You are on page 1of 6

Transportation Modeling and Data Needs for Fast Charging Electric

Vehicles
Dylan Dean Thomas Ortmeyer Lei Wu
ECE Department Life Fellow, IEEE Senior Member, IEEE
Clarkson University ECE Department ECE Department
Potsdam, NY 13699 USA Clarkson University Clarkson University
deand@clarkson.edu Potsdam, NY 13699 USA Potsdam, NY 13699 USA
ortmeyer@clarkson.edu lwu@clarkson.edu


Abstract—Electric Vehicle (EV) technology is advancing distribution system. The best way to identify the impact of
rapidly. Battery storage technology is a prime contributor to EVs on the distribution system is to utilize an accurate
vehicle range increases, charging time reductions, and cost transportation model to determine the most likely charging
reductions. These technology improvements are predicted to
times and methods of charging for EVs. This strategy is not
rapidly change the popularity and usage of EVs. These advances
will cause the need for comprehensive models of EV usage and new, it has been the basis of numerous papers. In fact, most
charging, and the resulting requirements on the electric power papers that analyze the impacts of EV charging on a
system. This paper discusses transportation models for this new distribution system primarily focus on Level 1 and Level 2
generation of EVs, and predicts the characteristics of charging [1, 2, 3, 4]. Some briefly cover impacts of Level 3
commercial rapid (Level 3) charging stations at a point where but not in any particular depth [1].
these stations would become commercially viable. The paper also
Some research papers model the activity of EVs
discusses uncertainties in the modeling and data needs for the
development of improved models. unrealistically, randomly distributing the charging behavior of
a population of EVs throughout the course of the day without
Index Terms-- Electric vehicles, rapid charging, taking into account documented travel tendencies. The
transportation models. resulting model would prove unreliable if it were used for
power system planning activities. A common practice to
I. INTRODUCTION develop a transportation model is to use data from the
Electric vehicle (EV) technology is rapidly developing. National Household Travel Survey (NHTS) in conjunction
The increases in vehicle range and reductions in charging with probability functions to determine the likely charging
time and vehicles costs are forecast to result in increasing times for EVs [3, 5, 6]. Some studies utilize empirical
numbers of EVs in the near future. EVs present unique localized travel data to serve as a basis for their transportation
challenges to the power system, and it is important to know models [7, 8, 9]. This method for creating a transportation
future EV charging characteristics in order to promote the model is logical and may well be the best method for
integration of these new loads on the grid. accurately predicting the most likely arrival and departure
This paper presents an EV transportation model based on times for vehicles in addition to likely daily travel distances.
data from the National Household Travel Survey. The paper The travel distances are useful for developing an accurate
predicts the need for commercial Level 3 (quick) vehicle representation of state of charge (SOC) upon arrival to a
charging stations in a representative suburban area at the charging unit. An accurate depiction of SOC for a population
point when there are sufficient EVs to make these stations of vehicles is pivotal for accurately modeling the load from a
commercially viable. The paper includes two case studies. charging event. Some models take into account the likelihood
The first predicts the electric power requirements of a of EVs departing more than once per day and potentially
commercial Level 3 charging station. The second case study charging more than once per day using various charging
predicts the split between Level 2 and Level 3 charging that methods over the course of a day, be it Level 1, 2, or 3 [1, 6].
will take place for a projected suburban population of EVs. Some models only focus on the impacts from Level 1 and
The paper also discusses uncertainties in the transportation Level 2 charging [10]
and charging models, and identifies areas where improved Models must also be developed for the EVs and their
data is needed. charging systems. Some papers choose arbitrary values for
battery capacity and vehicle range in their analyses [3, 5, 7].
II. BACKGROUND A more accurate model can be developed by using readily
available data on current EVs like the Nissan Leaf or Chevy
There has been much work done to determine the best
Volt [1, 9, 10]. The chargers used and their associated
method by which to model the future tendencies of EVs in the
specifications are a critical aspect of developing an accurate
transportation system and to quantify their impacts on the
load model. There are numerous charging specifications used

