IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO.

3, SEPTEMBER 2013 1255

The Optimal System Design of the Online Electric

Vehicle Utilizing Wireless Power
Transmission Technology
Young Dae Ko and Young Jae Jang, Member, IEEE

Abstract—The Online Electric Vehicle (OLEV) is an innovative

electric transportation system developed by the Korea Advanced
Institute of Science and Technology (KAIST), Daejon, Korea,
which remotely picks up electricity from power transmitters
buried underground. Unlike a conventional electric vehicle that
requires significant recharging downtime, the battery in the OLEV
can be charged while the vehicle is in motion. Selected as one of
“the 50 Best Innovations of 2010” by TIME Magazine, the OLEV
is considered as a potential solution for the next-generation electric
public transportation system in South Korea. The prototype of the
OLEV has been developed, and the commercialization process is
now in progress. One of the main tasks to achieve the successful
commercialization of the system is to determine economically how
to allocate the power transmitters on the given routes and how to
evaluate the right battery capacity for the vehicle. The allocation
of the power transmitters and the size of the battery capacity
directly affect the initial infrastructure cost. In this paper, we
first introduce the system design issues of the mass transportation
system operating with OLEV. We then present a mathematical
model and an optimization method to allocate economically the
power transmitters and to determine the battery capacity of the
OLEV-based mass transportation system. The particle swarm
optimization (PSO) algorithm is used as the solution method for
the optimization problem. Numerical problems with sensitivity Fig. 1. System configuration of the OLEV System (above) and a prototype of
analysis are presented to show the validity of the mathematical the OLEV system operating at the KAIST campus (below).
model and solution procedure.
Index Terms—Electric vehicle, optimization, particle swarm vehicles, there are still technical and economic limitations,
optimization (PSO), transportation system design, wireless power mainly caused by the weight and size of the battery, the cost
transmission. of the battery, the long battery charging time, and the limited
availability of charging stations. The Online Electric Vehicle
I. I NTRODUCTION
(OLEV) has been developed to overcome the limitations of
A. Introduction to the KAIST OLEV the current generation of battery-powered electric vehicles. The
OLEV shown in Fig. 1 is a battery-powered transportation sys-
W ITH INCREASING concern over global warming and
energy issues, electric vehicles are being considered
as an alternative to vehicles with an internal combustion en-
tem that remotely picks up electricity from power transmitters
buried underground. The system utilizes the innovative wireless
charging technology developed by Korea Advanced Institute of
gine. Over the last couple of years, automotive makers and
Science and Technology (KAIST) in South Korea [1].
researchers have introduced numerous ideas for new electric
The OLEV mainly consists of vehicles and power trans-
transportation systems to meet the new market demand and
mitters. Note that, although the name OLEV literally refers
social and economic requirements. Although some automotive
to the vehicle, it actually includes the integrated system that
makers have already introduced commercial versions of electric
consists of a vehicle or a fleet of vehicles together with the
charging units, which are the power transmitters installed on
Manuscript received January 25, 2012; revised August 7, 2012; accepted
April 2, 2013. Date of publication May 1, 2013; date of current version the road. The vehicle is similar to the conventional electric
August 28, 2013. This work was supported by the National Research Foun- vehicle, i.e., an electric motor is used as a main engine and
dation of Korea funded by the Korean Ministry of Education, Science, and a battery pack is installed to supply the energy to the motor.
Technology through the Basic Science Research Program under Grant 2011-
0015075. The Associate Editor for this paper was F.-Y. Wang. However, unlike conventional electric vehicles that require a
The authors are with Korea Advanced Institute of Science and Technology, cable connection between the car and the charger or power
Daejeon 305-701, Korea (e-mail: [email protected]). outlet when the battery needs to be recharged, the battery in the
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. OLEV can be wirelessly charged from the transmitters using
Digital Object Identifier 10.1109/TITS.2013.2259159 the noncontact charging mechanism. The power transmitter

1524-9050 © 2013 IEEE

1256 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 3, SEPTEMBER 2013

ments and research progress since Maxwell’s equation are

described in [4].
The recent advancements in wireless power transfer tech-
nologies have enabled various commercial applications, partic-
ularly in small electronic devices such as mobile phones and
small personal devices. Wireless power transfer technology can
be applied to charge the batteries in such equipment without
the need for a wire connection. For example, Qualcomm in-
Fig. 2. Decision variables of the optimization problem. troduced a wireless charging solution for mobile phones with
resonant coupling [5]. Researchers at Kyoto University, Japan,
have recently demonstrated a mobile phone that transmits both
power and information via microwave connections [6]. For
systems that use only very low power, the concept of “energy
harvesting” has been proposed, which utilizes wireless power
transmission technology. Energy harvesting is the process by
which energy is directed, captured, and stored from external
sources [7]–[9]. Radio-frequency identification (RFID) is one
of the well-known applications that uses the concept of energy
harvesting. A review of the recent progress in wireless power
transfer technology for RFID can be found in [4].
The commercial application of wireless power transfer tech-
nology is not limited to small electronic devices. One emerging
application that utilizes such wireless charging technology is
Fig. 3. Power transmitter installed at the KAIST Campus. electric vehicles. In 2006, the team at Massachusetts Institute
generates a magnetic field to provide the vehicle with the of Technology (MIT), Cambridge, MA, USA, introduced a
required energy. Each power transmitter consists of an inverter resonant coupler that transmits a large amount of power (from
and inductive cables buried under the roadway, as shown in watts to kilowatts) over midlength distances at low frequencies
Fig. 2. An inductive cable installed for the prototype system [10]. WiTricity, which is a company developing commercial
developed at the KAIST campus is shown in Fig. 3. The applications with MIT’s solution, is already introducing prod-
length of the inductive cable varies depending on the road ucts such as wireless charging stations for car parks, garages,
condition and power requirements. Multiple transmitters are and beyond [11]. Daimler (which owns Mercedes-Benz and
installed on the route to supply enough power to the OLEV. Smart) is teaming up with Conductix-Wampfler and Röchling
One of the main benefits of wireless charging systems is that Automotive KG to develop wireless charging solutions [12].
no cable connection is required for charging. Moreover, the Hino Motors and Showa Aircraft in Japan applied inductive
wireless technology developed at KAIST enables vehicles to coupling in wireless power transfer to an electric bus called the
be charged while they are in motion, thus significantly reducing inductive power transfer electric vehicle. The bus operated in
the charging downtime. an experimental environment from April 13, 2009 to April 27,
Recognized as one of the 50 Best Innovations of 2010 by 2009 [13].
TIME Magazine [2], the OLEV is now considered a potential There have been some research studies on the application of
solution for the next-generation electric public bus system in wireless power transfer technologies for electric vehicles [14],
Seoul, Korea [3]. A commercialization process is in progress [15]. The KAIST OLEV utilizes a resonant coupling technique
at the time of writing of this paper. One of the main tasks for wireless power supply. Power from the 60-Hz supply is
for achieving successful commercialization of the system is to converted to a frequency of 20 kHz by the inverter, and a current
determine how to allocate economically the power transmitters of 200 A flows through the power transmitter; 60 kW of power
on the road and how to determine the right battery capacity. may be wirelessly transferred from the power transmitters with
With given routes and bus operation policies, transmitting 80% efficiency. Huh et al. [16] introduced the technological
enough energy through the power transmitters to the bus is the development of the inductive power supply system adopted in
critical design variable. The allocation of the power transmitters the OLEV. The power supply, power transmitter, and wireless
and size of the battery capacity also directly affect the initial transfer mechanisms in the OLEV were discussed in the paper.
infrastructure cost. Suh et al. [1] presented the design concept of the OLEV
with axiomatic design theory. They also briefly discussed the
economic benefit of the OLEV compared with conventional
B. Background Research and Applications of the
electric vehicles. The wireless charging mechanism in the
Wireless Power Transfer Technology
OLEV was introduced in detail in [17] and [18]. These studies
The history of wireless power transfer technology started specifically considered the process of efficiency improvement
when Maxwell’s equation was first introduced in 1862. The for the wireless charging mechanism used in the OLEV. The
equation is essentially the first theoretical foundation of wire- history of the OLEV development process at KAIST was de-
less power transfer technology. The technological develop- scribed in [19]. More detailed information about the mechanical
KO AND JANG: OPTIMAL SYSTEM DESIGN OF OLEV UTILIZING WIRELESS POWER TRANSMISSION TECHNOLOGY 1257

