4
`
`IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 3, NO. 1, MARCH 2015
`
`Wireless Power Transfer for Electric
`Vehicle Applications
`
`Siqi Li, Member, IEEE, and Chunting Chris Mi, Fellow, IEEE
`
`Abstract— Wireless power transfer (WPT) using magnetic
`resonance is the technology which could set human free from
`the annoying wires. In fact, the WPT adopts the same basic
`theory which has already been developed for at
`least 30
`years with the term inductive power transfer. WPT tech-
`nology is developing rapidly in recent years. At kilowatts
`power level, the transfer distance increases from several mil-
`limeters to several hundred millimeters with a grid to load
`efficiency above 90%. The advances make the WPT very
`attractive to the electric vehicle (EV) charging applications in
`both stationary and dynamic charging scenarios. This paper
`reviewed the technologies in the WPT area applicable to
`EV wireless charging. By introducing WPT in EVs, the obstacles
`of charging time, range, and cost can be easily mitigated. Battery
`technology is no longer relevant in the mass market penetration
`of EVs. It is hoped that researchers could be encouraged by
`the state-of-the-art achievements, and push forward the further
`development of WPT as well as the expansion of EV.
`(EV),
`Index Terms— Dynamic
`charging,
`electric vehicle
`inductive power transfer (IPT), safety guidelines, stationary
`charging, wireless power transfer (WPT).
`
`I. INTRODUCTION
`
`F OR energy, environment, and many other reasons, the
`
`electrification for transportation has been carrying out for
`many years. In railway systems, the electric locomotives have
`already been well developed for many years. A train runs on a
`fixed track. It is easy to get electric power from a conductor rail
`using pantograph sliders. However, for electric vehicles (EVs),
`the high flexibility makes it not easy to get power in a similar
`way. Instead, a high power and large capacity battery pack is
`usually equipped as an energy storage unit to make an EV to
`operate for a satisfactory distance.
`Until now,
`the EVs are not so attractive to consumers
`even with many government incentive programs. Government
`subsidy and tax incentives are one key to increase the market
`share of EV today. The problem for an electric vehicle is
`nothing else but
`the electricity storage technology, which
`requires a battery which is the bottleneck today due to its
`unsatisfactory energy density, limited life time and high cost.
`
`Manuscript received February 2, 2014; revised April 6, 2014; accepted
`April 18, 2014. Date of publication April 23, 2014; date of current ver-
`sion January 29, 2015. Recommended for publication by Associate Editor
`J. M. Miller.
`S. Li is with the Department of Electrical Engineering, Kunming Uni-
`versity of Science and Technology, Kunming 650500, China (e-mail:
`lisiqi@kmust.edu.cn).
`and Computer
`C. C. Mi
`is with the Department of Electrical
`Engineering, University of Michigan, Dearborn, MI 48128 USA (e-mail:
`chrismi@umich.edu).
`Color versions of one or more of the figures in this paper are available
`online at http://ieeexplore.ieee.org.
`Digital Object Identifier 10.1109/JESTPE.2014.2319453
`
`In an EV, the battery is not so easy to design because of
`the following requirements: high energy density, high power
`density, affordable cost, long cycle life time, good safety,
`and reliability, should be met simultaneously. Lithium-ion
`batteries are recognized as the most competitive solution to
`be used in electric vehicles [1]. However, the energy density
`of the commercialized lithium-ion battery in EVs is only
`90–100 Wh/kg for a finished pack [2].1 This number is so poor
`compared with gasoline, which has an energy density about
`12 000 Wh/kg. To challenge the 300-mile range of an internal
`combustion engine power vehicle, a pure EV needs a large
`amount of batteries which are too heavy and too expensive.
`The lithium-ion battery cost is about 500$/kWh at the present
`time. Considering the vehicle initial investment, maintenance,
`and energy cost, the owning of a battery electric vehicle will
`make the consumer spend an extra 1000$/year on average
`compared with a gasoline-powered vehicle [1]. Besides the
`cost issue, the long charging time of EV batteries also makes
`the EV not acceptable to many drivers. For a single charge,
`it takes about one half-hour to several hours depending on
`the power level of the attached charger, which is many times
`longer than the gasoline refueling process. The EVs cannot
`get ready immediately if they have run out of battery energy.
