Method to Assess the Power Quality Impact of Plug-in Hybrid …apic/uploads/Research/sample6.pdf · 2013-08-13 · collective impact of PEVs. In response to this need, this paper

1

Abstract—As the sales of plug-in electric vehicles (PEVs) have

substantially increased in recent years, utility companies are

becoming concerned with the associated impacts on the power

quality (PQ) of distribution systems. As a result, there is an

urgent demand for techniques that can model and assess the

collective impact of PEVs. In response to this need, this paper

proposes a probabilistic PQ assessment model of PEVs, using a

Monte Carlo simulation method. The key challenge faced by this

technique is to determine a realistic PEV model, considering its

random behavior, market growing trends, in addition to different

chargers electrical characteristics. The result is a harmonic

analysis technique suited for extensively studying the PQ impact

of the growing deployment of electric vehicles, which is of great

interest to utility companies, since it may subsidize their network

planning and maintenance decisions.

Index Terms—Distribution power system, harmonic analysis,

Monte Carlo simulation, plug-in hybrid electric vehicle (PHEV).

I. INTRODUCTION

RANSPORTATION makes up a substantial portion of

global air pollution and oil consumption. In recent years,

the desire to reduce air pollution and reliance on oil has

resulted in increased reception of plug-in and pure electric

vehicles (PEV) [1]-[3].

As PEV draws power through power electronic circuits

from the grid when being charged, utility companies are

concerned with the potential power quality (PQ) impact of the

mass adoption of the vehicles [4]-[5]. For example, PEVs may

inject harmonics into utility systems, creating higher harmonic

distortions. The high level of unbalanced charging load may

cause problems such as neutral current rise. Therefore, there is

an urgent need for techniques that can model and assess the

collective impact of PEVs.

As will be shown later, the above problem cannot be

analyzed using some form of deterministic approaches based

on “average” or “typical” model and data. A Monte Carlo

simulation-based method is proposed. This is because PEV

chargers can be plugged into utility system at any time. A

utility feeder will supply a set of randomly connected PEVs

with diverse harmonic characteristics and charging duration.

This work was supported by NSERC, Canada and FAPESP, Brazil.

C. Jiang, R. Torquato, D. Salles and W. Xu were with the Department of

Electrical and Computer Engineering, University of Alberta, Edmonton, AB

T6G 2V4, Canada (e-mail: [email protected]; [email protected]) when this

research was carried out.

The locations of the EV being charged are also random. For

studying the impact of PEV, a main challenge, therefore, is to

model the random charging behavior. This behavior is in turn

affected by the driving habits of the PEV owners. In addition,

the PQ impact of PEV cannot be assessed in isolation, as other

home appliances can also produce PQ disturbances, whose

operating pattern may have some correlation with that of the

PEVs.

Based on the Monte Carlo simulation method developed in

[6], this paper presents techniques to model the random

operation of PEV chargers and to integrate the models into the

simulation method of [6] for system wide PQ impact

assessment. With this approach, PEVs are treated as one of the

(large) “home appliances”. As a result, the combined and

relative impact of various home appliances can be determined.

It is worthwhile to point out that the sales of plug-in hybrid

electric vehicles grow much faster than the pure electric

vehicles due to longer driving range, lower battery costs and

faster recharging times [3], [7]. The models and results

presented in this paper are related to the plug-in hybrid. Many

of the concepts and models proposed can be modified for pure

EVs once their mass-market versions mature and are accepted

by consumers. In addition, the study assumes that PEV owners

charge their cars at homes.

The remainder of the paper is organized as follows: Section

II explains the overall strategy of Monte Carlo simulation.

Section III presents a set of models to characterize the random

behaviors of PEV. Section IV shows a model of PHEV

consumer adoption trends, which is needed to predict the PQ

impact of PEV in the future. The electric model of PEV and

utility system are provided in Section V. Sample case study

results are presented in Section VI.

II. GENERAL SIMULATION SCHEME

Over the years, some works have addressed the PQ impact

of EV integration, which greatly contributed to the authors

understanding on the subject. Reference [8], one of the pioneer

works on the subject, identifies the harmonic currents

produced by electric vehicle chargers. However, it considers

all chargers clustered on a single bus and, therefore, harmonic

currents add in phase. Such conclusion is not true for a more

realistic disperse chargers scenario. In addition, only one

charger characteristic is considered. Harmonic impacts

produced by five different types of chargers are assessed on

[9], without considering, however, the vehicle plug-in

behavior and penetration characteristics. The authors have still

Method to Assess the Power Quality Impact of

Plug-in Hybrid Electric Vehicles (V1.0)

Chen Jiang, Student Member, IEEE, Ricardo Torquato, Student Member, IEEE, Diogo Salles,

Member, IEEE, Wilsun Xu, Fellow, IEEE

T

2

expanded their work on [10] to include EV random plug-in

time and charging duration. But the proposed method only

addresses EV connection impact on HV/MV substation,

limiting penetration level to, at most, one EV per MV/LV

transformer.

References [11]-[12] derive EV random impact on system

voltage and current harmonics using an analytical approach.

This approach presents some limitations, such as: (a) to

provide reliable results (converged average and standard

deviation), at least 6 vehicles must be connected to network;

(b) the impact of chargers with different characteristics (e.g.,

power demand and harmonic spectrum) simultaneously

connected to the network is not considered. On [13], a more

detailed EV model is proposed, but random characteristics of

plug-in instant, charging time and EV location are not taken

into account. Vehicle penetration level, for instance, is

regarded as equally distributed within all LV consumers. It

may lead to inaccuracies since, on a realistic scenario, the

number of EVs is different for each household and, therefore,

must be randomly distributed.

These approximations considered on previous works make

the EV model less realistic and unsuitable for PQ analyses.

This paper takes a step further and proposes a model capable

of considering all main factors that may influence EV random

PQ impact on network. The proposed model is able to evaluate

the combined PQ impact of EV and home appliance random

behaviors, which has not been included on any previous work.

In addition, previous papers have not emphasized the network

modeling, which must be multiphase since charger and home

appliance connections may be either phase-to-neutral or phase-

to-phase. This paper provides a detailed description of primary

and secondary networks, including the multiphase house

equivalent circuit for harmonic studies. The assessment model

here proposed is suitable for both primary and secondary

networks analyses. The inclusion of such details may bring

new results to the analysis, such as the PQ impact on a single

house; especially due to home appliance and EV multiphase

interaction.

A. Overview of the study methodology

Before presenting the simulation technique procedure, it is

important to intuitively explain why this paper adopts a Monte

Carlo simulation approach for modeling the daily behavior (or

activities) of PEVs, as follows.

In a realistic scenario, a PEV can connect to a power

system for charging up at any time. As a result, a utility feeder

will supply a set of randomly connected EVs at any given

time. Furthermore, these vehicles will go off the grid at

different time as each of them has different charging duration.

The locations of the EV being charged in a feeder are also

random. In order to determine the power quality impact of

these vehicles, a methodology that can model these random

factors must be developed. It is impossible or unrealistic to

predict the PQ impact using some form of deterministic

approaches based on “average” or “typical” data. For example,

there is no such thing as the “average” locations of EVs in a

feeder for harmonic assessment.

The methodology adopted by this paper is the so-called

Monte Carlo simulation. The basic idea of this methodology is

to create numerous plausible scenarios of EV activities in a

feeder for a given instant of time, say at 12:04pm. One of the

scenarios, for example, represents the case where EV-A is

connected to location X and EV-B to location Y etc. At the

instant of 12:04pm, EV-A battery is charged to m% (i.e., SOC

= m%) and EV-B is at n%. Once such a scenario is created, all

load activities become known and are deterministic. Harmonic

power flow studies can then be conducted to determine the

harmonic levels associated with this particular scenario. In this

paper, one such study is called one Monte Carlo run. A Monte

Carlo simulation involves the creation of thousands of

plausible scenarios for 12:04pm or thousands of Monte Carlo

runs. Therefore, thousands of harmonic studies are performed.

The average of the harmonic results among all scenarios

should represent the most likely results expected at 12:04pm.

The remaining problem becomes how to create the multiple

plausible scenarios for the instant of 12:04pm. This is done by

considering the probability of occurrence of the various

factors. For example, if 10% houses are found to have PHEV

based on consumer trend, 10% of the randomly selected

houses in the feeder model will be flagged as the candidate

locations for PHEV connection. For each house, the time of

PHEV connection and its charging duration depend on the

driving habit. There is also a probability curve for the habit.

For example, if the probability curve shows that there is a 10%

chance that the car will be charged at 12:04pm, 10% the

scenarios created will have a car being charged at that

particular house location. By considering various probabilistic

factors, multiple scenarios can be created. Naturally, these

scenarios will satisfy or have included the probabilistic

characteristics of the various factors.

The above multi-scenario simulations or Monte Carlo runs

can be conducted for every minute (or other resolution) over a

24 hour period. Thus, in this case, there are 1440 snapshots for

one day. Since each minute has one set of statistical averages,

the variation of the harmonic impact over a 24 hour period is

estimated. In order to provide more succinct indices to

characterize the PEV impact, the results over a 24 hour period

can be further condensed by selecting their 95% probability

values, i.e. the values that will not be exceeded for the 95% of

the time over a 24 hour period. With this approach, the impact

of PEV can be understood from one value for each PQ index

of concern.

However, to include the PEV on the analysis, several

challenges must be solved. Such challenges may be divided

into: PEV random behavior characteristics, consumer adoption

pattern, and electrical characteristics. The following sections

are devoted to further address these concerns. In spite of

focusing PHEVs, the proposed methodology is applicable to

any PEV.

3

III. BEHAVIOR CHARACTERISTICS OF A PHEV

When attempting to model a PHEV for PQ studies, the first

step is to identify the factors that influence its random network

connection behavior throughout 1 day. PHEV connection

characteristics may be described by two variables: PHEV

charging start time and charging duration. Both are further

described in this section.

