You are on page 1of 8

Design and Analysis of a 200 kW Dynamic

Wireless Charging System for Electric Vehicles


Lingxiao Xue, Veda Galigekere, Gui-jia Su, Rong Zeng, Mostak Mohammad
Emre Gurpinar, Shajjad Chowdhury, Omer Onar
Building and Transportation Science Division
Oak Ridge National Laboratory
Oak Ridge, USA
xuel@ornl.gov, galigekerevn@ornl.gov, sugj@ornl.gov, zengr@ornl.gov, mohammadm@ornl.gov,
gurpinare@ornl.gov, chowdhuryms@ornl.gov, onaroc@ornl.gov

Abstract—Dynamic wireless charging of electric vehicles can in the range of 300 to 400 V. Both inductor-capacitor-capacitor
significantly alleviate or eliminate range anxiety while reducing (LCC) and series tuning are widely used in wireless power
the required on-board battery capacity. To achieve an electric transfer (WPT) systems and can be used on both transmitter
vehicle charging balance with minimal infrastructure costs,
power transfer levels around 200 kW are required. In this paper, and receiver sides. LCC tuning on the primary side leads to
a system architecture with a modularized power electronics and a constant current source behavior on the primary-side coil
optimized power transfer couplers was adopted to enable efficient at resonance, and the coil current is independent from load
power transfer. A DC/DC converter was used for secondary- and coupling coefficient variations [5]. LCC tuning provides a
side charging control of the battery. The system performance constant current into the rectifier independent of the DC-link
was validated in an experimental setup at 120 kW with 91.31%
efficiency from the DC input on the primary side to the vehicle voltage, leading to a runaway over-voltage situation that will
battery. eventually destroy the rectifier and/or capacitor. The only way
Index Terms—dynamic wireless charging, electric vehicles, to prevent this situation is to physically disconnect the vehicle
transportation electrification coupler from the resonant network. This results in its own set
of challenges and additional design constraints placed on the
I. I NTRODUCTION vehicle coupler to safely de-energize the unloaded resonant
Dynamic wireless charging of electric vehicles (EVs) can network.
significantly alleviate or eliminate range anxiety while reduc- In contrast, a series tuning method exhibits a voltage
ing the required on-board battery capacity [1]–[3]. Current source–like behavior at resonance. Therefore, the peak voltage
state-of-the-art light-duty dynamic wireless EV charging pro- is limited. With series tuning, the vehicle DC-link capacitor
totypes are limited in power level to approximately 20 kW. will charge to the peak voltage induced in the vehicle coupler,
This limitation requires a significant portion of the roadway to which is approximately 1,019 V. With 1,200 V rated semicon-
be electrified to achieve a charge-balanced mode of operation. ductors and DC-link capacitors, this over-voltage condition is
For light-duty vehicles to achieve a charge-balancing mode well within the design margins. Therefore, series tuning at the
of operation on primary roadways with minimal infrastructure secondary side was selected.
costs, power transfer levels in the range of 200 kW are required Without secondary control, the high-power DWPT creates
[4]. This paper demonstrates the analysis and design consider- grid power pulsations [7] and charges the battery with a
ations of a dynamic wireless power transfer (DWPT) system pulsating current. To address these issues, a DC/DC converter
architecture to efficiently transfer power, and the experimental was added after the rectifier stage to conduct the battery
validation results are presented. charging control. The overall power architecture is shown in
Fig. 1.
II. OVERALL P OWER A RCHITECTURE D ESIGN
The overall DWPT system assumes an 800 V DC voltage III. W IRELESS C OUPLER D ESIGN AND A NALYSIS
input, either from a rectified output from the AC grid or
directly fed by a local DC grid, and charges a light-duty battery For stationary wireless charging applications, WPT coils
have been designed with the objective of optimizing the power
This manuscript has been authored by UT-Battelle, LLC, under contract transfer efficiency over a small range of misalignment. For the
DE-AC05-00OR22725 with the US Department of Energy (DOE). The US dynamic charging scenario, the net energy transferred over
government retains and the publisher, by accepting the article for publication,
acknowledges that the US government retains a nonexclusive, paid-up, irre- the entire overlap window as the vehicle coil moves over the
vocable, worldwide license to publish or reproduce the published form of this primary coil must be considered. Additionally, for dynamic
manuscript, or allow others to do so, for US government purposes. DOE will charging applications, the ground or primary coil must be
provide public access to these results of federally sponsored research in ac-
cordance with the DOE Public Access Plan (http://energy.gov/downloads/doe- embedded on a roadway lane and withstand much harsher
public-access-plan). conditions than in stationary parking lot charging.
Fig. 1. Overall power electronics architecture of the high-power dynamic wireless charging system.