`To overcome this, what the owners would most likely do is
`to find any possible opportunity to plug-in and charge the
`battery. It really brings some trouble as people may forget
`to plug-in and find themselves out of battery energy later on.
`The charging cables on the floor may bring tripping hazards.
`Leakage from cracked old cable, in particular in cold zones,
`can bring additional hazardous conditions to the owner. Also,
`people may have to brave the wind, rain, ice, or snow to plug-
`in with the risk of an electric shock.
`The wireless power transfer (WPT) technology, which can
`eliminate all the charging troublesome, is desirable by the
`EV owners. By wirelessly transferring energy to the EV, the
`charging becomes the easiest
`task. For a stationary WPT
`system, the drivers just need to park their car and leave. For a
`dynamic WPT system, which means the EV could be powered
`while driving; the EV is possible to run forever without a stop.
`Also, the battery capacity of EVs with wireless charging could
`be reduced to 20% or less compared to EVs with conductive
`charging.
`Although the market demand is huge, people were just
`wondering whether the WPT could be realized efficiently at
`
`1Although lithium ion battery can achieve up to 200 Wh/kg for individual
`cells, the battery pack requires structure design, cooling, and battery manage-
`ment systems. The over energy density of a battery pack is much lower than
`the cell density.
`
`2168-6777 © 2014 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.
`
`
`
`Apple v. GUI Global Products
`IPR2021-00471, 472, 473
`GUI Ex. 2028
`
`

`

`LI AND MI: WPT FOR EV APPLICATIONS
`
`5
`
`a reasonable cost. The research team from MIT published a
`paper in Science [3], in which 60 W power is transferred at
`a 2-m distance with the so called strongly coupled magnetic
`resonance theory. The result surprised the academia and the
`WPT quickly became a hot research area. A lot of interesting
`works were accomplished with different kinds of innovative
`circuit, as well as the system analysis and control [4]–[9]. The
`power transfer path can even be guided using the domino-form
`repeaters [10], [11]. In order to transfer power more efficiently
`and further,
`the resonant frequency is usually selected at
`MHz level, and air-core coils are adopted.
`When the WPT is used in the EV charging, the MHz
`frequency operation is hard to meet the power and efficiency
`criteria. It is inefficient to convert a few to a few hundred
`kilowatts power at MHz frequency level using state-of-the-
`art power electronics devices. Moreover, air-core coils are too
`sensitive to the surrounding ferromagnetic objects. When an
`air-core coil is attached to a car, the magnetic flux will go
`inside the chassis causing high eddy current loss as well as
`a significant change in the coil parameters. To make it more
`practical in the EV charging, ferrite as a magnetic flux guide
`and aluminum plate as a shield are usually adopted in the coil
`design [12]. With the lowered frequency to less than 100 kHz,
`and the use of ferrite, the WPT system is no different from
`the inductive power transfer (IPT) technology which has been
`developed for many years [13]–[39]. In fact, since the WPT is
`based on the nonradiative and near-field electromagnetic, there
`is no difference with the traditional IPT which is based on
`magnetic field coupling between the transmitting and receiving
`coils. The IPT system has already been proposed and applied
`to various applications, such as underwater vehicles [32]–[34],
`mining systems [16], cordless robots in automation production
`lines [36]–[39], as well as the charging of electric vehicles
`[13], [14], [25]–[27].
`Recently, as the need of EV charging and also the progress
`in technology,
`the power transfer distance increases from
`several millimeters to a few hundred millimeters at kilowatts
`power level [12], [14], [40]–[60]. As a proof-of-concept of a
`roadway inductively powered EV, the Partners for Advance
`Transit and Highways (PATH) program was conducted at
`the UC Berkeley in the late 1970s [14], [54]. A 60 kW,
`35-passanger bus was tested along a 213 m long track with
`two powered sections. The bipolar primary track was supplied
`with 1200 A, 400 Hz ac current. The distance of the pickup
`from the primary track was 7.6 cm. The attained efficiency was
`around 60% due to limited semiconductor technology. During
`the last 15 years, researchers at Auckland University have
`focused on the inductive power supply of movable objects.
