project report on energy transitions in Canada

Contents lists available at ScienceDirect

Applied Energy

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

Electrification of road transportation with utility controlled charging: A case study for British Columbia with a 93% renewable electricity target

Victor Kellera,⁎, Jeffrey Englisha,b, Julian Fernandeza,c, Cameron Wadea, McKenzie Fowlera, Sven Scholtysika, Kevin Palmer-Wilsona, James Donalda, Bryson Robertsond, Peter Wilda, Curran Crawforda, Andrew Rowea

a Institute for Integrated Energy Systems, University of Victoria, PO Box 1700 STN CSC, Victoria, BC V8W2Y2, Canada b B.C. Hydro, 333 Dunsmuir St, Vancouver, BC V6R 5R4, Canada c University of British Columbia, Clean Energy Research Centre, 2360 East Mall, Vancouver, BC V6G 1Z3, Canada d School of Civil and Construction Engineering, Oregon State University, Kearney Hall, 1491 SW Campus Way, Corvallis, OR, USA

H I G H L I G H T S

• Abatement cost of electrification of road vehicle fleet range from 14 to 21 $/tCO2e. • Electricity cost is up to 9% higher in scenarios with transport electrification. • Eliminating renewable target negates 60% of GHG benefits of transport electrification. • Use of UCC in half of available fleet may decrease generation capacity needs by 7%.

A R T I C L E I N F O

Keywords: Electrification of transport Battery electric vehicles Renewable energy Utility controlled charging

A B S T R A C T

To mitigate emissions from the electricity and transportation sectors, large scale deployment of renewable en- ergy generators and battery electric vehicles are expected in the coming decades. However, adoption of these technologies may exacerbate issues related to mismatch of electricity supply and demand. In this study, we utilize a hybrid capacity expansion and dispatch model to quantify grid impacts of the conversion of the entire road vehicle fleet to electric vehicles by 2050. We examine impacts of policies, such as targeting a renewable energy penetration of 93%, using British Columbia as a case study. Scenarios making use of utility controlled charging of vehicles to balance supply and demand are further analyzed. Results show that although electrifying the entire road vehicle fleet will require generation capacity to increase by up to 60%, relative to a scenario without electrification, levelized cost of electricity only increases by 9% in the same scenario due to availability of low cost generation options such as wind and solar. We also find that a 93% renewable energy target leads to carbon abatement costs 30% lower than a scenario where this policy is removed. Further use of utility controlled charging reduces total system capacity up to 7%.

1. Introduction

The electricity and transportation sectors respectively account for 25 and 14% of global anthropogenic emissions as of 2010 [1]. As de- mand in both sectors is expected to continue to grow over the coming decades [2], increased use of low-carbon energy supplies, such as wind and solar PV (photovoltaic), and electric vehicles (BEVs) are necessary to mitigate emissions. At large penetrations, these technologies impact grid operations due to issues such as excess generation [3], lack of flexibility [4], and increased localized peak demands due to coincident

vehicle charging [5]. As a result, there is a need to understand how simultaneous integration of variable renewable energy (VRE) supplies and BEV vehicles may affect electrical system structure and costs.

Globally, the electricity sector has made significant progress in implementing renewable energy, with its generation growing more than 30% in the last five years [6], however this may generate load balancing issues [4]. Renewable energy generation in the U.S.A. in- creased nearly 50% over the last 5 years and forecasts suggest strong growth to 2050 [7]. As described elsewhere, high penetrations of re- newable supplies can be challenging due to the need for additional

https://doi.org/10.1016/j.apenergy.2019.113536 Received 23 March 2019; Received in revised form 4 June 2019; Accepted 11 July 2019

⁎ Corresponding author. E-mail address: [email protected] (V. Keller).

Applied Energy 253 (2019) 113536

Available online 16 July 2019 0306-2619/ © 2019 Elsevier Ltd. All rights reserved.

T

system flexibility [8] and mechanisms such as storage to ensure load- balancing [3]. Curtailment of total wind generation grew from virtually zero in 2007 to 2–4% by 2013 in U.S. jurisdictions [9] and British Columbia (BC) experienced a 1.5 TWh increase in surplus energy from 2006 to 2016 as a result of increased run-of-the-river capacity [10].

Excess electricity supply may at times lead to financial loss to the electricity system operator or to generator owners [3]. Although VRE generators typically decrease system cost as they may replace fuel based generation at zero marginal cost, when energy is curtailed a financial loss is incurred by the generator owner as fixed and capital costs are amortized over a lower amount of generation [3]. From the perspective of the system operator, if the electricity is “must take” it may require system operators to sell it at times of low or negative values. This has been observed in markets where negative prices associated with excess wind power are realized due to minimum generation constraints [10]. Alternatively, contractual obligations may require the system operator to pay generators to curtail energy, as exemplified in Germany, where in 2015 wind generators were paid an average of €53/MWh to curtail their generation [3] These factors highlight the need for careful plan- ning of electricity system evolution to manage future increases in re- newable penetration.

Flexibility requirements for electricity systems are being challenged by more than the addition of VRE generators; new demands are ex- pected due to the conversion of internal combustion engine (ICE) ve- hicles to BEVs. BEV sales increased by 70% from 2014 to 2015, fol- lowed by an additional 40% growth from 2015 to 2016, at which point the global EV stock reached 2 million vehicles [6]. With decreasing battery costs and recent announcements made by several jurisdictions such as France, U.K. [11] and B.C., where targets have been put in place to phase out ICE vehicles before mid-century [12], the demand for BEVs is expected to experience strong growth in coming decades. As a result, careful system planning is necessary to balance future supply and de- mand.

Numerous studies have been carried out in recent years to de- termine the impact that electrification of the transportation sector may have on the electricity system. Madzhrov et al. use a unit commitment model to study a hypothetical country with a generic energy mix and the European average number of vehicles per capita [13]. The authors demonstrate that this system would only be able to serve up to a 10% penetration of passenger EVs with decentralized charging and no ca- pacity expansion. Kelly et al. developed a model to simulate fleet average electricity consumption based on driving pattern data [14]. The authors found that for the hypothetical case where 50% of the passenger vehicle sector is converted to plug-in hybrid electric vehicles (PHEVs), peak electricity demands could increase by as much 12%, in the worst case. Rosler et al. conducted a long term optimization study of the European system focusing on electrification of the transport sector through use of battery electric or fuel cell technology [15]. The authors found that in the scenario where BEVs have fully replaced conventional passenger vehicles by mid century, the electricity production in the year 2060 is over 50% higher than 2010. Graabak et al. studied the adoption of 100% electrified passenger transportation in the Nordic system by 2050 [16]. The authors found that energy demand increases by 7.5% as a result of electrification. Further, uncontrolled “dumb” charging leads to load increase during periods of peak energy demand such as early morning or evening peaks. Schill et al. studied the grid impacts of converting over 10% of the passenger vehicle sector to battery electric in Germany by 2030 [17]. Even though demand from electric vehicles only represents 1.5% of total electricity demand, peak load increases by up to 5.5%. The authors concluded that while energy requirements from electrifications of vehicles is not a concern over the short term, the impact on peak loads should be considered closely by policy makers.

Utility controlled charging (UCC) has been suggested as a method to manage the intermittency of VREs, lessen the effects of the peak de- mands issues mentioned above and to enable larger shares of vehicle

electrification. UCC allows for the utility to control how much energy is fed into an electric vehicle that is connected to the grid at a given time, allowing it to shift demand by a number of hours.