A. PHEV charging start time

The charging start time determines when vehicles will start

being charged during the day. Since all PHEVs will not begin

charging simultaneously, it is more suitable to treat this

variable as a random variable based upon daily charge

strategy. Charging strategy is the way PHEV owners plan (or

forced to plan) to charge their batteries every day. In

following, three possible charging strategies are introduced.

1) Uncontrolled charging

When there is no plan, it is called uncontrolled charging,

which means that PHEVs could start charging any time during

the day. In this strategy, most people will immediately charge

their vehicles to prepare them for the next trip just when they

arrive home from work. Obviously, uncontrolled charging

tendency is strongly associated with people’s travelling

behavior. In this case, most PHEVs are plugged in and start

charging around 18:00 when people usually come back home.

However, a uniform distribution with a narrow range around

18:00 is more close to reality, as shown in Fig. 1. So, start time

pattern for uncontrolled charging is:

19 b 18,a b,xa ,1

)(

ab

xf (1)

2) Controlled charging

Once PHEV adoption rate reaches a certain level, because

of the coincidence with early evening system peak,

uncontrolled charging plan could significantly increase this

peak. In such case, both new peak generation capacity and

burden on transmission and distribution will be of concern.

Therefore, it is likely that utilities would use either Time-Of-

Use (TOU) pricing or direct control methods such as in-home

delay devices to shift PHEV charging load to off-peak time.

TOU pricing is used in most regions, which leads people to

postpone charging after 9:00 p.m. in order to minimize their

electricity bills. As a result, (2) is applied instead of (1) for

controlled charging.

24 b ,12a b,xa ,1

)(

ab

xf (2)

3) Smart charging

The last strategy is smart charging which is implemented by

incorporating smart technology into the charging system and

distribution grid. This strategy not only maintains the evening

peak constant but also can be advantageous to utilities. For

example, reference [14] uses Advanced Metering

Infrastructure (AMI) together with PHEV control unit and

remote switches to imply stagger charging which limits the

charging based upon pre-determined power levels

communicated through the grid. This method will help smooth

the PHEV charging load seen by service transformer,

especially for high PHEV penetration. As a result, no

additional peaks due to PHEV will be created on transformer

load, either in the early evening or midnight. Considering the

complexity of employing various smart charging optimization

algorithms, a simple distribution of start time to represent all

smart charging strategies is adopted in this paper, as illustrated

in Fig. 1 [15]. In this case:

3 ,1 ,e1

)(

2

2

1

x

xf (3)

0 2 4 6 8 10 12 14 16 18 20 22 240

20

40

60

80

100

Time (h)P

robabili

ty (

%)

Uncontrolled Controlled Smart

Fig. 1. Charging start time pattern for each charging strategy.

B. PHEV charging duration

Once the driver arrives at home, the PHEV battery will

have a remaining energy quantified here by the factor state of

charge (SOC). The state of charge (SOC) is the equivalent of a

fuel gauge for the battery pack in an electric vehicle and the

units of SOC are percentage points (0% = empty; 100% =

full). The factor SOC is random and its associated pdf is based

upon daily charge and travel pattern, as explained below.

To determine the SOC of each PHEV, it is essential to

know how deep it was discharged during the day, which is

directly related to the distance it traveled. From general

vehicles driving pattern, a probability distribution of daily

distance driven has been derived in [8]. The distribution is

found to be log-normal type, but with zero probability at all

negative distances. The mean of the distribution is 34.2 miles

and the standard deviation is 21.1 miles, at the year of 1983.

The probability function is as follows:

0,2

1)(

2

2

2

ln

mem

mdm

(4)

where

2][

][1ln

2

1][ln

ME

MVarME (5)

2

2

][

][1ln

ME

MVar

(6)

where the random variable m is the daily distance travelled in

miles, E[M] and Var[M] stand for the mean value and

variance of m. According to updated information provided by

4

[16], daily average vehicle miles travelled is 33 miles. To

update our estimation, the standard deviation is scaled to 20.4

miles with constant ratio. Fig. 2 shows the probability density

function of daily distance driven.

0 20 40 60 80 1000

0.5

1

1.5

2

2.5

3

Daily Travel Distance (miles)

Pro

babili

ty D

ensity (

%)

Fig. 2. Probability density function of PHEV daily distance driven.

Having the traveled distance (m), the battery state of charge

(SOC) at the beginning of charging can be determined by:

Rm

RmR

mR

SOC

,0

0%,100 (7)

where R is the all-electric range of the PHEV in miles.

Then, the next step is to find the charging duration (D) from

the SOC value determined above. This can be calculated from

the PHEV battery nameplate parameters, which includes

battery capacity, charger level and initial SOC, as follows:

P

DODSOCCD

1 (8)

where C is battery capacity (kwh); DOD stands for the depth

of discharge and it determines the fraction of power that can

be withdrawn from the battery (%); P is the power rating of

charger (kW), which is determined by charger level; and is

the efficiency of charger (%).

IV. CONSUMER ADOPTION OF PHEVS

In order to model charger loads in a network and anticipate

their impact in the future, the penetration level of chargers

should be estimated. In other words, the number of PHEVs per

household in the network should be predicted. Several

organizations have provided estimations for PHEV penetration

level for the near future. As a case in point, a report by Oak

Ridge National Laboratory in 2006 [17] estimates that PHEV-

20 vehicles (which have 20-mile all electric range) will have a

base case market potential of over 25% of sales for the entire

car and light-truck market in 2018. EPRI reports [18]-[19]

have similar analysis results for 2010 to 2030 PHEV market

share.

The number of cars sold every year in the US is about 20

million. So, considering the PHEV market share level for year

2010 to 2030 as well as the assumption of average 12 years

vehicle lifespan, the total number of PHEV for 2010 to 2030 is

shown in Fig. 3(a) [17], [20]. According to the linear

regression analysis over the data from 1990 to 2008 available

in [20], the total projected number of passenger vehicles for

2010 to 2030 is calculated and plotted in Fig. 3(b).

0

20

40

60

80

100

120

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025

2026

2027

2028

2029

2030

Tota

l Num

ber

of P

HEV

s (m

illio

ns)

Year

(a)

0

50

100

150

200

250

300

350

400

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025

2026

2027

2028

2029

2030To

tal N

umbe

r of

Pas

seng

er

Vehi

cles

(mill

ions

)

Year

(b)

Fig. 3. PHEV market trend data for 2010 to 2030. (a) Total estimated

number of PHEVs. (b) Total estimated number of passenger vehicles.

Assuming that the average number of vehicles per

household is 2, the penetration level across households can be

calculated as follows:

PVHiN

iNiPHEVH

PV

PHEV )(

)()( (9)

where PHEVH(i) is the number of PHEVs per household at

year i, NPHEV is the total number of PHEVs shown in Fig. 3(a),

NPV is the total number of passenger vehicles shown in Fig.

3(b) and PVH is the average number of vehicles per

household, which is assumed equal to 2. From (9), the final

expected number of PHEVs per household can be estimated

and it is shown in Fig. 4. From this figure, the average number

of PHEVs per household in 2020 is 0.2, which shows good

consistency with the estimation in [14]. Given the number of

houses connected to each transformer, the number of PHEVs

fed by each transformer can be simply calculated using Fig. 4.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.72

01

0

20

11

20

12

20

13

20

14

20

15

20

16

20

17

20

18

20

19

20

20

20

21

20

22

20

23

20

24

20

25

20

26

20

27

20

28

20

29

20

30

Year

PHEV per Household

Fig. 4. Total estimated number of PHEVs per household for 2010 to 2030.

Fig. 4 information is essential to determine PHEVs

locations for a 24-hour Monte Carlo simulation. The following

procedure is used: for each residential customer (x), a

randomly generated number (rx), uniformly distributed

between 0 and 1 is compared to the corresponding PHEVs

penetration rate. If rx is less than the penetration rate, then a

PHEV is assigned to that house.

V. ELECTRICAL MODELS

In order to assess PHEV power quality impacts on a

network, it is essential to determine the electrical

characteristics of PHEV chargers. This information data will

form the technical base to develop a harmonic producing

electrical model for the chargers. In addition, a power

distribution system multiphase model is also provided.

A. PHEV charger harmonic model

According to the Society of Automotive Engineers (SAE)

all EVs produced by automakers in North America must

follow SAE J1772 standard [21]. Based on voltage and power

levels, three levels of charging are identified in SAE J1772,

5

which are shown in Table I. Level 3 chargers are mainly

intended for commercial and public applications and,

therefore, are not considered in this paper. However, the

proposed model can be extended to include these chargers

once their electrical and usage data are available.

TABLE I

PHEV CHARGING CHARACTERISTICS DEFINED IN SAE J1772 [21].

Charger Type Input Voltage Maximum Power (kW)

Level 1 120 VAC 1.440

Level 2 208-240 VAC 11.50

Level 3 208-240 VAC 96.00

Level 3 (DC) 208-600 VDC 240.0

As PHEV chargers use switched-mode power supplies, they

can inject harmonic currents into the supply system. This is the

main power quality concern for the PHEV. Harmonic

characteristics of one Level 1 and one Level 2 chargers have

been measured, and Table II presents their harmonic spectra.

TABLE II

HARMONIC SPECTRA OF LEVEL 1 AND LEVEL 2 CHARGERS (AVERAGE VALUES

DURING CHARGING).

Charger Level 1 Level 2

Harmonic

Order Mag. (%) Mag. (%)

H1 100.0 100.0

H3 9.125 8.864

H5 3.230 2.452

H7 0.947 0.891

H9 1.520 0.911

H11 1.329 0.870

According to measurements, Level 1 and Level 2 chargers

can be considered as nonlinear loads. Therefore, they may be

modeled by constant power load for the fundamental

frequency and by a current source at the harmonic frequencies

[22].