`Their recent achievement in designing pads for the stationary
`charging of EV is worth noting. A 766 mm × 578 mm
`pad that delivers 5 kW of power with over 90% efficiency
`for distances about 200 mm was reported [48], [55]. The
`achieved lateral and longitudinal misalignment tolerance is
`250 and 150 mm, respectively. The knowledge gained from
`the on-line electric vehicle (OLEV) project conducted at the
`Korea Advanced Institute of Science and Technology (KAIST)
`also contributes to the WPT design. Three generations of
`OLEV systems have been built: a light golf cart as the first
`
`Fig. 1. Typical wireless EV charging system.
`
`generation, a bus for the second, and an SUV for the third.
`The accomplishment of the second and the third is noteworthy:
`60 kW power transfer for the buses and 20 kW for the
`SUVs with efficiency of 70% and 83%, respectively; allowable
`vertical distance and lateral misalignment up to 160 mm and
`up to 200 mm, respectively [56], [57]. In the United States,
`more and more public attention was drawn to the WPT since
`the publication of the 2007 Science paper [3]. The WiTricity
`Corporation with technology from MIT released their WiT-
`3300 development kit, which achieves 90% efficiency over a
`180 mm gap at 3.3 kW output. Recently, a wireless charging
`system prototype for EV was developed at Oak Ridge National
`Laboratory (ORNL) in the United States. The tested efficiency
`is nearly 90% for 3 kW power delivery [53]. The research
`at the University of Michigan–Dearborn achieved a 200 mm
`distance, 8 kW WPT system with dc to dc efficiency as high
`as 95.7% [61]. From the functional aspects, it could be seen
`that the WPT for EV is ready in both stationary and dynamic
`applications. However, to make it available for large-scale
`commercialization, there is still abundant work to be done on
`the performance optimization, setup of the industrial standards,
`making it more cost effective, and so on.
`This paper starts with the basic WPT theory, and then
`gives a brief overview of the main parts in a WPT system,
`including the magnetic coupler, compensation network, power
`electronics converter, study methodology, and its control, and
`some other issues like the safety considerations. By introduc-
`ing the latest achievements in the WPT area, we hope the
`WPT in EV applications could gain a widespread acceptance
`in both theoretical and practical terms. Also, we hope more
`researchers could have an interest and make more brilliant
`contributions in the developing of WPT technology.
`
`II. FUNDAMENTAL THEORY
`A typical wireless EV charging system is shown in Fig. 1.
`It includes several stages to charge an EV wirelessly. First,
`the utility ac power is converted to a dc power source
`by an ac to dc converter with power factor correction.
`
`Apple v. GUI Global Products
`IPR2021-00471, 472, 473
`GUI Ex. 2028
`
`

`

`6
`
`IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 3, NO. 1, MARCH 2015
`
`Fig. 2. General two-coil WPT system.
`
`where I1 and I2 are the root mean square value and ϕ12 is
`the phase difference between ˙I1 and ˙I2. The active power
`
`transfer from the primary side to the secondary side can be
`expressed as
`
`(3)
`
`(4)
`
`(cid:3) ˙I
`
`∗2
`
`∗1
`
`2
`
`P12 = ωM I1 I2 sin ϕ12.
`The system shown in Fig. 2 can transfer active power in
`both directions. In the analysis below, we assume the power
`is transferred from L1 to L2. When ϕ12 = π/2, which means
`˙I1 leads ˙I2 by a quarter cycle, the maximum power can be
`transferred from L1 to L2.
`The total complex power goes into the two-coil system is
`˙S = ˙S1 + ˙S2
`(cid:3) ˙I
`˙I2 + ωM ˙I1
`˙I1 + ωM ˙I2
`(cid:2)
`(cid:2)
`+ j
`= j
`ωL2
`ωL1
`(cid:3)
`(cid:2)
`+ 2M I1 I2 cos ϕ12
`= j ω
`+ L2 I 2
`L1 I 2
`.
`1
`Therefore, the total reactive power goes into the two-coil
`system is
`
`(cid:2)
`
`L1 I 2
`1
`
`2
`
`(cid:3)
`
`.
