Due to increased concerns for rising oil prices and environmental problems, various solutions have been proposed for solving energy problems through tightening environmental regulations such as those regarding CO
_{2}
reduction. Among them, Electrical Vehicles (EVs) are evaluated to be the most realistic and effective approach. Accordingly, research and development on EVs and charging infrastructures are mainly proceeding in developed countries. Since EVs operate using electric energy form a battery, they must be connected to the power system to charge the battery. If many EVs are connected during a short time, power quality problems can occur such as voltage sag, voltage unbalance and harmonics which are generated from power electronics devices. Therefore, when EVs are charged, it is necessary to analyze the effect of power quality on the distribution system, because EVs will gradually replace gasoline vehicles, and the number of EVs will be increased. In this paper, a battery for EVs and a PWM converter are modeled using an ElectroMagnetic Transient Program (EMTP). The voltage sag and unbalance are evaluated when EVs are connected to the distribution system of the Korea Electric Power Corporation (KEPCO). The simulation results are compared with IEEE standards.
1. Introduction
The Kyoto protocol of 2005 made CO
_{2}
reduction compulsory for developed countries. The Republic of Korea has invested in R&D and the commercialization of EVs with aims for mass production in 2011, and for comprising 10% of the world green car market share by 2015, which would guarantee that green cars would comprise up to 10% of domestic compact vehicles by 2020
[1]
.
Since EVs are powered by batteries, charging is necessary. In order to charge the battery, EVs should be connected to the power system. However, EVs are charged randomly depending on driver’s patterns, driving purposes, the time of days, the day of the week, the season, and so on. Consequently, various power quality problems such as voltage sag, harmonics, and voltage unbalance could occur in the power system. Among the power quality problems, voltage sag is the most frequent event. Hence, prior to the mass production of EVs, it is necessary to analyze the power quality problems and to make arrangements through simulations when EVs are connected to the power system.
Various studies have already been performed, including; the analysis of harmonics due to the connection of EVs
[2]
, and of the impact of plugin electric vehicle loading
[3
,
4]
, the impact of transformers
[5]
, the active and reactive power demand
[6

9]
, the power quality and voltage unbalance
[10]
, loss reduction using charging control
[11
,
12]
, frequency control
[13]
, and of EV chargers characteristics
[14]
. In these studies, however, EVs were not connected to an actual power system and numerical power quality analysis was not performed clearly. In this paper, voltage sag and unbalance occurs upon the connection of EVs are analyzed based on actual traffic volume and average driving time in distribution system of KEPCO. Simulations are performed on various penetration levels, and the results are compared with IEEE standards.
2. Connection of EV
EV charging can be divided into slow charging and rapid charging
[15]
. Generally, slow charging systems for households are operated at 230V on a single phase of the LV level. In contrast, rapid charging systems are operated at 350400V on the three phase of MV level. This paper considers a slow charging system only. The single line diagram of a slow charging network is shown in
Fig. 1
.
Configuration of Slow Charging Network
 2.1 Assumptions
In this paper, we make the following assumptions:

① The number of EVs connected to the power system is proportional to the load capacity.

② Charging is started promptly after driving is stopped.

③ While EVs are charging, other components of the power system do not affect power quality.

④ Characteristics of weekends and holidays are not considered.

⑤ A slow charging system will complete the battery charging within 2 hours.

