This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

Economic Analysis of the Dynamic Charging

Electric Vehicle
Seungmin Jeong, Student Member, IEEE, Young Jae Jang, Member, IEEE, and Dongsuk Kum, Member, IEEE

Abstract—A wireless charging or inductive charging electric

vehicle (EV) is a type of EVs with a battery that is charged
from a charging infrastructure, using a wireless power transfer
technology. Wireless charging EVs are classified as stationary
or dynamic charging EVs. Stationary charging EVs charge
wirelessly when they are parked, and dynamic charging EVs
can charge while they are in motion. The On-Line Electric
Vehicle (OLEV T M ) developed at the Korea Advanced Institute
of Science and Technology is an example of a commercially
available dynamic charging transportation system. Numerous
studies have reported that one of the benefits of dynamic
charging is that it allows smaller and lighter batteries to be
used, due to frequent charging using the charging infrastructure
embedded under roads. In this paper, we quantitatively analyze
the benefits of dynamic charging with an economic model of
battery size and charging infrastructure allocation, using a
mathematical optimization model. Particularly, we analyze by
how much battery size can be reduced and what the cost saving Fig. 1. The OLEV system in Gumi City, South Korea
of reducing the battery size is with the model. We also show
that the dynamic charging can be beneficial to battery life.
Index Terms—Electric vehicle, system optimization, wireless transferred to a secondary pickup installed in the moving
power transfer, road charging vehicle, dynamic charging electric vehicle. It is known that the dynamic charging reduces the
vehicle. high initial cost of EVs by allowing smaller batteries [4].
Moreover, the charging infrastructure, called power track, is
I. I NTRODUCTION effectively used, because a large number of vehicles can use
the same road segments equipped with the charging system.
Wireless charging and inductive charging electric vehicles The On-Line Electric Vehicle (OLEV T M ) developed at
(EVs) are a new type of EVs that remotely charge their batter- the Korea Advanced Institute of Science and Technology
ies using a wireless charging technology. Unlike conventional (KAIST) is a commercially available dynamic charging trans-
plug-in EVs, the wireless charging EVs do not require power portation system. The first commercial version of the OLEV
cable connection for charging. system was introduced in 2009 and used in a trolley system
Wireless charging EVs are classified as either stationary in the Seoul Grand Park. The second one was developed
or dynamic charging EVs. One of the first commercially for shuttle buses on the KAIST campus, where it has been
available EVs using stationary charging was the General in operation since 2012 [5]. Gumi City, one of the biggest
Motor EV1 [1], for which the primary transducer was a industrial cities in South Korea, deployed two OLEV buses
paddle and the secondary transducer a vehicle charger port. on a metro bus line in 2013, shown in Figure 1. There are
The topology of the charging system was introduced in [2] plans for expansion to more bus lines in the city. Sejong
and [3]. Stationary charging transfers energy more efficiently City, a new Government Capital City of South Korea, plans
than dynamic charging due to better alignment between the to deploy an OLEV-based public transportation system to
transducers. serve existing bus lines [6].
Dynamic charging EVs, which are also referred to as Numerous studies have reported that one of the benefits
move-and-charge or road-way-charging EVs, can charge of dynamic charging is that it allows smaller and lighter
while vehicles are in motion. The power from a primary batteries to be used, due to frequent charging using power
source embedded below the pavement’s surface is wirelessly tracks embedded under roads [2][7][8][9][10][11]. However,
The authors are with the Korea Advanced Institute of Science and Tech- none of these studies analyzed by how much battery size
nology (KAIST), 291 Daehak-ro, Daejeon, 350-701 South Korea. Email: could be reduced and what the cost saving of reducing
Young Jae Jang ([email protected]), (+82) 42-350-3130. the battery size was. A true quantitative benefit analysis of
This research was supported by the Basic Science Research Program
through the National Research Foundation of Korea (NRF), funded by the dynamic charging has not yet been reported. It is difficult to
Ministry of Education (NRF-2013R1A1A2009076). perform a quantitative analysis of dynamic charging because

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

We summarize the related research in Section II. Then we

introduce the OLEV system’s architecture and operational is-
sues in Section III. We present an optimization issues for the
dynamic charging EV system with power dynamics, energy
flow, and battery life prediction models in Section IV. The
economic design model for the dynamic charging EV system
is developed in Sections V. We implement our model with a
numerical case study in Section VI, using data acquired from
the OLEV bus system in Gumi City. We show that dynamic
charging EV system has meaningful benefits, specifically
compared to the stationary charging EV system. Sensitivity
analyses presented in the section also provide insight into
the dynamic charging EV system. Section VII summarizes
Fig. 2. (a) Shallow, frequent charging contrasted with (b) deep, infrequent the paper and proposes future research directions.
charging

