IEEE PES ISGT ASIA 2012 1569576261
The Impact of Discharging Electric Vehicles on
the Distribution Grid
Shao yun Ge, Long Wang*, Hong Liu, Liang Feng
Abstract—Distribution grid will be impacted to some extent
when large scale electric vehicles (EV) are discharging
simultaneously. Firstly, the factors related to the discharging
power of EVs are discussed. Secondly, the probability density
statistic models of discharging time and discharging duration are
built according to the traffic pattern of X city in China and the
national household travel survey of America in 2001. Thirdly, the
probability equation of single EV’s discharging power at each
hour in one day is derived. After that, the mathematical
expectation of above probability equation is solved by Monte
Carlo simulation method. Finally, the impacts of large scale
discharging EVs on the daily load of X city are analyzed. The
results provide the theoretical foundation for the application of
large scale EVs in X city.
Key Words--Electric Vehicle; Monte Carlo; Distribution Grid;
Discharging; Probability Density Function; Statistic Model;
Smart Grid;
I. INTRODUCTION
I n the world where energy conservation, environment
protection and low carbon emission are growing concerns,
the development of new transportation tools has taken on an
accelerated pace, and the electric vehicle (EV) appears as a
preference.
The number of EVs will reach 60 million in year 2030
according to the Electric Vehicle Development Strategy
Research Predicting Report published by Ministry of Industry
and Information Technology [1]. Assuming that the EV
number in X city will reach 0.2 million and each EV generates
5kW active power, and then the total power these EVs provide
is 1000 MW. As a result, the EV discharging power accounts
for about 10 percent of the maximum daily load of X city. So
the impact of discharging EVs on the distribution grid has to
be discussed before the large scale EVs are simultaneously
being connected.
Most research works have focused on harmonic pollutions,
loading profiles, and charging profiles for these electric
vehicles. Different EV charging equipments are analyzed and
harmonic metering models of different chargers are built in Li
Na [2]. The charging and discharging characteristics of plug in
electric vehicle (PHEV) based on transient state are discussed
The work was supported by National Natural Science Foundation of China
(51107085), National 863 Program (2011AA05A115) and the National Basic
Research Program of China (973 Program) (2009CB219703).
S. Y. GE, L. WANG*(Corresponding author), H. LIU and L. FENG are
with Key Laboratory of Smart Grid (Tianjin University), Ministry of
Education, Nankai District, Tianjin 300072, China. (wangling @ tju.edu.cn,
xiaoxiao_1302 @ 163.com )
1
and their statistic formulas are derived in Yang Hong ming [3].
The EV using model is built after analyzing human travel
pattern and car using statistical results in Bradley T [4]. The
statistical model of EVs’ power demand based on the national
household travel survey (NHTS) of America in 2001 is built in
Tian Li ting [5]. The basic effects that PHEVs will have on the
grid based upon their characteristics are covered by Hadley [6].
The general results about the impacts of PHEVs on
distribution grid are determined by the number of vehicles and
vehicle demand profile are given in [7]. A time coordinated
optimal flow model for integrating PHEVs and vehicle-to-grid
(V2G) in order to minimize power loss is suggested in Acha et
al. [8]. A quadratic and dynamic programming model for
assessing the impacts of charging PHEVs on the distribution
grid of Belgium is developed in Clement et al. [9]. The
impacts of charging EV on the daily load curve are examined
in [10] and [11]. All the above researches have neglected to
investigate the impact of large scale discharging EVs on the
daily load of distribution grid.
The paper is organized in the following way: Section II
introduces the daily driving distance statistical pattern based
on national household travel survey of America in 2001 and
the human daily driving pattern in X city. Section III discusses
the probability density statistic models of discharging time and
discharging duration. Section IV gives the probability
equation of single EV’s discharging power at each hour in one
day and solves the mathematical expectation of above
probability equation by Monte Carlo simulation method.
Section V discusses the impact of large scale discharging EVs
on the daily load of X city. Section VI concludes the whole
work be done in this paper.
II. DAILY DRIVING STATISTICAL PATTERN
Following are the assumptions to be used for building the
probability density statistic models of discharging time and
discharging duration in Section III based on the surveying
results about the daily driving pattern in X city.
A. Assumptions of Daily Driving Statistical Pattern
(i) Battery capacity of each EV is Cbattery = 30 kWh, and the
general energy consumption is W = 0.15 kWh/km [12].
(ii) The driving speed is constant, Vspeed = 25 km/h.
(iii) The discharging power is constant, Pdis = 5 kW.
(iv) The distance the EV covers from home to workplace is
equal with the distance back home from workplace.
(v) The EV immediately discharges when it reaches the
workplace.
 (vi) The EV is charged fully at home and discharged Where μ D = 3.20 , σＤ = 0.88 , 0 < t ≤ 4 .
leaving enough power for back home from workplace. Equation (3) represents the probability density of the time
(vii) The EV begins to be driven at 8:00 am, and driven the EV consumes from home to workplace. The maximum
back home in the afternoon. time the EV used for reaching the workplace is 4 hours. The
B. The Probability Density Equation of Daily Driving daily probability density value of discharging time between 8
Distance am and 12 am is equals with the value calculated by equation
(3), but its value is zero at other time during the day.
According to the national household travel survey of
1
America in 2001 [13], [14], we could get the daily driving
distance probability density equation (1).
1 (ln x − μ D ) 2 (1) 0.8
f dis tan ce ( x) = exp[− 2 ]
xσ D 2π 2σ D
Where μ D = 3.20　, σＤ = 0.88 , 　
0 < x ≤ 2.