and electrical design issues and the wireless power transfer tech-
nology used in the OLEV can be found in [20]–[22] and [23].
The papers aforementioned explained the wireless power
transmission issues and hardware design issues of the OLEV,
focusing on the hardware mechanism of the wireless power
transmission in the OLEV. However, there has been little re-
search on the OLEV from the perspective of a road–vehicle
integrated transportation system. The mathematical model and
formulation presented in this paper are based on the ideas
proposed in [24] and [25].

C. Goal and Organization of the Paper

In this paper, we consider an OLEV-based mass transporta-
tion system in which multiple identical buses service a fixed Fig. 4. OLEV system. Charging and depleting the energy in the battery.
route. The OLEV is an integrated road–vehicle system, and (a) A vehicle is operating on the road where no transmitter unit is installed,
the size of the battery in the vehicles and the number of (b) A vehicle is operating on the road where the transmitter unit is installed.
transmitters installed on the road directly affect the overall
performance and cost of the transportation system. The main
purpose of this paper is to introduce the system issues of the
OLEV-based transportation system, which we illustrate with a
mathematical model. Another goal of this paper is to describe
the method for determining the key design variables of the
transportation system—the battery size and allocation of the
power transmitters—which are shown in Fig. 2. Section II
describes the system modeling issues of the OLEV. The detailed
mathematical model describing the dynamic behavior of the
Fig. 5. OLEV shuttle bus servicing in the Seoul Grand (Amusement) Park.
OLEV system is presented in Section III. The optimization
problem and the solution algorithm using the particle swarm while recharging the battery. As a result, the OLEV does not
optimization (PSO) are explained in Section IV. A case with need to carry a heavy and bulky battery. The disadvantage of
numerical solutions and system sensitivity issues are discussed the OLEV is the initial investment cost. Compared with the
in Section V. The summary and further research directions are existing plug-in electric vehicle system, the OLEV requires a
described in Section VI. significant initial investment to build the power transmitters.
However, this initial cost can be offset by the reduced battery
II. O NLINE E LECTRIC V EHICLE size and increased operating time by eliminating the recharging
downtime of the vehicles. Investigating this tradeoff is critical
A. Wireless Charging System in the commercialization of the OLEV system. More specif-
The wireless power transfer system for the OLEV is com- ically, there is a tradeoff between the capacity of the battery
posed of an inverter, an inductive cable, a pickup device, a and allocation of the power transmitters. If the vehicle has a
capacitor, a battery, and a motor, as shown in Fig. 4. Power battery with a large capacity, fewer transmitters are needed.
from the 60-Hz supply is converted to a frequency of 20 kHz by However, a vehicle with a small battery may require more
the inverter, and a current of 200 A flows through the inductive frequent recharges; thus, more transmitters should be installed.
cable. The magnetic flux generated from the cable is collected As the costs of the battery and power transmitters account for a
by the pickup device to generate dc power for the vehicle motor. significant portion of the total investment cost of the OLEV, the
A more detailed description of the system’s power transfer tradeoff should be carefully analyzed. To position optimally the
mechanism and hardware configuration can be found in [16]– transmitters and determine the battery capacity, the energy con-
[18], [26], and [26]. When a vehicle is operating on a road sumption rate of the vehicle and the charging rate of the power
with a power transmitter installed, as shown in the lower part transmitters should be quantitatively modeled. Furthermore, the
of Fig. 4, the power pickup device installed underneath the model should consider the physical constraints in installing the
vehicle remotely collects electricity from the transmitter and power transmitters. For example, a road on a bridge or on an
distributes the power either to the motor or to the battery or overpass may not have enough space for power transmitters.
both, depending on the power requirement of the motor and The quantitative model evaluating the optimal battery size and
the charging level of the battery. However, when the OLEV is the placement of the power transmitters should consider these
running where no power transmitter unit is installed, the motor physical constraints.
in the vehicle uses the power from the battery. This situation is
shown in the upper part of Fig. 4. B. Practical Motivation of the Optimization Model
The wireless power transmission technology enables the This paper is motivated by the first commercial OLEV
motor to receive power remotely while the vehicle is operating system operating in Seoul Grand Park, as shown in Fig. 5.
on the road; hence, the vehicle does not need to be idle The city government of Seoul has recently adopted the OLEV
1258 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 3, SEPTEMBER 2013