978-1-5090-0814-8/16/$31.00 ©2016 IEEE


Authorized licensed use limited to: b-on: Instituto Politecnico de Lisboa. Downloaded on January 16,2024 at 20:14:59 UTC from IEEE Xplore. Restrictions apply.
by the EV charging industry, including the EPRI-NEC and chargers are becoming faster and more powerful.
SAEJ1772 standards that can be used to accurately represent The impact of EV charging on the distribution system is
the charging load for Level 1 and Level 2 [1, 2, 3]. The generally well studied, although sometimes with less accurate
standards can be observed in Table 1 below. The EV model transportation models, and often times without significant
and the transportation model can be described by a attention paid with regards to potential impacts of Level 3
combination of factors that include vehicle type, charging charging. It is important in the scope of this study for the
power level, start time, the miles driven, and associated SOC. future impact of EV charging to be understood for power
system planning and potential necessitation of an increase in
Table 1. Charging Standards for EV Chargers robustness of distribution transformers and feeders.

Standard Charging Level III. MODEL DEVELOPMENT


1 2 3 In order to study the impact of EV charging on the power
EPRI-NEC 120VAC, 240VAC, 480VAC, distribution system, it must be known when and where
15A (12A), 40A, 3-Phase, vehicles will charge, vehicle charger characteristics, and SOC
of the vehicle at the beginning and end of the charge cycle.
1.44kW 1-Phase 60-150kW With this knowledge, a regional model can be developed that
SAEJ1772 120VAC, 208-240VAC, 208-600VAC, will predict the cumulative EV charging characteristics.
12A, 1-Phase 32A, 1- 400A, 3- In order to develop this regional model, the following data
1.44kW Phase, Phase, is needed:
6.66- >7.68kW x Number and type of EVs in the region
7.68kW x A stochastic model of the driving characteristics for
these vehicles, including
o Vehicle departure model from home
Most papers that analyze the effects of charging on a o Miles per day
distribution system use models developed independently or o Trips per day
those developed by EPRI or IEEE [10]. Although this is an o Vehicle return home model
accurate way to analyze the impact on a distribution system as x Charging cycle, including charging start time and
it considers the spatial characteristics of the distribution charging duration
feeder, using a geographically specific distribution system can x Type and location of charging
give more pointed and more relevant results, which is crucial
in the scope of this study. In most documents, this study When there are sufficient EVs in the region under study,
included, the EVs are placed arbitrarily around the spatial stochastic models will provide data that is sufficient for
geography of a distribution feeder. Most papers utilize distribution system studies.
different methods to determine the impact of EV charging on A. Electric Vehicle Population
a distribution system, including Monte Carlo analysis and The first step in the development of the Level 3 charging
daily load flows resulting from the increased EVs penetration. station load profile model is identifying the two key variables.
Many papers conclude that transformer thermal overloading, The first variable in the model is the total population of
phase imbalance, voltage regulation, power losses, vehicles in an area under study. This study focusses on mid-
transformer degradation, and harmonic distortion are the size urban and suburban counties or compact geographic
primary impacts of EV charging on a distribution system [1, areas with the total personal vehicle populations of about
2, 4]. 140,000 vehicles within a relatively compact area. Census
With respect to Level 3 charging, due to the higher peak studies show that this type of area is relatively common.
load from the charger, the effect on 3-phase distribution lines The second variable in the model is the penetration rate of
and 3-phase transformers is exacerbated and becomes a EVs for the study. The intent of this study is to study the
greater focus of concern when compared to lower level characteristics of commercial Level 3 charging stations after
charging. The addition of these chargers to the system poses the EV fleet has developed a viable commercial market for
less risk for system imbalance and single phase distribution these chargers. It is predicted that this would occur in a
transformer overloads. Additionally, the peak load of Level 3 140,000 vehicle county at a 10% EV penetration level. It is
chargers can temporally coincide with the distribution system expected that the market would respond to higher numbers of
peak load, which gives further emphasis to distribution EVs in the region by adding additional Level 3 charging
equipment concerns [1, 2]. It is important, when conducting a stations.
study on the impact of EV charging on the distribution
system, to consider a multitude of factors when developing a B. Travel Model
model to most accurately emulate the potential impact of The temporal distribution for vehicle trip departure times in
EVs. It should also be noted that the potential impacts of EVs the US is used to develop the load profile for Level 3
are increasing as the battery capacity of EVs is increasing and charging stations. The distribution has remained largely