⑥ All EVs are connected to phase

⑦ 1 hour is assumed as 0.1 second in the simulation.
We assume that the number of EVs is proportional to the load capacity because the load capacity is closely related to residential density. EVs will be charged promptly after driving is stopped. For this, we consider the actual traffic volume and average driving time. The average driving time in the Republic of Korea is 1 hour is applied in this paper
[1]
. Since this paper focuses on the impact of the EV charging, it is assumed that other components on the power system do not affect power quality. And, we consider only weekdays. EVs will be completely charged within 2 hours by considering the average driving distance, fuel efficiency and capacity of the charger. Next, in order to consider the worst case, all EVs are connected to phase A. Finally, we regard 1 hour as 0.1 second to reduce the simulation time.
 2.2 Simulation system[16]
We model the distribution system in one part of Seoul, Republic of Korea, as shown in
Fig. 16
of the Appendix. This system consists of “X S/S ~ Y D/L” and “X S/S ~ Z D/L”. The total length of this system is 1.5[km], and consists of twosteptype poles and a neutral line. Electric power is delivered to 13 loads in the upper portion and 11 loads in the bottom portion. The total active and reactive loads are 28.6[MW] and 13.9[Mvar], respectively. In this paper, only part 2 is considered. Because according to
[1]
, actual traffic volume is surely required to estimate the number of EVs. However, it is very hard to get an actual traffic volume data in part 1. Because the part 2 only covers one certain area. On the other hand, part 1 covers two nearby areas. Therefore, if we consider part 1, it could cause uncertainties on analysis. It means when we regard part 1, it is obvious that certain amount of traffic volume have intersection between neighboring areas. Thus, uncertainties would occur. That is the main reason why we only consider part 2 area. For more detailed explanation, assume that the set A and B represents the actual traffic volume of each areas in part 1. To calculate the traffic volume of part 1 (A∪B), we should know the number of actual traffic volume of A, B and its intersection (A∩B, i.e. the number of EVs which drive from A to B and vice versa). However, if we do not know the exact number of traffic volume in intersection, the uncertainties would occur. On the other hand, part 2 include only one area. Therefore, there is no intersection and uncertainties. That is to say, more accurate analysis can be performed by considering only the part 2 area.
3. Case Study
 3.1 Setting for the number of EVs
The number of EVs is an essential part of analyzing voltage sag and unbalance. In this paper, it is estimated according to the following process.

① Set the Penetration Level.

② Calculate the number of EVs on each load (L1L11) based on the load capacity.

③ Analyze the system connection time based on actual traffic volume.

④ Estimate the number of EVs connected to the power system at each time by considering the load capacity and traffic volume.
The Republic of Korea established a scenario that EVs will comprise up to 10% of the total compact vehicles (about 65,000) by 2020
[1]
. In 2012, the actual number of compact vehicles of the X S/S ~ Z D/L area is 8961
[17]
. Thus, it is expected that there will be approximately 896 EVs in this area. The number of EVs on each load process could be calculated using Eqs. (1) and (2).
where,