technology for the shuttle buses circulating in the amusement imize the total cost of the system. Here, we first define the
park as a test case before applying the technology to an actual notations with the model descriptions, and then, we explain the
mass transportation system in the city [28]. The OLEV-based modeling assumptions.
shuttle buses are running at a speed between 20–40 km/h
on the circular route, with a total length of 2.85 km. The A. Model Descriptions and Notations
authors found two major issues in the current deployment
of the OLEV-based shuttle buses. First, the capacity of the The proposed model is based on the actual OLEV system
battery in the shuttle bus was determined without considering operating in Seoul Grand Parks, as shown in Fig. 5.
the allocation of the power transmitters. We found that when 1) The vehicle is operating on a fixed route, where there are
the shuttle system was developed, the vehicle design and the multiple stopping points to load and unload passengers.
route design were considered in separate design processes. In The total length of the route is denoted by L.
the vehicle design process, the vehicle hardware specifications 2) The route information, such as the location of the stations,
were determined, including the capacity of the battery. Once the stopping points, and the degree of slope (hill profile) are
vehicle design process was complete, the route was designed, given.
including the allocation of the power transmitters. There is 3) The k numbers of identical OLEVs are operated sepa-
a close performance correlation between the capacity of the rately. The number of vehicles k is given. The power
battery and the allocation of the power transmitters. In addition, transmitter can provide enough power even when there
the costs of the battery and the power transmitters account for a are multiple vehicles operating on the same inductive
significant portion of the total cost of the system. Therefore, we cable.
believe that a more cost-efficient solution can be provided by 4) Each vehicle is equipped with one battery unit, and
analyzing together the battery capacity and the allocation of the
the battery capacity is identical for all the vehicles. The
power transmitter. In this paper, we introduce a mathematical
cost of the battery is proportional to the capacity. The
model of the OLEV-based transportation system circulating on
battery cost per unit capacity (in kilowatthours) is denoted
a single fixed route. There are multiple vehicles in the system,
by cbat .
and they are traveling in one direction. The loading/unloading
stations are also predetermined in the route. We also assume 5) The N numbers of power transmitters are installed
that the vehicle operates with a predefined velocity profile; thus, separately.
the speed of the vehicle at a specific location on the route 6) Let xoi and xfi indicate the starting point and end point
is deterministically defined. We also consider the attitude on of the power cable of the ith power transmitter unit,
the route. With this information, the required energy for the respectively. The cable length of the ith transmitter yi is
vehicle at a specific point can be evaluated. The mathematical simply xfi − xoi . These values should satisfy
modeling constructed in this paper is based on the actual
xoi < xfi , for i = 1, . . . , N
OLEV model deployed in Seoul Grand Park. Although the
modeling approach and optimization are particularly designed xfi < xoi+1 , for i = 1, . . . , N − 1
for the OLEV in Seoul Grand Park, they could be utilized as a 
N
fundamental building block for more complicated models to be yi ≤ L.
developed in the future. Another motivation of this paper is to i
provide an analytical foundation for emerging intelligent trans-
portation systems that integrate the vehicle, the driver, and the With these notations, allocating a power transmitter in
road (the general environment). The advancement of in-vehicle this paper refers to determining the starting xoi and end
information systems and accurate GPS technology opens the xfi points of the power cable of the ith transmitter. The
door for the efficient operation of transportation systems, such total number of transmitters N is also a decision variable.
as better fleet management and real-time scheduling of public 7) There is a limit on the length of the transmitter cable
transportation [28]–[30]. Note that the OLEV is an existing due to the power supply limit of the inverter. The limit
vehicle–driver–road integrated system. There is strong potential is denoted by Lmax
c , and the constraint is expressed as
for further development of the OLEV system if the system
can effectively utilize advanced in-vehicle information and real- yi ≤ Lmax
c for i = 1, . . . , N.
time traffic data. For example, traffic and location information
may enable efficient real-time energy control of the OLEV, i.e., 8) The installation cost of one transmitter consists of the cost
optimal charging decisions could be made in real time based of the inverter cinv [unit cost per inverter] and the cost of
on the energy requirement to the destination, traffic conditions, the transmitter cable ccable [unit cost per cable length],
and other real-time data. The analytical model proposed in which is proportional to the length of the cable (this cost
this paper could be extended to create an analytical building includes the construction and equipment costs). Then, the
block for the OLEV system utilizing the advanced in-vehicle installation cost of the power transmitter i is described by
information system. the function denoted by ci , such that

III. O PTIMIZATION M ODEL D EVELOPMENT c(i) = cinv + ccable · yi .

The purpose of developing a mathematical model of the 9) The maximum battery capacity of the OLEV is Imax . This
OLEV system is to construct an optimization problem to min- value is determined by the optimization problem.
KO AND JANG: OPTIMAL SYSTEM DESIGN OF OLEV UTILIZING WIRELESS POWER TRANSMISSION TECHNOLOGY 1259