A number of studies have evaluated the impact that employing UCC on the passenger vehicle segment may have on the electricity system. Li et al. used an hourly multi-region unit commitment model of China in the year 2030 to evaluate how the use of UCC may impact the elec- tricity system with a 30% penetration of battery electric vehicles in the passenger vehicle sector [18]. The authors found that employing UCC leads to higher emissions than scenarios with static charging as demand is shifted to hours when low cost, high emitting technologies, i.e. coal, are the marginal generators. Similar results were found by Hedegaard et al, who modelled the Northern European system to 2030, with the passenger vehicle sector achieving a 53% penetration of BEVs by the end of the model period [19]. Vehicles were assumed to be able to meet peak demand and employ UCC or “smart charging”, allowing their demand to be time shifted. The results showed that in some jurisdic- tions, use of UCC leads to increased wind penetration, while in others, such as Germany and Denmark, use of UCC leads to higher usage of coal generation. Prebeg et al. conducted long term optimization of the Croatian energy system to 2050 with 100% penetration of battery electric or plug-in hybrid vehicles [20]. In the study, a maximum of 25% of the vehicle fleet is committed to vehicle to grid applications at any one point it time. The authors found that shifting vehicle demand to the night time, when demand is lower, can keep peak demands from increasing, leading to overall cost savings. Weis et al. studied the im- pact of UCC given a 10% penetration of plug-in hybrid electric vehicles (PHEV) in the NYISO system [21]. The authors found a 54–73% PHEV integration cost reduction by making use of UCC. Lyon et al, studied the impact of demand shifting of PHEVs on the MISO and PJM independent system with a 60% penetration of BEVs in the passenger vehicle sector [22]. Although the use of UCC led to billions in savings, this re- presented less than a 1% reduction from a scenario employing un- controlled charging. Wolinetz et al. studied the impact of adoption of utility controlled charging on varying portion of the passenger vehicle fleet in British Columbia and Alberta, Canada [23]. They found that UCC could lead to a capacity requirement reduction of up to 8% com- pared to a scenario where UCC was not used. Further, UCC was found to modestly reduce wholesale electricity prices (0.7%). These studies show the restricted extent to which UCC can make a contribution to lowering the cost of integration of BEVs with the electricity system; however, they only consider the passenger vehicle segment and partial elec- trification of the fleet. The implications of broader transport elec- trification that includes freight and transit – often representing a larger portion of the transportation sector energy demand – have not been examined thoroughly. To the knowledge of the authors, only Taljegard et al. have attempted to quantify the combined effect of widespread electrification of passenger vehicles and freight recently [24]. The au- thors evaluated the electricity system impact of electrification of the road transportation fleet in Northern Europe and Germany, including passenger vehicles and the freight sector. Although the study provided valuable insight into system expansion needs, the authors did not quantify UCC cost benefits per vehicle and did not employ varying penetration levels of UCC.

Here, we study the impacts on the electricity system of electrifica- tion of the entire road vehicle fleet, including passenger vehicles, light, medium and heavy duty freight and transit. A bottom-up linear pro- graming model is used to determine the optimal generation capacity and dispatch to meet an exogenous demand. As a case study, the pro- vince of B.C. is considered due to its high share of freight in comparison to passenger vehicles, its aggressive targets for electrification of trans- portation, and existing low-carbon generation mixture. Scenarios evaluate varying adoption levels of UCC and the impact of enforcing a renewable portfolio standard aiming to achieve a 93% renewable pe- netration. Further, the value of UCC per vehicle-year is further quan- tified.

V. Keller, et al. Applied Energy 253 (2019) 113536

2

Methods are described in Section 2. Section 3 provides data on costs and other parameters used in the model. Section 4 details the results, followed by a discussion in Section 5. Finally, Section 6 provides the conclusions.

2. Methods

Section 2.1 provides an overview of the modelling approach. Section 2.2 provides details on model platform, technology assump- tions, and temporal structure. Section 2.3 provides details on trans- portation forecasts.

2.1. Model overview

A linear programing optimization tool is used to compare pathways for the electricity and transportation sectors. As a case study, the B.C. system is modelled. Although the B.C. system has connections to Alberta and to the United States, in this work it is modelled as an iso- lated system for simplicity. The model period analyzed spans from 2015 to 2055, meeting demands for conventional electricity and electricity for BEVs.

As shown in Fig. 1, demands are met by generators, which in turn consume resources to operate. Conventional electricity demand and transportation demand are exogenously defined by annual quantity and their temporal profiles. Technologies are defined by capital cost, fixed and variable costs (FOM & VOM), resource consumption rate (effi- ciency), lifetime and production profile in the case of variable renew- able energy (VRE) generators such as wind and solar. Resources are also assigned a cost and may further be defined by a finite annual or model period maximum availability.

Each year is further subdivided into representative days with re- presentative hours to capture seasonal and daily variability of demand and renewable output. More detailed information is given in Section 2.3.

System optimization is performed using the Open Source Energy Modelling System (OSeMOSYS) [25,26]. OSeMOSYS is a linear pro- graming tool used for capacity expansion and dispatch of energy sys- tems. It has been used in a variety of studies ranging from expansion of hydroelectric systems in Bolivia [27], flexibility requirements to meet high levels of VRE in Ireland [28], and adoption of biomass retrofitted units to replace stranded coal assets in Alberta [29]. Eqs. (1) and (2) describe the key mathematical formulations of the model. Eq. (1) is the objective function, which minimizes total discounted cost over the model period. Eq. (2) ensures that production of each fuel must be greater or equal to its demand plus its use in any intermediate process.

∑ + + −Minimize CC OC EC SV y t r

y t r y t r y t r y t r , ,

, , , , , , , , (1)

∀ ≥ +Production Demand Use,y l f r y l f r y l f r y l f r, , , , , , , , , , , , (2)

where CC stands for capital investment costs, OC represents operational costs (fixed and variable), EC stands for emissions costs, and SV is the salvage value of remaining technologies at the end of the model period. Production stands for the generation of a particular fuel type, Demand

stands for the demand of a fuel type and Use refers to intermediate use of fuels as input for other processes. The subscripts y t r l f, , , , represent year, technology, region, time step, and fuel, respectively.

The model is also subject to constraints to ensure that enough ca- pacity is built to meet demand and a prescribed reserve margin, that technology capacity limits are not violated, and that carbon emissions limits are enforced, when applicable, among others. A thorough de- scription and full mathematical formulation of the model can be found in [25]. Changes in transmission and distribution capacity are not considered, in the current version.

2.2. Technology options

Available technologies fall into two categories: electricity gen- erators and vehicles. Electricity generators are the technology options that meet the electricity demand, as shown in Fig. 2.

Hydroelectric generators are subdivided into storage hydroelectric (hydro) and run-of-the-river (ROR) hydroelectric. A portion of the en- ergy available from storage hydro is considered as “must run”, in other words, part of it must operate according to seasonal constraints of water inflow into the system [10], the remaining portion of storage hydro, or flexible hydro, may be dispatched at any point in the year, as long as the total annual energy budget is respected. ROR hydro operates si- milarly to the must run portion of storage hydro, where minimum generation values are assigned depending on time of the year.

Additional renewable supplies include wind, solar photovoltaic (PV), and geothermal. Wind and solar have pre-specified generation profiles representing regional resources in B.C. Wind is separated into three regions, the Peace region, the North Coast (NC) region and the Kelly Nicola (KN) region [30]. Profiles for the three wind regions are based on the CanWEA study on wind integration in Canada, 35% TRGT scenario, actual data, where the largest site, by capacity, in each region is selected [31]. The solar generation profile is based on data from PV Watts, for the Cranbrook region. Geothermal and biomass are con- sidered dispatchable generators. Generation from wind, solar PV and hydro ROR is considered as “must take”. In other words, these three generator types cannot curtail generation if the energy is not required at a given time step; however, no monetary penalty is applied to excess electricity generation.

Thermal generation options include open cycle gas turbines (CCGT), combined cycle gas turbines (CCGT) and combined cycle gas turbines with carbon capture and storage (CCGT-CCS). All three generator types consume the same fuel – natural gas. OCGT has a lower capital cost and lower efficiency, used for peaking demand, while CCGT is commonly used at higher capacity factors. CCGT-CCS is similar to CCGT; however, its capital cost is higher and efficiency is lower, with the benefit of a 90% reduction to its carbon intensity.

Some electricity generators can satisfy reserve margin requirements. In addition to the electricity demand, a capacity reserve constraint is also present to ensure firm resource adequacy requirements are met. A reserve margin of 14% of peak annual demand is required [32]. Storage hydro, geothermal, biomass and thermal generators may contribute 100% of their capacity to the reserve margin. In accordance with utility

Fig. 1. Schematic representation of the model. Exogenous demands are met by generators that incur capital, operational and fuel (when applicable) costs. Model calculates optimal capacity mix and dispatch that leads to lowest system cost.

V. Keller, et al. Applied Energy 253 (2019) 113536

3

estimates for regional coincidence with demand, wind contributes 26% of its capacity to the reserve margin, while ROR hydro only contributes 10% of its capacity, equivalent to its lowest annual generation divided by its capacity [30]. Solar PV does not contribute to the reserve margin. All generators are further assigned a lifetime, a fuel consumption rate, maximum annual output, and when applicable, a CO2 emission rate. Values are provided in Section 3.