Probability
0.6
0.03

0.4
0.025

0.02 0.2
probability

0.015
0
02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 24:00
charging time(h)
0.01
Fig. 2 the probability density distribution of discharging time

0.005 Following results could be got from Fig. 2. All the

probability values are 0 except the values at the time between
0
8 am and 12 am. About 80% EVs will reach the workplace in
0 20 40 60 80 100 120 140 160 180 200 an hour, and only few numbers of EVs reach the workplace
daily driving distance(km)
Fig. 1 the probability density distribution of daily driving distance
more than 3 hours. In other words, about 80% EVs will start to
discharge at 9:00 am, and very few EVs start to discharge after
Fig. 1 shows us that most people drive their EVs at about 11 am.
20km per day, and very few people will drive more than 80km B. The Probability Density of Discharging Duration
per day.
The probability density of discharging duration will be
affected by the following 4 factors, battery capacity of each
III. MODELS OF DISCHARGING TIME AND DISCHARGING
EV, the general energy consumption per km, the discharging
DURATION
power, and the distance the EV covers daily.
The probability density statistic models of discharging time C − WLD
and discharging duration are being derived according to Tdisch arg e = battery (4)
equation (1) and the 7 assumptions listed in Section II. Pdis

A. The Probability Density of Discharging Time Where Tdisch arg e represents the discharging duration and
The probability density of discharging time will be affected LD is the distance the EV covers daily.
by the speed the EV drives and the distance between home and Equation (4) shows us that the probability density of
workplace. discharging duration has a linear relationship with the
0.5L D probability density of daily driving distance. But the
Trun = (2) probability value of discharging duration will be decreased
Vspeed when the probability value of daily driving distance is
Where Trun is the time the EV consumes from home to increased.
workplace, LD is the distance the EV covers daily. Equation (5) would be derived with equation (1), equation
Equation (2) shows the relation between the probability (4), and other 3 variables.
density of the time the EV consumes from home to workplace 1 (ln(6 − t ) − ln 0.03 − μ D ) 2 (5)
f (t ) = exp[− ]
and the probability density of the daily driving distance. Tdisch arg e
(6 − t )σ D 2π 2σ D
2

With equation (1) and equation (2), it is easily to get

Where μ D = 3.20 , σＤ = 0.88 , 0 ≤ t < 6 .
equation (3) as follows.
Equation (5) represents the probability density statistical
1 (ln t − ln 0.02 − μD )2 (3)
fTrun (t ) = exp[− 2
] model of daily discharging duration.
tσ D 2π 2σ D