the entire spectrum of the charging state is nonlinear, as

shown in Fig. 6, the charging state within Ihigh and Ilow ,
which is the actual state range in the OLEV, is almost a
straight line. Therefore, the charging rate, denoted by Pcs ,
is assumed to be constant. The linear energy charging and
discharging assumption is widely used in electric vehicle
modeling [31], [32].
3) Regenerative braking—an energy recovery mechanism
that slows a vehicle down by converting its kinetic
Fig. 6. Charging and depleting energy in the battery of the OLEV system. energy into electric energy and charges the battery
simultaneously—is neglected in the model. This is a
10) The state of the charge level of the OLEV at time t is reasonable assumption because the OLEV operates at a
presented by I(t). relatively low speed, and the energy generated by the
11) Let Ihigh and Ilow be the maximum and minimum charge- regenerative braking is negligible.
able battery capacities, respectively, of the OLEV during 4) The OLEVs are driven following a fixed and predefined
its operation. They satisfy the following: speed profile for the route, expressed by the function
Ihigh = α · Imax V (x). That is, at a point x, we can predict that the vehicle
is running at a speed V (x). In an actual operational
Ilow = β · Imax condition, drivers may not follow the predefined profile
exactly, and there is some variability in the vehicle speed.
where 0 ≤ α ≤ 1, and 0 ≤ β ≤ α. These α and β are However, our recent investigation of the actual driving
constants and given. data from the OLEV in Seoul Grand Park has indicated
12) The state-of-charge (SOC) level at the initial starting that the drivers followed the velocity instructions fairly
point is Ihigh , and the remaining battery capacity of the well [33]. Therefore, when operating in a special envi-
OLEV during operation should always be greater than ronment such as the Seoul Grand Park, a fixed velocity
Ilow . Therefore profile is a reasonable assumption.
Ilow ≤ I(t) ≤ Ihigh .
C. Power Requirement
13) The charge/supply rate of electricity per unit time is In this paper, we consider k vehicles operating on a fixed
always greater than the electricity consumption rate per route. The required power for the vehicles at a specific point x
unit time of the OLEV in operation, and our model on the route, denoted by Pte (x), is approximated with Fte (x)
reflects this fact. and V (x), which are the required force and velocity for point x.
14) There is one base station on the route, which is used as a Again, V (x) is the velocity profile, where a vehicle is expected
parking area when the OLEVs are idle. We assume that to run at V (x) at point x on the route. The required force at
there is a power transmitter installed in the base station, each point is decomposed into the five applied forces: rolling
and it is the starting point of the route. Therefore, once resistance force Frr , air resistance force Fad , hill climbing force
a vehicle completes a round trip, it moves to the base Fhc , linear acceleration force Fla , and wheel acceleration force
station, and the battery is fully charged before it begins Fwa . In general, the magnitude values of both air resistance
the next trip. and wheel acceleration are relatively small; thus, we assume
that these two forces can be neglected. As a consequence, the
B. Modeling Assumptions driving force can be described as
The following assumptions are considered in the model: Fte = Frr + Fhc + Fla . (1)
1) The battery cost is proportional to the maximum capacity Note that the acceleration force can be calculated from the
of the battery Imax , hence cbat · Imax . This is a reason- velocity profile, and the hill climbing force can be calculated
able assumption because the battery pack used in the from the route information (hill profile). The power and force
electric vehicle consists of multiple battery cells. The ca- requirements, i.e., Pte (x) and Fte (x), respectively, are func-
pacity of the battery is normally determined by the num- tions of the placement x. For convenience, we convert these
ber of cells in the battery pack; thus, the cost of variables into functions of time t, i.e., Pte (t) and Fte (t). Our
the battery is linearly proportional to the number of cells. primary focus in this modeling is to understand the battery
2) The SOC level is linearly proportional to the charging charging behavior on the route. The charging rate is a function
time, as long as the level of the charge is between of time; hence, it is more convenient to describe the variables as
Ihigh and Ilow , as shown in Fig. 6. This assumption a function of time rather than a function of displacement. This
is based on the mechanism of the charging controller conversion can be achieved using the velocity profile equation
in the OLEV. The charging limits are set to maximize V (x) and the definition of the velocity v(t) = dx/dt; thus,
the battery life. Therefore, if the vehicle is traveling on dx/dt = V (x). Required power at time t is then expressed as
a route with a power transmitter installed, the level of
charge in the battery will increase. Note that, although Pte (t) = Fte (t) · V (t). (2)
1260 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 3, SEPTEMBER 2013

We introduce the required battery power at time t denoted IV. O PTIMIZATION M ODEL AND S OLUTION A LGORITHM
by Pbat (t). In general, electricity from the battery is served to
A. Optimization Model Development
the electric motor through several separate processes, such as
gears and controllers, and there are further energy dissipation Our goal is to allocate the power transmitters and to deter-
in the form of sound, light, or heat. As a result, some degree mine the battery size so that the vehicles operate at a minimum
of efficiency loss is inevitable. We define the efficiency loss cost while maintaining the energy level of the batteries Imin ≤
η and assume that it is a constant. Then, the battery power I(t) ≤ Imax . To allocate optimally the power transmitters for
requirement is then described as the OLEV operating on fixed routes, we develop a mathematical
model with the objective of minimizing the total cost. The
Pbat (t) = Pte (t)/η. (3) objective function is presented such that
 
This approximation is widely used in industry [34], and the 
N  
f
numerical value of the efficiency can be evaluated from the min k · Fp (Imax ) + N · cinv + ccable · xi − xi . (7)
o

experiment. i=1

The first term in (7) indicates the cost of the batteries for a k
D. Battery Power Behavior number of vehicles operating on the route. There are k vehicles,
and each vehicle is equipped with one identical battery. The
Our primary decision variables are the battery size Imax and
cost of the battery is a function of the maximum capacity Imax ,
the allocation of the power transmitters on the route denoted
which is described by the function Fp . The cost monotonically
by xoi and xfi , where i = 1, 2, . . . , N . As explained earlier, it
increases with the maximum capacity of the battery. The second
is more convenient to describe the battery power dynamics as
and third terms represent the cost of the transmitters. Each
a function of time rather than as a function of displacement.
transmitter unit consists of an inverter and a power cable;
If the vehicle is traveling at a predefined speed, following the
therefore, if we install one transmitter unit, there is also a cost
specified speed profile V (x), we can calculate the location of
for the inverter and the cable. There are N transmitter units,
the vehicle at a specific time t. At this point, we introduce the
and each transmitter needs to have one inverter. Therefore,
following variables: the arrival time at the starting point of the
the terms for the inverter cost are multiplied by N . Note that
ith power transmitter toi , and the arrival time at the ending point
the cost of the cable linearly increases with the length of the
of ith power transmitter, tfi . Once we know these two arrival cable. As a result, the cost of each cable unit is described as
time variables, we can evaluate the decision variables, xoi and ccable (xfi − xoi ). The third term sums up the cost of the cables.
xfi from the following relationship: Before we construct the constraint equations for the opti-
j mization model, consider the case in which a vehicle is about to
ti leave an area where a power transmitter i is installed. From our
xji = V (t)dt for j ∈ {o, f } notion, we can say that, at this particular moment, the vehicle
0 is located at xfi , and the time at which the vehicle is at this
and i = 1 . . . N. (4) point is denoted by tfi . The level of charge in the battery at this
moment is expressed as I(tfi ). Once the vehicle starts traveling
Let us express the dynamic behavior of the battery. If the on the route where no transmitter is installed, the energy in
vehicle is traveling where no power transmitter is installed, the the battery will be consumed at the rate of Pbat (t), which is
level of energy in the battery will be decreased; then, the rate of the power consumption rate of the vehicle at time t. When the
the energy level in the battery is described as vehicle arrives at the starting point of the next transmitter at
dI(t) time toi+1 , the total energy consumption from point xfi to point
= −Pbat (t). (5) xoi+1 , which can also be described by the time stamps between
dt
tfi and toi+1 , is evaluated by integrating Pbat (t) over tfi and toi+1 .
Again, Pbat (t) is the required battery power at time t. That This quantity should be larger than the minimum charge level
is, the energy in the battery is consumed at the rate of Pbat . Ilow ; hence, we can state the following constraint:
However, if the vehicle is traveling where a power transmitter o
  i+1
t
is installed, energy will be added to the battery, and at the same
f
time, some energy will be consumed by the motor. Following I ti − Pbat (t)dt ≥ Ilow . (8)
our assumption, the charging rate, which is denoted by Pcs , is
tfi
assumed to be constant and given. Then, the level of energy in
the battery at time t is expressed as
If the vehicle keeps traveling from point xoi+1 , the level of