Vehicle demand is divided into five sub-sectors: heavy-freight, medium-freight, light-freight, passenger vehicles and transit. Each ve- hicle type is prescribed a demand, fuel consumption rate and lifetime. Capital costs for vehicles are not considered; however, fuel consump- tion for vehicles is accounted for. Each vehicle type is assumed to have a battery electric counterpart. This allows for a comparison of variable costs for transportation with and without electrification. Fuel con- sumption for all vehicle types is provided in Section 3.

2.3. Temporal structure

The model period spans from 2015 to 2055. A clustering algorithm is used to reduce computational effort while maintaining temporal ac- curacy. The temporal clustering method is based on the work by Nahmmacher et al. [33] and similar to that employed by Palmer-Wilson et al. [34] and Keller et al. [35]. Clustering analysis using BC profiles for generation and load results in ten representative days per year. Each representative day represents a cluster of days with similar demand profiles, wind capacity factors, and solar capacity factors. Each day is subdivided into 8 representative hours. Electricity demand and wind and solar capacity factor per region is assigned for each representative hour based on historical values for the province.

To capture minimum generation requirements of storage hydro, representative days are assigned from one of three “seasons”. Each model year is comprised of four days from the “off-freshet” season (August to April), three days from the “mid-freshet” season (May and July) and three days from the “peak-freshet” season (June). Additional information on temporal structure can be found in supplemental ma- terial.

2.4. Transportation demand

The forecast for transport electricity demand in terms of annual energy requirement by sub-sector is shown in Fig. 3. Conventional electricity demand projection (Elec) is shown in blue (excludes trans- portation), HD stands for heavy-freight, MD stands for medium-freight, LD is light-freight, passenger refers to personal use vehicles and transit

includes buses, trains or any other type of government operated transportation. The forecast excludes air transport and marine trans- port; which represent a small portion of transportation demand in the province. Transportation demand forecast is based on exponential re- gression of the past 20 years of demand for each sub-sector [36]. Ve- hicle energy consumption data is provided in Section 3.

In scenarios with electrification of vehicles, it is assumed that 100% of new vehicles entering the stock are electric by 2040, in accordance with recent announcements made by U.K., France [11] and British Columbia [12]. This transition of new vehicles from conventional to electric is assumed to increase linearly from zero starting in 2030. Vehicles are assumed to have a 10-year lifetime such the total stock is fully electrified by 2050.

Charging profiles are uncertain and a subject of research. As no data for charging profiles is currently available for BC, charging profiles are modelled in accordance with previous studies. Demand for passenger vehicles is akin to residential charging, as demonstrated by Lojowska et al. [37] and Schey et al. [38]. Commercial use vehicles have been found to have similar charging profiles to personal use vehicles, al- though demand peaks were found to occur slightly earlier in the day [39]. We assume a scenario where charging for personal vehicles and commercial use vehicles is coincident. Demand for Freight-heavy and transit fleets are assumed to be distributed uniformly throughout a day thereby appearing as a baseload, consistent with Keller et al. [35]. Due

Fig. 2. Schematic representation of the modeled energy systems including energy sources, technologies, currencies and services.

0

20

40

60

80

100

120

El ec

tr ic

it y

d em

an d

[ TW

h ]

Year

Elec HD

MD LD

Passenger Transit

Fig. 3. Electricity forecast including conventional demand and vehicle elec- trification. Vehicle stock is assumed to be fully electric by 2050.

V. Keller, et al. Applied Energy 253 (2019) 113536

4

to limited data on the charging profile for transit, it is assumed to have an identical profile to HD-freight. Further, individual vehicles are not explicitly modelled. Rather, a fleet-average demand is imposed, as shown in Fig. 4.

An alternative charging profile is examined by defining a daily profile representing a utility controlled charging scenario (UCC). The UCC profile is described in Section 3.7.

3. Data

In this section, input data for costs, efficiencies and lifetime of electricity generators is presented. As costs for vehicles are not ac- counted for, only their energy consumption is considered.

3.1. Electricity demand

Total electricity demand is subdivided into transportation and standard (or conventional) electricity demand. Transportation demand is described above in Section 2.4. Standard electricity demand is based on BC Hydro’s “Electric Load Forecast, 2012, Reference scenario” [40]. As demand growth from 2012 to 2018 has been slower than forecasted, a factor of 3/4 is applied to match realized demand growth. The de- mand forecast also includes a component representing electrification of vehicles which is removed to avoid double counting.

3.2. Technologies

Cost, lifetime and heat rate of electricity generators is primarily based on the U.S. Energy Information Administration (EIA) – Cost and Performance Characteristics of New Generating Technologies, Annual Energy Outlook 2018, unless otherwise stated [41].

As wind generation costs are dependant on region, wind costs are based on regionally specific estimates provided by BC Hydro [30], but updated to reflect recent cost reductions. The three lowest cost wind regions from the report are used: Peace, North Coast and Kelly Nicola. The Peace region is further divided into three parts, a low cost, medium cost and high cost portion. The North Coast region is divided into a low cost and a high cost portion. Capital cost is calculated based on the report’s unit energy cost and the capacity factor per region, as described in Section 2.2. Further, a learning rate is applied to capital costs of wind generators, decreasing linearly to 2055, consistent with the work of English et al. [26]. Solar generators are subject to learning rates con- sistent with Keller et al. [35]. Capital costs in 2015 and 2055, effi- ciency, and lifetimes are listed in Table 1. Further information on the temporal structure of wind can be found in supplemental material. Solar temporal data is described in Section 2.2.

3.3. Vehicles

Vehicle fuel consumption is based on demand by sub-sector (Section 2.4) and fuel consumption by vehicle type. Fuel consumption by vehicle type is based on Natural Resources Canada’s Comprehensive Energy Use Database [36]. As Natural Resources Canada does not provide an energy consumption forecast, fuel efficiency gains are based on the same rate as EIA’s Annual Energy Outlook 2018 [42]. Vehicle fuel consumption per kilometer for fossil fuel technologies is summarized in Table 2. Vehicle efficiency is assumed constant past 2030.

It is important to note that there are two technology options for passenger vehicles; passenger cars and passenger trucks. To keep the model from selecting only the most efficient type, a minimum annual market share of 44% for passenger trucks is enforced, reflecting current shares [36]. Similarly, annual market shares of 41% and 59% for medium-freight gasoline and medium-freight diesel is enforced, re- spectively, in accordance with the current stock mix.

Energy consumption for BEVs is based on the Argonne National Laboratory’s GREET model [43], and summarized in Table 3. “Car – EV conventional weight” values are used for passenger cars. Values for “electric SUV” are used for passenger trucks and light-freight, as these two sub-sectors are not available in the model. “Refuse truck” values are used for medium-freight. Transit is taken as an average between light-freight and medium-freight. Heavy-freight BEV and heavy-freight fuel cell values are taken from ICCT [44].

3.4. Fuel costs

All fuel costs are based on the EIA AEO 2017 [45]. Fuel costs for OCGT, CCGT and CCGT-CCS are taken from EIA AEO’s electricity generators for Pacific region. Gasoline, diesel and compressed natural gas for transportation costs are taken from the EIA AEO data for transportation for the same region. As the EIA forecast only goes to 2050, data for the last five years is extrapolated assuming a constant growth rate equal to the average of the prior ten years. Additional in- formation on fuel costs is available on Supplemental material.

3.5. Residual capacity

The model residual capacity entails the current capacity of elec- tricity generators by type and their respective expected decom- missioning dates. Storage hydro and ROR hydro have decommissioning dates past 2055, hence the initial capacity remains for the entire model period [26]. Wind capacity is based on [26] and assumed to be located at the peace region. All initial 630 MW of capacity are present in the system until 2032, at which point generators start being decommis- sioned with current capacity fully retired by 2040. Current biomass capacity decreases from 500 MW in 2015 to 40 MW by 2030 and is fully retired by 2045. CCGT capacity decreases from 500 MW in 2015 to 88 MW in 2033, later decreasing to zero by 2045, based on commis- sioning dates of the four existing generators and an expected lifetime of 30 years [26,46,47].