II. P REVIOUS RESEARCH

the battery size is directly related to the installation of power As already noted, wireless charging has been applied in
tracks on the EV route. For instance, let us consider public EVs. The first such concept was presented by Bolger in
transportation with dynamic charging such as metro electric 1978 and involved picking up electric energy transferred
buses. If power tracks are installed in many places, the buses from a source embedded in a roadway. An inductive charger,
can charge frequently with small batteries. If not many power consisting of a primary coil beneath the roadway, generated
tracks are available, relatively larger batteries are needed for a magnetic field. A power pick-up device in the vehicle
the buses. The battery size and power track allocation have received and converted the magnetic energy into electrical
to be considered together to determine the true benefit of a power [13]. The major issue in wireless charging for EVs is
dynamic charging EV system. efficiency. The large air gap between the grounded primary
In this paper, we analyze the benefit of the dynamic coil and the secondary pick-up coil in the EV reduces the
charging EV system using data collected from the OLEV charging efficiency. Many studies have focused on improving
buses currently operating in Gumi City, South Korea. We the charging efficiency across the air gap. Esser et al.
propose an economic design model for the dynamic charging [7] achieved 92% charging efficiency using a 0.2 mm air
EV system. The economic model offers the optimal battery gap, taking into consideration of the filters, converters, and
size and allocation of power tracks considering the battery rectifiers in an EV. Wu et al. [14] presented a new inductive
life while minimizing the total cost. power pick-up device with more advantages than traditional
The battery life is specifically considered because we versions. Budhia et al. [15] also proposed a new induc-
hypothesize that dynamic charging can improve the lifetime tive power transfer system design for improving efficiency
of an EV’s battery, compared with stationary deep-cycling and economics. Other studies have presented improvements
charging. Cost saving using dynamic charging is based on in the wireless (or inductive) power transfer technology
the following idea. A shallow, frequent charging (Figure 2 [16][17][18][19].
(a)) for the lithium-type battery is known to be better for the The OLEV system is one of the first commercialized
battery life than a deep, infrequent charging (Figure 2 (b)) dynamic charging EV systems. Many studies have presented
[12]. Dynamic charging with a number of installed power the development of the KAIST OLEV and comprehensively
tracks allows the battery to be charged on a shallow, frequent described the hardware systems of the OLEV, such as
charging. In other words, installing more power tracks will [20][21][22] and [10]. They describe the wireless power
improve the battery life. However, more power tracks will transfer technology for the OLEV to accomplish higher
obviously require a greater initial investment. The goal of efficiency for the dynamic charging. Although the dynamic
the optimization is to find the optimal allocation of power charging technology has been improved enough to be com-
tracks and the optimal battery size considering this trade-off. mercialized, little research on economic system design issues
We aim to answer following research questions with the for the dynamic charging EVs has been investigated.
proposed economic models. This paper deals with the economic design of the dynamic
• By how much can dynamic charging reduce the size of charging EV system. We define that the system design is to
an EV’s battery, compared with stationary charging? determine the system design variables which are the size of
• What is the optimum battery size and power track the battery and the allocation of power tracks. Recently, the
allocation for a given route, for a dynamic charging EV importance of the system design for the dynamic charging EV
system? has raised due to commercialization of the dynamic charging
• Can dynamic charging increase the life of an EV’s technology in EV applications. Ko and Jang [23] presented a
battery? non-linear optimization model which minimizes the battery

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

and charging infrastructure costs for the OLEV system.

They first introduced the economical issues for designing the
dynamic charging system. Their approach aimed to determine
the battery size and the allocation of power tracks. Jang et
al. [24] discussed a benefit of the dynamic charging from
the perspective of system utilization, and proposed economic
model frameworks for the OLEV system design. Jang et al.
[25] also dealt with the analogous problem using a mixed-
integer programming. Although they have dealt with the
system design issues on the OLEV system, their goals are to
provide mathematical optimization algorithms with solution
approaches.
Research on the wireless charging EV with the considera-
tion of the battery life is limited. Pantic et al. [26] presented a Fig. 3. Architecture of the OLEV bus system
heuristic approach for determining the placement of wireless
charging infrastructures on the route. Their multi-objective
model relates the battery life to the peak output power. That
is, it is assumed that a high peak output power negatively
affect to the battery life.
In this paper, we focus on a cost benefit analysis of the
dynamic charging EV system with the consideration of the
quantitative battery life. We numerically compare the benefit
of dynamic charging to that of stationary charging.

III. O N -L INE E LECTRIC V EHICLE (OLEV)

A. OLEV architecture
Figure 3 illustrates an architecture of the OLEV system.
Each vehicle contains a pick-up device, battery, regulator, and
motor. A charging unit is a power track, which consists of an
inverter and an inductive track. In the OLEV system, 60 Hz
power supplied from the public electric grid is converted to a
frequency of 20 kHz through the inverter. From the inverter,
a high current flows through the coiled inductive track placed
beneath the road. The high current induces a magnetic flux
toward a surface of the road and the magnetic flux is Fig. 4. Route (upper) and driving cycle (lower) of the Gumi OLEV buses
transmitted to the bottom of the vehicle while the vehicle
passes over. The power is then generated by converting the
electromagnetic field into current at the pick-up device. The The upper part of Figure 4 depicts a servicing route of
remotely collected electricity is distributed to the motor, to the Gumi OLEV buses. There are 48 stations on the round
the battery via the regulator, or both, depending on the power trip and the total trip length is 34.448 km. There are east-
requirement of the motor and the battery’s energy level. The bound and west-bound routes. A bus leaves a base station (at
more detailed system architecture can be found in [24]. the west end) and travels on the east-bound route. Once it
reaches the station at the east end (a turning point), it rests
B. OLEV operations in Gumi City for about 10 minutes. It then travels back to the base station
on the west-bound route. The lower part of Figure 4 shows
The economic analysis presented in this paper is based
an example of velocity profiles collected from an OLEV
on the operational data collected from the OLEV system
bus in Gumi City. It takes about 2 hours, including about
in Gumi City. The OLEV system in Gumi City is the first
10 minutes of resting time at the turning point. Technical
implemented dynamic charging EV system in a public transit
characteristics of the OLEV system implemented in Gumi
bus system. To verify that the OLEV system is a practicable
City are summarized in Table I.
option for an eco-friendly public transportation, Gumi City
and KAIST have worked together since 2013 to install the
OLEV system in an existing service line. As of Spring 2015, IV. S YSTEM DESIGN MODEL OF THE DYNAMIC
two OLEV buses have been in service to demonstrate the CHARGING EV SYSTEM
feasibility of the OLEV technology, and more buses are In this section, we present the power dynamic and energy
planned to be deployed gradually. flow models of the dynamic charging EV to develop the