RLn: Rate of load capacity of Ln

CLn: Load capacity of Ln

Ct: Total load capacity

SLn: The number of EVs which are connected to Ln

St: Total number of EVs

n: 1~11
The number of EVs connected to each load
The number of EVs connected to each load
Charging start time of EVs
Based on these equations, the number of EVs connected to the power system for a day is shown in
Table 1
.
In order to set the charging start time of EVs at each load, the actual traffic volume of a simulation system
[18]
is used and depicted in
Table 2
. The traffic volume is highest during commute times (8h, 17h, and 18h) and noon.
According to the second assumption of section 2.1, EV charging would start 1 hour later than the actual traffic volume. Therefore, the number of EVs connected to the system at each time can be calculated using Eq. (4).
where,
: The number of EVs which are connected to Ln at each time
S_{Ln} : The number of EVs connectedto Ln
R_{h} : The rate of EVs at each time
Fig. 2
shows the charging start time of EVs in case of the 10% penetration level. From the
Tables 1
and
2
, the number of EVs connected to the system considering actual traffic volume and load capacity in the X S/S ~ Z D/L area is shown in
Table 3
. It should satisfy Eq. (5).
Traffic volume and ratio of the simulation system
Traffic volume and ratio of the simulation system
where,
S_{t} : Total EVs
S_{h} : The number of EVs which start charging at each time
 3.2 Model of EVs using EMTP/ATPDraw
In this paper, the driving gear of vehicles is neglected, because we just focus on the case of EVs charging. The harmonics filter, PWM AC/DC converter
[19]
, and Liion battery for EVs
[20]
are modeled using EMTP/ATPDraw. It is represented in
Fig. 3
.
EV charging interface
The number of EVs connected to the power system considering time and load capacity (Penetration Level: 10%)
The number of EVs connected to the power system considering time and load capacity (Penetration Level: 10%)
EV models (Penetration level: 10%)
If the penetration level is 10%, 23 models of EVs are required, and a “Time Controlled Switch” is used to connect to the power system. The name of the input node of each EV model shown in
Fig. 4
should correspond to the name of a switch node. For example, according to the
Table 3
, 9 EVs are connected to L11 of the system at 8 AM. In order to simulate this case, the name of the switch operated at 8 AM on L11 should coincide with ev_9 as shown in
Figs. 4
and
5
.
Switch setting for system connection on L11 (Penetration level: 10%)
4. Simulation Results
 4.1 Voltage sag
EVs could be regarded as a load and its charging point is concentrated. Hence, if many EVs are connected to the power system, voltage sag would occur. Voltage sag could be the most frequent power quality problem. IEEE standard 11591995
[21]
define voltage sag as in
Fig. 6
.
Classification of voltage magnitude used in IEEE Std.11591995
Single line diagram on the simulation system
A simple diagram of the modeled system and charging interface is shown in Fig 7. In order to consider the voltage drops and losses of distribution lines, we calculate the complex power and voltage at node “n” as shown in Eqs. (6) and (7).
where,
S_{n} : Complex power at node n
S_{#1} : Complex power at node #1
S_{#2} : Complex power at node #2
Z_{LV D/L #1} : Line impedance of LV D/L #1
Z_{LV D/L #2} : Line impedance of LV D/L #2
I_{1 }: Current of node #1
I_{2 }: Current of node #2
V_{n} : Voltage at node n
V_{#1} : Voltage at node #1
N : Turn ratio of LV Transformer
In the equation (6),
Z_{LV D/L}
_{#1}
×
I
_{1}
^{2}
and
Z_{LV D/L}
_{#2}
×
I
_{2}
^{2}
terms indicate the line losses by LV D/L #1 and #2, respectively. And the voltage drops on LV D/L #1 can be calculated by the latter term of Eq. (7). As shown in Eq. (7),
V
_{#1}
is varied according to the connection of EVs. Therefore we measured the voltage
V
_{1}
on L1~L11.
Fig. 8
shows the RMS voltage measured on L4 with the 10% of penetration level. If EVs are not connected the voltage is 244.63V. However, the voltage is rapidly reduced to 240.83V (0.964p.u.) when EVs are connected.
RMS value of voltage measured on L4 (Penetration level: 10%)
RMS value of voltage measured on L1~L11 (Penetration level: 10%)
Fig. 9
represents the RMS voltage measured at L1~L11 expressed as a perunit system. Generally, most cases show that voltage decreases expressly in commuting times and at noon. In other words, more EVs are stopped and connected to the system for charging at these times (i.e. 7~10 AM, 12 PM and 15~18 PM) than others. But voltage does not exceed the voltage sag limit when the penetration Level is 10% as shown in
Fig. 9
.
For more specific analysis, the minimum values from the results of
Fig. 9
are depicted in
Fig. 10
. It is shown that the voltage decrease when EVs were connected to the power system and the decreasing rate was proportional to the number of connected EVs.
In this area, if the penetration level of EVs is 10% the voltage does not drop under the voltage sag limit defined by the IEEE standard. Thus, we simulate three cases of penetration levels and the results are shown in
Fig. 11
.
In
Fig. 10
