary information
and Level 3) as follows. The charging rate is determined by the section for the full details. Fig. 13 in the supplementary information
equipment used by the vehicle owner (for home charging) or the public section depict the impact of temperature on the vehicle’s range). Due to
charging service provider. the ambient temperature variations, the implied average fuel con-
sumption of the EV fleet listed in Table 1 increased from 0.18 kWh/km
• Home charging: Level 1: 1.44 kW, Level 2: 6 kW (manufacturer specification) to 0.23 kWh/km.
• Public charging: Level 2: 6 kW, Level 3: 50 kW
3 4
For an EV, the fuel economy is measured as the electricity consumption per The uncoordinated charging method is referred to as “uncontrolled charging”
kilometer driven. in some recent publications. For example, [11].

5
G. Doluweera, et al. Applied Energy 268 (2020) 114961

Daily Event

Individual EV
No
Vehicle driving today? EV fleet
Yes Input
Day temperature Calculate seasonal impact
Sample daily driving distance and
calculate distance driven on electric
Day of week
Sample home arrival time and home
departure time (next day)

Vehicle can reach No Do public fast charging


home?
Yes
Calculate home arriving state of charge
If valley filling
Calculate energy need
If delayed
Charging equipment Choose charging power level
If dumb Delay charging
Define charging start time
Parking duration Calculate charging duration
Calculate initial SOC (next day) Aggregate energy need
Calculate charging
All EVscurve
calculated Valley filling algorithm

Aggregate charging curves


Calculate the average EV load curve
Alberta’s EV number Calculate Alberta’s EV load curve

Fig. 2. Calculation procedure of an EV charging curve.

2.3. Scenarios In all scenarios, we assume the electricity generation system to


follow the forecast made by the AESO’s in its 2017 long term outlook
The analysis is developed around six scenarios that are summarized [29]. Policies and programs in Alberta’s Climate Leadership Plan [4],
in Table 2. These scenarios are characterized by the level of EV pene- such as phasing out of coal-fired power plants, the renewable energy
tration and their charging method. The scenario “No EV” models a target, and the carbon pricing system, are incorporated into the model.
future where there are no EVs are used for passenger transportation. The results of the electricity system simulation enable hourly,
This scenario is used as the reference scenario, and other scenarios are weekly, monthly, and yearly evaluation of the electricity system. Under
compared to determine the impacts of using EVs for passenger trans- each scenario, we aggregate simulation results to construct time series
portation. Given there are some EVs in the passenger transportation data of annual electricity generation mix, the electricity imports, and
vehicle fleet already, it is unreasonable to assume a future without any exports from neighboring grids, GHG emissions, and EV fleet electricity
EVs in Alberta. However, in the reference scenario, the relative share of demand curve.
EVs is negligible to make any system-wide changes. The other five
scenarios (EV20-UC, EV30-UC, EV40-UC, EV30-DL, and EV30-VF)
3. Results and discussion
model a future where the EV penetration level (i.e., the share of EVs in
the passenger transportation vehicle fleet) in 2031 is 20% (Low pene-
In this study, we assess the GHG emissions reduction potential and
tration), 30% (Medium penetration), or 40% (high penetration). Those
impacts on the electricity system operations due to the adaptation of
five scenarios are henceforth referred to as “EV scenarios”. As listed in
EVs for passenger transportation in Alberta. By using the hybrid si-
Table 2, the other factor that distinguishes the scenarios is the charging
mulation model and the scenarios described in Section 2, we estimate
method.
the magnitude of additional electricity that needs to be produced by the
The five EV scenarios provide insights into the impacts of EV fleet
electricity generation system in Alberta to satisfy passenger transpor-
size on energy requirements (Scenarios EV20-UC, EV30-UC, and, EV40-
tation energy demand. We use the energy requirements under different
UC) and impacts of the way consumers charge the EV on the peak de-
scenarios to estimate the GHG emissions reduction potential. We also
mand conditions experienced by the electricity system (Scenarios EV30-
estimate the impacts of alternative home charging techniques on the
UC, EV30-DL, and, EV40-VF).
demand profiles and temporal balancing of the electricity supply and
demand. Our analysis time period is 2020–2031. We take the last year
Table 2 of the analysis period to be 2031 as it is the year immediately after
Summary of scenarios.
Alberta’s first major emissions reduction target year of 2030. Further-
Scenario Name EV penetration in 2031 Charging method more, by 2031, the retirements of coal-fired generating units will be
completed, and the power system would see one full year of operations
No EV 0 (Reference scenario) –
without coal-fired electricity.
EV20-UC 20% (Low) Uncoordinated
EV30-UC 30% (Medium) Uncoordinated
EV40-UC 40% (High) Uncoordinated
3.1. Energy and emissions impacts of level of EV penetration
EV30-DL 30% (Medium) Delayed
EV30-VF 30% (Medium) Valley filling
In the five EV scenarios, ICE vehicles are replaced by EVs at a

6
G. Doluweera, et al. Applied Energy 268 (2020) 114961

Reference electricity Supply source: Net imports Natural Gas Coal Renewables
demand

EV charging electricity demand by scenario


No EV
100000
EV20−UC EV30−UC EV40−UC EV30−DL EV30−VF

5000

Electricity supply (GWh)


4000

75000 3000

2000
Electricity supply (GWh)

1000

0
50000
2020 2031 2020 2031 2020 2031 2020 2031 2020 2031

"Reference electricity demand" pertains to the electricity consumed in Alberta under "No EV" scenario. In that case,
no EVs are used for passenegr transportation. "EV charging electricity demand" pertains to the electricity
consumed for passenger transportation under the five EV scenarios. Under a given EV scenario, the total annual
25000 electricity supply is equal to the summation of Reference electricity demand and EV charging electricity demand.