dI(t) −Pbat (t) + Pcs , if I(t) < Imax charge will increase as it is now moving on the path with
= (6) a transmitter installed, and the battery receives the energy
dt 0 if I(t) = Imax .
wirelessly from the transmitter, i.e., i + 1. The rate of energy
Note that there is an upper charging limit in the battery, i.e., increase is Pcs , and the amount of charge added to the battery is
Imax . When the energy level reaches this upper limit, no charge proportional to the time that the vehicle spends traveling above
will be added. the transmitter. As a result, the motor will consume the charge
KO AND JANG: OPTIMAL SYSTEM DESIGN OF OLEV UTILIZING WIRELESS POWER TRANSMISSION TECHNOLOGY 1261

at the rate of Pbat (t), whereas at the same time, the charge
will be added at the rate of Pcs as the vehicle travels above the
transmitter. However, when the battery charge level reaches its
maximum, i.e., Ihigh , the current from the pickup device will
be cut off to prevent the battery from overcharging. Therefore,
when the vehicle arrives at the end of the transmitter i + 1, the
level of charge I(tfi+1 ) is expressed as
⎧
⎪
⎪ tf
  ⎨   i+1
I tfi+1 = min Ihigh , I tfi − Pbat (t)dt
⎪
⎪
⎩ t f
i

 
+ Pcs tfi+1 − toi+1

i = 1, . . . , N − 1. (9)

Considering the fact that, in the current OLEV system, the

battery is to be fully charged before it begins the round trip at
the starting point x = 0, we can set up the following constraint
Fig. 7. Solution procedures for the optimization problem with PSO.
equations:

t1
o B. Solution Procedures
Ihigh − Pbat (t)dt ≥ Ilow (10) The optimization problem consists of the objective function
0 described in (7) and the constraints expressed from (10)–(18).
⎧ f
To solve the optimization problem, we selected PSO as the
  ⎪
⎨ t1 solution algorithm because it is very robust for solving com-
I tf1 = min Ihigh , Ihigh − Pbat (t)dt plex nonlinear optimization problems. Moreover, PSO does
⎪
⎩ not require that the optimization problem be differentiable.
0
Therefore, it is the best candidate for our problem, which

  contains nondifferentiable equations. PSO is a metaheuristic
+ Pcs tf1 − to1 (11) algorithm introduced in [35]. The PSO algorithm was moti-
vated by the social behavior of flocks of birds or schools of
o fish. Compared with other metaheuristic optimization methods,
  i+1
t
the main advantage of the PSO algorithm is its simplicity; it
I tfi − Pbat (t)dt ≥ Ilow , for i = 1 . . . , N − 1 (12) requires fewer initial parameters, and its searching mechanism
tfi is more intuitive. Other applications of PSO can be found
⎧ in [36]. In PSO, the particle refers to a certain candidate
⎪
⎪
f

  i+1
t solution. Each particle has its own position (solution), velocity,
  ⎨
I tfi+1 = min Ihigh , I tfi − Pbat (t)dt and best position (solution). To find the best solution to a
⎪
⎪ given problem, the particles move around in the search space
⎩ tfi according to the given mathematical formula for their current
 positions and velocities. Each particle’s velocity is calculated
 
+ Pcs tfi+1 − toi+1 and updated at each step. All particles share their information
about their current positions, best positions, and velocities, so
that they eventually find the best global solution. More detailed
for i = 1, . . . , N − 1 (13) descriptions of PSO can be found in [35].
to The solution procedure for the optimization problem with
   N
PSO is shown in Fig. 7 and is described as follows:
I tfN − Pbat (t)dt ≥ Ilow (14)
1) Multiple particles, which are candidate solutions, are ran-
T
domly generated in the feasible solution space. As we do
xfi − xoi ≤ Lmax , for i = 1, . . . , N (15) not know the optimal solution at this point, the particles
c
are randomly distributed. The position of a particle s at
xoi < xfi , for i = 1, . . . , N (16) the iteration step n is denoted by λs (n).
2) Evaluate the current fitness value with the objective func-
xfi < xoi+1 , for i = 1, . . . , N − 1 (17) tion in the optimization problem.
  3) Compare the fitness value at the current iteration with the
f tji = xji , for i = 1, . . . , N, j = {o, f }. (18) previous best and update the individual (pbest) and group
1262 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 3, SEPTEMBER 2013

best (gbest). Note that pbest is the best value achieved by TABLE I
I NPUT PARAMETERS FOR THE N UMERICAL E XAMPLE
the individual particle so far, and gbest is the best value
of the group.
4) Evaluate the velocity, i.e., direction and magnitude, de-
noted by vs (n) for each particle, and update the position
of the particles with the velocities from the following
equations:

λs (n + 1) = λs (n) + vs (n + 1) (19)
vs (n + 1) = ω · vs (n)
+ α1 · rand1 · (pbests − λs (n))
+ α2 · rand2 · (gbest − λs (n)) . (20)

The values α1 and α2 are the acceleration parameters, and

their values are typically set as α1 + α2 = 4. The detailed
parameter settings can be found in [35]. Parameter ω
is the inertia constant, which is set by the following
equation:
ωmax − ωmin
ω = ωmax − · n.
nmax
With this equation, the particles initially move freely
throughout the search space to find the global optimum.
As the particles reach the optimal value, their velocities
are reduced to define the search more narrowly [37].
In (20), rand1 and rand2 are uniform random numbers Fig. 8. Planned velocity profile of the route used for the case problem.
between 0 and 1.
5) Repeat steps 2 to 4 until the termination criteria are met. TABLE II
S LOPE P ROFILE OF THE ROUTE U SED FOR THE C ASE P ROBLEM