2
 1 composed of the probability density function of discharging
time and discharging duration, td is the discharging duration, t
represents the current time, and Pt0 .is the discharging power
0.8
at time t0.
From Section III we will find out that the probability
probability

0.6 density of discharging time and discharging duration is

0.4
0.2
0
0 1 2 3
discharging duration(h)
Fig. 3 the probability density distribution of discharging duration
Fig. 3 shows that about 90% EVs could discharge for
longer than 5 hours at workplace, and very few EVs discharge
less than 3 hours. Compared Fig. 3 with Fig. 1, we could find
out the longer the distance EVs covers, the shorter the duration
they discharge. Most of the EVs travel in short distance, so
most of them discharge with long duration. The reason behind
this result is that the battery capacity of each electric vehicle is
constant according to the assumption (i) listed in Section II.
IV. EXPECTATION OF SINGLE EV’S DISCHARGING POWER
In order to examine the discharging state at time t0, here
we introduce the random variable ξ . If the EV is discharging
at time t0, we set ξ =1, else let ξ =0. Then we will get equation
(6) and equation (7) as follows.
P(ξ = 1) = Fdis (t < t 0 , t + t d ≥ t 0 ) + Fdis (t > t 0 , t + t d ≥ t 0 + 24) (6)
P(ξ = 0) = 1 − P(ξ = 1)
And the single EV’s discharge power at time t0 is calculated
by equation (8).
Pt = P(ξ = 1) Pdis
0
5 used as a theoretical reference for the application of large
4
Load / kW
dependent on each other, it’s very difficult to derive its
analytical equation, let alone calculating its probability
analytical solution. So here we introduce the Monte Carlo
simulation method.
Here we calculate the single EV’s daily discharging power
at each hour in one day by the Monte Carlo simulation method
based on equation (8), and the results are showed in Fig. 4.
Fig. 4 shows that the maximum discharging power
4 5 6 appears at 10:00 am, and its value is about 4.7kW. Compared
Fig. 4 with Fig. 2, we will find out that the appearance time of
the maximum discharging power is 1 hour after the
appearance time of the maximum probability value of
discharging time. The reason behind is that the maximum
discharging power is added by all the discharging power
during the discharging duration before it. In another word, it’s
caused by the probability density distribution of discharging
duration.
V. THE IMPACT OF LARGE SCALE DISCHARGING EVS ON THE
DAILY LOAD OF X CITY
The impact of large scale discharging EVs on the daily
load of X city is shown in Fig. 5.
The blue line in Fig. 5 is the load curve with large scale
discharging EVs connected, and the orange line is the power
demand curve without EVs.
Figure 5 indicates that the impact of large scale
discharging EVs on the daily load of X city is mainly between
8:00 am and 13:00 pm. With large scale discharging EVs
(7)
connected to the distribution grid, the total power demand of
the other appliances will be reduced hugely. The discharging
EVs could also shave the load peak and make the load curve
(8) smoother if a good strategy is provided. These results could be
scale EVs in X city.
10500

3 9500
Load/MW

2
8500

7500

0 Original Load
02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 24:00
Withdischarging Electric Vehicles
charging time(h)
Fig. 4 the expectation of single EV’s discharging power 6500
02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 24:00
time(h)
Where Fdis is the joint probability density function