Scenarios are chracterized by the EV penetration level in 2031 and EV charging method as follows: "No EV" = No
EVs are used; "EV20−UC, EV30−UC, and EV40−UC" = EV penetration level is 20%, 30%, and 40% respectively.
Uncoordinated EV charging; EV30−DL = 30% EV penetration and delayed charging; EV30−VF = 30% EV
penetration and valley filling charging.
0 EV penetration in 2020 is 5% under EV20−UC scenario; 8% under EV30−UC, EV30−DL, and EV30−VF scenarios;
11% under EV40−UC scenario.
2020 2031

Fig. 3. Electricity generation in Alberta by technology in 2020 and 2031 under different scenarios. Power generation technology classification: Natural gas = natural
gas fired combined cycle, simple cycle, and cogeneration power plants; Coal = Sub-bituminous coal fired power; Renewables = Wind, hydropower, solar photo-
voltaic, and biomass fired power plants; Net imports = difference between imports and exports.

constant annual rate to reach the EV share assumed by each scenario by consistent with the observations made by Li et al. [11] and Huo et al.
2031. We use the No EV reference scenario and five EV adoption sce- [12] in their work modeling the impacts of EV charging in China, where
narios to identify the energy and GHG emissions impact of adaptation the electricity generation is dominated by coal-fired electricity gen-
of EVs for passenger transportation in Alberta. Fig. 3 depicts the total eration. The marginal cost of coal-fired generation is low. Therefore, as
electricity generation in Alberta by a source under all scenarios we long as there are coal-fired generating units in the power plant fleet, it
assess in the years 2020 and 2031. is uncertain whether there is any GHG emissions reduction benefit in
Fig. 3 explicitly shows the electricity generated to satisfy EV char- adapting EVs for transportation.
ging demand (Referred to as “EV charging electricity demand” in Fig. 3) All coal-fired units in Alberta are scheduled for retirement by 2030.
and electricity consumed in the rest of the economy of Alberta (Re- According to the power generation forecast we used for the analysis
ferred to as “Reference electricity demand” in Fig. 3). As observed from [29], the coal-fired generating units are mostly replaced by natural gas-
Fig. 3, compared to the No EV reference case, the increase in electricity fired generating units. Therefore, in 2031, most of the EV charging
demand to satisfy the passenger transportation energy demand under electricity is produced by legacy and new natural gas-fired generating
all three EV penetration levels is marginal. The electricity demand in- units.
creases in 2031 under EV penetration levels of 20%, 30%, and 40% is In the simulation model, electricity trade with neighboring pro-
2.6%, 4%, and 5.3% respectively. The relatively lower increase in vinces is constrained by the capacity of existing interties. Compared to
electricity demand for EV charging is primarily due to the higher effi- the current electricity imports by Alberta, the increase in net imports
ciency of the EVs. For example, under the assumed weather and driving due to EV charging by 2031 is in the range of 14% to 27%. The increase
conditions in Alberta, the average fuel economy of EVs is 0.83 MJ/km in net imports is well below the capacity of Alberta’s current electricity
(which is equivalent to 0.23 kWh/km). Under the same conditions, the interconnections with three neighboring jurisdictions, so it is assumed
average fuel economy of a conventional ICE vehicle is 2.21 MJ/km.5 that these imports will continue using existing infrastructure, and the
The EV charging electricity demand is satisfied primarily by in- emissions associated with these imports are included in the total GHG
creased electricity generation from conventional fossil fuel-fired power emissions estimates. The impact on electricity prices due to the increase
plants and imports (see Fig. 3). Before the retirement of coal-fired in imports should be assessed in future work.
generating units, EV charging electricity is produced mainly by coal- The total transportation energy requirement under EV30-UC, EV30-
fired generating units (for example, in 2020). This observation is DL, and EV30-VF scenarios are the same. Therefore, the EV charging
electricity requirement does not vary by the charging method.
However, the charging method changes the EV charging start time. In
5
2.21MJ/km is equivalent to the weighted average fuel economy of the the case of EV30-DL and EV30-VF scenarios, the charging start time is
private passenger vehicle stock of Alberta in 2016. Passenger vehicle stock moved away from the peak demand period. As such, EV charging
consists of cars and light trucks. The value is estimated using the data obtained electricity under those two scenarios are mostly produced by generating
from NRCan [5].