A. Circular and Double-D Comparison coil current to a 750 A peak was determined to be 1.5 µH
from Eq. 1.
Because of the naturally large misalignment in the direction
of the vehicle travel, shielding of the vehicle chassis is an P
M= (1)
important consideration. Initial design work indicated that the wo Ipri Isec
circular coupler topology had better performance character-
where wo = 2πfo , and Ipri and Isec are the primary and
istics than the double-D topology. However, the shielding
secondary coil currents, respectively.
requirements greatly affected the performance of the circu-
lar coupler. The initial circular coupler power dropped by
more than 50% when the vehicle-length shielding was added. C. Finite Element Analysis
The double-D coupler power was almost unaffected by the
The transmitter dimensions are 1.78 m in length and 0.85
shielding. Design modifications to accommodate shielding
m in width. The receiver coil is 0.66 m in length and width.
significantly decreased the circular coupler advantages. Even
Figure 2 shows the transmitter and receiver coil dimensions,
after solving the shielding issue, the sensitivity of the circular
along with the finite element analysis (FEA) simulation data
topology was still a concern regarding the many uncertainties
points. Figure 3 shows the mutual inductance M acquired
in a practical system. Therefore, the robustness of double-D
through FEA simulations over the range of x = −152 cm
topology took priority over the reduced mass of the circular
to x = 152 cm and y = −76 cm to y = 76 cm, where
topology.
the origins of transmitter and receiver coils are aligned at
There are additional practical reasons to prefer the double- x = y = 0. The air gap between the transmitter and receiver
D topology for high-power DWPT systems. To reduce the coil was considered to be 25.4 cm.
coupler voltage drop, multiple litz wires can be run in parallel
As depicted in Fig. 3, the mutual inductance between the
to form the coupler winding. In the circular topology, the
transmitter and receiver coil was the highest at the central limb
innermost wire would have a lower inductance than the
(variations in x when y = 0). It reduced with the increase
outermost wire. This inductance imbalance would cause a
in misalignment along the y-direction. The FEA simulation
current imbalance that would reduce the system performance.
value of mutual inductance was close to 1.7 µH, which meets
In the double-D topology, the multiple litz wires are transposed
the design target of mutual inductance at 1.5 µH. The output
during the winding process, resulting in better inductance and
power capability could therefore be determined and is plotted
current balance between wires.
in Fig. 4, assuming a vehicle DC-link voltage of 800 V and
primary-side LCC tuning and secondary-side series tuning.
B. Optimized Double-D Coil Design Thermal constraints also limit the number of ampere-turns
An elongated transmitter coil comparable to the length of on the ground coupler and maximum flux density in the
a light-duty EV was developed, along with a 200 kW vehicle ferrites. Figure 5 shows the simulated steady-state temperature
receiver optimized to fit underneath a light-duty EV. The width distribution on the pavement surface and through the roadway
of the transmitters had an upper limit of the lane width. This cross-section assuming an ambient temperature of 55°C. The
allows for deploying the DWPT by square precast concrete design was constrained to limit the maximum hot-spot tem-
blocks or provides accommodation for expansion joints if perature of 100°C. This simulation assumes 100% up-time for
the couplers were directly buried. All ground- and vehicle- the ground coupler.
side electromagnetic coupler designs were constrained to fit Figure 6 shows a map of the critical regions for electromag-
within the smallest area bounded by the tires of the vehicle. netic field emissions just above the surface of the pavement.
The surface of the ground couplers were assumed to be Half of the 12 ft wide lane is depicted. The transmitter meets
buried 3 in. beneath the surface of the pavement. Assuming a the ICNIRP and SAEJ2954 specifications for pacemaker safety
minimum ground clearance of 6 in. for the vehicle couplers, of 15 µT rms at the lane boundary. This indicates that high-
the total coupler surface-to-surface air gap was 9 in. The power DWPT systems can be safely implemented in dedicated
mutual inductance required to transfer Po = 200 kW at a highway lanes without causing safety concerns for adjacent
nominal operating frequency of fo = 85 kHz and limit the vehicles.
Fig. 2. 200 kW WPT transmitter and receiver coil dimensions with FEA simulation points.

Fig. 5. Thermal analysis of DWPT system embedded in concrete with ambient


temperature of 55°C for half of the lane (6 ft width).

Fig. 3. Variation of mutual inductance as a function of x and y misalignment


for the 200 kW DWPT system.