`
`(5)
`
`Q = ω
`+ 2M I1 I2 cos ϕ12
`+ L2 I 2
`For a traditional transformer, the reactive power represents
`the magnetizing power. Higher magnetizing power brings
`higher copper and core loss. To increase the transformer
`efficiency, the ratio between the active power and reactive
`power should be maximized. The ratio is defined by
`
`(cid:4)(cid:4)(cid:4)(cid:4)
`
`(cid:4)(cid:4)(cid:4)(cid:4)
`
`(cid:6)
`
`k
`
`L1
`L2
`
`I1
`I2
`
`L2
`L1
`
`I2
`I1
`
`2
`
`x
`
`f (ϕ12) = |P12|
`|Q| =
`ωM I1 I2 sin ϕ12
`+ ωL2 I 2
`+ 2ωM I1 I2 cos ϕ12
`ωL1 I 2
`(cid:5)
`(cid:5)
`1
`1−cos2 ϕ12
`1−cos2 ϕ12
`=
`= k
`+(cid:6)
`x + 1
`+ 2k cos ϕ12
`+2k cos ϕ21
`(6)
`
`Then, the dc power is converted to a high-frequency ac to
`drive the transmitting coil through a compensation network.
`Considering the insulation failure of the primary side coil,
`a high-frequency isolated transformer may be inserted between
`the dc-ac inverter and primary side coil for extra safety and
`protection. The high-frequency current in the transmitting coil
`generates an alternating magnetic field, which induces an ac
`voltage on the receiving coil. By resonating with the secondary
`compensation network, the transferred power and efficiency
`are significantly improved. At last, the ac power is rectified
`to charge the battery. Fig. 1 shows that a wireless EV charger
`consists of the following main parts:
`1) the detached (or separated, loosely coupled) transmitting
`and receiving coils. Usually, the coils are built with
`ferrite and shielding structure, in the later sections, the
`term magnetic coupler is used to represent the entirety,
`including coil, ferrite, and shielding;
`2) the compensation network;
`3) the power electronics converters.
`The main difference between a wireless charger and a
`conventional conductive or wired charger is that a transformer
`is replaced by a set of loosely couple coils. To give a quick
`idea of the WPT principle, the coil and the compensation
`network are pulled out separately, as shown in Fig. 2, where
`L1 represents the self-inductance of the primary side transmit-
`ting coil and L2 represents the self-inductance of the receiving
`coil; ˙I1 and ˙I2 are the current in the two coils;
`˙U12 is the
`˙U21 is the voltage in the primary
`
`voltage in the secondary coil that is induced by the current
`in the primary side coil.
`coil that is induced by the current in secondary side coil due
`to coupling, or mutual inductance between the primary and
`secondary coils. S1 and S2 are the apparent power goes into
`L1 and L2, respectively. S3 and S4 are the apparent power
`provided by the power converter. S12 and S21 represent the
`apparent power exchange between the two coils. The form of
`the compensation network is not specified. The characteristics
`of the compensation network will be discussed later.
`As shown in Fig. 2, neglecting the coil resistance and
`magnetic losses, we can calculate the simplified form of
`exchanged complex power from L1 to L2
`˙S12 = − ˙U12
`= − j ωM ˙I1
`˙I
`˙I
`= ωM I1 I2 sin ϕ12 − j ωM I1 I2 cos ϕ12
`˙S21 = − ˙U21
`= − j ωM ˙I2
`˙I
`˙I
`= −ωM I1 I2 sin ϕ12 − j ωM I1 I2 cos ϕ12
`
`where π/2 < ϕ12 < π
`x =
`
`(cid:7)
`
`L1
`L2
`
`I1
`I2
`
`> 0
`
`k is the coupling coefficient between L1 and L2.
`f (ϕ12), we solve the
`To achieve the maximum value of
`following equations:
`f (ϕ12) = 0,
`
`∂ 2
`∂ 2ϕ12
`
`f (ϕ12) < 0
`
`(7)
`
`∂
`∂ϕ12
`
`(1)
`
`(2)
`
`∗2
`
`∗1
`
`∗2
`
`∗1
`
`Apple v. GUI Global Products
`IPR2021-00471, 472, 473
`GUI Ex. 2028
`
`

`

`LI AND MI: WPT FOR EV APPLICATIONS
`
`7
`
`(cid:5)
`
`L1 L2 I1
`
`(10)
`
`=
`
`We have
`U12 = I2(R2 + RLe) = ωM I1 = ωk
`where R2 is the secondary winding resistance and RLe is the
`equivalent load resistance.