7
G. Doluweera, et al. Applied Energy 268 (2020) 114961

units within the province, reducing net imports. In 2031, the reduction VF scenarios not shown in Fig. 4(b). Emissions under those two sce-
in net imports due to coordinated charging is replaced by natural gas- narios closely follow (only about 0.3% lower) the GHG emissions under
fired generating units. the EV30-UC scenario. The red dotted line in Fig. 4(b) refers to the GHG
Renewable power such as wind and solar power cannot be used to emissions in Alberta in 2005 in the two sectors assessed in this paper.
cover the additional power demand, because their electricity produc- We estimate the GHG emissions reduction potential in 2031 under
tion is driven by resource availability, which does not necessarily co- different scenarios compared to Alberta’s year 2005 emissions. To gain
incide temporally with EV charging demand. Furthermore, due to the insights about the sectoral emissions reduction potential, we also esti-
zero marginal cost of renewable electricity generating units, they are mate and depicted in Fig. 4(c) the relative emissions reduction com-
dispatched first in the merit order. As such, when available for power pared to the emissions from the Provincial electricity sector and pas-
generation, renewable electricity generators generally satisfy the non- senger transportation sector in the year 20058. Emissions reduction
EV charging electricity demand. potential under all six scenarios in 2031 is also summarized in Table 3.
By utilizing the hybrid simulation model along with the EV fleet- The use of EVs for passenger transportation leads to a decrease in
level charging curves, we estimate the GHG emissions attributable to the combustion of conventional transportation fuels. EVs do not emit
EV electricity consumption at each hour of the simulation period. This GHGs at the point of use, and therefore, an adaptation of EVs leads to a
is done by determining the GHG emissions of the electricity supply at decrease of GHG emissions in the passenger transportation sector,
each hour and multiplying it by the hourly EV charging demand. We which is proportional to the EV penetration in each scenario. For ex-
then aggregate hourly emissions to determine annual GHG emissions. ample, a market share of 40% EVs in 2031 leads to a 40% emission
Furthermore, the emission intensity of ICE vehicles is assumed to be reduction (3.32MtCO2eq) compared to GHG emissions in the No EV
constant at 148.3 gCO2eq per kilometer in this period.6 That means we scenario. However, the net GHG reduction potential would be lower
do not assume improvements in the fuel economy of the internal than that because EV charging leads to higher GHG emissions from the
combustion engine-based vehicles over the time horizon in this ana- electricity sector due to the increase in fossil fuel (mainly natural gas)
lysis. This assumption may appear to be conflicting with the gradual fired electricity generation. In other words, utilizing EVs for passenger
increase in fuel economy of ICE vehicles due to retiring vehicles being transportation moves emissions from the transportation sector to the
replaced by new vehicles with higher fuel economy. For example, the electricity sector under the assumed conditions (Fig. 4(b)). The net
Corporate Average Fuel Economy (CAFE) standard requires new pas- emissions reduction depends on the travel demand, vehicle efficiency,
senger vehicles sold in North America in 2025 to have at least a 30% weather conditions, and the GHG emissions intensity of the electricity
higher fuel economy than 2020 [45]. However, we assume the emission supply in EV charging time periods.
intensity of ICE vehicles constant at 148.3 gCO2eq due to several rea- The net GHG emissions in 2031 under all five EV adaptation sce-
sons. Changes to fleet average fuel economy and carbon intensity narios is about 9% below Alberta’s 2005 total GHG emissions (see
changes gradually, as retiring vehicles are replaced by new high fuel- Table 3). This would contribute to Alberta’s 2030 emissions reduction
efficient vehicles. Fleet average carbon intensity reduction in Alberta target of 30% below 2005 levels. As observed in the No EV scenario, the
could be much slower as passenger transportation is dominated by majority of the emissions reductions (8.5 percentage points) would be
light-truck (commonly known as sport-utility vehicles, SUV). For ex- achieved in the electric power sector. The two main drivers of the lower
ample, in 2016, 55% of the passenger-kilometers driven through pri- GHG emissions from the electricity sector are the phasing out of coal-
vate vehicles were through light-trucks [5]. Light trucks have lower fuel fired electricity generation and renewable energy integration by 2030.
economy (hence, higher carbon intensity)7. Even if the whole ICE based Both of these measures are proposed as part of Alberta’s Climate Lea-
passenger vehicle fleet in Alberta adopted minimum CAFE standard dership Plan [4].
requirements over the 2020–2030 period, the fleet average carbon in- An examination of GHG emissions under the six scenarios we assess
tensity would be 148 gCO2eq per kilometer. However, the adaptation shows that compared to the No EV scenario, additional GHG emissions
of other alternative vehicles such as hybrid vehicles and the im- reductions achievable through adaptation of EVs for passenger trans-
plementation of climate change mitigation plans proposed by the fed- portation is marginal. For example, at 40% EV penetration, the total
eral government could potentially lower the number [1]. A detailed GHG emissions are only about 1MtCO2eq less than the No EV scenario
assessment of the fleet average fuel economy of the ICE vehicle fleet is emissions. A higher level of EV penetration reduces emissions from the
retained for future work. passenger transportation sector by a proportional amount but increases
Fig. 4 summarizes the results of GHG emissions impacts assessment electricity sector emissions. For example, compared to the No EV sce-
in the period 2020–2031. As discussed in Section 1, Alberta’s 2030 GHG nario, EV penetration levels of 20%, 30%, and 40% would increase
emissions reduction target is 30% below 2005 emissions. Therefore, we electricity sector emissions by 4%, 6%, and 8%, respectively. GHG
compare the emissions under different scenarios with Alberta’s 2005 emissions from electricity generation increase because the EV charging
emissions. According to the 2019 Canadian National GHG Inventory electricity demand is satisfied primarily by natural gas-fired power
[3], Alberta’s total, electricity sector and passenger transport sector plants in 2031 (See Fig. 3).
GHG emissions in 2005 was 231.1 MtCO2eq, 48.8 MtCO2eq, and 9.2 The net GHG emissions reductions achievable by electrification of
MtCO2eq respectively (see Fig. 4(a)). passenger transportation depends on the GHG emissions per unit dis-
Fig. 4(b) shows the GHG emissions from the electricity sector and tance traveled by the EV that displaces a conventional vehicle. This will
passenger transportation sector in Alberta in the period 2020–2031. We be driven by the marginal generating units that would satisfy the EV
split the emissions from the electricity sector between the EV charging charging electricity demand. For example, the simulation results show
and the rest of the province electricity demands (i.e., electricity demand that in 2031, approximately 85% to 93% of the time, the marginal
for the rest of the economy). GHG emissions under EV30-DL and EV30- electricity generating unit that satisfied the EV charging demand was a
natural gas-fired combined cycle (NGCC) or a natural gas-fired simple
cycle (NGSC) unit. If the marginal generator was an NGCC unit that has
6 an emission intensity of 370gCO2eq/kWh, an EV would emit 85
Estimated using an average fuel economy of 2.21MJ/km and gasoline GHG
gCO2eq/km (43% lower than an ICE vehicle). In the case of an NGSC
intensity of 2.31kgCO2eq/L [3].
7
In 2016 the average carbon intensity of the light-truck (passenger vehicles)
fleet in Alberta was 170 gCO2eq/km. A gasoline fueled new light-truck that
8
meets minimum CAFE standard for model year 2020 would have a carbon in- For passenger transportation we only focus on GHG emissions from pas-
tensity of 166.7gCO2eq/km. A similar vehicle in model year 2025 would have a senger transportation using personal vehicles. This is consistent with the sec-
carbon intensity of 138.25 gCO2eq/km. toral boundary used for the current analysis.