Fundamental power flow results are used to calculate

harmonic current injections of nonlinear loads, in order to

perform harmonic power flow calculation. Hence, the EV

charger current injection is calculated using (10) [22], where

spectrum refers to charger typical harmonic current spectrum,

and 11I is the fundamental current injected by the charger

on the network.

spectrum

spectrumh

hI

III

1

1

spectrumspectrumhh

h

11

(10)

Current phase angle correction is important since charger

harmonic currents may cancel with other nonlinear appliances

currents. Such cancellation effect may potentially reduce the

overall harmonic impact of EV chargers.

Due to their input voltage rating, Level 1 chargers are

connected from phase to neutral (120 V), while Level 2

chargers should be connected from phase to phase. Hence, one

may notice that Level 2 chargers do not increase network

voltage imbalance, unlike Level 1 chargers.

B. Network model

This subsection will describe both primary and secondary

multiphase network models used for PHEV PQ assessment

studies.

In North America, the most common power distribution

system is the multigrounded neutral (MGN) distribution

system. The primary feeder delivers the electrical power from

the source at the substation to the customers at various

locations through secondary distribution systems. The

secondary system connects several houses to the service

transformer. Loads (home appliances and PHEV chargers) of

each house are connected between phases and neutral or

between phases.

For studying the harmonic impact on the secondary system,

the primary system is modeled as an equivalent circuit. The

service transformer is modeled explicitly and the loads

(individual houses) are modeled as equivalent circuits in the

form of one house per equivalent circuit. The multiphase

equivalent model to study the secondary system is shown in

Fig. 5 [23].

Neutral

House #1

THV

THZ

MGNZ

House #N

Service

TransformerPrimary System Secondary System

ia ia

ib ib

ZabZab

RcRcRT

ZNZN

Za Za

Zb Zb

iabiab

ZA

ZB

ZA

ZB

+120 V

-120 V

Fig. 5. Schematic model to study multiphase secondary distribution systems.

The houses are represented by residential loads, which are

supplied with phase-to-neutral 120V (modeled by Za, ia and Zb,

ib) and phase-to-phase 240V (modeled by Zab, iab). ZN is the

neutral impedance. Linear residential loads are modeled as

constant power loads at the fundamental frequency and as

impedance (Za, Zb and Zab) at harmonic frequencies. The

nonlinear residential loads including PHEVs are modeled as

constant power loads at the fundamental frequency and as

current sources (ia, ib and iab) at harmonic frequencies. Hence,

with such model, interaction between vehicles and appliances

is taken into account.

If the main target consists on studying the primary system

impacts, a secondary system equivalent must be calculated and

attached to the multiphase primary network, as shown on Fig.

6. The portion of the network inside the dashed rectangle

represents one or more single-phase service transformers

connected between each phase of the primary conductor and

neutral conductor, and each service transformer could be

represented by the secondary system configuration shown in

Fig. 5. Table III presents the system parameters of the primary

and secondary test systems.

6

Therefore, one may conclude that the methodology

proposed in this paper is suitable for both primary and

secondary system PHEV PQ impact analyses.

Primary System

Rgs RgnRgn Rgn Rgn

Section 2Section 1 Section N. . .Vsub Zsub

One or more

service transformers

Fig. 6. Schematic model to study multiphase primary distribution systems.

TABLE III

BASE CASE SYSTEM PARAMETERS.

Base Case System Parameters Values

Primary

System

Supply system voltage 14400 V (@ 60 Hz)

Substation pos. sequence impedance 0.688 + j2.470 ohms

Substation zero sequence impedance 0.065 + j2.814 ohms

Substation grounding (Rgs) 0.15 ohms

MGN grounding resistance (Rgn) 15 ohms

Grounding span of MGN neutral 75 m

Feeder length 15 km

Feeder conductor type 4 - 336.4 ACSR

Service

Transformer

Voltage (VH/VL) rating 14400/120 V

KVA rating 37.5 kVA

Impedance 2 %

Resistance 1.293 %

Grounding resistance (RT) 12 ohms

Secondary

System

Customer grounding resistance (RC) 1 ohm

Neutral impedance (ZN) 0.55 + j0.365 ohm/km

Phase impedance (ZA and ZB) 0.21 + j0.094 ohm/km

Num. of houses for each transformer 10

Distance between houses 20 m

C. Summary

From the general characteristics of PHEVs previously

presented, we can identify 6 main factors, outlined below, that

influence the time instant the PHEV connects to the grid, the

time instant it goes off the grid (or charging duration) and the

harmonic current behavior during charging. These factors can

be either deterministic or random and compose the Monte

Carlo simulation technique. The deterministic factors are

parameters (i.e., previously known) and the random factors are

modeled as probability distribution functions (pdfs). The main

characteristics of PHEVs are:

1. Charging strategy – deterministic;

2. Charging start time – random;

3. PHEV state of charge (SOC) – random;

4. PHEV penetration level – deterministic;

5. PHEV location in a residential system – random;

6. Type of chargers – semi-deterministic.

As each PHEV charging load is modeled as constant power

for fundamental frequency and constant current source for

harmonic frequency, the integration of PHEV charging load is

achieved by connecting the model at the certain node (house)

of the secondary system shown in Fig. 5 during corresponding

charging duration. The detailed procedure to simulate PHEVs

integrating with secondary system is presented on Fig. 7. In

order to simplify the flowchart, only 1 day simulation is

presented. The overall Monte Carlo simulation method

consists on repeating this algorithm several times until

converged results are achieved.

Assign daily usage pattern for

home appliances according to [6]Determine PHEV locations

on network according to

penetration levels

k=0

nPHEV = # of PHEVs on network

k < nPHEV?

Determine charger

type for PHEVk

Determine charging

start time and charging

duration for PHEVk

k = k + 1

Yes

Not = 0

t < 24h?

Determine house equivalent

circuits on secondary network

for fundamental frequency:

PA, QA; PB, QB; PAB, QAB

Solve fundamental

power flow

Determine house equivalent

circuits on secondary network

for harmonic frequency:

ZA, iA; ZB, iB; ZAB, iAB

Solve harmonic

power flow

Determine charging

strategy for PHEVk

Store results

t = t + 1min

Yes

Output 24-hour

results

PHEV characterization

No

PQ impact assessment algorithm

Fig. 7. Flowchart for PHEV PQ impact analysis on secondary networks.

Primary system PQ impact assessment may be performed

using the same flowchart with an additional step. Before

performing power flow solution, once house circuit parameters

have been determined, the equivalent circuit seen from

primary network must be calculated for each MV/LV

transformer.

VI. CASE STUDIES

This section illustrates two case studies that can be

investigated under the proposed PHEV modeling method.

PHEVs are connected on the secondary network previously

described.

A. Case Study 1: overall PHEV harmonic contribution

For this study, the base case scenario is defined as

integrating mixed charger types under uncontrolled charging

strategy with 30% PHEV penetration level into the system.

Mixed chargers include Level 1 and Level 2 chargers at a ratio

1:1. This case is used to give an overview of the situation

when PHEVs penetrates into system at the worst charging

strategy and most common chargers composition are used. The

objective of this case study is to compare the impacts of PHEV

with those of other nonlinear home appliances. The “30%

PHEV” base case is compared to three other scenarios:

a. “NO PHEV”: This scenario considers only home

appliances and the respective penetration rates refer to

year 2011 market data. There are no PHEVs in the

system;

b. “CFL 2015”: All home appliances loads remain the same

7

as the “NO PHEV” scenario, except that the penetration

of CFL appliances refers to year 2015 market data.

c. “PC 2015”: All home appliances loads remain the same as

the “NO PHEV” scenario, except that the penetration of

personal computer appliances refers to year 2015 market

data.

Fig. 8 shows a sample result, which is the 95% probability

values over the harmonic voltage profile at the metering point

averaged among all homes connected to the sample secondary

system. This figure indicates that even with 30% penetration

level, the harmonic distortion caused by PHEV is not

significant since both phase and neutral harmonic voltage

levels are comparable to those of “Pure Load” case study.

Moreover, from “CFL 2015” and “PC 2015”, the increasing

penetration of CFL and PC loads should be more of concern to

utilities in terms of harmonic distortion in secondary systems

compared to increasing usage of PHEVs. The significant

increase on fundamental and RMS neutral voltage is mainly

caused by load imbalance between the two phases, due to

integration of Level 1 chargers.

3rd 5th 7th 9th 11th 13th 15th THDv0

1

2

3

4

5

Harmonic Order #

IHD

V (

%)

Pure Load

30% PHEV

CFL 2015

PC 2015

(a)

1st 3rd 5th 7th 9th 11th 13th 15th RMS0

0.2

0.4

0.6

0.8

Harmonic Order #

Voltage (

V)

Pure Load

30% PHEV

CFL 2015

PC 2015

(b)

Fig. 8. Comparison between harmonic currents produced by PHEVs, CFLs

and PCs. (a) 95% index for phase voltage. (b) 95% index for neutral voltage.

B. Case Study 2: charger type impact

In the second case study, the charger types are changed in

order to identify their impact on the secondary system. All

system parameters remain the same as base case described on

Case Study 1, except charger types. Fig. 9(a) shows the 95%

index associated to the average phase harmonic voltages, for

different chargers types. Fig. 9(b) shows the 95% index of the

average neutral to ground voltage.

Once again, the proposed PHEV modeling methodology is

able to identify that the different charger types have little

impact on network harmonic distortions. On the other hand,

Level 2 chargers cause no neutral voltage rise, since they are

phase-to-phase connected; unlike Level 1 chargers, which are

phase-to-neutral connected and, therefore, increase load

imbalance level.

3rd 5th 7th 9th 11th 13th 15thTHDv0

0.5

1

1.5

2

2.5

Harmonic Order #

IHD

V (

%)

Level 1

Level 2

Mixed

(a)

1st 3rd 5th 7th 9th 11th13th15thRMS0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Harmonic Order #

Volta

ge (

V)

Level 1

Level 2

Mixed

(b)

Fig. 9. Power quality impact of different charger types. (a) 95% index for

phase voltage. (b) 95% index for neutral voltage.