Fig. 6. Map of critical electromagnetic field emission regions on the surface


of the roadway; (green) below 15 µTrms, (yellow) between 15 and 27 µTrms;
and (red) above 27 µTrms.

IV. M ODULAR P OWER E LECTRONICS S UBSYSTEM D ESIGN


The power architecture of the proposed high-power WPT
system as shown in Fig. 1 and comprises multiple H-bridges.
The power electronic subsystems and their corresponding
Fig. 4. Variation of output power as a function of x and y misalignment for operating conditions for the 200 kW+ dynamic EV charging
the 200 kW DWPT system.
system are as follows:
1) Primary-side inverter: 800 V input, 200 kW, 85 kHz frequency above 1 MHz is required. The B58033I9505M001
2) Secondary-side rectifier: 800 V input, 200 kW, 85 kHz 5 µF, 1,300 V PLZT capacitor with a self-resonant frequency
3) Secondary-side DC/DC converter: 800 V input, 400 V greater than 1 MHz and high ripple current capability was
output, 200 kW, 50-100 kHz selected as the decoupling capacitor for the power electronics
These conditions infer that the key requirement is a high- building block.
voltage (800 V), high-frequency (85 kHz) inverter, rectifier,
and DC-DC converter. In the proposed approach, the systems
were devolved into optimally sized power electronic building
blocks based on the available semiconductor modules.
A. Power Semiconductor Module Selection
Based on the operating conditions, three possible SiC-
phase leg modules were identified: CAB450M12XM3 (1,200
V, 450 A), CAB400M12XM3 (1,200 V, 400 A), and
CAS325M12HM2 (1,200 V, 325 A). Based on simulation and
analytical studies, the primary-side high-frequency inverter
has the worst-case operating conditions, so loss analysis was
performed for the three modules considering a single mod- Fig. 7. Simulated decupling capacitor current, voltage, and spectrum of the
ule per leg and two paralleled modules per leg. TABLE I current for safe 250 kW operation.
shows that having two modules in parallel leads to slightly
better efficiencies. More importantly, with the single-module
C. Modular High-Current Printed-Circuit-Board Design
approach, each of the modules was close to the thermal
power dissipation limit, rendering the single-phase approach A modular approach was developed to construct the power
infeasible. CAS325M12HM2 (1,200 V SiC-phase leg module electronics, as shown in Fig. 8. The building block consisted
from Wolfspeed with 5 nH package inductance) was selected of the low-inductance SiC power module CAS325M12HM2,
for the power electronics building block (phase leg module) vertical commutation-loop bus bars, built-in electromagnetic
for the primary-side high-frequency inverter, secondary-side interference shielding, localized gate drivers, a liquid-cooled
high-frequency rectifier, and secondary-side DC-DC converter. cold plate, and low-inductance PLZT decoupling capacitors.
One building block contained one H-bridge (or two half
bridges). The switching current and electromagnetic interfer-
TABLE I ence noise were well confined within each building block,
L OSS ANALYSIS OF POWER SEMICONDUCTOR MODULES FOR 200 K W+ so higher power can be easily achieved by horizontal daisy-
OPERATION AT 85 K H Z
chain connections of the building blocks at the DC+ and DC-
CAB450M12XM3 CAB400M12XM3 CAS325M12HM2 terminals.
Modules in parallel 1 2 1 2 1 2
Total loss (kW) 4.144 3.879 2.571 2.025 2.299 1.821 Figure 9 also shows the clean switching waveforms achieved
Inverter efficiency 0.983 0.984 0.989 0.991 0.990 0.992 by low loop inductance. The resulting voltage overshoot was
measured to be 85 V over the 750 V DC link. This overshoot
provides an ample margin to the maximum device rating of
B. Decoupling Capacitor Selection 1,200 V. The heavy-copper PCB-based bus bar temperature
High-frequency operation (85 kHz) resulted in compact only rose to 40.7°C when the single building block delivered
WPT couplers, and faster switching transition times of 85 kW.
SiC modules resulted in high-efficiency operation. A high- Figure 10 shows the computer-aided design model of the
performance decoupling capacitor should be optimally se- primary-side high-frequency power electronic inverter de-
lected and placed as near to the module as possible with signed using two of the designed building blocks. The overall
minimal impedance between the capacitor and the module to dimensions are 305.6 × 290 × 71 mm, and the volume is
supply the energy required for the high-frequency switching 6.29 L, resulting in a power density of 31.78 kW/L. The
(85 kHz) with short switching transitions (less than 50 ns). theoretically estimated efficiency is 99%. The primary-side
A detailed simulation study was conducted to estimate the inverter block was practically realized using two building
magnitude and frequency component of the current ripple that blocks in an H-bridge setup. Four building blocks were used
the decoupling capacitor should be able to supply to achieve to build the secondary-side rectifier and DC/DC converter.
efficient and safe 85 kHz operation. As indicated in Fig. 7, Four half-bridges were used as the rectifier, and the other
the decoupling capacitor must be able to supply a large ripple four half-bridges were configured as a four-phase interleaved
current (around 300 A peak) for 250 kW operation. The figure buck converter for charging control. Figure 11 shows the
also shows considerable current at the higher frequency range, computer-aided design model of the designed secondary-side
as well including 5.5 A at 1.02 MHz. This current indicates power electronics, including the high-frequency rectifier and
that a high-frequency operation capacitor with a self-resonant the four-phase DC-DC converter with the filter components.
Fig. 11. Rectifier/DC-DC combo unit using four building blocks.