8
G. Doluweera, et al. Applied Energy 268 (2020) 114961

Sector: Transportation Electricity (EV charging) Electricity (Other)

(a) (b)
Emissions in 2005 No EV EV20−UC EV30−UC EV40−UC
60 60
GHG emissions (MtCO2eq)

GHG emissions (MtCO2eq)


40 40

20 20

0 0

2005 2022 2025 2028 2031 2022 2025 2028 2031 2022 2025 2028 2031 2022 2025 2028 2031
(c)
20 Figure (a) depict the GHG emissions in Alberta's electricity sector and passenger
1.49% transportation sector (using private vehicles) in 2005. Alberta's 2030 GHG emissions
4.34% 5.77% 7.2%
reduction target is 30% below 2005 emissions. The electricity sector and the passenger
transportation sector GHG emissions in 2005 were 48.8MtCO2eq, and 9.2MtCO2eq
(MtCO2eq below 2005 emissions)
GHG emissions reduction in 2031

respectively. Other sectors in the province emitted 173.1MtCO2eq, making the total GHG
15
emissions 231.1MtCO2eq.

The four sub−figures in figure (b) depict the GHG emissions in the
electricity sector and passenger transportation sector in the period 2020−2031 under the
10
32.6% "No EV" scenario and three EV scenarios where EVs are charged in an uncoordinated
30.6% 29.6% 28.6% manner. The three scenarios differ by the level of EV penetration in 2030. The GHG
emissions under the scenarios EV30−DL or EV30−VF are similar to the emissions under the
scenario EV30−UC. Those three scenarios have identical EV penetration levels (i.e., 30% in
5
2030) but differ by the charging method.The dashed red lines in each sub−figure refer to the
emissions in the two sectors in 2005. Green dashed lines refer to an emissions level 30%
bellow 2005 emissions.

0 Figure (c) depicts the GHG emissions reduction below 2005 levels in 2031
in the electricity sector and the passenger transportation sector under "No EV" scenario and
No EV EV20−UC EV30−UC EV40−UC
three EV scenarios with uncoordinated EV charging. The percentage values within the bars
correspond to emissions reduction in each of the sectors relative to GHG emissions in 2005
Contributing sector: Transportation Electricity
in the corresponding sector.

Fig. 4. GHG emissions impacts under different scenarios (no EV = 0 EVs by 2030), (EV20-UC = EVs 20% of fleet by 2030), EV30-UC = 30% of fleet by 2030), EV40-
UC = 40% of fleet by 2030).

Table 3
Summary of electricity demand and GHG emissions under different scenarios in 2031.
Scenario

No EV EV20-UC EV30-UC EV40-UC EV30-DL EV30-VF

EV penetration level in 2020 (%) 0 5 8 11 8 8


EV penetration level in 2030 (%) 0 20 30 40 30 30
Charging method Uncoordinated Uncoordinated Uncoordinated Delayed Valley filling
Total GHG emissions (MtCO2eq) 38.2 37.7 37.5 37.2 37.6 37.6
GHG emissions reduction (% of passenger transportation sector and electricity sector 34.1 35.0 35.3 36.0 35.2 35.2
emissions in 2005)
GHG emissions reduction (% of Alberta's total 2005 emissions) 8.6 8.8 8.9 9 8.8 8.8
EV charging electricity demand (GWh) 2500 3750 5000 3750 3750
Peak electricity demand (GW) 13.3 14.2 14.7 15.2 13.4 13.4
New generation requirement beyond No EV scenario (MW) 350 1050 1750 0 0

unit being the marginal generator (GHG intensity = 500gCO2eq/kWh), lower the GHG emissions from electricity generation in Alberta. For
the effective GHG emissions of an EV would be 115 gCO2eq/km (still example, if the electricity supply comes from low or zero GHG emissive
22% lower than the average carbon intensity of the ICE vehicle fleet). sources—for example, from renewable energy sources—emissions re-
The resulting effective GHG intensity, although lower than conven- ductions will increase dramatically.
tional ICE vehicles, would not achieve the targeted 30% reduction;
however, higher reductions are achievable through further actions to

9
G. Doluweera, et al. Applied Energy 268 (2020) 114961

Scenario: No EV EV30−UC EV30−DL EV30−VF

EV charging electricity demand (Home charging) Total Alberta electricity demand

1.5 14
Electricity demand (GW)

1.0 13

0.5 12

0.0
11
08:00 20:00 08:00 20:00 08:00 20:00 08:00 20:00 08:00 20:00 08:00 20:00
Time of the day (hours)

Fig. 5. EV charging curves and its impact on Alberta’s electricity demand profile at 30% EV Penetration in 2031. The figures show the charging curves and net system
electricity demand in a period of arbitrarily selected three consecutive days. All three EV scenarios shown here have the same EV penetration level of 30% by 2030
but differ by charging method. Scenario EV30-UC uses uncoordinated EV charging. The charging method employed in EV30-DL and EV30-VF is “Delayed Charging”
and “Valley Filling” respectively.