, as the penetration level is increased by 5% intervals from 10% to 20%, the maximum and average decrements are 0.035 p.u. and 0.016 p.u., respectively. In the case of the 10% and 15% penetration levels, voltage does not drop under the voltage sag limit of IEEE standard. But, in the case of the 20% penetration level, the voltage decreases under 0.9p.u. at 15h and from 17h to 19h.
The minimum value of voltage measured on L1~L11 (Penetration Level: 10%)
The minimum voltage magnitude according to penetration levels
 4.2 Voltage unbalance
Due to the long charging time and single phase connection, the slow charger for EVs leads to the voltage unbalance in the power system. Voltage unbalance can cause some power quality problems in the power system such as reverse torque and heating of a rotating machine, and irregular harmonics
[22]
. Therefore, it is necessary to analyze the voltage unbalance. Generally, voltage unbalance can be evaluated by using symmetrical transformations as shown in Eqs. (8~10).
where,
V_{0} : Zero sequence voltage
V_{1} : Positive sequence voltage
V_{2} : Negative sequence voltage
Positive sequence( ① ), Negative sequence( ② ) voltage and uf( ③ ) wave according to the time on L9 (Penetration Level: 10%)
u_{f} results measured on L1~L11 (Penetration Level: 10%)
Voltage unbalance is also has a limitation according to the IEEE Standard
[21]
. u
_{f}
was the negative sequence unbalance factor represented by using the ratio of the positive sequence to the negative sequence. And it is limited to 2%. In this paper, to use the method applied in previous study
[22]
, symmetrical transformation is performed through Eqs. (8) and (9) by measuring the 3 phase instantaneous voltage at L1~L11. Based on the calculated results, the negative sequence unbalance factor is analyzed.
Fig. 12
shows the voltage waveform of the positive sequence, negative sequence, and u
_{f}
measured at L9 in the 10% penetration level.
In
Fig. 12
, the voltage unbalance occurred at L9 in the most time, but it exceeded the limit (2%) at 12 and 16h only. Therefore, in order to evaluate the voltage unbalance, the values of uf calculated at L1 ~ L11 are represented in
Fig. 13
.
In
Fig. 13
, it is shown that the voltage unbalance exceeds the limit at L4, L9 and L11 which is connected with many vehicles compared with other load points.
In contrast to the case of voltage sag, the voltage unbalance exceeds the limit a total 8 times with a penetration level of 10%.
Fig. 14
shows the maximum values of u
_{f}
at each time and
Fig. 15
represents the simulation results with 10%, 15%, and 20% penetration levels
The Maximum value of uf measured on L1~L11
u_{f} magnitude according to penetration levels
Figs. 13
and
14 s
how that the voltage unbalance exceeds the limit at most times. On average, u
_{f }
is increased by 0.836% as the penetration level rises in 5% intervals from 10% to 20%.
5. Conclusion
The voltage sag and unbalance incurred by connecting EVs to actual distribution system of KEPCO in one part of Seoul, Republic of Korea were analyzed. EVs are connected to the power system at varying times 11 points. The number of EVs is estimated based on actual traffic volume and average driving time. Actual traffic volume data or the number of EVs should be investigated. EMTP/ATPDraw is used and the results are compared with IEEE standards. While voltage sag exceeds the limit at 20% penetration level, the voltage unbalance exceeds the limit at 10% penetration level. From the results of various penetration levels, it is also shown that voltage unbalance is more frequent than voltage sag. Accordingly, to maintain the supply reliability in the distribution system and to prepare the propagation of EVs, compensation devices and algorithms for reducing the voltage sag and unbalance should be applied.
Acknowledgements
This work was supported by the Human Resources Development of the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korea government Ministry of Trade, Industry and Energy (No. 20124010203300)
BIO
SoonJeong Lee He received his B.S. degree in Depart ment of Electrical and Electronics Engineering from Kangwon National University, 2010 and M.S. dgree in College of Information and Communication Engineering from Sungkyunkwan University, South Korea, 2012 respectively. At present, he is working for his Ph.D. course in Sungkyunkwan University. His research interests are power quality, power system transient analysis, power system protection and electric vehicle.
JunHyeok Kim He received B.S degree in School of Electrical an d Computer Engineering from Sungkyunkwan University, 2012. At present, he is working on his MS course in Sungkyunkwan University. His research interests include power system transients, protection and stability with electric vehicle.
DooUng Kim He received B.S degree in School of Electrical and Computer Engineering from Sungkyunkwan Uni versity, 2012. At present, he is working on his MS course in Sungkyunkwan University. His research interests include power system transients, DC distribution system and electric vehicle.
HyoSang Go He received his B.S in School of Department of Physics Chungbuk National University, Korea, 20 10. At present, he is working on his MS course in Sungkyunkwan Uni versity. His main research interests are power system transients, protection and power quality.