3
 Fig. 5 the impact of large scale discharging EVs on the daily load of X city His major is urban power system programming and
automation of distribution system. And moreover he is
a professor in Department of power system and its
automation in Tianjin University and has published
VI. CONCLUSIONS more than 20 papers in IEEE. In addition, he has
This paper builds the probability density statistic model of accomplished two Natural Scientific Funds of China
and received the second prize of the Natural Scientific
discharging time and discharging duration, and analyzes the and Technological Progress.
statistical discharging characteristics of large scale EVs and
their impact on the distribution grid.
The total power demand of the grid will be reduced
hugely with large scale discharging EVs connected. Also, the Long Wang received master degrees from School of
discharging EVs could shave the load peak and make the load computer science and software in Tianjin Polytechnic
curve smoother if a good strategy is provided. These results University. He is currently pursuing Doctoral degree
in School of electric engineering and automation in
provide the theoretical foundation for the application of large Tianjin University.
scale EVs in X city. He is mainly interested in smart grid and the
The next step of this research is to investigate the optimal impact of elctric vehicle on the distribution grid.
control strategy of large scale electric vehicles to shave the
load peak and reduce peak-valley ratio.
Hong LIU received master and Ph.D. degrees from
School of electric engineering and automation in
VII. REFERENCES Tianjin University in 2005 and 2009, respectively.
[1] Zhang Wen liang, Wu Bin and Li Wufeng, et al. “Discussion on He is a lectuer of electrical engineering in Tianjin
development trend of battery electric vehicles in china and its energy University with research interests in urban power
supply mode”. Power Supply Technology, 2009, 33(4): 1-5. system programming and decision support theory on
[2] Li Na, Huang Mei. “Analysis on Harmonics Caused by Connecting power system. In addition, he has been supported by
Different Types of Electric Vehicle Chargers with Power Network”. Natural Scientific Funds of China and received the
Power Supply Technology, 2011, 35(1): 170-174. second prize of the Natural Scientific and
[3] Yang Hong ming, Xiong Luo cheng, Liu Bao ping. “Analysis of Technological Progress.
Charging and Discharging for Plug-in Hybrid Electric Vehicles”.
Journal of Electric Power Science and Technology, 2010, 25(3): 8-12.
[4] Bradley T, Quinn C. “Analysis of plug-in hybrid electric vehicle utility Liang Feng received bachelor degree and master
factors”. Journal of Power Sources, 2010, 195(16): 5399-5408. degree from School of electric engineering and
[5] Tian Li ting, Shi Shuang long, Jia Zhuo. “A Statistical Model for automation in Tianjin University in 2008 and 2010,
Charging Power Demand of Electric Vehicles”. Power System respectively. He is currently pursuing Doctoral degree
Technology, 2010, 34(11):126-130. in School of electric engineering and automation in
[6] Hadley SW. “Impact of plug-in hybrid vehicles on the electric grid”. Tianjin University.
Technical report ORNL/TM-2006/554. Oak Ridge National Laboratory, He is mainly interested in smart grid and the
2006. elctric vehicle charging station planning.
[7] Denholm P, Short W. “An evaluation of utility system impacts and
benefits of optimally dispatched plug-in hybrid electric vehicles”.
Technical report. National Renewable Energy Laboratory, 2006.
[8] Acha S, Green T,Shah N. “Effects of optimized plug-in hybrid vehicle
charging strategies on electric distribution network losses”. In: PES,
editor. Conference and exposition on transmission and distribution.
Prague: IEEE, 2010: 1–6.
[9] Clement K, Haesen E, DriesenJ. “Stochastic analysis of the impact of
plug-in hybrid electric vehicles on the distribution grid”. In: Electricity
distribution, CIRED 2009. 20th International conference on electricity
distribution; 2009,: 1–4.
[10] Putrus G, Suw. A, Johnston P. “Impact of electric vehicles on power
distribution networks”. IEEE Vehicle Power and Propulsion Conference,
Dearborn, USA, 2009 : 827-832 .
[11] Wynne J. “Impact of plug-in hybrid electric vehicles on California’s
electricity grid”. North Carolina： Nicholas School of the Environment
of Duke University, 2009．
[12] Han Hai ying, “The study on the Control Strategy of V2G Participating
Peak Regulation and Frequency Regulation of the Grid”, Ph.D.
dissertation, Dept. Electrical Eng., Beijing jiaotong Univ., 2011.
[13] Taylor M J, Alexander A. “Evaluation of the impact of plug-in electric
vehicle loading on distribution system operations”. IEEE Power &
Energy Society General Meeting, Calgary, Canada, 2009: 1-6.
[14] Vyas A, Santini D. “Use of national surveys for estimating ‘full’ PHEV
potential for oil use reduction”. 2008-07-21. http://www.transport-
ation.anl.gov/pdfs/HV/525.pdf.

VIII. BIOGRAPHIES

Shaoyun GE received master degrees from School of electric engineering and

automation in Tianjin University and acquired his Ph.D. from Hong Kong
University.