3.2. Impact of EV charging on Alberta’s electricity system operations temporally coincide with none EV charging peak electricity demand
(i.e., the peak demand of the No EV scenario). That increases the ef-
As shown in Fig. 3, the additional electrical energy required to sa- fective peak demand seen by Alberta’s electric power system. The
tisfy the EV charging demand under all penetration levels considered in combined capacity of the generating unit outlook made by the AESO
this analysis is marginal, up to 5% in 2031, suggesting the additional [29] is not sufficient to satisfy the net peak demand under all EV pe-
generating capacity requirements beyond No EV scenario. However, an netration levels we modeled when uncoordinated EV charging is em-
examination of simulation results shows that uncoordinated EV char- ployed. Therefore, additional power generating units are required.
ging exacerbates the peak demand conditions experienced by the This observation is consistent with the work published by Muratori
electricity system, leading to an increase in electricity generation ca- [20] or de Tena and Pregger [13]. By modeling residential-scale elec-
pacity requirements. In this section, we quantify the magnitude of the tricity distribution systems, Muratori [20] shows that even at low EV
increase in peak demand. We also demonstrate how alternative char- penetration levels, uncoordinated charging can lead to peak demand
ging methods can be used to avoid the rise in peak electricity demand. conditions that the existing distribution system is not able to handle. De
Conventional ICE vehicles can be refueled in fueling stations within Tena and Pregger [13] also find out that in the case of Germany un-
a fairly short period of time. EVs, however, take more time to charge, controlled charging can lead to a significant increase of the residual
and it is plausible that most users will choose to charge the vehicles at demand peak. In our work, we observe similar reliability issues at the
home using residential electricity supply. In fact, this is one con- overall generation system level.
venience of using EVs [46]. Under the uncoordinated EV charging We find that at 30% penetration, with uncoordinated EV charging,
scenarios (EV20-UC, EV30-UC, and EV40-UC), we assume that charging the additional power generating capacity is equivalent to three 350 MW
starts almost immediately after a vehicle returns home at the end of the natural gas-fired generating units. Under 20% and 40% penetration, the
day. Examination of travel patterns shows that home arrival generally new capacity requirement is equivalent to one and five 350 MW natural
occurs in early evening hours, where the electric power system ex- gas-fired generating units, respectively. See Table 3 for a summary of
periences peak demand conditions. Electric power systems—Alberta’s net peak electricity demands and equivalent new generating capacity
included—still do not have large scale storage capacity, and therefore, requirements under all six scenarios. These additional units will es-
power system operators must dispatch power generating units whose sentially be peaking power plants and operate at a lower capacity
combined power production must be equivalent to the electricity de- factor. This increases the overall cost of operations of Alberta’s power
mand at any given time. The total power demand that can be satisfied is system. Although not explicitly modeled, it is plausible that electricity
constrained by installed generating capacity and electricity intertie transmission and distribution systems, too, will experience capacity
capacities. Due to uncoordinated EV charging, electric Alberta’s electric constraints if uncoordinated EV charging is employed.
power system may experience challenges in matching the electricity The aforementioned adverse impacts can be mitigated by the
supply and demand. In this section, we demonstrate this challenge. adaptation of alternative charging methods. These include employing
Furthermore, we also show how alternative changing methods can be relatively simple options such as time-delayed chargers or more so-
used to mitigate that challenge. phisticated adaptive charging methods that would move EV charging
Fig. 5 shows the electricity demand data under the scenarios No EV, into periods with lower electricity demand. We demonstrate this
EV30-UC, EV30-DF, and EV30-VF over a period of arbitrarily selected through simulations under EV30-DL (Delayed charging) and EV30-VF
three consecutive weekdays. The EV penetration level under all three (Valley Filling charging) scenarios9. These two scenarios would defer
EV scenarios is 30% by 2030. The left facet of Fig. 5 shows the ag- EV charging into low demand periods. As shown in Fig. 5, under those
gregated EV charging demand curves. The right facet shows the com- scenarios, the peak EV charging demand does not coincide with the
bined electricity demand seen by Alberta’s electricity system in the
same period. This includes the EV charging demand as well as the
electricity consumed in the rest of the economy of Alberta. 9
Adaptation of “Valley Filling” type charging method would require com-
As seen from Fig. 5, when EVs are charged in an uncoordinated munications with the power system and adapt to conditions to avoid coincident
manner (i.e., EV30-UC scenario), the peak EV charging demand peak demands.