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

TABLE I
T ECHNICAL SPECIFICATIONS OF THE G UMI C ITY OLEV SYSTEM

Specification Value
Inverter power capacity 100 kW
Inverter frequency 20 kHz
Power track current 200 A
Pick-up power efficiency 80 %
Pick-up air gap 20 cm
EMF exposure level < 10 mG
Battery size 100 kWh

economic design model of the dynamic charging EV system.

The battery life prediction model adopted in this paper is
also presented.
Fig. 5. Design variables in OLEV system design optimization [27]
A. Optimization issues and model definitions for the dynamic
charging EVs
As a battery and power track costs account for a significant by E max . The length of each power track is determined
proportion of the total investment cost, economically evaluat- by the length of the inductive track and can vary. Figure
ing these parameters is key to the success of commercializing 5 illustrates the design variables for optimization. We need
a dynamic charging EV system. The goal of optimization is to to identify the number of power tracks required and the
determine the most economical battery size and power track length of each power track. We let the base station be the
allocation by simultaneously considering the cost factors and reference point. Any point on the route is described by the
power requirements. We consider the design of the dynamic travel distance x from the reference point. As the route is
charging EV system, particularly as it applies to a mass circular, the starting point where (x = 0), and end point
transportation. More specifically, the OLEV system model where (x = L) both indicate the base station. We consider
here refers to a mass transportation system operating under a vehicle making one circular trip (a service). The vehicle
the following conditions: begins the service at x = 0 and completes it at x = L. As
shown in Figure 5, the start and end points of the ith power
• Identical buses serve passengers along a predefined
track are denoted by xsi and xei , respectively. These values
route.
indicate the distance measured from the base station. Suppose
• Buses follow the regulated velocity and the required
that N power tracks have been allocated. Then, i = 1, . . . , N .
driving cycle is predetermined.
Note that N is one of design variables. The optimization
• There is one base station, at which the buses are idle
model aims to minimize the total cost finding the optimal
when they are not in service, and each service begins
design variables E max , and (xsi , xei ) for i = 1, . . . , N by
and ends at the base station.
satisfying the requirement that the vehicle serves the route
• Once a bus completes a service, it remains idle at the
without any service interruption.
base station for a certain amount of time that we call
We define t as a continuous variable indicating the vehicle
the resting or dwell time.
travel time measured from the base station. Let v(t) be the
• During the base station resting time, the battery is
required driving cycle of the vehicle over the route. Then
charged fully before service resumption.
we can easily project the displacement variables onto the
These operational rules of the model are based on the temporal variables. With v(t), we define a function, fts ,
actual operation of the OLEV system in Gumi City and also which relates the temporal and displacement variables
based on other previous research on the OLEV system design, Z t
including [24] and [25]. The final condition stipulates that the x = fts (t) = v(τ )dτ. (1)
bus must stop at the base station for enough to fully recharge 0
its battery, and thus assumes that there is a charger at the The displacement x, which is the distance from the reference
base station. As the bus stops at the base station for a long point, is a function of the time at which the vehicle reaches
time, there is no reason not to install a charging infrastructure that point. As it is a monotonically increasing function, there
[24][25]. exists an inverse function. If we know a displacement value,
we can evaluate the temporal value
B. System design variables and parameters −1
t = fts (x). (2)
As mentioned above, a battery size and power track
allocation are two sets of design variables. The vehicles in the With the function in (2), we can easily calculate the times
system each have an identical battery size, which is denoted when a bus is at the start and end points of the ith power

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

track, denoted by tsi and tei , respectively.

We denote E(t) as the amount of energy in the battery at
time t. The energy level should be within a lower and upper
limits during travel

E low ≤ E(t) ≤ E high , (3)

where E low and E high are the lower and upper limits of the
battery level, respectively. These values have the following
relationship

E high = h × E max
E low = l × E max (4)
0 < l < h < 1, Fig. 6. Battery cycle-to-failure versus the depth of discharge gap [34]

where l and h are the lower and upper limit coefficients.