For the sake of space, primary network impacts due to

PHEV connection are not presented on this paper. However,

the proposed methodology is able to address such impact by

calculating an equivalent secondary system and attaching it to

the primary network, as shown on Fig. 6.

C. Justification of results using the equivalent CFL concept

It is useful to compare the harmonic injection levels of the

PHEV charger with other major home appliances or consumer

devices using the total equivalent CFL index concept proposed

in [24]. The equivalent CFL index is introduced to provide a

quantitative comparison of the harmonic effects of the

appliances. This index quantifies each appliance in terms of its

harmonic effect expressed as the number of CFLs it is

equivalent to.

Table IV shows the values of the total equivalent CFL

index for one PHEV charger and other key home appliances.

One can observe that, in terms of harmonic effects to the

system, one PHEV is equivalent to approximately 10 CFLs or

1.5 desktop PCs. Since most homes will have CFLs and PCs

while the PHEV penetration level is only 0.2 per home by year

2020, one can draw a preliminary conclusion that the PHEV’s

harmonic impact will be significantly less than that of the

CFLs or PCs. TABLE IV

COMPARING THE HARMONIC EFFECT OF PHEVS AND HOME APPLIANCES.

Appliance type Operating

Power [W] Power Ratio Equivalent CFL

CFL 15 1 1.00

Electronic-Ballast

Fluorescent 18 1 1.23

Dryer 4500 300 2.13

LCD monitor 40 3 2.35

Furnace 500 33 3.49

Fridge 1200 80 4.34

Desktop PC 100 7 6.53

Microwave oven 1200 80 24.30

PHEV Charger 1355 90 9.77

VII. CONCLUSIONS

This paper presented a power quality impact assessment

model for plug-in EVs. This model considers realistic

deterministic and random variables that influence the daily

activities of plug-in EVs, including charging strategies,

charging start time, charging duration, market penetration,

vehicle connection point and charger electrical data. A Monte

Carlo simulation method has been employed to identify PHEV

impacts both on fundamental and harmonic frequencies. Its

usefulness has been demonstrated by some case studies, which

identified that the EV charger has negligible impact on

network voltage harmonic distortion, when compared to home

appliances such as CFL and PC. In addition, Level 2 charger

do not increase voltage imbalance, unlike Level 1 charger,

which is phase-to-neutral connected. Fundamental system

losses, service transformer overloading and neutral current

may also be addressed by the proposed technique. Findings

arisen from such studies are of great importance for the utility,

8

on its network planning and reinforcement analyses.

Furthermore, the proposed model may also be applied to

investigate the impacts of future standard developments, such

as new harmonic distortion limits; the use of PEVs as energy

storage devices, under the Vehicle to Grid (V2G) concept; and

the impact of integrating charger and PEV electric motor drive

converters. The latter concept, for instance, may reduce EV

harmonic impacts, with the expense of increasing rated power

demand.

VIII. REFERENCES

[1] R. Lache, “Vehicle electrification, more rapid growth; steeper price

declines for batteries,” DeutscheBank Global Markets Research, 2010.

[2] Texas Transportation Institute, Strategic Solutions Center (2011).

Electric Vehicle Market Penetration: An Examination of Industry

Literature. Austin, TX. [Online]. Available: http://tti.tamu.edu/group

/stsc/files/2010/11/BP-2010-01-Electric-Vehicles-_Updated-Final__edit

ed-_TB_.pdf.

[3] A. K. Srivastava and B. Annabathina, “The challenges and policy

options for integrating plug-in hybrid electric vehicle into the electric

grid,” The Electricity Journal, vol. 23, no. 3, pp. 83–91, Apr. 2010.

[4] Power Quality Requirements for Plug In Vehicle Chargers - Part 1:

Requirements, SAE Standard J2894/1, 2010.

[5] C. C. Chan and Y. S. Wong, “Electric vehicles charge forward,” IEEE

Power and Energy Magazine, vol. 2, no. 6, pp. 24- 33, Nov. 2004.

[6] D. Salles, C. Jiang, W. Xu, W. Freitas, and H. E. Mazin, “Assessing the

collective harmonic impact of modern residential loads—Part I:

Methodology,” IEEE Trans. Power Del., vol. 27, no. 4, pp. 1937-1946,

Oct. 2012.

[7] K. Clement-Nyns, E. Haesen, and J. Dreisen, “The impact of charging

plug-in hybrid electric vehicles on a residential distribution grid,” IEEE

Trans. Power Syst., vol. 25, no. 1, pp. 371-380, Dec. 2009.

[8] J. A. Orr, A. E. Emanuel, and K. W. Oberg, “Current Harmonics

Generated by a Cluster of Electric Vehicles Battery Chargers,” IEEE

Trans. PAS, vol. PAS-101, no. 3, pp. 691-700, Mar. 1982.

[9] J. A. Orr, A. E. Emanuel, and D. G. Pileggi, “Current Harmonics,

Voltage Distortion, and Powers Associated with Battery Chargers Part I:

Comparisons Among Different Types of Chargers,” IEEE Trans. PAS,

vol. PAS-101, no. 8, pp. 2703-2710, Aug. 1982.

[10] J. A. Orr, A. E. Emanuel, and D. G. Pileggi, “Current Harmonics,

Voltage Distortion, and Power Associated with Electric Vehicle Battery

Chargers Distributed on the Residential Power System,” IEEE Trans.

Ind. Applications, vol. IA-20, no. 4, pp. 727-734, Jul./Aug. 1984.

[11] P. T. Staats, W. M. Grady, A. Arapostathis, and R. S. Thallam, “A

Statistical Method for Predicting the Net Harmonic Currents Generated

by a Concentration of Electric Vehicle Battery Chargers,” IEEE Trans.

Power Del., vol. 12, no. 3, pp. 1258-1266, Jul. 1997.

[12] P. T. Staats, W. M. Grady, A. Arapostathis, and R. S. Thallam, “A

Statistical Analysis of the Effect of Electric Vehicle Battery Charging

on Distribution System Harmonic Voltages,” IEEE Trans. Power Del.,

vol. 13, no. 2, pp. 640-646, Apr. 1998.

[13] J. C. Gómez, and M. M. Morcos, “Impact of EV Battery Chargers on

the Power Quality of Distribution Systems,” IEEE Trans. Power Del.,

vol. 18, no. 3, pp. 975-981, Jul. 2003.

[14] S. Shao, M. Pipattanasomporn, and S. Rahman, “Challenges of PHEV

penetration to the residential distribution network,” in Proc. 2009 IEEE

Power & Energy Society General Meeting, pp.1-8.

[15] K. Qian, C. Zhou, M. Allan, and Y. Yuan, “Modeling of load demand

due to EV battery charging in distribution systems,” IEEE Trans. Power

Syst., vol. 26, no. 2, pp. 802-810, May 2011.

[16] M. Kintner-Meyer, K. Schneider, and R. Pratt (2007). Impacts

Assessment of Plug-in Hybrid Electric Utilities and Regional U.S.

Power Grids Part 1: Technical Analysis. PNNL Report. Richland, WA.

[Online]. Available: http://www.ferc.gov/about/com-mem/wellinghoff/5-

24-07-technical-analy-wellinghoff.pdf.

[17] S. W. Hadley, A. Tsvetkova, “Potential Impacts of Plug-in Hybrid

Electric Vehicles on Regional Power Generation,” ORNL/TM-

2007/150, Jan 2008.

[18] EPRI, “Environmental Assessment of Plug-in Hybrid Electric Vehicles:

Nationwide Greenhouse Gas Emissions,” Palo Alto, CA, Tech. Rep.

1015325, Jul. 2007.

[19] EPRI, “Environmental Assessment of Plug-in Hybrid Electric Vehicles:

United States Air Quality Analysis Based on AEO-2006 Assumptions

for 2030,” Palo Alto, CA, Tech. Rep. 1015326, Jul. 2007.

[20] U.S. Department of Transportation Federal Highway Administration

(2003). Journey to Work Trends in the United States and its Major

Metropolitan. Washington D.C. [Online]. Available:

http://www.fhwa.dot.gov/ctpp/jtw/contents.htm.

[21] C. B. Toepfer, “SAE Electric Vehicle Conductive Charge Coupler,”

Society of Automotive Engineers, U.S., Tech. Rep. SAE J1772, 2001.

[22] “Task force on harmonics modeling and simulation, modeling and

simulation of the propagation of harmonics in electric power networks–

part I: concepts, models, and simulation techniques,” IEEE Trans.

Power Del., vol. 11, no. 1, pp. 452–465, Jan. 1996.

[23] H. E. Mazin, E. E. Nino, W. Xu, and J. Yong, “A study on the harmonic

contributions of residential loads,” IEEE Trans. Power Del., vol. 26, no.

3, pp. 1592-1599, Jul. 2011.

[24] A. B. Nassif, J. Yong, W. Xu, and C. Y. Chung, “Indices for

comparative assessment of the harmonic effect of different home

appliances,” European Transactions on Electrical Power, Feb. 2012.

IX. BIOGRAPHIES

Chen Jiang (S’09) received the B.Eng degree in Electric Engineering and

Automation from Huazhong University of Science and Technology (HUST),

Wuhan, China, in 2008, and the M.Sc degree in Power Engineering and

Power Electronics from University of Alberta, Edmonton, Canada, in 2012.

Since 2012, he has worked as a Power System Planning Engineer at North

China Power Engineering Co., Ltd of China Power Engineering Consulting

Group. His main research interests are power system planning issues and

power quality.

Ricardo Torquato (S’11) obtained the B.Sc. degree in electrical

engineering from the University of Campinas, Campinas, Brazil in 2011,

where he is pursuing a M.Sc. degree. His research interests are power quality,

analysis of distribution systems and distributed generation.

Diogo Salles (S’04-M’12) received the B.Sc., M.Sc. and Ph.D. degrees,

all in electrical engineering, from the University of Campinas, Campinas,

Brazil, in 2006, 2008 and 2012, respectively. Currently, he is a Post-Doctoral

Researcher at the University of Campinas. From 2010 to 2012, he was a

Visiting Doctoral Scholar at the University of Alberta, Edmonton, AB,

Canada. His research interests focus on power quality, harmonics and power

disturbance data analysis.