`By defining the quality factor of
`the
`two coils,
`Q1 = ωL1/R1, Q2 = ωL2/R2, the transferred efficiency can
`be expressed as
`η=
`I 2
`2 RLe
`RLe
`
`1 R1 + I 22 R2 + I 22 RLe
`(11)
`(R2+RLe )2
`+ R2 + RLe.
`I 2
`
`k2 Q1 Q2 R2
`By defining a = RLe/R2, we obtain the expression of
`efficiency as a function of a
`η(a) =
`
`sin ϕ12 =
`
`,
`
`x
`
`(8)
`
`.
`
`(cid:3)2
`
`x
`
`and the solutions are
`(cid:8)(cid:9)
`(cid:9)(cid:10)1 − 4k2
`cos ϕ12 = − 2k
`x + 1
`(cid:2)
`x + 1
`When k is close to 1, it is a traditional transformer. In this
`case, if ˙I2 is an induced current by ˙I1, x will be close to 1.
`Thus, cos ϕ12 ≈ −1. The phase difference between ˙I1 and ˙I2
`f (ϕ12) is
`is nearly 180°. While for WPT, k is close to 0.
`maximized at sin ϕ12 = 1, at which point the transferred power
`is also maximized. The phase between ˙I1 and ˙I2 is around 90°
`
`instead of 180°. Hence we can see the difference between the
`tightly and the loosely coupled coils.
`The degree of coupling affects the design of the compensa-
`tion network. Taking the series–series topology as an example,
`there are two ways to design the resonant capacitor. One way
`is design the capacitor to resonate with the leakage inductance
`[46], [62] which could achieve a higher f (ϕ12). Another way
`is to resonate with the coil self-inductance [27], [41], [63]
`which could maximum the transferred power at a certain coil
`current. When the coupling is tight with a ferrite, like k > 0.5,
`it is important to increase f (ϕ12) to achieve better efficiency.
`In this case, resonate with the coil self inductance, which
`makes ϕ12 = π/2 and lowers f (ϕ12), is not recommended.
`Otherwise the magnetizing loss may significantly increase.
`When the capacitor resonates with the leakage inductance, it
`is like the leakage inductance is compensated. This makes the
`transformer perform as a traditional one and increases f (ϕ12).
`However, the overall system does not work at a resonant
`mode. When the coupling is loose, like k < 0.5, which is
`the case for the EV wireless charging, usually the capacitor
`is tuned with the self inductance to make the system working
`at a resonate mode to achieve maximum transferred power
`at a certain coil current. In this case, most of the magnetic
`field energy is stored in the large air gap between the two
`coils. The hysteresis loss in the ferrite is not so relative
`to the magnetizing power. However, the loss in the copper
`wire is proportional to the square of the conducting current.
`To efficiently transfer more power at a certain coil current, the
`
`induced current ˙I2 should lag ˙I1 by 90°. Since the induced
`voltage ˙U12 on the receiving coil lags ˙I1 by 90°, ˙U12 and ˙I2
`resistive characteristic seen from ˙U12 at the frequency of ˙I1.
`At the meanwhile, the primary side input apparent power S3
`should be minimized. At cos ϕ12 = 0, the complex power ˙S1 is
`˙S1 = j ωL1 I 2
`+ ωM I1 I2.
`(9)
`the primary side compensation network should
`Ideally,
`cancel the reactive power and make S3 = ω0 M I1 I2, where
`ω0 is the resonant frequency. From the above analysis, we see
`for a certain transferred power, it is necessary to make the
`secondary side resonant to reduce the coil volt-ampere (VA)
`rating, which reduces the loss in the coils; and to make the
`primary side resonant to reduce the power electronics converter
`VA rating, which reduces the loss in the power converter.
`Therefore, we transfer power at the magnetic resonance.
`With the above analysis, we can calculate the power transfer
`efficiency between the two coils at the resonant frequency.