V. N UMERICAL EXAMPLE
A. Computational Results
The proposed solution procedure was programmed in C++
language and was run on a PC with an AMD Athlon II X3 450
microprocessor under Windows Vista. For the road condition,
battery size, and distance values, we utilized the information
from the existing route in the Seoul Grand Park shown in Fig. 5. TABLE III
The total distance is 2.85 km, and the maximum allowable R ESULT OF THE N UMERICAL E XAMPLE
battery capacity of the OLEV is 20 kW. We limited the battery
size to 20 kW due to the space constraint of the vehicle. The
input parameters for the case are shown in Table I. The planned
velocity, which is the target velocity for the vehicles, and the
uphill/downhill profiles are shown in Fig. 8 and Table II, respec-
tively. Due to the data ownership issue, we modified the actual
data used in the OLEV at the Seoul Grand Park. However, the
input values used to present the results in this paper are not
very different from the actual system. The algorithm reliably
generated the results within a few seconds. The solution is
summarized in Table III. The expected SOC over the course the route. The second, third, and fourth transmitters are lo-
of the route is depicted in Fig. 9. Note that the SOC is evaluated cated at 1140.3–1194, 1782.8–1831, and 2153 m–2182.7 m,
using the linear model proposed in [31] and [32], and the state respectively. The optimal battery capacity is 11 kW, and the
of health of the battery is not considered in the SOC evaluation, total cost is 64025.4, which is the minimum figure possible
as proposed in [38]. while satisfying the constraints. There are four stopping points
The optimal solution indicates that four transmitters should (stations), excluding the base station, on the route. We discov-
be allocated. The first transmitter is located between 425 ered that the optimal solution allocated all four transmitters
and 660.6 m from the starting point of the vehicles on near the stop points (station). For instance, the first station is
KO AND JANG: OPTIMAL SYSTEM DESIGN OF OLEV UTILIZING WIRELESS POWER TRANSMISSION TECHNOLOGY 1263

Fig. 9. Expected charge level with the optimized result.

located at 647.0 m, and the first transmitter is installed between

425 and 660.6 m. This result intuitively makes sense for the
following reasons. The vehicles slow down as they approach Fig. 10. Decision variables and total cost versus various battery costs per unit
the stations and then eventually stop. Note that the amount of capacity.
charging is proportional to the charging time. That is, given the
same length of a transmitter cable, more charging is possible
when the vehicles are moving slowly along the transmitter cable
because they spend more time on the transmitters. Therefore, it
is cost effective to allocate the transmitters to the areas where
the vehicles are moving slowly.

B. Sensitivity Analysis
To understand the system behavior of the OLEV system, a
sensitivity analysis was performed.
We first investigated the system behavior of the OLEV
system with a varying battery cost. In this analysis, the number
of vehicles in the OLEV system is five (k = 5), and this
number is fixed. We assume that the largest battery that the
Fig. 11. Decision variables and total cost versus various cable installation cost
OLEV system can equip in each vehicle is 20 kWh. An upper per unit meter.
limit on the battery size is needed in the analysis to reflect
reality; without this upper bound, the optimization problem out any additional charge. An interesting behavior is captured
would give an infinite capacity of the battery as an optimal when the battery cost increases from 640 to 720. In this case, the
solution if the battery cost was 0. Typically, the capacity of total cable length decreases, although the battery size decreases.
the battery is proportional to the volume of the battery. Due This case can be explained by the number of transmitters and
to the vehicle’s space limitation, the size of the battery is also their allocation. Note that the number of transmitters increases
limited. We varied the battery cost from 0 to 800 in increments from four to five when the cost increases from 640 to 720. The
of 80. That is, 11 different cases were evaluated. The results optimization problem calculates that it would be cost effective
are shown in Fig. 10. The red line (dotted line) and the blue to add one more transmitters to five instead of increasing the
line (solid line) represent the optimal values for the total cable length of the cables with the four existing transmitters. With five
length and battery capacity, respectively, for each battery cost. transmitters, the optimization model can allocate them to better
The total cable length in the figure is the sum of all of the locations and eventually reduce the total length of the cable.
cable lengths installed on the route. The values shown on the This case clearly shows the nonlinear behavior of the system
left y-axis are for the battery capacity, and those on the right and the reason why the lines do not monotonically increase or
y-axis are for the total cable length. The numerical values decrease as the input parameters change.
shown above the red line indicate the number of transmitters Fig. 11 shows the results of the sensitivity analysis when the
N . In general, as the cost of the battery increases, the optimal cable cost per unit length is varied from 0 to 120 in increments
battery size decreases and the total cable length increases, i.e., of 12. In general, the battery size increases as the cable cost
more transmitters are installed, and longer cables are installed increases. As the cable cost increases, the cable cost becomes
in response to the increasing battery cost. Note that, when the the dominant factor in the total cost. Therefore, the optimization
cost of the battery is zero, the optimization recommends the model tries to seek the solution that minimizes the total length
largest battery available, which is 20 kW. Note that the opti- of the cable either by installing more transmitters with a short
mal solution still recommends the installation of transmitters, cable or by increasing the battery capacity. The figure also
although the battery cost is 0. This is because a 20-kW battery, shows that the rate of decrease in the total length of the cable
which is the largest allowable size for the vehicle, is not big diminishes. We conjecture that there is an absolute minimum
enough to hold the required energy to complete the circuit with- length of cable required for the route, and this is around 250 m.
1264 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, VOL. 14, NO. 3, SEPTEMBER 2013

Fig. 12. Decision variables and total cost versus unit inverter cost.
Fig. 13. Decision variables and total cost versus number of operating OLEVs.