receiver coil was very small, and the DC capacitor voltage on


the on-board side and the output power were almost zero. As
Fig. 8. PCB bus bar layout and building block assembly. The PCB bus bar the vehicle approached, the DC capacitor voltage increased
can symmetrically support two power modules (left and right). The building following the increase of the mutual inductance. When the
block assembly only shows a single power module connection. DC capacitor voltage reached 450 V, the secondary DC/DC
converter activated, and the output power ramped up and stabi-
lized at the targeted value of approximately 200 kW. After the
vehicle drove away, the DC capacitor voltage decreased; when
the voltage was below 450 V, the secondary DC/DC converter
stopped transferring power to the battery. Thus, during the
charging process, the operation status of the secondary DC/DC
converter only depended on the DC capacitor voltage, and
no wireless communication was required to coordinate the
operations of the primary-side inverter and secondary DC/DC
converter.
Fig. 9. Measured metal-oxide-semiconductor field-effect transistor voltage
and current for a double pulse test with 750 V DC link voltage and gate
signal set to obtain a turn-off current of 244 A.

More details of this SiC building block implementation for the


receiver-side unit can be found elsewhere [6].

Fig. 10. DC/AC inverter units using two building blocks.


Fig. 12. Primary-side DC/AC inverter input and output voltage (V) and current
(A) over a dynamic charging event on one 200 kW transmitter coil.
V. S YSTEM -L EVEL S IMULATION FOR DWPT C HARGING
The coupler parameters obtained from FEA simulations
VI. E XPERIMENTAL R ESULTS
were lumped as impedance networks that vary according to
different alignment conditions. The impedance networks were The transmitter and receiver coils shown in Fig. 2 were
integrated to the PLECS simulation platform to facilitate the assembled using 3C95 ferrite bars, AWG 2–equivalent Litz
simulation of the dynamic wireless charging system. The wire for the transmitter, AWG 4–equivalent Litz wire for
simulation results are shown in Fig. 14 and Fig. 15. The phase the receiver, and additively manufactured wire holders. The
shift angle of the primary inverter was set to a constant of 180°, transmitter coil will be assembled, and MasterFlow concrete
and the secondary DC-DC converter was used to track the will be poured over it in a controlled environment to build
charging current reference. At the beginning of the simulation, a precast concrete block embedded with transmitter coil. The
the mutual inductance between the transmitter coil and the prototypes of 200 kW ground-side transmitter and receiver
coils are shown in Fig. 16 and Fig. 17, respectively. The
long lead length shown in Fig. 16 was designed so that it
can traverse the length of the electrified lane to connect to
the ground-side power electronics assembly placed by the
roadway. The full system setup is shown in Fig. 18.

Fig. 13. Zoomed-in waveform of the primary-side DC/AC inverter input and
output voltage (V) and current (A) over a dynamic charging event on one
200 kW transmitter coil.
Fig. 16. Fabricated 200 kW transmitter coil.

Fig. 14. Secondary-side DC-DC converter input voltage, rectifier input current
(A), DC link capacitor current (A), filter inductor current (A), and output Fig. 17. Fabricated 200 kW receiver coil.
power (W). Note the smooth output power transfer profile.

Fig. 18. Test setup of the 200 kW WPT system.