`
`should be in phase. The secondary side should have a pure
`
`1
`
`+2
`a+ 1
`+ 1
`a
`k2 Q1 Q2
`The maximum efficiency is obtained by solving the follow-
`ing equations:
`
`.
`
`(12)
`
`1
`+ 1
`
`a
`
`(cid:3)2
`
`∂ 2
`∂ 2a
`
`η(a) < 0.
`
`(13)
`
`η(a) = 0,
`∂
`∂a
`The maximum efficiency
`ηmax =
`k2 Q1 Q2
`1 + (cid:5)
`(cid:2)
`1 + k2 Q1 Q2
`(cid:3)1/2.
`is achieved at aη max = (cid:2)
`1 + k2 Q1 Q2
`In [64], the maximum efficiency is also derived based on
`several different kinds of compensation network. The results
`are identical and accord with the above results. The analysis
`here does not specify a particular compensation form. It can be
`regarded as a general formula to evaluate the coil performance
`and estimate the highest possible power transfer efficiency.
`In EV wireless charging applications, the battery is usu-
`ally connected to the coil through a diode-bridge rectifier.
`Most of the time,
`there is some reactive power required.
`The reactive power can be provide by either the coil or the
`compensation network like a unit-power-factor pickup. The
`battery could be equivalent to a resistance Rb = Ub/Ib, where
`Ub and Ib is the battery voltage and current, respectively.
`If the battery is connected to the rectifier directly in a series-
`series compensation form, the equivalent ac side resistance
`could be calculated by Rac = 8/π 2 · Rb. Thus, a battery
`load could be converted to a resistive load. The Rac equation
`is different for different battery connection style, like with
`or without dc/dc converter, parallel or series compensation.
`Most of the time, the equivalent Rac could be derived. Some
`typical equivalent impendence at the primary side is given in
`paper [42]. By calculating the equivalent ac resistances, the
`above equations could also be applied to a battery load with
`rectifier.
`For stationary EV wireless charging, the coupling between
`the two coils is usually around 0.2. If both the sending and
`receiving coils have a quality factor of 300, the theoretical
`maximum power transfer efficiency is about 96.7%. More
`efficiency calculations under different coupling and quality
`factors are shown in Fig. 3.
`
`Apple v. GUI Global Products
`IPR2021-00471, 472, 473
`GUI Ex. 2028
`
`

`

`8
`
`IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 3, NO. 1, MARCH 2015
`
`Fig. 3. Theoretical maximum transfer efficiency between two coils.
`
`III. MAGNETIC COUPLER DESIGN
`
`To transfer power wirelessly, there are at least two magnetic
`couplers in a WPT system. One is at the sending side, named
`primary coupler. The other is at the receiving side, named
`pickup coupler. Depending on the application scenarios, the
`magnetic coupler in a WPT for an EV could be either a
`pad or a track form. For higher efficiency, it is important
`to have high coupling coefficient k and quality factor Q.
`Generally, for a given structure, the larger the size to gap
`ratio of the coupler is, the higher the k is; the thicker the wire
`and the larger the ferrite section area is, the higher the Q is.
`By increasing the dimensions and materials, higher efficiency
`can be achieved. But this is not a good engineering approach.
`It is preferred to have higher k and Q with the minimum
`dimensions and cost. Since Q equals ωL/R, high frequency is
`usually adopted to increase the value of Q. The researchers at
`Massachusetts Institute of Technology (MIT) used a frequency
`at around 10 MHz and the coil Q value reached nearly
`1000 [3]. In high power EV WPT applications, the frequency
`is also increased to have these benefits. In Bolger’s early
`design, the frequency is only 180 Hz [13]. A few years later,
`a 400 Hz frequency EV WPT system was designed by System
`Control Technology [14]. Neither 180 Hz nor 400 Hz is high
`enough for a loosely coupled system. Huge couplers were
`employed in the two designs. Modern WPT system uses at
`least 10 kHz frequency [15]. As the technical progress of
`power electronics, 100 kHz could be achieved [65] at high
`power level. The WiTricity Company with the technology from
`MIT adopts 145 kHz in their design. In the recent researches
`and applications, the frequency adopted in an EV WPT system
`is between 20 and 150 kHz to balance the efficiency and
`cost. At this frequency, to reduce the ac loss of copper coils,
`Litz wire is usually adopted.