In other words, even if a battery with the largest capacity is

ticular case, the optimization model managed to find an optimal
installed, the vehicle needs to have at least 250 m of cable to
solution by adding one more transmitter instead of increasing
make one round trip.
the cable length with the existing number of transmitters.
Fig. 12 shows the optimal cable length and battery size when
Transmitters are then rearranged so that the system requires less
the inverter cost is varied. Remember that each transmitter
cable than the case in which there are eight vehicles.
consists of an inverter and a cable; thus, the number of inverters
in the OLEV system is identical to the number of transmitter
units. The numerical values shown along the cable length in the
VI. C ONCLUSION AND F UTURE R ESEARCH
figure indicate the optimal number of inverters for each case.
As the inverter cost increases, the optimal number of inverters In this paper, we have introduced the electric vehicle system
decreases. As shown in the figure, the total length of the cable called OLEV with the wireless power transfer technology de-
either increases or decreases, and there is a particularly sharp veloped by KAIST. The OLEV is an innovative electric trans-
increase when the cost of the inverter increases from 3000 to portation system that remotely picks up electricity from power
4000. This is because, as the cost of the transmitter increases, transmitters buried underground. The OLEV mainly consists of
the number of transmitters is reduced, and the effect of this vehicles and power transmitters installed under the road. As the
reduced number of transmitters is offset by assigning longer battery in the vehicle can be remotely charged while in motion,
cables for the remaining transmitters. That is, the optimization the vehicle does not need to stop at the charging station, and
model responds to the increased cost of the inverter by allocat- recharging downtime is significantly reduced. We logically an-
ing fewer transmitters with longer cables. Conversely, the cable alyzed the tradeoff between the battery size and the positions of
length is reduced when the inverter cost reduces from 5000 to the power transmitters on the route. In particular, we considered
6000. In this case, as the number of transmitters is reduced the OLEV system with multiple vehicles circulating on a single
from four to three, the size of the battery increases from 11 route, such as the OLEV currently operating in Seoul Grand
to 14 kW. As the battery capacity is increased, the vehicle is Park. We constructed an analytical model to imitate the battery
able to travel with less charging time, and the cable is reduced. charging and discharging behavior of the OLEV with reason-
There is another interesting case shown in the figure. Note that able assumptions and then presented the optimization problem
both the cable length and the battery size increase as the cost to find the minimum total investment cost while satisfying the
of the inverter increases from 8000 to 9000. In this case, the operational requirements. We considered the battery size and
number of transmitters is reduced to two. We conjecture that the total number of power transmitters and their allocation as
this transmitter count is too small to satisfy the constraints, and decision variables. We presented the solution algorithm with the
both the cable length and the battery size increase dramatically PSO technique to solve the optimization problem. We validated
with only two transmitters. our model and algorithm with a numerical solution for a case
Fig. 13 shows the system behavior in response to the number with numerical parameters similar to the actual values of the
of vehicles. As more vehicles are required, the cost of the bat- OLEV in Seoul Grand Park. It was proven that the algorithm
tery becomes the dominant factor in the total cost of the OLEV reliably generated solutions. We also performed a sensitivity
system. Therefore, as the number of vehicles increases, the analysis to show the model with our intuition. In this paper,
optimization seeks optimal solutions with less battery capacity. we have assumed that the vehicles follow a regulated vehicle
As the battery capacity gets smaller, either longer cables or profile. This assumption may work for the vehicles operating
more transmitters are needed. Hence, the red line representing in a controlled environment, such as the OLEV in Seoul Grand
the total length of the cable and the numerical values indicating Park. However, if the method is applied to a system operating
the number of transmitters show trends. However, this upward in a normal traffic environment, velocity variability should be
trend is not always true. When the number of vehicles changes considered. We suggest that future research will incorporate the
from eight to nine, the cable length reduces because, in this par- velocity variation issue.
KO AND JANG: OPTIMAL SYSTEM DESIGN OF OLEV UTILIZING WIRELESS POWER TRANSMISSION TECHNOLOGY 1265