Wilsun Xu (M’90-SM’95-F’05) obtained the Ph.D. from the University

of British Columbia, Vancouver, in 1989. Currently, he is a Professor and a

NSERC/iCORE Industrial Research Chair at the University of Alberta. His

current research interests are power quality and information extraction from

power disturbances.

1

Abstract — This paper presents a probabilistic harmonic

simulation method to study the power quality impact of electric

vehicles (EV). One of the main contributions of this paper is the

development of a set of probabilistic models for the vehicles, such

as the models for random charging time and charging duration.

The PQ impact is studied using a multiphase presentation of the

power system. This is due to the need to model different type of

chargers and to determine the PQ impact on neutrals. Extensive

case studies have been conducted using the proposed methods

based on measured EV data. The results reveal that the current

versions of plug-in hybrid EV have insignificant harmonic impact

on power systems. However, the Level 1 charger can increase the

neutral to earth voltage at homes which could lead to increased

stray voltage incidents.

Index Terms— Power quality, harmonics, electric vehicles.

I. INTRODUCTION

RANSPORTATION makes up a substantial portion of

global air pollution and oil consumption. In recent years,

the desire to reduce air pollution and reliance on oil has

resulted in increased reception of plug-in and pure electric

vehicles (PEV) [1]-[3].

As PEV draws power through power electronic circuits

from the grid when being charged, utility companies are

concerned with the potential power quality (PQ) impact of the

mass adoption of the vehicles [4]-[5]. For example, PEVs may

inject harmonics into utility systems, creating higher harmonic

distortions. The high level of unbalanced charging load may

cause problems such as neutral current rise. Therefore, there is

an urgent need for techniques that can model and assess the

collective impact of PEVs.

As will be shown later, the above problem cannot be

analyzed using some form of deterministic approaches based

on “average” or “typical” model and data. A Monte Carlo

simulation-based method is proposed. This is because PEV

chargers can be plugged into utility system at any time. A

utility feeder sees a set of randomly connected PEVs with

diverse harmonic characteristics and charging duration. The

locations of the EV being charged are also random. A main

challenge, therefore, is to model the random charging behavior

This work was supported by NSERC, Canada and FAPESP, Brazil.

C. Jiang, R. Torquato, D. Salles and W. Xu were with the Department of

Electrical and Computer Engineering, University of Alberta, Edmonton, AB

T6G 2V4, Canada (e-mail: [email protected]; [email protected]) when this

research was carried out.

of the PEV. This behavior is in turn affected by the driving

habits of the PEV owners. In addition, the PQ impact of PEV

cannot be assessed in isolation, as other home appliances can

also produce PQ disturbances, whose operating pattern has

some correlations with that of the PEVs.

Based on the Monte Carlo simulation method developed in

[6], this paper presents techniques to model the random

operation of PEV chargers and to integrate the models into the

simulation method of [6] for system wide PQ impact

assessment. With this approach, PEVs are treated as one of the

(large) “home appliances”. As a result, the combined and

relative impact of various residential harmonic producers can

be determined.

It is worthwhile to point out that the sales of plug-in hybrid

electric vehicles grow much faster than the pure electric

vehicles due to longer driving range, lower battery costs and

faster recharging times [3], [7]. The models and results

presented in this paper are related to the plug-in hybrid

(PHEV). Many of the concepts and models proposed can be

modified for pure EVs once their mass-market versions mature

and are accepted by consumers. In addition, the study assumes

that PEV owners charge their cars at homes. The proposed

method can be extended to charging-station based PEV

refueling systems.

The remainder of the paper is organized as follows: Section

II explains the overall strategy of Monte Carlo simulation.

Section III presents a set of models to characterize the random

behaviors of PEV. Section IV shows a model of PHEV

consumer adoption trends, which is needed to predict the PQ

impact of PEV in the future. The electric model of PEV and

utility system are provided in Section V. Sample case study

results are presented in Section VI.

II. GENERAL SIMULATION SCHEME

A. Overview of published works

Over the years, various works have investigated the PQ

impact of EVs, which greatly contributed to our understanding

on the subject. Reference [8], one of the pioneer works on the

subject, identified the harmonic currents produced by electric

vehicle chargers. However, it considers all chargers clustered

on a single bus. The methods and results cannot be applied to

the more realistic, disperse charging scenario. Harmonic

impacts produced by five different types of chargers are

assessed on [9], without considering the vehicle plug-in

behavior and penetration characteristics. The authors further

expanded their work on [10] to include EV random plug-in

Method to Assess the Power Quality Impact of

Plug-in Electric Vehicles (Final Version)

Chen Jiang, Student Member, IEEE, Ricardo Torquato, Student Member, IEEE, Diogo Salles,

Member, IEEE, Wilsun Xu, Fellow, IEEE

T

2

time and charging duration. But the proposed method only

addresses EV connection impact on HV/MV substation,

limiting penetration level to, at most, one EV per MV/LV

transformer.

References [11]-[12] derive EV random impact on system

voltage and current harmonics using an analytical approach.

This approach presents some limitations, such as: (a) to

provide reliable results (converged average and standard

deviation), at least 6 vehicles must be connected to network;

(b) the impact of chargers with different characteristics (e.g.,

power demand and harmonic spectrum) simultaneously

connected to the network is not considered. On [13], a more

detailed EV model is proposed, but random characteristics of

plug-in instant, charging time and EV location are not taken

into account. Vehicle penetration level, for instance, is

regarded as equally distributed within all LV consumers. In

addition, many of the above studies have not modeled the

secondary, “two-phase” networks. So the impact of EV on the

secondary network cannot be evaluated properly.

B. Proposed Monte-Carlo simulation methodology

Build on the above works and the Monte-Carlo simulation

concept of [6], this paper analyzes the various factors that

must be considered for the PQ impact assessment of PEV and

proposes methods to include such factors.

In a realistic scenario, a PEV can connect to a power

system for charging up at any time. As a result, a utility feeder

will supply a set of randomly connected EVs at any given

time. Furthermore, these vehicles will go off the grid at

different time as each of them has different charging duration.

The locations of the EV being charged in a feeder are also

random. In order to determine the power quality impact of

these vehicles, a methodology that can model these random

factors must be developed. It is impossible or unrealistic to

predict the PQ impact using some form of deterministic

approaches based on “average” or “typical” data.

The methodology proposed by this paper is the Monte

Carlo simulation technique. The basic idea is to create

numerous plausible scenarios of PEV activities in a feeder for

a given instant of time, say at 12:04pm. One of the scenarios,

for example, represents the case where PEV-A is connected to

location X and PEV-B to location Y etc. At the instant of

12:04pm, PEV-A battery is charged to m% and PEV-B is at

n%. Once such a scenario is created, all load activities become

known and are deterministic. Harmonic power flow studies

can then be conducted to determine the harmonic levels

associated with this particular scenario. A Monte Carlo

simulation involves the creation of thousands of plausible

scenarios for 12:04pm. Therefore, thousands of harmonic

studies are performed. The average of the harmonic results

among all scenarios should represent the most likely results

expected at 12:04pm.

The remaining problem becomes how to create the multiple

plausible scenarios for the instant of 12:04pm. This is done by

considering the probability of occurrence of the various

factors. For example, if 10% houses are found to have PEV

based on consumer trend, 10% of the randomly selected

houses in the feeder model will be flagged as the candidate

locations for PEV connection. For each house, the time of

PEV connection and its charging duration depend on the

driving habit. There is also a probability curve for the habit.

For example, if the probability curve shows that there is a 10%

chance that the car will be charged at 12:04pm, 10% the

scenarios created will have a car being charged at that

particular house location. By considering various probabilistic

factors, multiple scenarios can be created. Naturally, these

scenarios will satisfy or have included the probabilistic

characteristics of the various factors.

The above multi-scenario simulations can be conducted for

every minute (or other resolution) over a 24 hour period. A

total of 1440 snapshots are produced for one day. In order to

provide more succinct indices to characterize the PEV impact,

the results over a 24 hour period can be further condensed by

selecting their 95% probability values, i.e. the values that will

not be exceeded for the 95% of the time over a 24 hour period.

With this approach, the impact of PEV can be understood

from one value for each PQ index of concern.

C. Key factors for inclusion in simulation studies

To achieve the above simulation goal, three groups of

factors must be modeled. The 1st group is the charging

behavior of PEV. This behavior is affected by the following

factors:

Charging strategy – deterministic and random;

Charging start time – random;

PEV charging duration – random.

The 2nd

group is the locations and density of the PEV in a

feeder. It is affected by the following factors:

PEV penetration level – deterministic;

PEV location in a residential system – random;

Type of chargers – semi-deterministic

The 3rd

group is the electric characteristics of the PEV and

power systems. Important factors are:

The harmonic model for PEV chargers;

The multi-phase model of the supply network;

The impact of other home appliances.

The following three sections will present specific solutions

to deal with the above factors.

III. CHARGING BEHAVIOR OF PEVS

This section will present methods to model the three factors

affecting the charging behavior of PEVs.

A. PEV charging start time

The charging start time determines when vehicles will start

being charged during the day. Since all PEVs will not begin

charging simultaneously, it is more suitable to treat this

variable as a random variable based upon daily charge

strategy. Charging strategy is the way PEV owners plan (or

forced to plan) to charge their batteries every day. There are

three type of charging strategies that need to be modeled.

3

1) Uncontrolled charging

Uncontrolled charging means that PEVs could start

charging any time during the day. In this strategy, most people

will immediately charge their vehicles to prepare them for the

next trip just when they arrive home from work. Obviously,

uncontrolled charging tendency is strongly associated with

people’s travelling behavior. In this case, most PEVs are

plugged in and start charging around 6:00 p.m. when people

usually come back home. However, a uniformly distributed

probability density function (pdf) with a narrow range around

6:00 p.m. is more close to reality, as shown in Fig. 1. It may be

mathematically described by (1) [14], where x represents the

random charging start time in hour; while a and b are the lower

and upper limits of hours of charging, respectively.