Authorized licensed use limited to: b-on: Instituto Politecnico de Lisboa. Downloaded on January 16,2024 at 20:14:59 UTC from IEEE Xplore. Restrictions apply.
unchanged since the 1980’s. The US temporal departure with the mean of 32.5 miles and a standard deviation of 12
distribution reflects the probability of an individual to depart miles. 32.5 miles is the mean value for daily distance traveled
on a trip through the course of a day. These trips could be by drivers in New York State according to the National
from home to work, work to grocery store, or from sporting Highway Travel Survey [8]. This distribution reflects the
event to home, etc. The temporal departure distribution was range of daily average miles per vehicle. A second
obtained from NHTS 2009 [11]. A graphical representation of distribution is used to account for the variation in miles driven
the distribution is shown in Figure 1. In addition, the survey for a specific vehicle from day to day. To account for these
data shows that on average in the United States, a given days, a second distribution is added. The mean for this
vehicle has 3.35 trips per day for a total of 32.7 vehicle miles distribution is assumed to be 90 miles with a standard
per day. deviation of 30 miles. The combination of these two
distributions yields the final distribution of daily miles driven,
as shown in Figure 2.

Figure 1. Temporal Departure Probability Distribution.

C. Level 3 Charging Profile


The load profile of a single charging vehicle is equally
important to the development of the Level 3 charging station
load profile model. Level 3 DC fast chargers charge lithium Figure 2. Distribution of Daily Miles Driven.
ion batteries in a manner that does not damage the battery.
Level 3 chargers will not charge a connected battery to more E. Charging Model
than 80% SOC. The reason for fast charging to only 80%
SOC is to protect the battery. To avoid battery damage, the At the time of this study, no official data has been
battery must be slowly charged for the final 20% of capacity. published regarding the impact of SOC and other factors on
an individual’s probability of charging, particularly in the
The Level 3 charging cycle charges at constant current for
case of extended range vehicles with fast charging capability.
the first portion of the charge, increasing the voltage until the
The charging decision is complex, and depends on
rated battery voltage is reached. At this point, the charging availability of a charger (both Level 2 and Level 3), time of
cycle switches to constant voltage, and the charging current day, future plans, vehicle state of charge, and other factors.
slowly decays as the SOC increases. The transition typically For the purpose of this study, the effect of SOC on the
occurs near 40% SOC. Level 1 and Level 2 chargers operate probability of using public Level 3 charging has been
with a similar profile, but will have the capability to charge estimated based on the limited data currently available. A
the battery up to full SOC without significant concerns on the summary table on the probability of using public charging
damage to battery. with relation to EVs’ SOC can be referenced in Table 2. It is
assumed that if the SOC of a vehicle is greater than 80%,
D. Daily Distance Profile
there is zero probability that the individual will utilize public
In addition to the temporal probability distribution for Level 3 charging. This is due to the fact that Level 3 chargers
departure, the daily distance driven and associated SOC for a will only charge the vehicle’s battery up to 80%, as fast
given sample must be taken into account. For each vehicle, a charging the battery past 80% risks damaging the battery. The
data point is created. For each data point, a value for daily probability of an individual to use public charging with an
miles driven is assigned based on a lognormal distribution of SOC below 20% is assumed to be high, because depleting
daily miles driven by drivers in New York State. The lithium ion EV batteries below 20% can damage the battery
distribution used in the analysis is comprised of a and effect longevity. The upper bound and lower bound for
combination of two distributions. The first distribution of charging level of an EV battery are about 95% and 20%
daily miles traveled is based on the lognormal distribution respectively for many manufacturers [12]. It is assumed that a