10
G. Doluweera, et al. Applied Energy 268 (2020) 114961

other non-EV charging peak demand. Therefore, we find that the net Table 4
peak demand under the two scenarios with alternative EV charging Impact of transportation demand growth on GHG emissions (EV30-UC
methods is essentially the same as that of the No EV scenario. There- Scenario).
fore, no additional power generating infrastructure would be needed to Annual transportation demand
satisfy the net Alberta electricity demand. The use of alternative char- growth
ging methods has the potential to reduce GHG emissions as well. For
0% 2% −2%
example, an adaptive charging system can communicate with the
electricity system and move EV charging to periods with lower GHG GHG emissions in 2031 (MtCO2eq) 37.5 40.3 35.4
emissions intensities such as periods with a higher level of renewable Relative magnitude of emissions reduction in 35.3% 30.5% 39.0%
electricity generation. 2031 (% GHG emissions in 2005 in the
electricity and transportation sectors)
Another benefit of coordinated charging is the reduction in elec-
tricity imports. As discussed in Section 3.1, coordinated charging does
not lower the total electricity demand for EVs. However, EV charging
electricity infrastructure requirements to facilitate EV adaptation.
demand is moved away from the system peak demand; a higher amount
Our simulation results show that due to the higher efficiency of EVs,
of EV charging demand can be satisfied with generating units within the
the increase in electricity demand is moderate—less than 6%—even at
province. For example, in 2031, under the EV30-DL scenario (i.e., de-
relatively high EV penetration levels of 40% by 2030. The absolute
layed charging), the net imports are reduced by about 30%. Reductions
growth in the electricity demand, however, is sensitive to the total
in electricity imports can potentially lower the price of electricity.
demand for transportation. The electricity demand will inevitably in-
crease with higher demand for transportation as well as at a higher
3.3. Impacts of changes in transportation demand
level of EV penetration.
However, depending on the charging system employed and user
Throughout the analysis, we assumed the total demand for pas-
behavior, usage of EVs could potentially add challenges to electricity
senger transportation to remain constant. The main reason for that
system operations by increasing the demand in periods with peak de-
assumption is that over the past ten years, the average demand growth
mand conditions. The worst condition would occur when households
for passenger transportation through private vehicles in Alberta has
recharge the EVs in an uncoordinated manner in residential electricity
been less than 1% [5]. In fact, some years have seen notable levels
distribution systems. Given the fact that most households tend to prefer
(5–9% reductions) of demand declines. However, it is plausible that the
to recharge their vehicles at home, distribution system operators should
demand will indeed grow over our analysis period. It is also plausible
explore options to reinforce the residential distribution systems as well
the overall demand for transportation through private vehicles declines
as facilitate the adaptation of controlled smart charging systems by, for
due to changes in behaviors such as higher use of public transit and
example, providing necessary control signals and facilitating the
ridesharing. Therefore, a detailed transportation demand forecast must
charger to communicate with the power system. Since electric power
be integrated into our simulation model. While retaining the develop-
systems are usually designed with a limited amount of reserve capa-
ment of a demand forecast for future research, we conducted a sensi-
cities, uncoordinated EV charging would put stress on electric power
tivity analysis by changing the demand growth by −2% and 2%. Our
systems in any jurisdiction.
assumption is that this range would capture any changes in average
GHG emissions reduction potential of the EV adaptation scenarios
demand growth over the analysis period. We only conducted a sensi-
we modeled is in the order of 9% below 2005 GHG emissions in Alberta
tivity analysis with the EV30-UC scenario. Table 4 summarizes the total
by 2031. Alberta’s current emissions reduction target for 2030 is 30%
GHG emissions, and the relative emissions reductions resulted from the
below 2005 levels. The majority of the emissions reductions—about 8.5
sensitivity analysis.
percentage points—are due to the electricity sector's GHG emissions
From the sensitivity analysis results, we find that reducing the
reduction measures such as phasing out of coal-fired electricity gen-
transportation demand by 2%, along with EV adaptation, can increase
eration and higher renewable energy integration. In a jurisdiction with
the overall emissions reductions by about further 2MtCO2eq. Similarly,
relatively low GHG intensive electricity generation systems (e.g., power
an increase in transportation demand will lower the emissions reduc-
systems with hydropower or nuclear power dominated power plant
tion benefit of electrification of transportation. Furthermore, demand
fleet), the GHG emissions reductions due to electrification of trans-
growth with uncoordinated EV charging will put further stress on the
portation would be much higher.
electric power system by increasing the peak demand to 15,230 MW,
In terms of energy demands and GHG emissions, we only focused on
which is about 15% higher than the No EV scenario, and 4% higher
the passenger transport sector and the electricity sector of Alberta. That
than EV30-UC scenario with no demand growth. Furthermore, a 2%
leaves other major GHG emissive sectors in Alberta, such as freight
growth in demand would increase the demand for imports by 35%
transport, building, and the industry sector. The current analysis has
compared to the EV30-UC scenario with no demand growth.
three key limitations that should be improved in future studies. First, a
detailed transportation demand forecast needs to be developed. The
4. Conclusions
forecast model should take into account the changes in demographic
conditions, economic growth, passenger behavior changes, multi-modal
In the analysis presented in this paper, we develop a hybrid simu-
nature of transportation choices, and transportation policies that are
lation model of the electricity system of Alberta. A stochastic vehicle
being proposed by different levels of governments. Second, in this
fleet model is developed and integrated into the electricity system si-
study, the only alternative transportation option we considered is bat-
mulation model to gain insights into the impacts of transitioning from
tery EVs. Impact of the adaptation of other alternative vehicle tech-
gasoline-fueled conventional vehicles to battery EVs for passenger
nologies such as hybrid vehicles, plug-in hybrid electric vehicles, and
transportation in Alberta. The model takes into account factors that
alternative fuel vehicles (e.g., compressed natural gas, hydrogen ve-
drive impacts associated with passenger transportation, including
hicles) should be investigated. In the case of EVs, future work should
driving behavior, diurnal and seasonal factors, weather conditions,
assess the impacts of emerging developments (e.g., autonomous
vehicle characteristics, evolving electricity infrastructure, and climate
driving, smart adaptive charging capabilities) and different classes of
policies. Using the model, we simulated six scenarios to assess the im-
vehicles (e.g., EVs that are comparable with light trucks in terms of
pacts of EV adaptation for passenger transportation focusing on the
cassis size/weight). Third, the evolution of the ICE vehicle fleet must be
level of penetration and charging methods employed. The main impacts
refined by taking into account regulatory mandated fuel economy
we assessed include future electricity demand, GHG emissions, and