Fig. 15. Zoomed-in waveforms of the secondary-side DC-DC converter The WPT system (excluding the DC/DC converter) was first
input voltage, rectifier input current (A), DC link capacitor current (A), filter tested separately; the tested waveforms are shown in Fig. 19.
inductor current (A), and output power (W). The two H-bridges for the DC/AC inverter achieved good
current sharing (194.7 and 176.6 A) at the primary side. The
receiver side achieved excellent current sharing (140.0 and
139.8 A). At 700 V input, the WPT system received 736 V
and 186 kW after the rectifier and achieved 93.27% efficiency.

Fig. 21. Power analyzer screenshot of the complete system test at 120 kW
including both WPT system and post DC/DC regulator.

The complete system was experimentally validated by testing


at 120 kW, which measured 91.31% efficiency.
Fig. 19. WPT waveforms tested at 700 V input and 736 V output at 186 kW.
ACKNOWLEDGMENT
The DC/DC converter was tested to the full 200 kW power. This research used resources available at the Power Elec-
The waveforms are shown in Fig. 20. The interleaved inductor tronics and Electric Machinery Research Facility located at the
current was clearly observed with a 90°phase shift between National Transportation Research Center, a US Department of
adjacent phases. Current balancing among all four phases Energy Energy Efficiency Renewable Energy User Facility
was very well maintained. The measured efficiency under this operated by Oak Ridge National Laboratory (ORNL). The
condition was 98.67%. authors would like to thank David Smith (ORNL), Burak
Ozpineci (ORNL), and Lee Slezak (US Department of Energy)
for their support of this work; Randy Wiles and Jon Wilkins for
building the coils and other mechanical fixtures; Larry Seiber
for building the power electronics hardware; and Cliff White
for assembling the DSP platform. This work is completed
under the US Department of Energy Vehicle Technologies
Office–funded High Power and Dynamic Charging of EVs
project.

R EFERENCES
[1] A. Foote, O. C. Onar, S. Debnath, M. Chinthavali, B. Ozpineci, and D. E.
Smith, “Optimal sizing of a dynamic wireless power transfer system
for highway applications,” in 2018 IEEE Transportation Electrification
Conference and Expo (ITEC), pp. 1-6, 2018.
[2] A. N. Azad, A. Echols, V. A. Kulyukin, R. Zane, and Z. Pantic, “Anal-
Fig. 20. DC/DC stage waveforms tested at 750 V input and 400 V output at
ysis,optimization, and demonstration of a vehicular detection system
182 kW.
intended for dynamic wireless charging applications,” IEEE Transactions
on Transportation Electrification, vol. 5, no. 1, pp. 147-161, 2019.
After combining the WPT and DC/DC converter, the overall [3] A. Zaheer, M. Neath, H. Z. Z. Beh, and G. A. Covic, “A dynamic EV
system tests were conducted. The test results at 120 kW are charging system for slow moving traffic applications,” IEEE Transac-
tions on Transportation Electrification, vol. 3, no. 2, pp. 354-369, 2017.
shown in Fig. 20; the overall efficiency reached 91.31%. [4] O. C. Onar, S. L. Campbell, L. E. Seiber, C. P. White, and M.
Chinthavali, “Vehicular integration of wireless power transfer systems
VII. C ONCLUSION and hardware interoperability case studies,” in 2016 IEEE Energy
Conversion Congress and Exposition (ECCE), pp. 1-8, 2016.
In this paper, we presented the design of a 200 kW dynamic [5] R. Tavakoli, A. Jovicic, N. Chandrappa, R. Bohm, and Z. Pantic,
wireless charging system for EVs. The experimental results “Design of a dual-loop controller for in-motion wireless charging of an
electric bus,” in 2016 IEEE Energy Conversion Congress and Exposition
show over 93.27% WPT efficiency from the DC input to 186 (ECCE), pp. 1-8, 2016.
kW DC output (after the rectifier). Considering a post–battery [6] L. Xue, V. Galigekere, E. Gurpinar, G. -j. Su and O. Onar, “Mod-
charging control stage, which was implemented by a validated ular Design of Receiver Side Power Electronics for 200 kW High
Power Dynamic Wireless Charging System,” 2021 IEEE Transportation
98.67% efficient four-phase buck converter, the overall system Electrification Conference and Expo (ITEC), pp. 744–748, 2021. DOI:
efficiency is expected to be approximately 92% at 200 kW. 10.1109/ITEC51675.2021.9490095.
[7] R. Zeng, V. P. Galigekere, O. C. Onar, and B. Ozpineci, “Grid Integration
and Impact Analysis of High-Power Dynamic Wireless Charging System
in Distribution Network,” in IEEE Access, vol. 9, pp. 6746–6755, 2021.
DOI: 10.1109/ACCESS.2021.3049186.

You might also like