Authorized licensed use limited to: b-on: Instituto Politecnico de Lisboa. Downloaded on January 16,2024 at 20:14:59 UTC from IEEE Xplore. Restrictions apply.
decreasing exponential trend will exist for the probability of have commercial Level 3 charging stations available on
utilizing public charging station, for an individual, as SOC demand at convenient locations. This implies that there will
increases from 0 to 100%. be businesses that provide multiple Level 3 charging stations,
It is assumed that the majority personal vehicles will have similar to the present gasoline fueling stations. There will be
access to Level 1 or Level 2 chargers at their residence. sufficient charging units available at these stations to handle
However, there will be a small but significant number of EVs the demand with at most occasional congestion. As Level 3
that do not have access to charging at home. It is predicted charging will continue to take longer than refueling a gasoline
that Level 2 chargers will be the preferred method of charging vehicle, it is anticipated that these stations will either have or
when available. It is also assumed that the SOC of an will be located near services that are attractive to the EV
individual vehicle influences the driver’s decision to charge
owners while they wait.
the vehicle. This assumption is made based on the idea of
Based on observations of existing fueling stations and
“range anxiety” [13]. Range anxiety is a term created to refer
to the fear of running out of charge, which poses more knowledge of the Level 3 charging times, an initial
difficulty than running out of fuel in an internal combustion assumption in this study was the Level 3 EV charging stations
engine vehicle. When one runs out of fuel in a gas vehicle, would have a minimum capability to serve 10 EVs
refueling is as simple as refilling with gas. The prospect of simultaneously.
recharging an EV on the roadside is a bit more daunting. In approximate numbers for the base case examined in this
Therefore, the probability of a sample or a group of samples paper, there will be 14,000 EVs in the region. With each of
to utilize public charging can be assumed to be a function of these charging once per day, and 20% of charges at Level 3,
the vehicles SOC. The lower the SOC, the more likely the that would be 2800 Level 3 charges per day for the region.
individual will utilize a Level 3 public charging station, rather 8.5 charges per charging unit per day is estimated to be a
than use a Level 2 charger on returning home. reasonable value that would allow cost recovery while
minimizing congestion. This rate would lead to 33 charging
Table 2. Probability of Using Public Level 3 Charging as a stations in the region under study. This is considered to be a
Function of State of Charge. sufficient number to allow convenient access to these stations
across the region. By nature, each station will have a distinct
State of Charge (SOC) Probability of Charging
load profile for each day. A representative load profile
10-20% 1.00
20-30% 0.70 predicted for one of these stations is shown in Figure 3.
30-40% 0.40
40-50% 0.20
50-60% 0.10
60-70% 0.10
70-80% 0.05
80-100% 0

The charging model used in this study assumes that:


x Home charging will take place with Level 2 chargers,
and will start right away when the vehicle has returned
home for the day in the late afternoon or evening.
x 20% of the charges will be on Level 3 chargers, and the
remaining 80% on Level 2 chargers.
These assumptions imply that the number of public Level 2
charging stations will decrease as advanced EVs and Level 3
chargers become prevalent. The 20%-80% split between
Level 3 and Level 2 charging is based on early Level 3
charging experience, and could change as the technology
matures. This model also does not directly address those EVs Figure 3. Representative Daily Load Profile of a Single Level 3
where it is not possible or not convenient to charge at home in Charging Station with 10 Charging Units and 85 Charging
the evening. The 20%-80% split should be adjusted Events.
accordingly in regions where it is known that a high
proportion of EV owners do not have access to chargers at
their residences. A Level 3 charging station will be consuming 3-phase
power at 480 V. Each station will have 10 individual units,
which are taken to be 50kW each in this study. The electricity
IV. LEVEL 3 CHARGING STATION CASE STUDY charges for each charging station used in this study include a
This study assumes that advanced EV owners will expect to demand charge and an energy charge, as shown in Table 3. A

Authorized licensed use limited to: b-on: Instituto Politecnico de Lisboa. Downloaded on January 16,2024 at 20:14:59 UTC from IEEE Xplore. Restrictions apply.
monthly flat demand rate will be used for the charging station, travel occurs, and that all Level 2 charges occur at home at
which changes with the amount of power consumed as well as the end of the day.
time of year. Similarly, a flat energy rate is considered, which The case study predicts that for this population of 14,000
changes with the amount of energy consumed as well as time vehicles, there will be a total of 26,257 charging events per
of year. day -- 12,976 of these Level 2 charges and 3281 of these
Level 3 charges. The study results predict that 3 vehicles will
Table 3. Electricity Demand and Energy Charges for the Study do 2 Level 3 charges in the day. The Level 3 charges are just
System. over 20% of the total number of charges. The distribution of
these charges is shown in Figure 4.
Charge une-September October-May The 14,000 vehicles travel a total of 569,300 miles in the
Monthly First 5 kW $1.26/kW $0.74/kW study day. Of these, 411,145 miles or 72% are recharged
Demand Above 5kW $12.89/kW $7.49/kW with Level 2 chargers, and 158,155 miles or 28% are
First recharged with Level 3 chargers. Figure 5 shows the
$0.06489/kWh $0.050/kWh
1250kWh distribution of charge-miles versus vehicle miles per day.
1250kWh-
Energy $0.02977/kWh $0.0287/kWh
30,000kWh
Above
$0.01499/kWh $0.0139/kWh
30,000kWh