`Besides the frequency, the coupling coefficient k is sig-
`nificantly affected by the design of the magnetic couplers,
`which is considered one of the most important factors in a
`WPT system. With similar dimensions and materials, different
`coupler geometry and configuration will have a significant
`difference of coupling coefficient. A better coupler design
`may lead to a 50%–100% improvement compared with some
`nonoptimal designs [48].
`
`Fig. 4. Main flux path of double-sided and single-sided coupler. (a) Double-
`sided type. (b) Single-sided type.
`
`A. Coupler in the Stationary Charging
`
`In a stationary charging, the coupler is usually designed in
`a pad form. The very early couplers are just like a simple
`split core transformer [19], [38], [56]. Usually this kind of
`design could only transfer power through a very small gap.
`To meet the requirements for EV charging, the deformations
`from spilt core transformers and new magnetic coupler forms
`are presented for large gap power transfer [12], [31], [37], [42],
`[47]–[50], [66]–[71]. According to the magnetic flux dis-
`tribution area,
`the coupler could be classified as
`the
`double-sided and single-sided types. For the double-sided
`type,
`the flux goes to both sides of
`the coupler
`[12],
`[31],
`[67]. A flattened solenoid inductor
`form is pro-
`posed in [12] and [67]. Because the flux goes through
`the ferrite like through a pipe,
`it
`is also called a flux-
`pipe coupler. To prevent
`the eddy current
`loss in the
`EV chassis, an aluminum shielding is usually added which
`bring a loss of 1%–2% [12]. When the shielding is added,
`the quality factor of a flux-pipe coupler
`reduces from
`260 to 86 [48]. The high shielding loss makes the double-sided
`coupler not the optimal choice. For the single-sided coupler,
`most of the flux exists at only one side of the coupler. As
`shown in Fig. 4, the main flux path flows through the ferrite in
`a single-sided coupler. Unlike the double-sided coupler having
`half of the main flux at the back, the single-sided coupler only
`has a leakage flux in the back. This makes the shielding effort
`of a single-sided type much less.
`Two typical single-sided flux type pads are shown in Fig. 5.
`One is a circular unipolar pad [47]. Another one is a rectan-
`gular bipolar pad proposed by University of Auckland, which
`is also named DD pad [48]. Besides the mechanical support
`material, a single-sided pad is composed of three layers. The
`top layer is the coil. Below the coil, a ferrite layer is inserted
`for the purpose of enhancing and guiding the flux. At the
`bottom is a shielding layer. To transfer power, the two pads
`are put closed with coil to coil. With the shielding layer,
`most of the high-frequency alternating magnetic flux can be
`confined in the space between the two pads. A fundamental
`flux path concept was proposed in the flux pipe paper [67].
`
`Apple v. GUI Global Products
`IPR2021-00471, 472, 473
`GUI Ex. 2028
`
`

`

`LI AND MI: WPT FOR EV APPLICATIONS
`
`9
`
`Fig. 5.
`(b) DD pad.
`
`Two typical single-sided flux type pads.
`
`(a) Circular pad.
`
`The flux path height of a circular pad is about one-fourth of
`the pad’s diameter. While for a DD pad, the height is about
`half of the pad’s length. For a similar size, a DD pad has a
`significant improvement in the coupling. The charge zone for
`a DD pad could be about two times larger than a circular pad
`with similar material cost. The DD pad has a good tolerant in
`the y-direction. This makes the DD pad a potential solution
`for the dynamic charging when the driving direction is along
`with the y-axis. However, there is a null point for DD pad in
`the x-direction at about 34% misalignment [48]. To increase
`the tolerant in x-direction, an additional quadrature coil named
`Q coil is proposed to work together with the DD pad, which
`is called DDQ pad [48], [49], [68]. With a DDQ receiving
`pad on a DD sending pad, the charge zone is increased to five
`times larger than the circular configuration. As the additional
`Q coil in the receiver side, the DDQ over DD configuration
`uses almost two times copper compared with the circular
`one [48]. A variant of a DDQ pad, which is called a new
`bipolar pad, was also proposed by University of Auckland
`[49], [50]. By increasing the size of each D pad and having
`some overlap between the two D coils, the new bipolar pad
`could have a similar performance of a DDQ pad with 25%
`less copper. With all the efforts, at 200 mm gap, the cou-
`pling between the primary and secondary pads could achieve
`0.15–0.3 with an acceptable size for an EV. Referred to Fig. 3,
`at this coupling level, efficiency above 90% could possibly be
`achieved.