11
G. Doluweera, et al. Applied Energy 268 (2020) 114961

improvements, the retirement of the existing vehicle stock and addition sector. That advantage, along with electrification, can leverage GHG
of new vehicles, and climate policies such as low carbon fuel standards. emissions reductions efforts in many jurisdictions.
The general simulation framework we developed has a high tem-
poral resolution to assess energy systems level impacts of transportation CRediT authorship contribution statement
electrification. Our intention was to develop a general simulation model
that is robust in terms of capturing systems level attributes but trans- Ganesh Doluweera: Conceptualization, Resources, Visualization,
parent to facilitate auditing and interpretations. The flexible and scal- Writing - original draft. Fabian Hahn: Investigation, Software, Formal
able model can be calibrated for any jurisdiction. The modeling fra- analysis, Data curation. Joule Bergerson: Writing - review & editing,
mework developed in this paper can leverage future studies to gain Supervision. Marco Pruckner: Conceptualization, Methodology,
insights, such as the consequences of large-scale electrification of end- Supervision.
use energy services as an option for GHG emissions management. The
interactions of the EV charging processes and the electricity sector, Declaration of Competing Interest
presented in this paper, demonstrate the importance of an energy sys-
tems-level EV impact study. Interest in adapting EVs for passenger The authors declare that they have no known competing financial
transportation is high in Canada and other countries. Therefore, the interests or personal relationships that could have appeared to influ-
current work can be adapted for other provinces in Canada or even ence the work reported in this paper.
other countries to assess the impact of EVs on the electric power sys-
tems to inform system planning. In addition to these national/regional Appendix A. Supplementary material
studies, future research could focus on several important extensions of
this work. For example, a study representing Alberta with higher spatial Supplementary data to this article can be found online at https://
resolution can give insight into the impact of EVs on large cities, doi.org/10.1016/j.apenergy.2020.114961.
transmission power flows, and congestion caused by a deficit of trans-
mission capacity. The impact of EVs on electricity prices and economics References
of peak load power plants, to cover increased peak load, should be
investigated. Environmental impact assessment could be extended to [1] ECCC. Progress towards Canada’s greenhouse gas emissions reduction target.
estimate the other environmental and health benefits such as an in- Environment and Climate Change Canada; 2019.
crease in urban air quality due to the reduction of air pollutants during [2] Statistics Canada. Population estimates, quarterly; 2017.
[3] ECCC. National Inventory Report: 1990 - 2017: Greenhouse Gas Sources and Sinks
the transition from ICEs to EVs. The high temporal resolution of the in Canada. Environment and Climate Change Canada; 2019.
model can facilitate electricity prices and market analyses. Moreover, [4] Government of Alberta. Climate Leadership Plan; 2018.
the investigation of different optimization-based charging algorithms [5] NRCan. Comprehensive Energy Use Database. Natural Resources Canada; 2019.
[6] Ferguson M, Mohamed M, Higgins CD, Abotalebi E, Kanaroglou P. How open are
(e.g., minimizing GHG emissions) could provide additional interesting Canadian households to electric vehicles? A national latent class choice analysis
insights. The stochastic agent-based vehicle aggregation model can be with willingness-to-pay and metropolitan characterization. Transp Res Part Transp
used to investigate vehicle fleets of different levels of granularity. The Environ 2018;58:208–24. https://doi.org/10.1016/j.trd.2017.12.006.
[7] Vypovska A. A Brief Overview of the Current State of the Electric Vehicle Market in
scope of this paper, however, is the evaluation of the impact of EV Canada. CERI Electr Rep 2018.
charging at the provincial level over a longer duration, and therefore [8] City of Edmonton. Electric Vehicle Strategy; 2019. https://www.edmonton.ca/city_
use of the average charging curve was sufficient. The stochastic model government/city_vision_and_strategic_plan/electric-vehicle-strategy.aspx [accessed
December 25, 2019].
can also facilitate to model impacts of changes in passenger behavior.
[9] City of Calgary. Electric vehicle strategy; 2019. https://www.calgary.ca/
The current interest in EVs for passenger transport is driven by Transportation/TP/Pages/Strategy/Electric-vehicle-strategy.aspx [accessed
many factors. These include concerns about energy security (e.g., crude December 25, 2019].
oil and refined petroleum product supply reliability) and concerns [10] NRCan. Electric Vehicle and Alternative Fuel Infrastructure Deployment Initiative.
Nat Resour Can; 2019. https://www.nrcan.gc.ca/energy-efficiency/energy-
about the environmental impacts of conventional vehicles such as efficiency-transportation/electric-vehicle-alternative-fuels-infrastructure-
urban air pollution and GHG emissions. Alberta is endowed with large deployment-initiative/18352 [accessed December 25, 2019].
hydrocarbon resources, and therefore, energy security is not a driving [11] Li Y, Davis C, Lukszo Z, Weijnen M. Electric vehicle charging in China’s power
system: Energy, economic and environmental trade-offs and policy implications.
factor for a transition to EVs. On the other hand, GHG emissions Appl Energy 2016;173:535–54. https://doi.org/10.1016/j.apenergy.2016.04.040.
management is a higher priority in the province’s energy policy dis- [12] Huo H, Zhang Q, Wang MQ, Streets DG, He K. Environmental Implication of Electric
course. Vehicles in China. Environ Sci Technol 2010;44:4856–61. https://doi.org/10.1021/
es100520c.
In the case of Alberta, as revealed by our simulation results, the [13] de Tena DL, Pregger T. Impact of electric vehicles on a future renewable energy-
technological change from internal combustion vehicles to EVs leads to based power system in Europe with a focus on Germany. Int J Energy Res
a shift from burning gasoline in the transport sector to burning fossil 2018;42:2670–85. https://doi.org/10.1002/er.4056.
[14] Hedegaard K, Ravn H, Juul N, Meibom P. Effects of electric vehicles on power
fuels in conventional power plants. Therefore, the magnitude of emis- systems in Northern Europe. Energy 2012;48:356–68. https://doi.org/10.1016/j.
sions reductions attributable to EV adaptation for passenger transport is energy.2012.06.012.
marginal as the electricity that displaces gasoline is anticipated to be [15] Bellocchi S, Gambini M, Manno M, Stilo T, Vellini M. Positive interactions between
electric vehicles and renewable energy sources in CO2-reduced energy scenarios:
produced primarily using natural gas by 2030. As such, to achieve
The Italian case. Energy 2018;161:172–82. https://doi.org/10.1016/j.energy.2018.
higher emissions reductions through EV adaptation requires deeper 07.068.
GHG emissions reductions from the power generation system. This re- [16] Jochem P, Babrowski S, Fichtner W. Assessing CO2 emissions of electric vehicles in
quires the implementation of current electricity sector carbon man- Germany in 2030. Transp Res Part Policy Pract 2015;78:68–83. https://doi.org/10.
1016/j.tra.2015.05.007.
agement policies and programs that lead to greater penetration of lower [17] Mccarthy RW, Yang C, Ogden J. Impacts of Electric-drive Vehicles on California’s
GHG emitting sources of electricity generation. Energy System; n.d.
The relevancy of that observation goes beyond the transportation [18] Weis A, Michalek JJ, Jaramillo P, Lueken R. Emissions and Cost Implications of
Controlled Electric Vehicle Charging in the U.S. PJM Interconnection. Environ Sci
sector and jurisdictional boundaries. In the current climate policy dis- Technol 2015;49:5813–9. https://doi.org/10.1021/es505822f.
course, the electrification of end-use energy services is seen as a tech- [19] Fischer D, Harbrecht A, Surmann A, McKenna R. Electric vehicles’ impacts on re-
nology option for economy-wide deep decarbonization. The effective- sidential electric local profiles – A stochastic modelling approach considering socio-
economic, behavioural and spatial factors. Appl Energy 2019;233–234:644–58.
ness of the electrification as a GHG mitigation option depends on the https://doi.org/10.1016/j.apenergy.2018.10.010.
rate and stringency of decarbonizing efforts in the electric power sector. [20] Muratori M. Impact of uncoordinated plug-in electric vehicle charging on re-
Compared to other economic sectors, there are many commercially sidential power demand. Nat Energy 2018;3:193–201. https://doi.org/10.1038/
s41560-017-0074-z.
ready technologies available for decarbonization of the electric power