19 b 18,a b,xa ,1

)(

ab

xf (1)

2) Controlled charging

Once PEV adoption rate reaches a certain level, because of

the coincidence with early evening system peak, uncontrolled

charging plan could significantly increase this peak. Therefore,

it is likely that utilities would use either Time-Of-Use (TOU)

pricing or direct control methods such as in-home delay

devices to shift PEV charging load to off-peak time. TOU

pricing is used in most regions, which leads people to

postpone charging after 9:00 p.m. in order to minimize their

electricity bills. As a result, model described in (2) is used to

simulate controlled charging.

24 b ,12a b,xa ,1

)(

ab

xf (2)

3) Smart charging

The last strategy is smart charging. This strategy not only

maintains the evening peak constant but also can be

advantageous to utilities. For example, reference [15] uses

Advanced Metering Infrastructure (AMI) together with PEV

control unit and remote switches to imply stagger charging

which limits the charging based upon pre-determined power

levels communicated through the grid. This method will help

smooth the PEV charging load seen by service transformer,

especially for high PEV penetration. As a result, no additional

peaks due to PEV will be created on transformer load, either in

the early evening or midnight. Considering the complexity of

employing various smart charging optimization algorithms, a

normal distribution of start time to represent all smart charging

strategies is adopted in this paper, as illustrated in Fig. 1 [16].

In this case, the charging start time is described by (3), where

x is the random plug-in instant; μ and σ are the average and

standard deviation of x, respectively.

3 ,1 ,e2

1)(

2

2

1

x

xf (3)

0 2 4 6 8 10 12 14 16 18 20 22 240

20

40

60

80

100

Time (h)

Pro

babili

ty (

%)

Uncontrolled Controlled Smart

Fig. 1. Charging start time pattern for each charging strategy.

B. PEV charging duration

When a driver arrives at home, the PEV battery has

remaining energy, which is quantified here by the factor called

state of charge (SOC). The state of charge (SOC) is the

equivalent of a fuel gauge for the battery pack in an electric

vehicle and the units of SOC are percentage points (0% =

empty; 100% = full). SOC is random and its associated pdf is

based upon daily charge and travel pattern, as explained

below.

To determine the SOC of each PEV, it is essential to know

how deep it was discharged during the day, which is directly

related to the distance it traveled. From general vehicles

driving pattern, a probability distribution of daily distance

driven has been derived in [8]. The distribution is found to be

log-normal type, but with zero probability at all negative

distances. The mean of the distribution is 34.2 miles and the

standard deviation is 21.1 miles, at the year of 1983. The

probability function is as follows:

0,2

1)(

2

2

2

ln

mem

mdm

(4)

where

2][

][1ln

2

1][ln

ME

MVarME (5)

2

2

][

][1ln

ME

MVar

(6)

where the random variable m is the daily distance travelled in

miles, E[M] and Var[M] stand for the mean value and

variance of m. According to updated information provided by

[17], daily average vehicle miles travelled is 33 miles. To

update our estimation, the standard deviation is scaled to 20.4

miles with constant ratio. Fig. 2 shows the probability density

function of daily distance driven.

0 20 40 60 80 1000

0.5

1

1.5

2

2.5

3

Daily Travel Distance (miles)

Pro

babili

ty D

ensity (

%)

Fig. 2. Probability density function of PHEV daily distance driven.

4

Using the traveled distance (m), the battery state of charge

(SOC) at the beginning of charging can be determined by:

Rm

RmR

mR

SOC

,0

0%,100 (7)

where R is the all-electric range of the PEV in miles.

The next step is to find the charging duration (D) from the

SOC value determined above. This can be calculated from the

PHEV battery nameplate parameters, which includes battery

capacity, charger level and initial SOC, as follows:

P

DODSOCCD

1 (8)

where C is battery capacity (kwh); DOD stands for the depth

of discharge and it determines the fraction of power that can

be withdrawn from the battery (%); P is the power rating of

charger (kW), which is determined by charger level; and is

the efficiency of charger (%).

IV. CONSUMER ADOPTION OF PHEVS

In order to model charger loads in a network and anticipate

their impact in the future, the penetration level of chargers

should be estimated. In other words, the number of PHEVs per

household in the network should be predicted. Several

organizations have provided estimations for PHEV penetration

level for the near future. As a case in point, a report by Oak

Ridge National Laboratory in 2006 [18] estimates that PHEV-

20 vehicles (which have 20-mile all electric range) will have a

base case market potential of over 25% of sales for the entire

car and light-truck market in 2018. EPRI reports [19]-[20]

have similar analysis results for 2010 to 2030 PHEV market

share.

The number of cars sold every year in the US is about 20

million. So, considering the PHEV market share level for year

2010 to 2030 as well as the assumption of average 12 years

vehicle lifespan, the total number of PHEV for 2010 to 2030 is

shown in Fig. 3(a) [18], [21]. According to the linear

regression analysis over the data from 1990 to 2008 available

in [21], the total projected number of passenger vehicles for

2010 to 2030 is calculated and plotted in Fig. 3(b).

0

20

40

60

80

100

120

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025

2026

2027

2028

2029

2030

Year

PH

EV

s (

mill

ion

s)

(a)

0

100

200

300

400

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025

2026

2027

2028

2029

2030

Year

Pa

sse

ng

er

Ve

hic

les (

mill

ion

s)

(b)

Fig. 3. PHEV market trend data for 2010 to 2030. (a) Total estimated

number of PHEVs. (b) Total estimated number of passenger vehicles.

Assuming that the average number of vehicles per

household is 2, the penetration level across households can be

calculated as follows:

PVHiN

iNiPHEVH

PV

PHEV )(

)()( (9)

where PHEVH(i) is the number of PHEVs per household at

year i, NPHEV is the total number of PHEVs shown in Fig. 3(a),

NPV is the total number of passenger vehicles shown in Fig.

3(b) and PVH is the average number of vehicles per

household, which is assumed equal to 2. From (9), the final

expected number of PHEVs per household can be estimated

and it is shown in Fig. 4. From this figure, the average number

of PHEVs per household in 2020 is 0.2, which shows good

consistency with the estimation in [15]. Given the number of

houses connected to each transformer, the number of PHEVs

fed by each transformer can be simply calculated using Fig. 4.

0

0.2

0.4

0.6

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025

2026

2027

2028

2029

2030

Year

PH

EV

per

Household

Fig. 4. Total estimated number of PHEVs per household for 2010 to 2030.

Fig. 4 is essential to determine PHEVs locations for a 24-

hour Monte Carlo simulation. The following procedure is

used: for each residential customer (x), a randomly generated

number (rx), uniformly distributed between 0 and 1 is

compared to the corresponding PHEVs penetration rate. If rx is

less than the penetration rate, then a PHEV is assigned to that

house. Once a house is selected as the PEV owner, it will stay

as the PEV location throughout the simulation.

According to the Society of Automotive Engineers (SAE)

all EVs produced by automakers in North America must

follow SAE J1772 standard [22]. Based on voltage and power

levels, three levels of charging are identified in SAE J1772,

which are shown in Table I.

TABLE I

PHEV CHARGER TYPES DEFINED IN SAE J1772 [22].

Charger Type Input Voltage Maximum Power (kW)

Level 1 120 VAC 1.440

Level 2 208-240 VAC 11.50

Level 3 208-240 VAC 96.00

Level 3 (DC) 208-600 VDC 240.0

Level 1 and 2 chargers are for consumer EVs. Their

distribution among the EV owners is treated as a sensitivity

study factor. Namely, PQ impact of different types of chargers

and their combinations are determined. Level 3 chargers are

mainly intended for commercial and public transportations

and, therefore, are not considered in this paper. However, the

proposed model can be extended to include these chargers.

V. ELECTRICAL MODELS

There are three factors to consider for the electrical models,

which are addressed in this section.

A. Harmonic model for PEV charger

The common method to model the harmonic characteristics

of a power electronic device such as the PEV charger is the

harmonic current source model, which is adopted here. This

model is deterministic, i.e. once a PEV charger is turned on, its

5

harmonic current output follows a typical harmonic current

spectrum (with reference to its supply voltage) regardless its

location in the system. The harmonic model is as follows [23],

where spectrum refers to charger typical harmonic current

spectrum, and 11I is the fundamental current injected by the

charger on the network. This current is determined from the

60Hz power flow results.

spectrum

spectrumh

hI

III

1

1

spectrumspectrumhh

h

11

(10)

The current phase angle correction shown above is

important since charger harmonic currents may cancel out

currents from other nonlinear appliances. Furthermore, it is

important to note that the current phase angle is not treated as

a random variable in the above equation. This is because the

PEV chargers always draw harmonic currents at approximately

the same phase angle with respect to the supply voltage. This

angle is therefore deterministic.

This research has obtained harmonic current spectrum data

from several PEVs. A sample results are shown in Table II and

Fig. 5. At the fundamental frequency, PEVs are treated as

constant power load.

TABLE II

HARMONIC SPECTRA OF PEV.

Charger Level 1 Level 2

Voltage 120V 240V

Harmonic Order Mag. (%) Mag. (%)

H1 100.0 100.0

H3 9.125 8.864

H5 3.230 2.452

H7 0.947 0.891

H9 1.520 0.911

H11 1.329 0.870

0 0.02 0.04 0.06 0.08 0.1-1.5

-1

-0.5

0

0.5

1

1.5

Charg

er

curr

ent

(pu)

time (s)

Level 1 charger

Level 2 charger

Fig. 5. PEV charger currents normalized by Level 1 60Hz current.

B. Network model

In North America, the most common primary power

distribution system is the multigrounded neutral (MGN)

system. This system contains three phase conductors and one

neutral conductor grounded at regular intervals (Fig. 6). The

secondary system connects to the primary system through a

single-phase 3-winding service transformer. The secondary

side contains 2 hot and 1 neutral conductors. Level 1 charger

connects between a hot and neutral conductor and Level 2

charger connects between the two hot conductors.