This assumption is based on the mechanism of the charging The ratio of released energy from the battery, the depth of
controller in the OLEV. The range between E high and E low discharge (DOD) is presented as
is set not to damage the battery and these values are set by
battery providers. DOD(t) = 1−SOC(t). (8)
The degree of energy fluctuation over time depends on the The trajectory DOD(t) represents the battery’s discharges
energy consumption and supply amounts. We define Pb (t) and is used to predict the battery degradation and the battery
as an energy consumption rate of the battery. This quantity life.
depends on the energy usage rate of the vehicle determined
by the velocity trajectory, road gradient, and use of peripheral
C. Battery life prediction model
devices such as the bus air-conditioner. The analytical model
for estimating the energy demand for the OLEV system and The battery life prediction in EV system has been the most
numerical validation of the model is explained in [23]. important research area to make decisions for commercializ-
We define Ps (t) as an energy supply to the battery at time ing EVs [30][31]. The battery life prediction model adopted
t. This quantity is actually the charging rate of the battery and in this paper accumulates losses of battery life by discharge
depends on power track allocation. We assume that E(t) is cycles [30][32]. This model is called fatigue model which is
linearly proportional to the charging time as long as the level one of the cycle counting approaches [30][33]. Although the
of the charge is between E high and E low . This assumption fatigue model is a high level approximation for the battery
is based on the mechanism of the charging controller in EV. life, it is widely used due to its simplicity and succinctness
If the vehicle is servicing on the route along power track [32]. Since we focus on the economic design of the dynamic
in place, the battery energy level will increase. Note that charging EV system considering the battery life, the simple
although the entire spectrum of the charging rate is nonlinear, fatigue model is suitable for predicting the battery life with
the charging rate within E high and E low , which represent the quick computations. The fatigue model assumes that every
actual utilizing range, is almost a straight line [23]. Therefore, discharge affects to the battery degradation. That is, the DOD
the charging rate, denoted by ps , is assumed to be constant. gap δ of the discharge cycle causes the battery degradation.
This linear energy charging and discharging assumption is To quantify the battery degradation caused by δ, the cycle-
commonly used in EV modelings [28][29]. Also, results to-failure which indicates the maximum number of cycles
of a numerical experiment supporting the assumption are versus a specific δ is used. Figure 6 shows the experimental
presented in [23]. Then, cycle-to-failure data from [34]. The curve fitted model is also
 plotted over the data in the figure, and the model is
ps , if a vehicle is on a power track −26.46·δ
Ps (t) = (5) +3.689·e−0.334·δ
0 , otherwise. fn (δ) = 102.302·e . (9)

The fluctuation of the battery energy level in this case is The loss of life (LoL) of the battery caused by a specific δ
described as can be estimated by the equation 1/fn (δ). The battery moves
toward to the end of its life, the LoL approaches to 1. In
dE(t)
= −Pb (t) + Ps (t). (6) the dynamic charging EV system, a number of discharges
dt can be shown in a DOD trajectory of one service. Thus, the
Generally, the battery energy state is expressed as ratio, the total losses of battery life by one service, LoLs is linearly
state of charge (SOC), and it is represented as accumulated as
Z t X 1
1 LoLs = . (10)
SOC(t) = SOC(0)− max {−Pb (τ )+Ps (τ )} dτ. (7) fn (δj )
E 0 ∀j

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

where δj is the DOD gap of j th discharge cycle in the DOD install a power track ($/each), and cv is the installation cost
trajectory of vehicle’s service. Therefore, the battery life can per length of each power track ($/meter). The cost of the
be represented by the maximum number of services with a battery is linearly proportional to the unit size. This linear
new battery, and denoted by Ns . cost approximation is widely used in industry [35]. Therefore,
1 the first term represents the total battery cost in the system. If
Ns = (11) the total service life of the dynamic charging EV system is Tt ,
LoLs
the battery will be replaced in accordance with its lifetime Tb .
If a vehicle serves Nd operations in a day, then the maximum The number of batteries required for each vehicle to cover
expected lifetime of its battery Tb is represented by the total service life is obtained by Tt divided by Tb . The
Ns second term and third term represent the total fixed cost and
Tb = . (12)
Nd the variable cost of the power tracks, respectively.
Let T be the time at which the vehicle reaches the base
V. E CONOMIC MODEL OF DYNAMIC CHARGING EV
station, and Lmax
pt be the maximum length of a power track.
We construct the optimization model in this section. Sup- Then, the economic model (EM) economically allocating
pose that a vehicle is about to leave the ith power track. With the power tracks and determining the battery size is then
our notation, this point is indicated as xei . The time at which described as follows:
the vehicle is at this point is indicated by tei . The level of
energy at tei is E(tei ). The vehicle now departs from the ith
power track, and continuously travels along a route with no N
Tt X
EM: Minimize k· cb ·E max · xei −xsi


power track until it reaches the next power track, at point +cf ·N +cv ·
xsi+1 . Again, the moment at which the vehicle arrives at the Tb i=1
next power track is tsi+1 . The following constraint must be subject to
satisfied. Z tsi+1 E(0) = E high (16)
E(tei ) − Pb (t)dt ≥ E low . (13) Z tsi+1
tei E(tei )− Pb (t)dt ≥ E low , for i = 0,. . . ,N (17)
tei
The first term in (13) is the energy level when the vehicle  Z tei+1
leaves the ith power track, and the second term indicates the
E(tei+1 ) e
Pb (t)dt+ps · tei+1 −tsi+1


= min E(ti )−
amount of energy consumed while the vehicle is traveling tei
the area with no power track. Therefore, (13) represents the 
amount of energy in the battery when the vehicle arrives , E high , for i = 0, . . . , N −1 (18)
at the beginning of the (i + 1)th power track. The energy
level needs to be greater than the lower energy limit of the E(T ) ≥ E low (19)
battery. In other words, the power track needs to be installed te0= 0, tsN +1 =T (20)
in such a way that the energy level is greater than the lower Z t∗
i
∗
limit. We now evaluate the amount of energy at the end of xi = v(t)dt, for ∗ = {s, e}, and i = 1, . . . , N (21)
0
the (i + 1)th power track, that is, the energy supply and
xsi < xei , for i = 1, . . . , N (22)
consumption between xei and xei+1 . The level of energy in
the battery at xei is E(tei ). Thus, the level of energy at xei+1 xei < xsi+1 , for i = 1, . . . , N −1 (23)
is described as follows: xei − xsi ≤ Lmax
pt , for i = 1, . . . , N (24)
s
0 ≤ xi ≤ L, for i = 1, . . . , N (25)
tei+1
( Z )
min E(tei )− Pb (t)dt+ps · tei+1 −tsi+1 , E high . (14)