3.6. Capacity constraints

All modelled technologies are allowed to expand their capacity to meet increasing demand and to replace existing capacity being de- commissioned. Capacity for all technologies is restricted to a maximum capacity (MC) and an annual maximum capacity investment (AMCI). MC is the maximum capacity a technology may reach. Gas and solar generators are assigned an unlimited MC. MC for remaining technolo- gies are primarily based on regional limits [32]. Storage hydro is en- forced an expansion of 1.1 GW, equivalent to a new reservoir currently under construction; however, beyond this, no further expansion of storage hydro is allowed. ROR hydro is allowed to expand from 5.5 GW to a maximum of 6 GW. Geothermal capacity has a maximum MC of

0 1 2 3 4 5 6 7 8 9

1 3 5 7 9 11 13 15 17 19 21 23

H o

u rl

y d

em an

d [

G W

h ]

Hour of day

MD + LD + Passenger

HD + Transit

Fig. 4. Example of the baseline charging profile for vehicles for a given model day for the year 2055.

V. Keller, et al. Applied Energy 253 (2019) 113536

5

1 GW. Biomass is allowed to expand to a maximum of 1.2 GW. Wind MC is based on regional supply curves and capacity factors [30]. The values are available on Table 4.

AMCI is defined as the maximum annual increase in capacity from a given year to the following. Storage and ROR hydro, wind and solar are not constrained by AMCI. Gas, geothermal and biomass generators AMCI are equal to 5% of average annual demand, consistent with Keller et al. [29] and Lyseng et al. [48].

3.7. Scenarios

A reference scenario along with an additional three scenarios are modelled, as seen in Table 5. The reference scenario (REF) assumes no vehicles are electrified, mandates a renewable portfolio standard (RPS) consisting of a minimum 93% share of electricity sourced from re- newable sources [49], and a carbon tax of 30 $/tCO2e (consistent with current provincial policy.) Two scenarios impose electrification of ve- hicles in the province, where all road vehicles studied are gradually converted to battery electric, as described in Fig. 3 and in Section 2.4. In the first electrification scenario (ELE-RPS), the renewable energy mandate is enforced. In the other electrification scenario, (ELE-N) the renewable energy mandate is removed. In both scenarios, BEVs follow the charging profile shown in Fig. 4.

The last scenario, UCC-X, is similar to the ELE-RPS scenario, but with a X% of the medium and light freight and passenger vehicle fleet participating in a UCC scheme. X varies between 10 and 50%. In this scenario, the utility controls the time of day when vehicles participating in the scheme are charged. The daily energy demand by sub-sector type is identical to the ELE-RPS scenario, but the utility may decide the time of day when a percentage of the demand is realized. Heavy duty freight and transit demands remain unchanged for all UCC scenarios, as it is assumed these transportation methods operate under strict schedules.

4. Results

Scenarios are compared based on generation capacity buildout, energy mix, share of emissions to 2055, cost, magnitude of excess supply and its temporal characteristics. The REF scenario is presented, followed by vehicle electrification (with and without renewable

Table 1 Summary of costs and generator assumptions by type.

Technology Capital cost 2015 [$/kW] Capital cost 2055 [$/kW] FOM [$/kW-yr] VOM [$/MWh] Heat rate [Btu/kWh] Lifetime [yr]

Hydroa 2898 2898 13.42 5.95 – 80 CCGT 982 982 11.11 3.54 6300 30 CCGT-CCS 2175 2175 33.21 7.08 7525 30 OCGT 680 680 6.87 10.81 9800 30 Biomass 3584 3584 112.15 5.58 13,500 20 Geothermal 5492 5492 119.87 40 Wind – Peace, High 4610 4385 47.47 – – 25 Wind – Peace, Med 2590 2463 47.47 – – 25 Wind – Peace, Low 1900 1807 47.47 – – 25 Wind – NC, High 5300 5041 47.47 – – 25 Wind – NC, Low 3100 2948 47.47 – – 25 Wind – KN 2220 2111 47.47 – – 25 Solar 2004 848 22.02 – – 25

a Hydro values apply to both storage hydro and ROR hydro. Based on [35].

Table 2 Fuel consumption data for all fossil based transportation technologies.

Fossil Technology Fuel consumption (GJ/thousand-km)

2015 2030

Passenger car 2.8 2.0 Passenger trucks 3.9 2.7 Light-freight 4.0 3.7 Medium-freight diesel 8.6 7.8 Medium-freight gasoline 7.9 7.4 Heavy-freight diesel 16.1 10.7 Heavy-freight natural gas 7.6 13.0 Transit 6.1 4.4

Table 3 Fuel consumption data for all electric based transportation technologies.

Alternative Technology Energy consumption (kWh/km)

2015 2030

Passenger car 0.23 0.18 Passenger trucks 0.3 0.24 Light-freight 0.3 0.24 Medium-freight 1.22 0.98 Heavy-freight BEV 2.93 2.13 Heavy-freight fuel cell 2.64 2.07 Transit 0.76 0.61

Table 4 Maximum capacity limits by wind region.

Region Maximum model capacity (GW)

Wind – Peace, High Unlimited Wind – Peace, Med 2.21 Wind – Peace, Low 6.67 Wind – NC, High 0.96 Wind – NC, Low 2.37 Wind – KN 3.33

Table 5 Summary of modelled scenarios.

Scenario Description

REF Vehicles remain fossil fuel dependent. Renewable energy requirement enforced ELE- RPS All road vehicles are gradually electrified. Charging profile for all vehicles is fixed. Renewable energy requirement enforced ELE-N Similar to ELE-RPS, but renewable energy mandate is removed UCC-X Similar to ELE-RPS, but X% of vehicles adopt UCC. X varies from 10 to 50%

V. Keller, et al. Applied Energy 253 (2019) 113536

6

mandate), and, finally, utility controlled charging.

4.1. Reference scenario

4.1.1. Capacity and generation Reference scenario results (no vehicle electrification) for system

expansion (capacity) and dispatch (energy) are shown in Fig. 5. From 2015 to 2055, the system remains dependent on hydroelectricity due to its low cost and flexibility. Storage hydro is expanded by 1.1 GW in the 2020s, providing an additional 5 TWh of energy annually. ROR hydro expands by 0.5 GWs in the 2020s, but due to its low capacity factor, only adds 0.7 TWh of annual energy generation. CCGT generation is expanded due to low cost and flexibility, but remains limited as a result of RPS requirements. OCGT capacity is expanded from zero in 2015 to 1.7 GW in 2055 primarily for system reliability, with a capacity factor of just 0.5% in the last 10 years of the model period. Combined, OCGT and CCGT reach the RPS limit of 7% of energy every year after 2033.

Geothermal capacity reaches 0.7 GW by 2055, which is 0.3 GW less than the capacity limit set exogenously. Although more expensive than wind and solar on a levelized cost of energy basis, geothermal is built due to its dispatchability and because it is able to contribute 100% of its capacity to the reserve margin requirement. Biomass capacity is even- tually retired past the mid 2030s whereas wind capacity is expanded until the late 2020s due to more favourable costs and to ensure the minimum renewable generation constraint of 93% set by the RPS is met. Starting in the early 2040s, wind capacity is replaced by solar due to decreasing PV system costs. Although wind can contribute a portion of its capacity to the reserve margin, the increasing difference between the capital cost of solar and wind in the last third of the model period makes solar the preferred option.

4.1.2. Emissions Total emissions, including electricity and transport, experience a

30% increase from 2015 to 2055, as shown in Fig. 6. Emissions from the electricity sector quadruple between this period, increasing from 0.5 MtCO2e per year in 2015 to just below 2 MtCO2e by 2055. This in- crease is due to the fact that currently the system does not reach its maximum allowable share of fossil generation allowed by the RPS (7%) and demand is low. However, increase in demand and flexibility re- quirements lead to an increase in gas generation.

Emissions from the transport sector experience a small decrease to the mid 2020s, before monotonically increasing to 2055. Although demand for vehicles increases from 2015 to 2055, expected gains in fuel efficiency lead to modest decreases in transport emissions to the mid 2020s. After this period, vehicle fuel efficiencies increase at a lower rate, and, combined with the increasing transport demand, lead to in- crease in fuel emissions.