For the case considered, a 10 charging unit station would


incur approximately $1.40 of demand charge for each EV
charging event. The cost of energy per 30-minute charge
would be under $1, bringing the total cost per charge for
electricity to approximately $2.30. With an expected installed
cost per 50-kW charging unit of $22,500, a 6% interest rate
and a 10 year lifetime, a station with 10 charging units should
recover $2500 per month to cover the capital costs. With
2550 charging events per month per station, this would be
about $1 per charge to cover the capital costs. While the Miles per day
station owner will have additional costs (and revenues), this
cost for electricity infrastructure and service for a Level 3 Figure 4. Distribution of charging behavior as a function of total
charging appears to be reasonable. That said, a Level 2 miles drive per day.
charging at the owner’s residence can be expected to be
somewhat less costly. With a residential rate of $0.075 per
kWh, a charge that draws 25 kWh from the grid would cost
$1.88.

V. VEHICLE CHARGING CASE STUDY


This case study involves 14,000 EVs driving in a compact
geographic area that is served by 33 commercial Level 3
charging stations, with each station having the capability to
charge 10 vehicles simultaneously.
This model presented in this paper can be solved with the
following procedure:
1. The vehicles are all considered to have a range of 100
miles.
Miles per day
2. All vehicles begin the day with fully charged.
3. The vehicle driving characteristics are given by Figure 2. Figure 5. Charge-mile comparison between Level 2 and Level 3
4. The vehicle charging characteristics are given in Table charging as a function of the vehicle miles per day.
2.
5. Vehicles that do a Level 3 charging within 5 miles of
VI. MODELING NEEDS
their daily total are considered not to do a Level 2
charge. This section discusses the areas of uncertainty in the
This case study does not consider the time of day that the modeling effort presented in this paper.

Authorized licensed use limited to: b-on: Instituto Politecnico de Lisboa. Downloaded on January 16,2024 at 20:14:59 UTC from IEEE Xplore. Restrictions apply.
x Extended Range EV Driving Characteristics. This paper on vehicle daily range and driver preference.
assumes that extended range EVs will be used very Finally, the paper describes the need for additional data
similarly to today’s gasoline powered vehicles. Research that is needed to improve the predictive capability of these
will be needed on these vehicles as they come on the models.
market to establish actual driving patterns.
x Day to Day Variation of Driving Range. There is VIII. ACKNOWLEDGMENT
significant data available on the range of daily average
driving range across a population, as discussed in Section This work was supported by the New York Energy Research
III.b. However, there is little data available that provides and Development Authority (NYSERDA), Project number
information on the day to day variation in range of an 39151. The authors gratefully acknowledge the contributions
individual driver (or vehicle) within that population. This of NYSERDA R&D Project Manager Greg Pedrick and his
will likely be a critical factor in the use of Level 3 chargers. colleagues on this research project.