12
G. Doluweera, et al. Applied Energy 268 (2020) 114961

[21] Hanemann P, Behnert M, Bruckner T. Effects of electric vehicle charging strategies demonstration. Appl Energy 2018;210:987–1001. https://doi.org/10.1016/j.
on the German power system. Appl Energy 2017;203:608–22. https://doi.org/10. apenergy.2017.07.069.
1016/j.apenergy.2017.06.039. [34] Darabi Z, Ferdowsi M. Aggregated impact of plug-in hybrid electric vehicles on
[22] Tagliaferri C, Evangelisti S, Acconcia F, Domenech T, Ekins P, Barletta D, et al. Life electricity demand profile. IEEE Trans Sustain Energy 2011;2:501–8. https://doi.
cycle assessment of future electric and hybrid vehicles: A cradle-to-grave systems org/10.1109/TSTE.2011.2158123.
engineering approach. Chem Eng Res Des 2016;112:298–309. https://doi.org/10. [35] Pouladi J, Sharifian MBB, Soleymani S. A new model of charging demand related to
1016/j.cherd.2016.07.003. plug-in hybrid electric vehicles aggregation. Electr Power Compon Syst
[23] Onat NC, Kucukvar M, Tatari O. Conventional, hybrid, plug-in hybrid or electric 2017;45:964–79. https://doi.org/10.1080/15325008.2017.1312719.
vehicles? State-based comparative carbon and energy footprint analysis in the [36] Qian K, Zhou C, Allan M, Yuan Y. Load model for prediction of electric vehicle
United States. Appl Energy 2015;150:36–49. https://doi.org/10.1016/j.apenergy. charging demand. 2010 Int. Conf. Power Syst. Technol. 2010. p. 1–6. https://doi.
2015.04.001. org/10.1109/POWERCON.2010.5666587.
[24] Kennedy C. Key threshold for electricity emissions. Nat Clim Change [37] Weiller C. Plug-in hybrid electric vehicle impacts on hourly electricity demand in
2015;5:179–81. the United States. Energy Policy 2011;39:3766–78. https://doi.org/10.1016/j.
[25] Pruckner M. Ein Simulationsmodell für den Energieumstieg in Bayern PhD-thesis enpol.2011.04.005.
Göttingen: Cuvillier Verlag; 2018. [38] Zhang X, Sun Y, Duan Q, Huang Y. The charging load model of electric vehicle
[26] Pruckner M, German R. The Impact of Electric Vehicles on the German Energy based on cloud model. In 2016 11th Int. Conf. Comput. Sci. Educ. ICCSE; 2016, p.
System. In Proc. 49th Annu. Simul. Symp., San Diego (CA, USA): Society for 415–8. https://doi.org/10.1109/ICCSE.2016.7581616.
Computer Simulation International; 2016, p. 1–7. [39] Han P, Jinmei W, Lu W. Electric Vehicle Charging Load Model Based on Diffusion
[27] Pruckner M. Modeling the impact of electrical energy storage systems on future Theory. Open Mech Eng J 2015;9:307–13. https://doi.org/10.2174/
power systems. In 2016 IEEE Electr. Power Energy Conf. EPEC; 2016, p. 1–7. 1874155X01509010307.
https://doi.org/10.1109/EPEC.2016.7771724. [40] Federal Highway Administration. 2009 National Household Travel Survey. U.S.
[28] Law AM. Simulation modeling and analysis. 5th ed. New York: McGraw-Hill Department of Transportation & Federal Highway Administration; 2018.
Education; 2015. [41] City of Calgary. Travel and activity surveys. The City of Calgary; 2011.
[29] AESO. 2017 Long-term Outlook. Alberta Electric System Operator; 2017. [42] Zhang R, Yao E. Electric vehicles’ energy consumption estimation with real driving
[30] CERI. Economic and environmental impacts of transitioning to a cleaner electricity condition data. Transp Res Part Transp Environ 2015;41:177–87. https://doi.org/
grid in Western Canada. Canadian Energy Research Institute; 2018. 10.1016/j.trd.2015.10.010.
[31] Zhang H, Song X, Xia T, Yuan M, Fan Z, Shibasaki R, et al. Battery electric vehicles [43] Yuksel T, Michalek JJ. Effects of Regional Temperature on Electric Vehicle
in Japan: Human mobile behavior based adoption potential analysis and policy Efficiency, Range, and Emissions in the United States. Environ Sci Technol
target response. Appl Energy 2018;220:527–35. https://doi.org/10.1016/j. 2015;49:3974–80. https://doi.org/10.1021/es505621s.
apenergy.2018.03.105. [44] Allen M. Electric range for the nissan leaf & chevrolet volt in cold weather.
[32] Ashtari A, Bibeau E, Shahidinejad S, Molinski T. PEV Charging Profile Prediction FleetCarma; 2013.
and Analysis Based on Vehicle Usage Data. IEEE Trans Smart Grid 2012;3:341–50. [45] AFDC. Vehicle Fuel Efficiency (CAFE) Requirements by Year. Alternative Fuels Data
https://doi.org/10.1109/TSG.2011.2162009. Center, U.S. Departmnet of Energy; 2013.
[33] DeForest N, MacDonald JS, Black DR. Day ahead optimization of an electric vehicle [46] US DOE. Charging at Home; 2019. https://www.energy.gov/eere/electricvehicles/
fleet providing ancillary services in the Los Angeles Air Force Base vehicle-to-grid charging-home [accessed December 23, 2019].

13

You might also like