4.1.3. Operation Excess supply of electricity happens in every year of the REF sce-

nario. As mentioned in Section 2.2, excess generation from wind, solar PV and hydro is not curtailed. The magnitude of excess supply varies by year, ranging from 0.7 to 3 TWh annually, and is primarily a seasonal phenomenon with roughly 30% of excess supply happening during the peak freshet period (June) and the remaining happening in the mid- freshet period (May and July). Although the magnitude of excess supply remains somewhat constant throughout the model period, typically between 1.7 and 2 TWh per year, the driver behind it changes over time. As shown in Fig. 7, excess supply happens early in the model period (2015) primarily due to excess supply of ROR hydro generation, minimum generation requirements from storage hydro, and wind con- tributing to a smaller degree. However, as seen in Fig. 7 (right), late in the model period (2055) minimum demand has increased, closely matching minimum generation requirements from ROR hydro and storage hydro. At this point, the minimum generation requirements of hydro combined with solar PV generation lead to excess supply during sunny hours of the day.

4.2. Electrification of transport

4.2.1. Capacity and generation Scenarios for electrification of transportation with (ELE-RPS) and

without (ELE-N) a renewable energy requirement are compared to the reference case. The impacts of vehicle electrification on generation capacity in the years 2015 and 2055 are summarized in Fig. 8. In the reference scenario, with no electrification of vehicles, capacity expands from 15.6 GW in year 2015 to 23 GW – an increase of nearly 50%. This

Fig. 5. Model results for REF scenario installed system capacity mix (left) and energy generation by source (right) for selected years.

Fig. 6. Total system emissions for REF scenario including electricity and transport sectors.

V. Keller, et al. Applied Energy 253 (2019) 113536

7

increased demand is associated with population growth and expansion of industry and commerce in the province [40]. To meet the RPS re- quirement, geothermal and solar PV account for 60% of the capacity increase.

Capacity growth is larger for the transport electrification scenarios. In the ELEC-RPS scenario, total installed capacity increases by a factor of 2.35 from 2015 values, or 60% higher than the REF scenario. By 2055 with 100% electric transport, wind capacity reaches 7.5 GW along with 5.5 from solar and 1 GW from geothermal. Combined, the installed capacity of these three generators nearly match the system capacity of 15.6 GW in 2015.

Transport electrification without the RPS requirement (ELE-N) leads to a reduction of nearly 5 GW of installed capacity compared to the ELE- RPS scenario. For ELE-N, wind capacity is zero in 2055 while CCGT capacity has increased. Further, both electrification scenarios show roughly 5 GW of OCGT capacity in 2055. However, as shown in Fig. 8 (right), OCGT is mainly present to back up solar PV, as capacity factors in both scenarios are close to 3%.

4.2.2. Cost and emissions Table 6 summarizes system cost (present value of total cost) and

cumulative emissions for the REF and vehicle electrification scenarios, ELE-RPS and ELE-N. Total system cost increases by 17 and 10% from reference scenario for the ELE-RPS and ELE-N, respectively. However, demand in the scenarios with electrification is 36% higher than in the REF scenario. The unit energy cost (UEC) is defined as the total system cost divided by total electricity generated. Compared to the reference case, UEC increases by 9% in the ELE-RPS scenario and 3% in the ELE-N scenario. Although the increase in electricity cost is lower in the ELE-N scenario than in the ELE-RPS scenario, so is the total emission reduc- tion.

The ELE-RPS scenario achieves a cumulative emission reduction of 260 MtCO2, or a 38% reduction relative to the REF scenario whereas the ELE-N scenario results in a 15% reduction from the REF scenario. Abatement costs are calculated by dividing the increase of total elec- tricity system cost by the emission decrease relative to REF scenario. At 14.2 $/tCO2 the ELE-RPS abatement cost is 30% lower than the ELE-N scenario. The system costs represent generation fixed and variable costs only; transmission and distribution or re-charging infrastructure costs are not accounted for.

Total combined electricity and transport system cost decreases with electrification of vehicles. Fig. 9 shows total costs for electricity system and transportation fuels. Although the electricity system cost is found to slightly increase with electrification of vehicles due to the demand growth, this cost increase is offset by the cost reduction associated with phasing out fossil fuels for transport. Model results show that a 17.13 $B and 18.66 $B total cost reduction would be achieved with the ELE-RPS and ELE-N scenarios, respectively, in reference to the REF scenario. However, as mentioned above, this cost reduction does not account for electricity transmission and distribution costs, nor incre- mental capital costs for electric vehicles.

4.2.3. Operation One of the consequences of the RPS requirement is that excess en-

ergy supply increases by a factor of 2.4 from the REF scenario by 2055. Fig. 10 shows a five year moving average of excess supply over the model period for the same three scenarios. As seen, excess electricity supply remains relatively constant for the REF scenario, varying be- tween 1.7 and 2 TWh per year, as discussed in the previous section. However, in the ELE-RPS scenario, where vehicles are electrified and the renewable standards are kept, excess electricity supply grows quickly in the 2040s, due to mismatch between electricity demand and

Fig. 7. Hourly demand (dotted line) and generation results for a mid-freshet day for 2015 (left) and 2055 (right).

Fig. 8. Total installed capacity (left) and generation by source (right) for year 2015 and for REF, ELE-RPS and ELE-N scenarios for year 2055.

V. Keller, et al. Applied Energy 253 (2019) 113536

8

solar PV generation. As the overall electricity must comply with the minimum 93% renewable standard and renewables generation is con- sidered “must take”, excess generation increases. However, in the ELE- N scenario, where electrification of vehicles happens, but the RPS standards are removed, excess generation decreases to zero by the late 2030s due to the increased use of natural gas. Excess generation starts increasing again in the early 2050s due to the increased share of solar PV. In the modelled system, there is no monetary penalty for any excess generation.

4.3. Utility controlled charging

In this section, model results for scenarios employing varying levels of utility-controlled charging (UCC-X) are shown. Results only include scenarios with RPS standards, as scenarios without the RPS were found not to vary significantly with UCC level.

4.3.1. Capacity and generation Implementation of UCC leads to a decrease in total installed capa-

city as well changes in the generation mix. Fig. 11 shows the difference in installed capacity with varying levels of UCC. Positive values

represent an increase in installed capacity from the ELE-RPS scenario, while negative values represent a decrease. In the UCC-10 scenario, total capacity decreases by 1 GW, or just less than 3% of the total ca- pacity of the ELE-RPS scenario. In the UCC-50 scenario, total capacity decreases 2.6 GW or just over 7% of the capacity of the ELE-RPS sce- nario.

UCC does not impact hydro and geothermal capacity, however, other technologies experiences capacity changes. In the UCC-10 sce- nario, capacity of OCGT, CCGT wind and solar decrease by allocating vehicle charging to times of wind and solar generation, decreasing overall capacity requirements. However, in higher UCC levels, OCGT, CCGT, biomass and wind capacities decrease and are partially replaced by solar PV. This can be attributed to vehicles being charged during the day when low cost solar electricity is available.

The decrease in OCGT capacity is due to the beneficial effects of increased penetration of UCC to manage demand during annual peak periods. As shown in Fig. 12, in the winter, when storage hydro gen- eration is lower, the system needs to deploy OCGT generation to meet peaks associated with vehicle charging demand (charging profile as shown in Fig. 4.) However, with higher levels of UCC, the system is able to displace some of this demand to other times of the day, when either flexible hydro or solar have available capacity to meet it.

4.3.2. Cost and emissions System cost is reduced relative to the ELE-RPS scenario when UCC is

implemented as shown in Table 7. The decreased installation capacity needs and shift in capacity type shown in Fig. 11 lead to decreases in system cost. Employing UCC in 50% of the available fleet leads to a system cost decrease of 3.5%. Although there is no significant change in emissions with the use of UCC, the lower system cost leads to lower carbon abatement cost; a decrease of up to 25% from the ELE-RPS scenario is achieved.

An annual cost reduction per vehicle is calculated by dividing non- discounted system cost difference for each UCC scenario and the ELE- RPS scenario by the number of participating vehicles and number of years the UCC scheme is utilized, results are shown in Fig. 13. Non- discounted costs are used to represent cost savings in 2015 values. The blue bars cost savings due to lower capital, O&M and fuel costs.

UCC leads to cost savings equivalent to up to $82/vehicle-year, in the best case. However, as UCC penetration increases, the value of

Table 6 Summary of system costs, unit energy costs, emissions, and abatement cost for REF and vehicle electrification scenarios. Abatement costs represent cost increase over REF scenario divided by emission decrease.