0 ≤ xei ≤ L, for i = 1, . . . , N (26)
max
tei E ≥0 (27)
Note that the battery energy level when the vehicle reaches
the end of the (i + 1)th power track should be less than the In the economic model, the constraints in (16) – (19)
upper limit of the battery, E high . If the energy amount is represent the battery energy state constraints. Specifically,
more than the upper limit, then the energy level would be the inequality in (17) represents the minimum energy re-
E high at xei+1 . quirement constraint, and (18) is the energy level when the
The objective function for the optimization model is de- vehicle leaves the end point of each power track. The physical
fined as follows: allocation constraints of the power tracks are defined in (20)
N – (24). The design variables and their boundary conditions
max Tt X
xei −xsi ,


k·cb ·E · + cf ·N + cv · (15) are defined in (25) – (27). Notations and definitions used in
Tb i=1 the economic model are summarized in Table II
where k is the number of vehicles, cb is the battery cost The EM can be solved using the meta-heuristics method,
per unit of energy capacity ($/kWh), cf is the fixed cost to and its solution algorithm is given in [23].

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

TABLE II TABLE V
N OTATION FOR THE ECONOMIC MODEL R ESULTS OF THE ECONOMIC MODEL (EM) WHEN k = 18

Notation Description Result Value

k Number of vehicles operating on the route Total cost $9,375,469
cb Battery cost per unit size ($/kWh) Battery size 50 kWh
cf Fixed cost of a power track ($/each) Battery life (service number) 1,809
cv Variable cost of a power track ($/m) Number of power tracks 7
N Number of power tracks Total power track length 617 m
xsi Start point of the ith power track 1st power track 1,970 - 1,974
xei End point of the ith power track 2nd power track 5,059 - 5,073
tsi Time at point xsi 3rd power track 6,503 - 6,636
tei Time at point xei 4th power track 16,792 - 16,825
E max Battery size (kWh) 5th power track 23,084 - 23,113
E(t) Energy state at time t 6th power track 24,708 - 24,740
E high Upper bound of the battery energy level 7th power track 26,923 - 27,295
E low Lower bound of the battery energy level Duration on power tracks 1,212 sec
Lmax
pt Maximum length of a power track
L Total length of the service route
T Total trip time for a sercive
Tt Total service life of the system are allocated with a 50 kWh battery. Each battery enables to
Tb Battery lifetime
run 1,809 services. The length of each power track is 4 m, 14
Pd (t) Energy consumption rate of the battery at t
ps Energy supply rate m, 133 m, 33m, 29 m, 32m, and 372 m, respectively. Thus,
the total length of power tracks is 617 m. Although the total
length of power tracks is only about 2 % of the total travel
distance, the total travel time on the power tracks is about
VI. N UMERICAL CASE STUDY 20 minutes which is about 16 % of the total service time.
In this section, we present a numerical case using op- Note that in the result, the power tracks are located around
erational data collected from the OLEV system currently bus stations including the turning point at which each bus
operating public electric transit buses in Gumi City. The stops for a while. Since the energy supplied from the power
system parameters including the vehicle specification and track is proportional to the time the bus spends on the power
the cost parameters are listed in Table III and Table IV, track, it is efficient to install the power tracks around the bus
respectively. These costs are close to the actual values of stops.
the OLEV system in Gumi City. The optimal total cost of the system is $9,375,469. As
we have already noted in Section V, the total cost includes
TABLE III both power track installation cost and the total battery cost.
S YSTEM PARAMETERS OF THE OLEV SYSTEM IN G UMI C ITY The power track installation costs $658,614 to install 7 power
Notation Description Value tracks with 617 m long. Whenever batteries end their life after
ps Energy supply rate (kW) 80 certain period, they have to be replaced by new batteries. For
h The SOC upper bound 0.800 10 years of operation, the total battery cost is $8,716,855.
l The SOC lower bound 0.500
Lmax
pt Maximum power track length (m) 1,000
L Total route length (m) 34,448 B. Comparison with stationary charging
Tt Total service life (years) 10
Economics of the dynamic charging EV system can be
clarified by comparing it with economics of the stationary
charging EV system. Stationary charging is the other charg-
TABLE IV ing method of wireless charging EVs in which vehicles need
C OST PARAMETERS to stop to charge the battery for a certain duration. For the
Notation Description Cost ($)
Gumi City case, there is a turning point where buses stop by
cb Battery cost per unit size 800
for 10 minutes in the middle of the service route. If a power
cf Fixed cost of a power track 50,000 track (or other charging devices) for stationary charging is
cv Variable cost of a power track 500 installed at the turning point, the buses utilize it when the
buses arrive at that point. If the stationary charging method
were used for the Gumi OLEV system, each bus in the
stationary charging should equip at least 100 kWh battery
A. Basic result which is twice larger than the optimal battery size of the
We first find a basic result of EM for the numerical case dynamic charging EV.
study. The result for k = 18, which is one of the target number Figure 7 shows the total cost comparison of the stationary
of buses in the route, is described in Table V. Table V shows charging system and dynamic charging system when 18
the optimal economic decisions. Note that seven power tracks buses are operated for 10 years. The total cost of stationary