x Proportional Use of Level 3 Chargers. The data on the


proportion of charges that will be made on Level 3 chargers IX. REFERENCES
is limited at this point of time. This proportion can be 1. Yilmaz, Murat, and Philip T. Krein. "Review of Battery Charger
expected to vary significantly from location to location due Topologies, Charging Power Levels, and Infrastructure for Plug-
to differences in driving patterns, housing patterns, and In Electric and Hybrid Vehicles." IEEE Transactions on Power
perhaps electricity rates. Electronics IEEE Trans. Power Electron., 28.5 (2013): 2151-
169. Web. 1 Jan. 2016.
x Number of EV Owners with No Access to Home Charging. 2. Giarratano, Alex M. Evaluating the Impact of EVs on Distribution
Feeders and the Potential for a Novel EV’ Charging Management
Data is needed to identify the number of EV owners that Strategy. MS Thesis, Clarkson University, 2013.
will not have the option to charge at home. These would 3. Darabi, Zahra, and Mehdi Ferdowsi. "Aggregated Impact of Plug-
include, for example, some apartment dwellers and those in Hybrid Electric Vehicles on Electricity Demand Profile." IEEE
who rely on street parking for their vehicles. Trans. Sustain. Energy Vol. 2 No. 4 (2011): pp. 501-08.
4. Liu, Ryan, Luther Dow, and Edwin Liu. "A Survey of PEV Impacts
x Impact of weather on EV range. There is a significant need on Electric Utilities." 2010 IEEE PES Innovative Smart Grid
Technologies Conference. (2011).
for reliable information on the impact of weather factors on 5. Wu, Qiuwei, A. H. Nielsen, Jacob Østergaard, Seung Tae Cha, F.
EV range. The primary impacts will be due to the effect of Marra, Yu Chen, and C. Træholt. "Driving Pattern Analysis for
extreme heat and cold on battery storage capability and Electric Vehicle (EV) Grid Integration Study." 2010 IEEE PES
efficiency, and the impact of passenger space heating and Innovative Smart Grid Technologies Conference Europe (ISGT
cooling on the range. It is expected that these factors will Europe) (2010).
6. Celli, G., G. Soma, F. Pilo, F. Lacu, and N. Natale. "Aggregated
have significant impact on EV adoption rates as well as on Electric Vehicles Load Profiles with Fast Charging Stations."
the usage patterns of commercial Level 3 charging stations. 2014 IEEE Power Systems Computation Conference, Aug.,
2014. Wroclaw, Poland.
VII. CONCLUSIONS 7. Neubauer, Jeremy, Aaron Brooker, and Eric Wood. "Sensitivity of
Battery Electric Vehicle Economics to Drive Patterns, Vehicle
In this paper, the development of a comprehensive EV Range, and Charge Strategies." Journal of Power Sources, Vol.
transportation model is discussed. There is a critical need for 209 No. 1 (July 2012): pp. 269-77.
8. Sioshansi, Ramteen, Riccardo Fagiani, and Vincenzo Marano.
accurate EV transportation models for the new generation of "Cost and Emissions Impacts of Plug-in Hybrid Vehicles on the
extended range, fast charging electric vehicles. These vehicles Ohio Power System." Energy Policy, Vol. 38 No. 11 (Nov.
will likely achieve significant market share in the foreseeable 2010). pp. 6703-6712.
9. Adornato, B., R. Patil, Z. Filipi, Z. Baraket, and T. Gordon.
future, and it is important to develop the supporting "Characterizing Naturalistic Driving Patterns for Plug-in Hybrid
infrastructure for these vehicles in a timely fashion. Electric Vehicle Analysis." 2009 IEEE Vehicle Power and
This paper identifies and discusses the element components Propulsion Conference. (Sept., 2009), Dearborn, MI. pp. 655-
660.
and data needs of a practical transportation model that would 10. Transportation Energy Data Book, Edition 34. Oak Ridge National
be useful for identifying the interactions between EV charging Laboratory. ORNL-6991. September, 2015.
and the electric power grid. 11. A. Santos, N. McGuckin, H.Y. Nakamoto, D. Gray, and S. Liss "
The transportation model has three primary elements: Summary of Travel Trends: National Household Travel Survey
2009." U.S. Department of Transportation. Federal Highway
x Vehicle characteristics Administration. Report FHWA-PL-11-022. June 2011.
x Daily driving range and time of travel 12. Park, Woo-Jae, Kyung-Bin Song, and Jung-Wook Park. "Impact of
Electric Vehicle Penetration-Based Charging Demand on Load
x Charging characteristics and usage for both Level 2 and Profile." Journal of Electrical Engineering and Technology, Vol.
Level 3 charging 8, No. 2 (2013): 244-51.
The paper presents two case studies. The first predicts the 13. Rauh, N., T. Franke, and J. F. Krems. "Understanding the Impact
of Electric Vehicle Driving Experience on Range Anxiety."
electric energy consumption of a 10 vehicle commercial Human Factors: The Journal of the Human Factors and
Level 3 charging station that is predicted by the stochastic Ergonomics Society, Vol. 57 No. 1 (2014). pp. 177-87.
driving model presented in this paper. The second case study
predicts the split between Level 2 and Level 3 charges based

Authorized licensed use limited to: b-on: Instituto Politecnico de Lisboa. Downloaded on January 16,2024 at 20:14:59 UTC from IEEE Xplore. Restrictions apply.

You might also like