Primary System

Rgs RgnRgn Rgn Rgn

Section 2Section 1 Section N. . .Vsub Zsub

One or more

service transformers

Fig. 6. Schematic model to study primary distribution systems.

It is clear that a single-phase or a three-phase equivalent

network model is not adequate for studying harmonics in such

a system. An equivalent model will make it impossible to

investigate harmonic impacts such as neutral voltage rise.

Consequently, a multiphase model of the network which

includes every conductor of the system shall be used.

Fortunately, reference [24] has developed a method to model

and calculate multiphase systems. It is adopted here.

With the multiphase network model, harmonic cancellation

effect caused by voltage phase differences at different points

of a feeder because of the branch impedances are included.

This is realized through Eqn. (10) where the impedance will

change 60Hz current angle 1.

If the study target is the primary system, an equivalent

circuit for the secondary system must be calculated and

attached to the multiphase primary network, as shown on Fig.

6. The portion of the network inside the dashed rectangle

represents one or more single-phase service transformers

connected between each phase of the primary conductor and

neutral conductor, and each service transformer could be

represented by the secondary system configuration shown in

Fig. 7. Table III presents the system parameters of the primary

and secondary test systems.

Vth

TR

AZ

BZ

NZ

1aZ

1bZ

1abZ

1AZ

1NZ

1BZ

maZ

mbZ

mabZ

mAZ

mNZ

mBZ

Phase A

Neutral

Phase B

House

#1

House

#m

PCC

Service panel

CR

CR

thZ 1BNZ

1ANZ

1ABZ

mABZmANZ

mBNZ

ABZANZ

BNZ

CFL TVPC +

LCD monitor

18m 12m 5m

10m

Branch

#1

Branch

#3

AV BVNV

Service

panel

Panel feeders

CR

EV charger

16m

Branch

#2Stove

Detailed house

PEV connection

ZMGN

Primary System

Equivalent

Fig. 7. Schematic model to study secondary distribution systems.

6

TABLE III

SYSTEM PARAMETERS.

Base Case System Parameters Values

Primary

System

Supply system voltage 14400 V (@ 60 Hz)

Substation pos. sequence impedance 0.688 + j2.470 ohms

Substation zero sequence impedance 0.065 + j2.814 ohms

Substation grounding (Rgs) 0.15 ohms

MGN grounding resistance (Rgn) 15 ohms

Grounding span of MGN neutral 75 m

Feeder length 15 km

Feeder conductor type 4 - 336.4 ACSR

Service

Transformer

Voltage (VH/VL) rating 14400/120 V

KVA rating 37.5 kVA

Impedance 2 %

Resistance 1.293 %

Grounding resistance (RT) 12 ohms

Secondary

System

Voltage source (Vth) 14400 V (@ 60 Hz)

Primary system impedance (Zth) 0.48 + j2.58 ohm

Grounding impedance (ZMGN) 0.885 + j0.580 ohm

Customer grounding resistance (RC) 1 ohm

Neutral impedance (ZN) 0.55 + j0.365 ohm/km

Phase impedance (ZA and ZB) 0.21 + j0.094 ohm/km

Num. of houses for each transformer 10

Distance between PCC and house 40-70 m

When studying the PEV impact on the secondary system

shown in Fig. 6, the primary system is modeled as an

equivalent circuit (Vth, Zth & ZMGN), whose parameters are

established by performing frequency scans at the location of

the service transformer. The equivalent voltage source Vth may

contain 60Hz voltage only. Sensitivity study has shown that

including harmonics in this voltage source will not change the

results significantly.

C. Modeling of Other Household Loads

The third factor that must be considered is the impact of

other household loads such as personal computers and

washers. Harmonics produced from such appliances may add

or cancel the harmonics produced by PEV. The household

loads also operate randomly. They are modeled according to

the method described in [6].

D. Summary

The detailed procedure to simulate PEV activities in the

secondary system is presented on Fig. 8. In order to simplify

the flowchart, only 1 day simulation is presented. The overall

Monte Carlo simulation method consists on repeating this

algorithm multiple times until converged results are achieved.

Primary system assessment can be performed using the

same flowchart with an additional step. Before performing

power flow solution, once house circuit parameters have been

determined, the equivalent circuit seen from primary network

must be calculated for each MV/LV transformer [6].

VI. CASE STUDIES

A number of case studies have been conducted to quantify

the impact of PEV. Due to space limitation, this section

illustrates two important findings. One is the harmonic impact

of PEV and the other is the neutral voltage impact. Both are

related to the secondary network.

Assign daily usage pattern for

home appliances according to [6]Determine PEV locations

on network according to

penetration levels

k=0

nPEV = # of PEVs on network

k < nPEV?

Determine charger

type for PEVk

Determine charging

start time and charging

duration for PEVk

k = k + 1

Yes

Not = 0

t < 24h?

Determine house equivalent

circuits on secondary network

for fundamental frequency:

PA, QA; PB, QB; PAB, QAB

Solve fundamental

power flow

Determine house equivalent

circuits on secondary network

for harmonic frequency:

ZA, iA; ZB, iB; ZAB, iAB

Solve harmonic

power flow

Determine charging

strategy for PEVk

Store results

t = t + 1min

Yes

Output 24-hour

results

PEV characterization

No

PQ impact assessment algorithm

Fig. 8. Flowchart for PHEV PQ impact analysis on secondary networks.

A. Case Study 1: Harmonic Impact

For this study, the base case scenario is defined as mixed

charger types under uncontrolled charging strategy with 30%

PHEV penetration level. Mixed chargers include Level 1 and

Level 2 chargers at a ratio of 1:1. This case represents the

worst charging strategy and most common charger

composition. The impact of PEV is also compared with those

of other nonlinear home appliances. This “30% PHEV” case is

compared to three other scenarios:

a. “Base Load”: This scenario considers only home

appliances and the respective penetration rates refer to

year 2011 market data. There are no PHEVs in the

system;

b. “CFL 2015”: All home appliances loads remain the same

as the “Base Load” scenario, except that the penetration

of CFL appliances refers to year 2015 market data [6].

c. “PC 2015”: All home appliances loads remain the same as

the “Base Load” scenario, except that the penetration of

personal computer appliances refers to year 2015 market

data [6].

Fig. 9 shows a sample result, which is the 95% probability

values over the harmonic voltage profile at the metering point

averaged among all homes connected to the sample secondary

system. This figure indicates that even with 30% penetration

level, the harmonic distortion caused by PHEV is not

significant since both phase and neutral harmonic voltage

levels are comparable to those of “Base Load” case study.

Moreover, the “CFL 2015” and “PC 2015” cases show that

CFL and PC Loads should be of more concern to utilities in

terms of harmonic distortion. The significant increase on

fundamental and RMS neutral voltage is mainly caused by

load imbalance between the two phases due to Level 1

chargers.

7

3rd 5th 7th 9th 11th 13th 15th THDv0

1

2

3

4

5IH

DV (

%)

Harmonic Order #

Base Load

30% PHEV

CFL 2015

PC 2015

(a)

1st 3rd 5th 7th 9th 11th 13th 15thRMS0

0.2

0.4

0.6

0.8

Vo

lta

ge

(V

)

Harmonic Order #

Base Load

30% PHEV

CFL 2015

PC 2015

(b)

Fig. 9. Comparison between harmonic voltages produced by PHEVs, CFLs

and PCs. (a) 95% index for phase voltage. (b) 95% index for neutral voltage.

B. Further Analysis on the PEV Harmonic Impact

The finding that the PEV has little harmonic impact on

power system is not surprising if we compare the harmonic

current characteristics of PEV with those of other home

appliances. For this comparison, the equivalent CFL index

proposed in [25] is used. This index quantifies the harmonic

current injection level of each appliance expressed as the

number of CFLs it is equivalent to.

Table IV shows the values of the total equivalent CFL

index for one PEV charger and other key home appliances.

One can observe that, in terms of harmonic effects to the

system, one PEV is equivalent to approximately 10 CFLs or

1.5 desktop PCs. Since most homes will have CFLs and PCs

while the PHEV penetration level is only 0.2 per home by year

2020, one can conclude that the PEV’s harmonic impact will

be significantly less than that of the CFLs or PCs. It is

worthwhile to point out the reason that a PEV generates fewer

harmonics: the PEV manufacturers have voluntarily adopted

the IEC 1000-3-2 and SAE J2894/1 harmonic limits [26]-[27].

C. Case Study 2: Neutral Voltage Rise

The charger types are changed in this study to identify their

impact on the secondary system. All system parameters remain

the same as base case described on Case Study 1 (30% PHEV

penetration), except charger types. Fig. 10(a) shows the 95%

index associated to the average phase harmonic voltages, for

different chargers types. Fig. 10(b) shows the 95% index of the

average neutral to ground voltage.

3rd 5th 7th 9th 11th 13th 15thTHDv0

0.5

1

1.5

2

2.5

Harmonic Order #

IHD

V (

%)

Level 1

Level 2

Mixed

(a)

1st 3rd 5th 7th 9th 11th13th15thRMS0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Harmonic Order #

Volta

ge (

V)

Level 1

Level 2

Mixed

(b)

Fig. 10. Power quality impact of different charger types. (a) 95% index for

phase voltage. (b) 95% index for neutral voltage.

The results show that the harmonic impact of different

chargers is comparable. However, Level 1 charger causes

noticeable rise of the neutral voltage at the fundamental

frequency. This is due to the fact that Level 1 charger draws

current from just one phase. Level 2 chargers cause no neutral

voltage rise, since they are phase-to-phase connected. This

finding is important since the rising neutral voltage may

increase the incidents of stray voltages at homes [28].

Unfortunately, most homes don’t have 240V supply at garage

so adopting level 2 charger involves cost of re-wiring home

circuits.