ChulHwan Kim He received his B.S. and M.S. degrees in Electrical Engineering from Sungkyunkwan University, South Korea, 1982 and 1984, respect ively. He received a Ph.D. degree in Electrical Engineering from Sungkyun kwan University in 1990. In 1990 he joined Cheju National University, Cheju, South Korea, as a fulltime Lecturer. He has been a visiting academic at the University of BATH, UK, in 1996, 1998, and 1999. Since March 1992, he has been a professor in the College of Information and Communication Engineering, Sungkyunkwan University, South Korea. His research interests include power system protection, artificial intelligence application for protection and control, the modelling/protection of underground cable and EMTP software.
EungSang Kim He received B.S degree in and M.S degrees in Electrical Engineering from Seoul National University of Science & Technology, 1988. He received his M.S and Ph.D. degree in Electrical Engineering from Soongsil University, 1991, 1997 re spectively. At present, he is working at KERI. His research interests include modeling and analyzing of distributed generation system, design, control and analysis of microgrid system.
SeulKi Kim He received B.S and M. s degrees in Electrical and Electronics Engineering from Korea University, 1998, 2000 respectively. He received his Ph.D. degree in Electrical and Electronics Engineering from Korea University, 2010. At present, he is working at KERI. His research interests include modeling and analyzing of distributed generation system, design, control and analysis of micro grid system. He received B.S and M.S neering versity, ing distributed micro
2014
“A Research of Charging Infrastructure for Electric Vehicle,”
Staats P. T.
,
Grady W. M.
,
Arapostathis A.
,
Thallam R. S.
1998
“A Statistical Analysis of the Effect of Electric Vehicle Battery Charging on Distribution System Harmonic Voltages,”
IEEE Transactions on Power Delivery
13
(2)
Kelly L.
,
Rowe A.
,
Wild P.
2009
“Analyzing the Impacts of Plugin Electric Vehicles on Distribution Networks in British Columbia,”
Electrical Power & Energy Conference
Taylor J.
,
Maitra A.
,
Alexander M.
,
Brooks D.
,
Duvall M.
2009
“Evaluation of the Impact of Plugin Electric Vehicle Loading on Distribution System Operations,”
Power & Energy Society General Meeting
Gómez J. Carlos
,
Mocrcos Medhat M.
2003
“Impact of EV Battery Chargers on the Power Quality of Distribution Systems,”
IEEE Transactions on Power Delivery
18
(3)
Shao Shengnan
,
Pipattanasomporn Manisa
,
Rahman Saifur
2012
“Grid Integration of Electric Vehicles and Demand Response With Customer Choice,”
IEEE Transactions on Smart Grid
3
(1)
Maitra Arindam
,
Taylor Jason
,
Brooks Daniel
,
Alexander Mark
,
Duvall Mark
2009
“Intergrating PluginElectric Vehicles with the Distribution System,”
International Conference on Electricity Distribution
Geske Martin
,
Stötzer Martin
2014
“Modeling and Simulation of Electric Car Penetration in the Distribution Power SystemCase Study,”
Modern Electric Power Systems
Qian Kejun
,
Zhou Chengke
,
Allan Malcolm
,
Yuan Yue
2011
“Modeling of Load Demand Due to EV Battery Charging in Distribution Systems,”
IEEE Transactions on Power Systems
26
(2)
Taylor Jason
,
Maitra Arindam
,
Alexander Mark
,
Brooks Daniel
,
Duvall Mark
2010
“Evaluations of Plugin Electric Vehicle Distribution System Impacts,”
Power and Energy Society General Meeting
Clement Kristien
,
Haesen Edwin
,
Driesen Johan
2009
“Stochastic Analysis of the Impact of PlugIn Hybrid Electric Vehicles on the Distribution Grid,”
International Conference on Electricity Distribution
ClementNyns Kristien
,
Haesen Edwin
,
Driesen Johan
2010
“The Impact of Charging PlugIn Hybrid Electric Vehicles on a Residential Distribution Grid,”
IEEE Transactions on Power Systems
25
(1)
Koichiro Shimizu
,
Taisuke Masuta
,
Yuyaka Ota
,
Akihiko Yokoyama
2010
“Load Frequency Control in Power System Using VehicletoGrid System Considering the Customer Convenience of Electric Vehicles,”
International Conference on Power System Technology
Papadopoulos P.
,
Cipcigan L. M.
,
Jenkins N.
,
Grau I.
2011
“Distribution Networks with Electric, Electric Vehicles Chargers Characterization: Load Demand and Harmonic Distortion,”
Electrical Power Quality and Utilisation
Trovão João P.
,
Pereirinha Paulo G.
,
Trovão Leonor
,
Jorge Humberto M.
2011
“Electric Vehicles Chargers Characterization: Load Demand and Harmonic Distortion,”
Electrical Power Quality and Utilisation
Park KeonWoo
,
Seo HunChul
,
Kim ChulHwan
,
Jung Changsoo
,
Yoo YeonPyo
,
Lim YongHoon
2009
“Analysis of the Neutral Current for TwoStepType Poles in Distribution Lines,”
IEEE Transactions on Power Delivery
24
1483 
1489
DOI : 10.1109/TPWRD.2009.2021031
Statics Korea
“http://www.index.go.kr”
Korea Transport Database
“http://www.ktdb.go.kr”
Kim DooUng
,
Kim JunHyeok
,
Go HyoSang
,
Seo Hun Chul
,
Kim ChulHwan
,
Kim Eungsang
2012
“Modeling of Single Phase PWM AC/DC Converter for EV using EMTP/MODELS,”
KIEE Summer Conference & General Meeting
Kim JunHyeok
,
Go HyoSang
,
Kim DooUng
,
Seo Hun Chul
,
Kim ChulHwan
,
Kim EungSang
2012
“Modeling of Battery for Electric Vehicle Using EMTP/ MODELS,”
IEEE Vehicle Power and Propulsion Conference
2009
“IEEE Recommended Practice for Monitoring Electric Power Quality,”
Meyer Jan
,
Hahle Stephan
,
Schegner Peter
,
Wald Carsten
2011
“Impact of electrical car charging on unbalance in public low voltage grids,”
Electrical Power Quality and Utilization conference