Scenario Total electricity system cost [$B] Unit Energy Cost [$/MWh] Cumulative Emissions [MTCO2] Abatement Cost [$/tCO2]

REF 21.9 21.8 691 – ELE-RPS 25.6 23.7 430 14.2 ELE-N 24.0 22.4 588 20.4

0

20

40

60

80

100

120

140

REF ELE-RPS ELE-N

C o

st [

$B ]

Electricity system

Transport fuel

Fig. 9. Total cost for electricity system and transport fuel (gasoline and diesel) by scenario. Results do not account for electricity transmission and distribution costs.

0

1

2

3

4

5

6

2015 2035 2055

A n

n u

al e

xc es

s el

ec tr

ic it

y su

p p

ly

[T W

h ]

Year

REF

ELE-RPS

ELE-N

Fig. 10. Five year moving average of excess energy generation by scenario.

-5

-4

-3

-2

-1

0

1

2 UCC-10 UCC-20 UCC-30 UCC-40 UCC-50

In st

al le

d c

ap ac

it y

d iff

er en

ce

fr o

m E

LE -R

P S

[G W

]

OCGT CCGT Bio Wind Solar

Fig. 11. Difference in installed capacity, in reference to ELE-RPS scenario, with varying levels of UCC.

V. Keller, et al. Applied Energy 253 (2019) 113536

9

additional unit of energy displaced diminishes. This leads to the di- minishing returns seen in Fig. 13. At a penetration of 50%, the value of UCC drops from $82 to $67/vehicle-year. It is important to note, however, that these costs do not account for transmission and dis- tribution system costs. Cost savings would likely be higher if these were considered.

4.3.3. Operation Dispatchable capacity decreases with UCC level due to decreased

seasonal peak demands. Model results show that although use of UCC leads to a modest annual peak demand reduction, off-freshet demand reduction is significant, as shown in Table 8, which summarizes peak demand by season for the year 2055 (only ELE-RPS and UCC-50 are shown.) As seen in Table 8, annual peak demand decreases by 0.4 GW when employing UCC at 50% of the available fleet, whereas, the off-

freshet peak is reduced by 2 GW. During the mid and peak-freshet seasons, backup capacity needs are lower, due to the high amounts of storage hydro and hydro ROR available; therefore, decreased peak de- mand in these periods has a reduced value to the system. In the off- freshet season hydro generators have a lower output, effectively low- ering their contribution towards capacity needs. Consequently, low- ering peak demand in the off-freshet season has a greater value to the system as it enables lower backup capacity installation.

5. Discussion

This work studies the impacts of electrifying all road transportation sub-sectors in the province of British Columbia and the effects in terms of capacity buildout and excess electricity generation to the electricity system. It is important to note, however, that all costs reported here only include the electricity system, and exclude transmission and dis- tribution system costs, vehicle costs and charging infrastructure costs. Further, this study only accounts for vehicles emissions associated with vehicle operation. Vehicle manufacturing emissions are not accounted for. It is important to note that all results of the current study are system specific and only fully applicable to BC. However, similar results could be found for regions with a similar hydroelectricity share such as Quebec, Northern Europe or South America if they were to apply a similar RPS.

Eliminating the RPS would decrease the accumulated carbon re- duction impact of electrification of transportation, leading to higher carbon abatement costs. Model results show that total system carbon emission reductions are achieved by electrification of transport when the RPS is eliminated. However, this effect is 60% lower, when com- pared to the scenario where the RPS is maintained. Although system capacity requirements decrease when eliminating the RPS, unit energy cost only increase by less than 6%. Removing the RPS would negate some of the benefits of electrification of the vehicle fleet as carbon abatement costs increase by 44%. If governments are serious about deep decarbonisation targets, combined measures in both the electricity and transportation sectors are necessary.

Electricity system cost increases by 17% in the ELE-RPS scenario in comparison to the REF scenario; however, this is due to larger demand for electric vehicles. When accounting for the additional electricity being generated, unit electricity cost only increases by 9% in the ELE-

Fig. 12. Hourly generation for ELE-RPS (left) and UCC-50 (right) scenarios for the same representative day in 2055. Area between line with black dots and line with blue crosses represent shifted demand resulting from UCC.

Table 7 Summary of system costs, unit energy costs, emissions, and abatement cost for REF, ELE-RPS and UCC scenarios.

Scenario Total electricity system cost [$B]

Unit Energy Cost [$/MWh]

Cumulative Emissions [MTCO2]

Abatement Cost [$/tCO2]

REF 21.9 21.8 691.0 – ELE-RPS 25.6 23.7 430.0 14.2 UCC 10% 25.3 23.5 430.0 13.2 UCC 20% 25.2 23.3 429.3 12.5 UCC 30% 25.0 23.2 429.0 11.8 UCC 40% 24.8 23.1 428.5 11.2 UCC 50% 24.7 22.9 427.9 10.6

0

10

20

30

40

50

60

70

80

90

UCC-10 UCC-20 UCC-30 UCC-40 UCC-50

A n

n u

al c

o st

r ed

u cti

o n

p er

v eh

ic le

[$

/v eh

ic le

- ye

ar ]

Infrastructure

Fig. 13. Annual equivalent cost savings resulting from implementation of UCC per participating vehicle.

Table 8 Peak generation by period for varying UCC levels for year 2055 in GW.

Off- freshet Mid-freshet Peak-freshet Annual peak

ELE-RPS 17.5 16.1 16.8 17.5 UCC-50 15.5 15.8 17.1 17.1

V. Keller, et al. Applied Energy 253 (2019) 113536

10

PRS scenario and 5.5% in the UCC-50 scenario. These results suggest that the use of renewable generators may enable an expansion of the electricity system to meet demand from electric vehicles, while keeping emissions low and maintaining electricity prices to a minimal increase from a reference scenario where electrification would not happen.

Results show a capacity expansion requirement from 2015 to 2055 of 134% in the ELE-RPS scenario to accommodate demand from electric vehicles. This value is shown to reduce to 117% employing UCC of 50% of the available fleet. Although this difference in system capacity be- tween the ELE-RPS and the UCC-50 scenarios is significant, a smaller cost difference between these scenarios is found, as system cost only differ by 3.3%. These results suggest that total system capacity expan- sion requirements may not be proportionally representative to system cost increase due to the falling costs of wind and solar PV generators.

Use of UCC leads to system benefits by reducing peak demand and shifting it to hours of low demand. The results of the study suggest that there are system benefits to employing UCC, by displacing demand from the early evening peaks to the middle of the day where generation from solar PV is abundant. In this context, our study is in agreement with Gnann et al, who found that use of UCC would displace peak night charging events to the middle of the day when electricity from solar PV is more abundant [50]. In contrast, the works of Schill et al. [17] and Li et al. [18] found that rather than promoting increased use of VRE, UCC predominantly displaced demand to hours where coal was the marginal generator, leading to lower emission benefits. These results suggest that these dynamics are system specific and careful consideration of each jurisdiction is necessary.

UCC leads to increased freshet peak demand. The results of the study find that the use of UCC resulted in decreased peak demand for the majority of the year. However, freshet peak demands increased with the use of UCC, which may be counterintuitive. Due to low cost of solar PV by mid-century, the system opts to increase the buildout of this generator type and displace demand in the freshet to mid-day, effec- tively increasing the peak demand in this season. By doing so, water is saved, which can then be used at other times when its value is higher. These results demonstrate that use of UCC does not necessarily reduce demand peaks, rather it leads to demand shifting that leads to lowering system cost.

Economic benefit per vehicle UCC is low. Wolinetz et al. [23] and Weis et al. [21] find benefits of $50 to $70/vehicle-year and $100/ vehicle-year, respectively. Although both studies only consider the passenger vehicle sector and with partial BEV penetration, 50% and 10% respectively, their results are similar to the maximum $82/vehicle- year found in the current study. In comparison, the average passenger vehicle in BC would be expected to consume close to $300–400 a year in terms of electricity for vehicle recharging, considering a 10,000 to 15,000 km annual range and electricity prices at $0.14/kWh [51]. A benefit of $82/vehicle-year, found for a vehicle participation rate of 10%, would result in a cost benefit of 20% to 30% of annual electricity refueling costs per vehicle. This ratio would be significantly lower for commercial vehicles or for higher vehicle participation rates. As a re- sult, this relatively low economic benefit of UCC may attract a limited number of participants, which highlights its limited capability as a demand side management application.