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

Fig. 9. Sensitivity analysis for decreasing power track cost

Fig. 7. Total cost comparison of a stationary and dynamic charging EV
systems
the other hand, if 20 buses are in service, 6 power tracks
with totally 740 m are needed to be installed with only 50
kWh size of the battery. Since the total battery cost becomes
the dominant cost in the system as the number of buses
increases, the economic model tries to find a solution in
which the buses can operate with smaller batteries. As the
battery size decreases, a greater energy supply is needed.
When the number of power tracks increases, the total length
of the power tracks also increases. This makes sense, as
the number of buses increases, which means an increasing
utilities of power tracks, the cost benefit of the power tracks
also increases.
Figure 9 illustrates the cost per vehicle and battery life
Fig. 8. Sensitivity analysis for increasing the number of buses as the power track cost is reduced. The x-axis indicates
the percent by which the power track cost is reduced from
its current cost. For example, a value of 90% indicates a
charging EV system is $11,837,737, whereas the dynamic 10% cost reduction from the current cost (100%). The bars
charging EV system needs $9,375,469 as we noted above. It indicate the cost per vehicle and how that cost is broken
is expected to reduce the total cost by 20.8 % when the down into the battery cost and power track cost. These costs
dynamic charging system would be implemented, instead are referenced on the left y-axis. The red line represents how
of the stationary charging system. Although the dynamic the battery life responds to the cost changes.
charging system requires more investments on the power
The rationale behind reducing the power track cost is as
tracks than the stationary charging system, more cost saving
follows. It is well known that the manufacturing cost of
can be achieved from the smaller batteries with extended
discrete products decreases as volume increases. According
battery life.
to Willcox [36], airplane production costs decrease by 10%
each time when the production volume doubles. Assum-
C. Sensitivity analyses ing similar behavior in bus production, in view of similar
Figure 8 shows the system behavior as the number of production volumes, if Korea were to produce 1,000 EV
operating buses increases from 1 to 20 in increments of 1. buses a year, the cost of the buses would reduce by 65%.
The optimal battery size is shown in the upper part of the Although this scenario is somewhat optimistic and the cost
graph. The lower part of the graph indicates the total length of the batteries is less likely to decrease, based on recent
of the installed power tracks. The numbers in the lower part battery costs, the cost of the power tracks is highly likely
of the graph indicate the optimal number of the power tracks. to decrease. The current power tracks are custom made and
For example, if 10 buses operate on the route, it is economical prototyping accounts for a significant portion of the total
to use 50 kWh battery for each bus and to install four power power track cost. A slight volume increase would reduce the
tracks with a total length of around 540 m along the route. cost. As the power tracks are manufactured in a country with
In Figure 8, if only one bus is in the system, it is most low labor cost, there is a good chance that the cost can be
economical to install only one power track at the turning reduced. Therefore, we analyze the case in which the power
point, but to operate with one large battery in the bus. On cost decreases. Note that in Figure 9, the total cost does