TABLE IV: COMPARING THE HARMONIC EFFECT OF PEVS.

Appliance type Operating

Power [W] Power Ratio Equivalent CFL

CFL 15 1 1.00

Dryer 4500 300 2.13

LCD monitor 40 3 2.35

Furnace 500 33 3.49

Fridge 1200 80 4.34

Desktop PC 100 7 6.53

Microwave oven 1200 80 24.30

PEV Charger 1355 90 9.77

This research has studied other PQ impact of PEV such as

metering error, losses and transformer overloading etc. Beside

the neutral voltage rise issue, no other significant impact was

found. It is also understandable that the harmonic impact of

PEV on the primary system is small. In theory, the PEV may

cause 60Hz overloading on the service transformers. Our study

shows that this is not a major issue if the PEV penetration

level up to 30% (i.e. year 2022 based on this paper market

data) is considered.

VII. CONCLUSIONS

This paper presented a power quality impact assessment

methodology for distributed plug-in EVs. Methods to model

various deterministic and random factors that influence the

charging activities of PEV are proposed. The factors include

charging strategies, charging start time, charging duration,

market penetration, vehicle connection point and charger

electrical data. A Monte Carlo simulation method has been

employed to identify PEV impacts at both fundamental and

harmonic frequencies. The usefulness of the method has been

demonstrated through case studies. The results show that PEV

chargers have negligible harmonic impact on power systems in

the near future (up to 2022). However, Level 1 charger may

cause the rise of neutral to earth voltage, which could lead to

stray voltage incidents. The proposed method can be used to

study the impact of other types of EVs such as pure-electric

type, different PEV market penetration scenarios and vehicle

to grid operation modes.

VIII. REFERENCES

[1] R. Lache, “Vehicle electrification, more rapid growth; steeper price

declines for batteries,” DeutscheBank Global Markets Research, 2010.

[2] Texas Transportation Institute, Strategic Solutions Center (2011).

Electric Vehicle Market Penetration: An Examination of Industry

Literature. Austin, TX. [Online]. Available: http://tti.tamu.edu/group

/stsc/files/2010/11/BP-2010-01-Electric-Vehicles-_Updated-Final__edit

ed-_TB_.pdf.

[3] A. K. Srivastava and B. Annabathina, “The challenges and policy

options for integrating plug-in hybrid electric vehicle into the electric

grid,” The Electricity Journal, vol. 23, no. 3, pp. 83–91, Apr. 2010.

[4] Power Quality Requirements for Plug In Vehicle Chargers - Part 1:

Requirements, SAE Standard J2894/1, 2010.

[5] C. C. Chan and Y. S. Wong, “Electric vehicles charge forward,” IEEE

Power and Energy Magazine, vol. 2, no. 6, pp. 24- 33, Nov. 2004.

8

[6] D. Salles, C. Jiang, W. Xu, W. Freitas, and H. E. Mazin, “Assessing the

collective harmonic impact of modern residential loads—Part I:

Methodology,” IEEE Trans. Power Del., vol. 27, no. 4, pp. 1937-1946,

Oct. 2012.

[7] K. Clement-Nyns, E. Haesen, and J. Dreisen, “The impact of charging

plug-in hybrid electric vehicles on a residential distribution grid,” IEEE

Trans. Power Syst., vol. 25, no. 1, pp. 371-380, Dec. 2009.

[8] J. A. Orr, A. E. Emanuel, and K. W. Oberg, “Current Harmonics

Generated by a Cluster of Electric Vehicles Battery Chargers,” IEEE

Trans. PAS, vol. PAS-101, no. 3, pp. 691-700, Mar. 1982.

[9] J. A. Orr, A. E. Emanuel, and D. G. Pileggi, “Current Harmonics,

Voltage Distortion, and Powers Associated with Battery Chargers Part I:

Comparisons Among Different Types of Chargers,” IEEE Trans. PAS,

vol. PAS-101, no. 8, pp. 2703-2710, Aug. 1982.

[10] J. A. Orr, A. E. Emanuel, and D. G. Pileggi, “Current Harmonics,

Voltage Distortion, and Power Associated with Electric Vehicle Battery

Chargers Distributed on the Residential Power System,” IEEE Trans.

Ind. Applications, vol. IA-20, no. 4, pp. 727-734, Jul./Aug. 1984.

[11] P. T. Staats, W. M. Grady, A. Arapostathis, and R. S. Thallam, “A

Statistical Method for Predicting the Net Harmonic Currents Generated

by a Concentration of Electric Vehicle Battery Chargers,” IEEE Trans.

Power Del., vol. 12, no. 3, pp. 1258-1266, Jul. 1997.

[12] P. T. Staats, W. M. Grady, A. Arapostathis, and R. S. Thallam, “A

Statistical Analysis of the Effect of Electric Vehicle Battery Charging

on Distribution System Harmonic Voltages,” IEEE Trans. Power Del.,

vol. 13, no. 2, pp. 640-646, Apr. 1998.

[13] J. C. Gómez, and M. M. Morcos, “Impact of EV Battery Chargers on

the Power Quality of Distribution Systems,” IEEE Trans. Power Del.,

vol. 18, no. 3, pp. 975-981, Jul. 2003.

[14] “Applied calculus on-line chapter: calculus applied to probability and

statistics. Section 2. Probability Density Functions and the Uniform

Distribution” [Online]. Available: http://www.zweigmedia.com/Real

World/cprob/cprob2.html.

[15] S. Shao, M. Pipattanasomporn, and S. Rahman, “Challenges of PHEV

penetration to the residential distribution network,” in Proc. 2009 IEEE

Power & Energy Society General Meeting, pp.1-8.

[16] K. Qian, C. Zhou, M. Allan, and Y. Yuan, “Modeling of load demand

due to EV battery charging in distribution systems,” IEEE Trans. Power

Syst., vol. 26, no. 2, pp. 802-810, May 2011.

[17] M. Kintner-Meyer, K. Schneider, and R. Pratt (2007). Impacts

Assessment of Plug-in Hybrid Electric Utilities and Regional U.S.

Power Grids Part 1: Technical Analysis. PNNL Report. Richland, WA.

[Online]. Available: http://www.ferc.gov/about/com-mem/wellinghoff/5-

24-07-technical-analy-wellinghoff.pdf.

[18] S. W. Hadley, A. Tsvetkova, “Potential Impacts of Plug-in Hybrid

Electric Vehicles on Regional Power Generation,” ORNL/TM-

2007/150, Jan 2008.

[19] EPRI, “Environmental Assessment of Plug-in Hybrid Electric Vehicles:

Nationwide Greenhouse Gas Emissions,” Palo Alto, CA, Tech. Rep.

1015325, Jul. 2007.

[20] EPRI, “Environmental Assessment of Plug-in Hybrid Electric Vehicles:

United States Air Quality Analysis Based on AEO-2006 Assumptions

for 2030,” Palo Alto, CA, Tech. Rep. 1015326, Jul. 2007.

[21] U.S. Department of Transportation Federal Highway Administration

(2003). Journey to Work Trends in the United States and its Major

Metropolitan. Washington D.C. [Online]. Available:

http://www.fhwa.dot.gov/ctpp/jtw/contents.htm.

[22] SAE Electric Vehicle Conductive Charge Coupler, SAE standard SAE

J1772, 2001.

[23] “Task force on harmonics modeling and simulation, modeling and

simulation of the propagation of harmonics in electric power networks–

part I: concepts, models, and simulation techniques,” IEEE Trans.

Power Del., vol. 11, no. 1, pp. 452–465, Jan. 1996.

[24] H. E. Mazin, E. E. Nino, W. Xu, and J. Yong, “A study on the harmonic

contributions of residential loads,” IEEE Trans. Power Del., vol. 26, no.

3, pp. 1592-1599, Jul. 2011.

[25] A. B. Nassif, J. Yong, W. Xu, and C. Y. Chung, “Indices for

comparative assessment of the harmonic effect of different home

appliances,” European Transactions on Electrical Power, Feb. 2012.

[26] Electromagnetic Compatibility – Part 3: Limits. Section 2: Limits for

harmonic current emissions (equipment input current ≤ 16 A per

phase), IEC Standard 1000-3-2, 1995.

[27] Power Quality Requirements for Plug-in Electric Vehicle Chargers,

SAE Standard SAE J2894/1, 2011.

[28] IEEE PES Stray-Voltage Working Group, "Voltages at Publicly and

Privately Accessible Locations", Trial Use Guide (Draft), 2009.

IX. BIOGRAPHIES

Chen Jiang (S’09) received the B.Eng degree in Electric Engineering and

Automation from Huazhong University of Science and Technology (HUST),

Wuhan, China, in 2008, and the M.Sc degree in Power Engineering and

Power Electronics from University of Alberta, Edmonton, Canada, in 2012.

Since 2012, he has worked as a Power System Planning Engineer at North

China Power Engineering Co., Ltd of China Power Engineering Consulting

Group. His main research interests are power system planning issues and

power quality.

Ricardo Torquato (S’11) obtained the B.Sc. degree in electrical

engineering from the University of Campinas, Campinas, Brazil in 2011,

where he is pursuing a M.Sc. degree. His research interests are power quality,

analysis of distribution systems and distributed generation.

Diogo Salles (S’04-M’12) received the B.Sc., M.Sc. and Ph.D. degrees,

all in electrical engineering, from the University of Campinas, Campinas,

Brazil, in 2006, 2008 and 2012, respectively. Currently, he is a Post-Doctoral

Researcher at the University of Campinas. From 2010 to 2012, he was a

Visiting Doctoral Scholar at the University of Alberta, Edmonton, AB,

Canada. His research interests focus on power quality, harmonics and power

disturbance data analysis.

Wilsun Xu (M’90-SM’95-F’05) obtained the Ph.D. from the University

of British Columbia, Vancouver, in 1989. Currently, he is a Professor and a

NSERC/iCORE Industrial Research Chair at the University of Alberta. His

current research interests are power quality and information extraction from

power disturbances.