R EFERENCES [26] S. Ahn, Y. Chun, and D. Cho, “Wireless power transfer technology in
on-line electric vehicle,” J. Korean Inst. Electromagn. Eng. Sci., vol. 11,
[1] N. Suh, D. Cho, and C. Rim, “Design of On-Line Electric Vehicle
no. 3, pp. 174–182, Sep. 2011.
(OLEV),” in Proc. CIRP Design Conf. Plenary Presentation, Nantes,
[27] Korea Advanced Inst. Sci. Technol. , (2011). Wireless Electric Vehicle
France, Apr. 2010, pp. 3–8.
in the Seoul Amusement Park, Daejeon, South Korea, KAIST News.
[2] “The 50 best innovations of 2010,” Time Mag., Nov. 2010. [Online]. Avail-
[Online]. Available: http://newspr.kaist.ac.kr/boards/view/campus/2033
able: http://www.time.com/time/specials/packages/0,28757,2029497,00.
[28] H. Cai and Y. Lin, “A roadside intelligent transportation systems data
html bus prototype for intelligent highways,” IEEE Trans. Intell. Transp. Syst.,
[3] “South korea’s innovative solution,” CNN World News Rep., Mar. 10,
vol. 9, no. 2, pp. 344–348, Jun. 2008.
2011.
[29] J. Du and M. J. Barth, “Next-generation automated vehicle location sys-
[4] N. Shinohara, “Power without wires,” IEEE Microw. Mag., vol. 12, no. 7,
tems: Positioning at the lane level,” IEEE Trans. Intell. Transp. Syst.,
pp. S64–S73, Dec. 2011. vol. 9, no. 1, pp. 48–57, Mar. 2008.
[5] K. Eaton, Qualcomms Vision of a Wireless Charging Future, Fast Com-
[30] J. Zhang, F.-Y. Wang, K. Wang, W.-H. Lin, X. Xu, and C. Chen, “Data-
pany, Harlan, IA, USA. [Online]. Available: http://www.fastcompany.
driven intelligent transportation systems: A survey,” IEEE Trans. Intell.
com/blog/kit-eaton/technomix/qualcomm-teases-vis
Transp. Syst., vol. 12, no. 4, pp. 1624–1639, Dec. 2011.
[6] N. Shinohara, “Development of rectenna with wireless communication [31] S. Mohrehkesh and T. Nadeem, “Toward a wireless charging for battery
system,” in Proc. 5th EUCAP, 2011, pp. 3970–3973.
electric vehicles at traffic intersections,” in Proc. 14th IEEE Int. Conf.
[7] N. Shinohara, T. Mitani, and H. Matsumoto, “Study on ubiquitous power
Intell. Transp. Syst., 2011, pp. 113–118.
source with microwave power transmission,” in Proc. Int. URSI Gen.
[32] A. Ashtari, E. Bibeau, S. Shahidinejad, and T. Molinski, “PEV charging
Assembly, 2005, pp. 1–4.
profile prediction and analysis based on vehicle usage data,” IEEE Trans.
[8] A. P. Sample and J. R. Smith, “Experimental results with two wireless
Smart Grid, vol. 3, no. 3, pp. 341–350, Mar. 2012.
power transfer systems,” in Proc. IEEE RWS, 2009, pp. 16–18.
[33] Y. J. Jang and Y. D. Ko, “Optimal design of the On-Line Electric Vehicle
[9] J. R. Smith, “Mapping the space of wirelessly powered systems,” pre- with various optimization methodology—Modeling analysis and cases,”
sented at the Proc. IMS Workshops, Lisbon, Portugal, 2010, WFB-3.
Korea Adv. Inst. Sci. Technol., Daejeon, Korea, Tech. Rep., 2011.
[10] A. Karalis, J. D. Joannopoulos, and M. Soljacic, “Efficient wireless non-
[34] J. Larminie and J. Lowry, Electric Vehicle Technology Explained.
radiative mid-range energy transfer,” Ann. Phys., vol. 323, no. 1, pp. 34–
Hoboken, NJ, USA: Wiley, 2004.
48, Apr. 2008. [35] J. Kennedy and R. Eberhart, “Particle swarm optimization,” in Proc. IEEE
[11] Kurzweil. (2012, Mar.) Dealing with range anxiety: Wireless highways
Int. Conf. Neural Netw., 1995, vol. 4, pp. 1942–1948.
to charge electric cars. Kurzweil AI Network. [Online]. Available:
[36] H. Hwang, Y. Ko, S. Yune, and C. Ko, “A closed-loop recycling system
http://www.kurzweilai.net/dealing-with-rangeanxiety-wireless-highways-
with a minimum allowed quality level on returned products,” Int. J. Ser-
to-charge-electric-cars vices Oper. Manage., vol. 5, no. 6, pp. 758–773, Jun. 2009.
[12] D. Cooper, Daimler Testing Wireless Charging on a Mercedes Benz in
[37] W. Su and M.-Y. Chow, “Performance evaluation of a phev parking station
a House from the Future, Engadget, Broadway, NY, USA. [Online].
using particle swarm optimization,” in Proc. IEEE Power Energy Soc.
Available: http://www.engadget.com/2011/12/08/daimler-testing-
Gen. Meeting, 2011, pp. 1–6.
wireless-charging-on-a-mercedes-benz-in-a-house/ [38] S. Piller, M. Perrin, and A. Jossen, “Methods for state-of-charge determi-
[13] HINO Motors, Ltd. (2007, Mar.). Hino Motors Annual Report, Tokyo,
nation and their applications,” J. Power Sourc., vol. 96, no. 1, pp. 113–
Japan. [Online]. Available: www.hino-global.com
120, Jun. 2001.
[14] T. Imura, T. Uchida, and Y. Hori, “Flexibility of contactless power transfer
using magnetic resonance coupling to air gap and misalignment for EV,”
World Elect. Veh. J., vol. 3, no. 2009, pp. 24–34, 2009.
[15] T. Imura, T. Uchida, and Y. Hori, “Basic experimental study on helical
antennas of wireless power transfer for electric vehicles by using magnetic Young Dae Ko received the B.S. degree in industrial
resonant couplings,” in Proc. IEEE Veh. Power Propulsion Conf., 2009, engineering and the Ph.D. degree in industrial and
pp. 936–940. systems engineering from Korea Advanced Insti-
[16] J. Huh, E. Park, G. Jung, and C. Rim, “High efficient inductive power tute of Science and Technology (KAIST), Daejeon,
supply system implemented for on line electric vehicles,” in Proc. Autumn Korea, in 2004 and 2011, respectively.
Conf. Korean Inst. Power Electron., 2009, pp. 105–110. He is currently a Postdoctoral Researcher with the
[17] S. Ahn, J. Pak, T. Song, H. Lee, J.-G. Byun, D. Kang, C.-S. Choi, E. Kim, Industrial Engineering and Management Research
J. Ryu, M. Kim, Y. Cha, Y. Chun, C.-T. Rim, J.-H. Yim, D.-H. Cho, and Institute, KAIST. His research interests include lo-
J. Kim, “Low frequency electromagnetic field reduction techniques for the gistics optimization, reverse logistics, green trans-
On-Line Electric Vehicle (OLEV),” in Proc. IEEE Int. Symp. EMC, 2010, portation, and revenue management.
pp. 625–630.
[18] S. Ahn, J. Y. Lee, D. H. Choi, and J. Kim, “Magnetic field design for low
EMF and high efficiency wireless power transfer system in on-line electric
vehicles,” in Proc. CIRP Des. Conf., 2011, pp. 233–239.
[19] S. Lee, J. Huh, C. Park, N.-S. Choi, G.-H. Cho, and C.-T. Rim, “On-line
electric vehicle using inductive power transfer system,” in Proc. IEEE Young Jae Jang (M’11) received the B.S. degree
Conf. ECCE, 2010, pp. 1598–1601. in aerospace engineering from Boston University,
[20] J. Shin, B. Song, S. Lee, S. Shin, Y. Kim, G. Jung, and S. Jeon, “Con- Boston, MA, USA, in 1997 and the M.S. degree in
tactless power transfer systems for On-Line Electric Vehicle (OLEV),” in mechanical engineering and operations research and
Proc. IEEE IEVC, 2012, pp. 1–4. the Ph.D. degree in mechanical engineering from the
[21] Y. Kim, Y. Son, S. Shin, J. Shin, B. Song, S. Lee, G. Jung, and S. Jeon, Massachusetts Institute of Technology, Cambridge,
“Design of a regulator for multi-pick-up systems through using current MA, USA, in 2001 and 2007, respectively.
offsets,” in Proc. IEEE IEVC, 2012, pp. 1–6. He was with Micron Technology, Inc., Manassas,
[22] G. Jung, B. Song, S. Shin, S. Lee, J. Shin, Y. Kim, and S. Jeon, “High VA, USA, where he was a Project Manager and led
efficient inductive power supply and pickup system for on-line electric a global initiative to improve the efficiency of their
bus,” in Proc. IEEE IEVC, 2012, pp. 1–5. manufacturing facilities located worldwide. While
[23] B. Song, J. Shin, S. Lee, S. Shin, Y. Kim, S. Jeon, and G. Jung, “Design working at Micron, he was also involved in activities supporting the company’s
of a high power transfer pickup for On-Line Electric Vehicle (OLEV),” in operational and strategic decision-making using quantitative modeling and
Proc. IEEE IEVC, 2012, pp. 1–4. analysis techniques. He is currently an Assistant Professor with the Department
[24] Y. J. Jang, Y. D. Ko, and S. Jeong, “Optimal design of the wireless of Industrial and Systems Engineering, Korea Advanced Institute of Science
charging electric vehicle,” in Proc. IEEE IEVC, 2012, pp. 1–5. and Technology (KAIST), Daejeon, Korea. He is also affiliated with the Cho
[25] Y. J. Jang, Y. D. Ko, and S. Jeong, “Creating innovation with systems Chun Shik Graduate School of Green Transportation, KAIST. His current
integration-road and vehicle integrated electric transportation system,” in research interests include stochastic modeling of complex systems and opti-
Proc. IEEE Int. SysCon, 2012, pp. 1–4. mizations in transportation and logistics systems.