The study also finds that electrification of vehicles may lead to a total combined electricity and transport system cost reduction. As is shown previously in Fig. 9, total the cost increase of the electricity system is offset by savings associated with phasing out use of gasoline and diesel. The current study does not account for transmission and distribution system expansion, rather it is a single region model with no transmission or distribution constraints. The cost savings may be con- sidered an upper bound for additional infrastructure costs, after which electrification would be more expensive than fossil transportation. Fu- ture work is necessary to evaluate broader system costs of electrifica- tion of the road transportation segment accounting for infrastructure changes.

The option of using hydrogen fuel cells for heavy-duty vehicles as a load shifting method by creating hydrogen from excess electricity was examined following the study in [35]. However, results showed that system costs increased from the ELE-RPS scenario by using fuel cell vehicles. This is in disagreement with our previous piece that found cost savings associated with using fuel cell heavy duty vehicles in Alberta, compared to battery electric vehicles [35]. The difference in the results is due to two features. Firstly, Alberta does not have the hydroelec- tricity resources present in British Columbia. As a result, additional flexibility has a much greater value to Alberta than in B.C. for managing variability. Secondly, Alberta has better solar resources than B.C. Use of electrolysers were found to have a high temporal correlation with solar generation in the previous study. The lower solar resource in B.C. would ultimately lead to higher hydrogen costs, making use of fuel cell ve- hicles less economical.

In the current study, the only considered charging profile for ve- hicles is found on Fig. 4. The profile used for passenger vehicles, light duty freight and medium duty freight assumes that most of the charging for these vehicles happens in the late evening hours, akin to home charging or charging after business hours. However, commercial char- ging has been demonstrated to lead to demand peaks happening earlier in the day [52]. If charging profiles focusing on commercial charging were to be used, it is likely that the value of UCC would be further decreased, as the majority of UCC displaced the demand to hours of high solar output.

5.1. Model limitations

Although the current study provides insights into electrification of vehicles for hydroelectric dominated jurisdictions, a number of limita- tions exist. The main limitations are the lack of ramping constraints and assuming that customers are willing to let the utility control the char- ging of their vehicles.

The electricity system cost does not account for ramping constraints. Due to the temporal structure of the model it would be difficult to implement ramping constraints. Therefore, the ramping ability of thermal generators such as biomass of CCGT may be overestimated. However, model results show limited use of both generator types. Biomass represents 3.5% of total generation at its peak, likely leading to minor deviation of results. Further, CCGT generators have been re- ported to be able to ramp 8% of full load per minute [53]. As a results it is unlikely, it is unlikely that this constraint would lead to a significant change in model results.

No ancillary services are considered. Ancillary services may include spinning reserves, voltage regulation and ramping capacity. These services are not considered in the current model for simplicity. At the time of submission of the current research piece, the authors have an- other research piece under review that explores these ancillary services requirements and demonstrates that BC has sufficient hydro resources to provide them. As a result, removing these services is not likely to lead to significant changes in the results. However, other techniques such as the use unit commitment modelling as shown in Dagoumas [54] and Koltsaklis [55] could be employed in the future to address this issue.

Modelled capacity credit contribution of wind generators is static. In real electricity systems, firm capacity credit of wind power decreases with wind penetration. However, due to the linear nature of the model, a variable capacity credit contribution is not possible to implement. As a result, a static contribution of 26% is applied, in accordance with BC Hydro’s IRP [30]. However, the results of previous studies suggest that a 26% contribution may be appropriate for the wind penetration values achieved [56] and capacity factors used in the current study [57].

6. Conclusion

Using the British Columbia system as a case study, we analyzed the impact on electricity generation capacity expansion, cost and emissions

V. Keller, et al. Applied Energy 253 (2019) 113536

11

associated with electrifying the entire road vehicle fleet. In addition to electrification of passenger vehicles, the model includes electrification of the entire freight and transit sectors, which have received little at- tention in the literature to date. Electricity system cost and unit energy cost increase resulting from vehicle electrification are also quantified. A capacity expansion and dispatch model spanning from 2015 to 2055 is used to analyze scenarios with and without a renewable portfolio standard, where a minimum of 93% of electricity generated in the province must be sourced from renewables. We further studied the impact of applying utility controlled charging on up to 50% of the vehicle fleet, in steps of 10% and quantified the value of this scheme per vehicle-year.

Results show that in scenarios with electrification of vehicles, ca- pacity expansion to 2055 is up to 60% higher than in a scenario where vehicle electrification does not take place. Although this may seem like a significant difference, model results show that unit energy cost only increases by 9%. Further, this value is found to decrease to 5% when UCC is used. These results demonstrate that electrification of the transport system can be carried out at low additional cost to the elec- tricity system.

Removing the renewable portfolio standard diminishes emission reduction benefits of electrification by 60%. Electrification of vehicles with the Renewable portfolio standard leads to emissions reduction of 260 MtCO2 over the model period, however, this value drops to 102 MtCO2 if the renewable portfolio standard is removed, which is equivalent to a reduction of 60%. This decrease in emission reductions diminishes the impact of electrification of vehicles, further increasing the carbon abatement cost by 47%. However, the scenario where the renewable portfolio standard is enforced leads to excess energy gen- eration over 6 times higher than the scenario without the renewable portfolio standard.

Use of utility controlled charging leads to a reduction in excess energy generation and reduction in required generation capacity, however, marginal impact diminishes with number of participating vehicles. Results show that use of utility controlled charging may de- crease capacity needs by up to 7%, in comparison to a scenario where the scheme is not employed, leading to a system cost decrease of 3%. However, due to the large number of vehicles participating, the value per vehicle is relatively low. In the best case, value is found to be $ 82/ vehicle-year, with a participation rate of 10% of eligible vehicles. However, when the participation rate increases to 50%, the value de- creases to just $ 67/vehicle-year. These results suggest that the low value per vehicle-year might lead to reluctance in the adoption of UCC, especially in the freight segment.

Funding

This work was supported by the by the Pacific Institute of Climate Solutions (PICS).

Declaration of Competing Interest

There are no conflicts of interest to declare.

Appendix A. Supplementary material

Supplementary data to this article can be found online at https:// doi.org/10.1016/j.apenergy.2019.113536.

References

[1] United States Environmental Protection Agency. Global Greenhouse Gas Emissions Data. [Online] [cited: 02 05, 2019] https://www.epa.gov/ghgemissions/global- greenhouse-gas-emissions-data.

[2] International Energy Agency. Transport, Energy and CO; 2009. [Online]. https:// www.iea.org/publications/freepublications/publication/transport2009.pdf.

[3] Niet Taco, et al. Valuing infrastructure investments to reduce curtailment. Energy

Strategy Rev. 2018;22:196–206. [4] Varone E, Ferrari M. Power to liquid and power to gas: an option for the German

Energiewende. Renew Sustain Energy Rev 2015;45(207–218):207–18. [5] Habib Salman, Kamran Muhammad, Rashid Umar. Impact analysis of vehicle-to-

grid technology and charging strategies of electric vehicles on distribution networks – a review. J Power Sources 2015;277:205–14.

[6] International Energy Agency. Tracking Clean Energy Progress 2017; 2017. [Online]. https://www.iea.org/publications/freepublications/publication/ TrackingCleanEnergyProgress2017.pdf.

[7] U.S. Energy Information Administration. Annual Energy Outlook; 2019. [Online] [Cited: 02 05, 2019] https://www.eia.gov/outlooks/aeo/pdf/aeo2019.pdf.

[8] Koltsaklis Nikolaos, Georgiadis Michael. A multi-period, multi-regional generation expansion planning model incorporating unit commitment constraints. Appl Energy 2015;158:310–31.

[9] National Renewable Energy Laboratory. Wind and Solar Energy Curtailemt: Experience and Practices in the United States. [Online] https://www.nrel.gov/ docs/fy14osti/60983.pdf.

[10] B.C. Hydro. Transmission Service Freshet Rate – Appendix D – BC Hydro System Conditions during Freshet and Associated Management Strategies; 2017.

[11] Sperling Daniel. Electric vehicles: approaching the tipping point. Bull Atom Sci 2018;74:11–8.