`
`B. Coupler in the Dynamic Charging
`The dynamic charging, also called the OLEVs [56] or
`roadway powered electric vehicles [14], is a way to charge
`the EV while driving. It is believed that the dynamic charging
`can solve the EVs’ range anxiety, which is the main reason
`limits the market penetration of EVs. In a dynamic charging
`system, the magnetic components are composed of a primary
`side magnetic coupler, which is usually buried under the road,
`and a secondary side pickup coil, which is mounted under an
`EV chassis. There are mainly two kinds of primary magnetic
`coupler in the dynamic charging. The first kind is a long track
`coupler [26], [31], [57], [70], [72]–[76]. When an EV with a
`
`Fig. 6. Top view of W-shape and I-shape track configuration.
`
`pickup coil is running along with the track, continues power
`can be transferred. The track can be as simple as just two
`wires [37], [77], or an adoption of ferrites with U-type or
`W-type [26], [56] to increase the coupling and power trans-
`fer distance. Further, a narrow-width track design with an
`I-type ferrite was proposed by KAIST [72], [73]. The dif-
`ferences between the W-type and I-type are shown in Fig. 6.
`For W-type configuration, the distribution area of the ferrite W
`determines the power transfer distance, as well as the lateral
`displacement. The total width of W-type should be about four
`times the gap between the track and the pickup coil. For I-type
`configuration, the magnetic pole alternates along with the road.
`The pole distance W1 is optimized to achieve better coupling at
`the required distance. The width of pickup coil W2 is designed
`to meet the lateral misalignment requirement. The relation
`between track width and transfer distance is decoupled and
`the track can be built at a very narrow form. The width for
`U-type and W-type is 140 and 80 cm, respectively [73]. For
`I-type, it could be reduced to only 10 cm with a similar power
`transfer distance and misalignment capacity. 35 kW power was
`transferred at a 200 mm gap and 240 mm displacement using
`the I-type configuration [73]. With the narrowed design, the
`construction cost could be reduced. Also, the track is far away
`from the road side, the electromagnetic field strength exposed
`to pedestrians can also be reduced.
`The problem of the track design is that the pickup coil only
`covers a small portion of the track, which makes the coupling
`coefficient very small. The poor coupling brings efficiency
`and electromagnetic interference (EMI) issues. To reduce the
`EMI issue, the track is built by segments [52], [70], [75]
`with a single power converter and a set of switches to power
`the track. The excitation of each segment can be controlled
`by the switches’ ON-OFF state. The electromagnetic field
`above the inactive segments is reduced significantly. However,
`there is always a high-frequency current flowing through the
`common supply cables, which lowers the system efficiency.
`The published systems efficiency is about 70%–80%, which
`is much lower than the efficiency achieved in the stationary
`charging.
`When each segment is short enough, the track becomes like
`a pad in the stationary charging, which is the other kind of
`the primary magnetic coupler. Each pad can be driven by an
`independent power converter. Thus, the primary pads can be
`selectively excited without a high-frequency common current.
`Also, the energized primary pad is covered by the vehicle. The
`electromagnetic field is shielded to have a minimum impact
`
`Apple v. GUI Global Products
`IPR2021-00471, 472, 473
`GUI Ex. 2028
`
`

`

`10
`
`IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 3, NO. 1, MARCH 2015
`
`TABLE I
`PRIMARY COMPENSATION CAPACITANCE
`
`Fig. 7. Four basic compensation topologies. (a) SS. (b) SP. (c) PS. (d) PP.
`
`to the surrounding environment. The efficiency and EMI
`performance could be as good as that in a stationary charging
`application. However, the cost to build a power converter for
`each pad is unaffordable. It is desired to use only one converter
`to drive a fe