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

not decrease significantly, as we use the case with 18 buses. [2] M. Yilmaz and P. T. Krein, “Review of battery charger topologies,
Eighteen batteries are needed, so the battery cost accounts for charging power levels, and infrastructure for plug-in electric and hybrid
vehicles,” Power Electronics, IEEE Transactions on, vol. 28, no. 5, pp.
a significant part of the total cost. Notice that as the cost drops 2151–2169, 2013.
from 70% to 60% of the current cost, the battery life increases [3] S. Choi, J. Huh, W. Lee, and C. Rim, “Asymmetric coil sets for wireless
substantially. As the cost of the power track decreases, the stationary ev chargers with large lateral tolerance by dominant field
analysis,” Power Electronics, IEEE Transactions on, vol. 29, no. 12,
optimal solution is to install more and longer power tracks, pp. 6406–6420, 2014.
giving the batteries frequent charges and improving battery [4] S. Lukic and Z. Pantic, “Cutting the cord,” IEEE Electrification
life. The additional cost benefits of the extended battery life Magazine, vol. 1, no. 1, pp. 57–64, September 2013.
[5] “An electric bus that wirelessly charges at each
such as environmental effects and battery disposal costs are stop,” Smart Planet, November 2012. [Online]. Avail-
not considered. If these indirect benefits are considered, the able: http://www.smartplanet.com/blog/bulletin/an-electric-bus-that-
cost benefit would be greater. wirelessly-charges-at-each-stop/
[6] “Korean electric buses go wireless,” The Wall
Street Journal, August 2013. [Online]. Avail-
VII. C ONCLUSION AND FUTURE RESEARCH able: http://blogs.wsj.com/korearealtime/2013/08/08/korean-electric-
In this paper, we introduced an economic model for the buses-go-wireless/tab/video/
[7] A. Esser, “Contactless charging and communication for electric vehi-
dynamic charging EV system with the consideration of the cles,” Industry Applications Magazine, IEEE, vol. 1, no. 6, pp. 4–11,
battery life. For dynamic charging EV system, both power 1995.
track allocation and battery size are the key design variables [8] D. A. Pedder, A. D. Brown, and J. A. Skinner, “A contactless
electrical energy transmission system,” Industrial Electronics, IEEE
determining the system’s performance. They also account for Transactions on, vol. 46, no. 1, pp. 23–30, 1999.
a significant portion of the total cost for dynamic charging [9] M. L. Kissin, G. Covic, and J. Boys, “Estimating the output power of
EVs. We validated this economic model with the real-world flat pickups in complex ipt systems,” in Power Electronics Specialists
Conference, 2008. PESC 2008. IEEE. IEEE, 2008, pp. 604–610.
data collected from the OLEV system operating in Gumi [10] J. Shun, S. Shin, Y. Kim, S. Ahn, S. Lee, G. Jung, S.-J. Jeon, and D.-H.
City, South Korea, which is one of the first commercialized Cho, “Design and implementation of shaped magnetic-resonance-based
dynamic charging EV systems. Furthermore, we compared wireless power transfer system for roadway-powered moving electric
vehicles,” IEEE Transactions on Industrial Electronics, vol. 61, no. 3,
the dynamic charging EV system with the stationary charging pp. 1179–1192, 2013.
EV system. We found that although more investments on dy- [11] W. Lee, J. Huh, S. Choi, X. Thai, J. Kim, E. Al-Ammar, M. El-
namic charging infrastructure would occur at the installation Kady, and C. Rim, “Finite-width magnetic mirror models of mono
stage, much more cost saving can be accomplished from the and dual coils for wireless electric vehicles,” Power Electronics, IEEE
Transactions on, vol. 28, no. 3, pp. 1413–1428, 2013.
extended battery life. It implies that the dynamic charging [12] N. Omar, M. A. Monem, Y. Firouz, J. Salminen, J. Smekens,
EV system is beneficial to both the reduction of the battery O. Hegazy, H. Gaulous, G. Mulder, P. Van den Bossche, T. Coosemans
size and extension of the battery life. Sensitivity analyses et al., “Lithium iron phosphate based battery–assessment of the aging
parameters and development of cycle life model,” Applied Energy, vol.
were also conducted to figure out the optimal designs for 113, pp. 1575–1585, 2014.
both increasing scale of the EVs and discounting the power [13] J. Bolger, F. Kirsten, and L. Ng, “Inductive power coupling for an
track costs. We concluded that the dynamic charging system electric highway system,” in Vehicular Technology Conference, 1978.
28th IEEE, vol. 28. IEEE, pp. 137–144.
is more beneficial for large scale EV systems and the battery [14] H. H. Wu, J. T. Boys, and G. A. Covic, “An ac processing pickup for
costs could be significantly reducible when the power track ipt systems,” Power Electronics, IEEE Transactions on, vol. 25, no. 5,
cost drops. pp. 1275–1284, 2010.
[15] M. Budhia, G. A. Covic, and J. T. Boys, “Design and optimization
Although there are some commercialized dynamic charg- of circular magnetic structures for lumped inductive power transfer
ing EVs, they are still in the very early stages of the com- systems,” Power Electronics, IEEE Transactions on, vol. 26, no. 11,
mercialization and their system designs have been focused pp. 3096–3108, 2011.
[16] C.-S. Wang, O. H. Stielau, and G. A. Covic, “Design considerations for
on proving the reliability of the systems rather than on a contactless electric vehicle battery charger,” Industrial Electronics,
economic values. Once these systems have been proven to IEEE Transactions on, vol. 52, no. 5, pp. 1308–1314, 2005.
be reliable by a few commercialized versions, the economic [17] G. A. Covic, J. T. Boys, M. L. Kissin, and H. G. Lu, “A three-
phase inductive power transfer system for roadway-powered vehicles,”
issues discussed in this paper will become critical. Industrial Electronics, IEEE Transactions on, vol. 54, no. 6, pp. 3370–
For the future research, it would be more challenging to 3378, 2007.
optimally design the dynamic charging EV system consider- [18] T. Imura and Y. Hori, “Maximizing air gap and efficiency of magnetic
ing stochastic driving cycle. In this paper, it is assumed that resonant coupling for wireless power transfer using equivalent circuit
and neumann formula,” Industrial Electronics, IEEE Transactions on,
all vehicles follow predetermined driving cycle. We did not vol. 58, no. 10, pp. 4746–4752, 2011.
consider the traffic and driving uncertainties which largely [19] S. Aldhaher, P. Luk, and J. Whidborne, “Electronic tuning of mis-
affect to the power and energy requirements of vehicles. The aligned coils in wireless power transfer systems,” Power Electronics,
IEEE Transactions on, vol. 29, no. 11, pp. 5975–5982, 2014.
economic model with uncertain driving behavior is a good [20] N. Suh, D. Cho, and C. Rim, “Design of on-line electric vehicle (olev),”
candidate for the future research. in Plenary Presentation, CIRP Design Conference, Nantes, France,
April 2010.
R EFERENCES [21] J. Huh, S. Lee, W. Lee, G. Cho, and C. Rim, “Narrow-width inductive
power transfer system for online electrical vehicles,” Power Electron-
[1] J. G. Hayes, “Battery charging systems for electric vehicles,” in ics, IEEE Transactions on, vol. 26, no. 12, pp. 3666–3679, 2011.
Electric Vehicles-A Technology Roadmap for the Future (Digest No. [22] S. Choi, B. Gu, S. Lee, W. Lee, J. Huh, and C. Rim, “Generalized
1998/262), IEE Colloquium on. IET, 1998, pp. 4–1. active emf cancel methods for wireless electric vehicles,” Power

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2424712, IEEE Transactions on Power Electronics