[12] Government of British Columbia. Office of the Premier. [Online] Provincial gov- ernment puts B.C. on path to 100% zero-emission vehicle sales by 2040, November 20, 2018. [Cited: December 3, 2018.] https://news.gov.bc.ca/releases/ 2018PREM0082-002226.

[13] Madzharov D, Delarue E, D'haeseleer W. Integrating electric vehicles as flexible load in unit commitment modeling. Energy 2014;65:285–94.

[14] Kelly Jarod, MacDonald Jason, Keoleian Gregory. Time-dependent plug-in hybrid electric vehicle charging based on national driving patterns and demographics. Appl Energy 2012;94:395–405.

[15] Rosler Hilke, et al. Electricity versus hydrogen for passenger cars under stringent climate change control. Sustain Energy Technol Assess 2014;5:106–18.

[16] Graabak Ingeborg, et al. Optimal planning of the Nordic transmission system with 100% electric vehicle penetration of passenger cars by 2050. Energy 2016;107:648–60.

[17] Schill Wolf-Peter, Gerbaulet Clemens. Power system impacts of electric vehicles in Germany: charging with coal or renewables. Appl Energy 2015;156:185–96.

[18] Li Ying, et al. Electric vehicle charging in China's power system: energy, economic and environmental trade-offs and policy implications. Appl Energy 2016;173:535–54.

[19] Hedegaard Karsten, et al. Effects of electric vehicles on power systems in Northern Europe. Energy 2012;48:356–68.

[20] Prebeg Pero, et al. Long-term energy planning of Croatian power system using multi-objective optimization with focus on renewable energy and intefration of electric vehicles. Appl Energy 2016;184:1493–507.

[21] Weis Allison, Jaramillo Paulina, Michalek Jeremy. Estimating the potential of controlled plug-in hybrid electric vehicle. Appl Energy 2014;115:190–204.

[22] Lyon Thomas, et al. Is “smart charging” policy for electric vehicles worthwhile? Energy Policy 2012;41:259–68.

[23] Wolinetz Michael, et al. Simulating the value of electric-vehicle–grid integration using a behaviourally realistic model. Nat Energy 2018;3:132–9.

[24] Taljegard M, et al. Impacts of electric vehicles on the electricity generation portfolio – a Scandinavian-German case study. Appl Energy 2019;235:1637–50.

[25] Howells Mark, et al. OSeMOSYS the open source energy modelling system an in- troduction to its ethos, structure and development. Energy Policy 2011;39:5850–70.

[26] English J, et al. Impact of electrical intertie capacity on carbon policy effectiveness. Energy Policy 2017;101:571–81.

[27] de Moura Pinto, Nikolaus Gustavo, et al. South America power integration, Bolivian electricity export potential and bargaining power: an OSeMOSYS SAMBA approach. Energy Strategy Rev 2017;17:27–36.

[28] Welsch Manuel, et al. Incorporating flexibility requirements into long-term energy system models – a case study on high levels of renewable electricity penetration in Ireland. Appl Energy 2014;135:600–15.

[29] Keller Victor, et al. Coal-to-biomass retrofit in Alberta –value of forest residue bioenergy in the electricity system. Renew Energy 2018;125:373–83.

[30] B.C. Hydro. Integrated Resource Plan. Appendix 3A-1. Resource Options Report Update; 2013.

[31] Canadian Wind Energy Association. Wind integration study – wind data. [Online] https://canwea.ca/wind-integration-study/wind-data/.

[32] B.C. Hydro. Integrated Resource Plan – Resource Options. [Online] https://www. bchydro.com/content/dam/BCHydro/customer-portal/documents/corporate/ regulatory-planning-documents/integrated-resource-plans/current-plan/0003-nov- 2013-irp-chap-3.pdf.

[33] Nahmmacher Paul, et al. Carpe diem: a novel approach to select representative days for long-term power system modeling. Energy 2016;112:460–1442.

[34] Palmer-Wilson Kevin, et al. Impact of land requirements on electricity system decarbonisation pathways. Energy Policy 2019;129:193–205.

[35] Keller Victor, et al. Electricity system and emission impact of direct and indirect electrification of heavy-duty transportation. Energy 2019;172:740–51.

[36] Natural Resources Canada. Comprehensive Energy Use Database – Transport Sector – British Columbia. [Online] http://oee.nrcan.gc.ca/corporate/statistics/neud/ dpa/menus/trends/comprehensive/trends_tran_bct.cfm.

[37] Lojowska Alicja, et al. From transportation patterns to power demand: stochastic modeling of uncontrolled domestic charging of electric vehicles. IEEE 2011:1–7.

[38] A first look at the impact of electric vehicle charging on the electric grid in the EV project. Schey, Stephen, Scoffield, Don and Smart, John. Los Angeles, California:

V. Keller, et al. Applied Energy 253 (2019) 113536

12

EVS26’; 2012. [39] Electricity demand of PHEVs operated by private households and commercial fleets:

effects of driving and charging behaviour. Smart, John, et al. Shenzhen, China: EVS- 25; 2010.

[40] B.C. Hydro. B.C. Hydro. Electric Load Forecast. [Online]; 2012. https://www. bchydro.com/content/dam/BCHydro/customer-portal/documents/corporate/ regulatory-planning-documents/integrated-resource-plans/current-plan/2012- electric-load-forecast-report.pdf.

[41] U.S. Energy Information Administration. Cost and Performance Characteristics of New Generating Technologies, Annual Energy Outlook; 2018. [Online] https:// www.eia.gov/outlooks/aeo/assumptions/pdf/table_8.2.pdf.

[42] Energy Information Administration. Annual Energy Outlook 2018. Transportation Sector Key Indicators and Delivered Energy Consumption. [Online] https://www. eia.gov/outlooks/aeo/data/browser/#/?id=7-AEO2018&cases=ref2018& sourcekey=0.

[43] Argonne National Laboratory. GREET Model. [Online] https://greet.es.anl.gov/. [44] The International Council on Clean Transportation. Transitioning to Zero-Emission

Heavy-Duty Freight Vehicles; 2017. [45] Energy Information Administration. Cost and Performance Characteristics of New

Generating Technologies, Annual Energy outlook 2017; 2017. [46] Capital Power Corporation. Generation portfolio. [Online] https://www.

capitalpower.com/generationportfolio/CA/Pages/IslandGeneration.aspx. [47] ATCO Power. Our Facilities. [Online] http://www.atcopower.com/Our-Facilities/

Our-Power-Technologies/Cogeneration/McMahon. [48] Lyseng Benjamin, et al. Decarbonising the Alberta power system with carbon

pricing. Energy Strategy Rev 2016;10:40–52. [49] Government of Brisith Columbia. Energy Efficiency Policy and Regulations.

[Online] https://www2.gov.bc.ca/gov/content/industry/electricity-alternative- energy/energy-efficiency-conservation/policy-regulations.

[50] Gnann Till, Klingler Anna-Lena, Kühnbach Matthias. The load shift potential of plug-in electric vehicles with different amounts of charging infrastructure. J Power Sources 2018;390:20–9.

[51] BC Hydro. Residential rates. [Online] https://app.bchydro.com/accounts-billing/ rates-energy-use/electricity-rates/residential-rates.html.

[52] Robinson AP, et al. Analysis of electric vehicle driver recharging demand profiles and subsequent impacts on the carbon content of electric vehicle trips. Energy Policy 2013;61:227–348.

[53] Gonzalez-Salazar Miguel Angel, Kirsten Trevor, Prchlik Lubos. Review of the op- erational flexibility and emissions of gas- and coal-fired power plants in a future with growing renewables. Renew Sustain Energy Rev 2018;82:1497–513.

[54] Dagoumas Athanasios, Koltsaklis Nikolaos. Review of models for integrating re- newable energy in the generation expansion planning. Appl Energy 2019;242:1573–87.

[55] Koltsaklis Nikolaus, Dagoumas Athanasios. State-of-the-art generation expansion planning: a review. Appl Energy 2018;230:563–89.

[56] Ibanez Eduardo, Daryanian Bahman, Stenclik Derek. Capacity value of Canadian wind and the effects of decarbonization. Ninth annual IEEE green technologies conference. 2017.

[57] Voorspools Kris, D’haeseleer William. An analytical formula for the capacity credit. Renew Energy 2006;31:45–54.

V. Keller, et al. Applied Energy 253 (2019) 113536

13