10

Electronics, IEEE Transactions on, vol. 29, no. 11, pp. 5770–5783, Young Jae Jang received his Ph.D. degree in me-
2014. chanical engineering from Massachusetts Institute
[23] Y. D. Ko and Y. J. Jang, “The optimal system design of the online of Technology (MIT) in 2007 and a double M.S.
electric vehicle utilizing a wireless power transmission technology,” degree in mechanical engineering and operations
IEEE Transactions on Intelligent Transportation Systems, vol. 14, research from MIT in 2001. He received a B.S
no. 3, pp. 1255–1265, 2013. degree in aerospace engineering from Boston Uni-
[24] Y. J. Jang, E. S. Suh, and J. W. Kim, “System architecture and versity in 1997. He is currently an Assistant Pro-
mathematical models of electric transit bus system utilizing wireless fessor in the Industrial and Systems Engineering
power transfer technology,” Systems Journal, IEEE, Early Access Department at the Korea Advanced Institute of
article, 2015. Science and Technology (KAIST), South Korea.
[25] Y. J. Jang, S. Jeong, and Y. D. Ko, “System optimization of the on- He is also affiliated with the Cho Chun Shik
line electric vehicle operating in a closed environment,” Computers & Graduate School of Green Transportation at KAIST. His current research
Industrial Engineering, vol. 80, pp. 222–235, 2015. includes the stochastic modeling of complex systems and optimizations in
[26] Z. Pantic, S. Bai, and S. M. Lukic, “Inductively coupled power transfer transportation and logistics systems. He has been involved in the KAIST
for continuously powered electric vehicles,” in Vehicle Power and On-Line Electric Vehicle (OLEV) project developing and commercializing
Propulsion Conference, 2009. VPPC’09. IEEE. IEEE, 2009, pp. the innovative wireless charging electric vehicle. The project was recognized
1271–1278. as the“50 Best Innovations of 2010” by TIME Magazine. His role in the
[27] Y. J. Jang and Y. D. Ko, “Optimal design of the on-line electric vehicle project is to develop the optimal energy management system to integrate
with various optimization methodology - modeling analysis and cases,” the vehicle system to the road traffic network. He has been published
Korea Advanced Institute of Science and Technology, Tech. Rep., numerous technical papers out of the OLEV technology. His recent work
2011. ”The Optimal Economic Design of the Wireless Powered Transportation
[28] S. Mohrehkesh and T. Nadeem, “Toward a wireless charging for battery System” was selected as the Best Paper in the 2013 International Conference
electric vehicles at traffic intersections,” in 2011 14th International on Intelligent Manufacturing and Logistics Conference. Dr. Jang is the
IEEE Conference on Intelligent Transportation Systems, 2011. Technical Program Chair of the 2014 IEEE Vehicular Technology Workshop
[29] A. Ashtari, EricBibeau, S. Shahidinejad, and T. Molinski, “Pev charg- on Emerging Technologies on Wireless Power. Before he joined KAIST, he
ing profile prediction and analysis based on vehicle usage data,” IEEE worked at Micron Technology, Inc., VA, USA, where as project manager,
TRANSACTIONS ON SMART GRID, vol. 3, no. 3, pp. 342–350, 2012. he led a global initiative to improve the efficiency of the manufacturing
[30] H. Wenzl, I. Baring-Gould, R. Kaiser, B. Y. Liaw, P. Lundsager, facilities located worldwide. While working at Micron, he was also involved
J. Manwell, A. Ruddell, and V. Svoboda, “Life prediction of batteries in activities supporting the company’s operational and strategic decision
for selecting the technically most suitable and cost effective battery,” making using quantitative modeling and analysis techniques.
Journal of power sources, vol. 144, no. 2, pp. 373–384, 2005.
[31] L. Lam and P. Bauer, “Practical capacity fading model for li-ion battery
cells in electric vehicles,” Power Electronics, IEEE Transactions on,
vol. 28, no. 12, pp. 5910–5918, 2013.
[32] D. U. Sauer and H. Wenzl, “Comparison of different approaches Dongsuk Kum received his Ph.D. degree in me-
for lifetime prediction of electrochemical systems – using lead-acid chanical engineering from the University of Michi-
batteries as example,” Journal of Power Sources, vol. 176, no. 2, pp. gan, Ann Arbor, in 2010. From 2010 to 2012, he
534–546, 2008. worked at the General Motors R&D Propulsion
[33] E. Schaltz, A. Khaligh, and P. O. Rasmussen, “Influence of bat- Systems Research Laboratory in Warren, MI as a
tery/ultracapacitor energy-storage sizing on battery lifetime in a fuel visiting research scientist. Since 2012, he has been
cell hybrid electric vehicle,” IEEE Transactions on Vehicular Technol- a faculty member at the Korea Advanced Institute
ogy, vol. 58, no. 8, pp. 3882–3891, 2009. of Science and Technology (KAIST) in Daejeon,
[34] W. C. Ltd., “Battery life(and death),” 2005. [Online]. Available: Korea. His research interests include modeling,
http://www.mpoweruk.com/life.htm control, and design of hybrid electric vehicles and
[35] J.-Q. Li, “Transit bus scheduling with limited energy,” Transportation autonomous driving systems.
Science, vol. Articles in Advance, no. 1, pp. 1–19, 2013.
[36] K. Willcox, “Cost analysis, mit course on aircraft systems engineer-
ing,” MIT, Tech. Rep., 2004.

Seungmin Jeong received his B.S. degree in

Industrial Engineering from Hanyang University
in 2011. He is currently pursuing the integrated
Master’s and Ph.D. degree program at the Korea
Advanced Institute of Science and Technology
(KAIST). His research topics include the mathe-
matical modeling of complex systems, system opti-
mizations, network flow, and logistics systems. He
has investigated in the project “Optimal Design and
Intelligent Operations for Public Mass Transporta-
tion Systems Utilizing Wireless Charging Electric
Vehicles” supported by the Basic Science Research Program through the
National Research Foundation of Korea (NRF). He has been published a
number of papers for the dynamic charging EV system design.

0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.