Core Structure and Electromagnetic Field Evaluation in WPT Systems for Charging Electric Vehicles

Abstract

The electromagnetic field (EMF) in a wireless power transfer (WPT) system needs to couple inductively between the primary and the secondary coils through a large air gap, thus giving the system a loosely coupled characteristic. Therefore, magnetically permeable material must be employed to improve the coupling and reduce leakage magnetic flux. However, adding an iron core increases the weight and introduces core loss as a new factor. In this paper, a WPT system model using a lumped circuit model is introduced. Moreover, the relationship between the relative permeability and the coupling coefficient in addition to the core amount (core thickness) and core loss are discussed. Three cores structure named: pot, slotted, and shaped bars cores are investigated using finite element method (FEM) software. Inspired by the investigation results, a new core structure using optimum shaped bars is proposed, the EMF level for reducing core loss in high-power transfer systems and in order to mitigate the EMF exposure to humans is intensively evaluated. The proposed core succeeded in reducing EMF and core loss by about 44% and 30%, respectively. The FEM software and physical prototype were used to validate the proposed optimum core structure. Results showed that 3.5 kW power transferred through a 20 cm air gap with 96% system efficiency(coil–coil).

1. Introduction

An electric vehicle (EV), which uses an electric motor for propulsion, replaces conventional vehicles that utilize the internal combustion engine (ICE). EVs outperform traditional vehicles in terms of high efficiency, less routine maintenance, and zero tailpipe emissions. However, the lack of reliable means of recharging EV batteries infrastructure has been a major issue in delaying their practical implementation. To address this problem, wireless power transfer (WPT) charging systems have been adopted for charging EVs either on-road [1] or in parking spaces [2]. Besides EV charging, WPT has been rapidly developed for powering many electronic devices such as medical implants and portable electronic devices such as cellular telephones, laptop computers, and toothbrushes [3].

WPT charging systems have many advantages: they are cordless, need less maintenance, and are safe even in rain conditions. Moreover, it is less prone to vandalism when compared to conductive charging systems [4]. However, WPT systems have some drawbacks; for example, when the air gap or coils are laterally misaligned, the system characteristics change significantly [5]. In addition, in WPT systems, the transformer architecture is different than its counterpart. In a conductive charging system, the transformer coils have a common metal core that creates a strong time-varying electromagnetic flux (EMF) linkage between the primary and the secondary coils. Thus, it leads to increases in the voltage induced in the secondary coil and reduces system losses. In WPT systems, the two coils are normally separated by a relatively large air gap, so the time-varying EMF generated by the primary coil needs to be coupled inductively through the large air gap with the secondary coil, giving the WPT system loosely coupled characteristics. To improve the coupling when the air gap increases, magnetically permeable material must be used [6]. The pad structure and core material have significant effects on system performance and misalignment tolerance between transmitting (Tx) and receiving (Rx) coils. The charger pads must meet many requirements, such as a low weight, a low EMF, a high efficiency, and a high durability. Recently, various studies have concentrated on coils and core geometries to increase the coupling coefficient $\left(k\right)$ and the quality factor $\left(Q\right)$ and to reduce the leakage magnetic flux in WPT systems regardless of core loss and core weight [7,8]. However, there are some studies describing the use of the optimal core thickness with core and coil shape to enhance system performance and misalignment tolerance [9,10,11].

A comparison between Litz wire and copper foil shows that copper foil provides less uniform magnetic flux inside the ferrite core, which reduces core loss, while the Litz wire has low AC resistance and the coupling coefficient is slightly higher [12]. However, the study does not take into account the power transfer amount and the number of turns (in the case of Litz wire), which are critical factors in WPT pad design. In the same context, three types of coil shapes are investigated [13], including the helix coil, the planar spiral coil, and the square helix coil in four-coil WPT systems where all coils have the same area. The result showed that the planar spiral coil has the lowest efficiency compared to helical and square helix, coils. Moreover, the helical coil was reported to have widest band width and longest air gap among the three types. However, the helical coil is not suitable for EV charging pads as it will increase pad thickness. In the same context, a cylindrical ferrite core with a helical coil are used as a transmitter in [14] to improve system efficiency. However, this type of core increases the pad thickness and weight by using a large amount of core, therefore becoming unsuitable for EV charging.

In [15], the optimum stepped core for transmitting Tx and receiving Rx dipole-coils was investigated. The air gap used in their study was very large (5 m), which rendered the power transferred and system efficiency insufficient. Moreover, the proposed core structure was unsuitable for EV charging systems. In [16,17,18], E-type, I-type, ultra-slim S-type, U-type, and W-type core structures were introduced and tested in a rail power supply for powering an on-line EV. Comparison between ferromagnetic sheet and copper sheet with the Tx coil was made in [19], revealing that use of a ferromagnetic sheet can increase the electromagnetic flux density generated by the Tx coil, whereas the use of a conductive sheet (copper sheet) can mitigate the electromagnetic flux density. A novel H-shaped core was proposed in [20] for a 1.5 kW WPT system to improve efficiency and increase misalignment tolerance. However, the core loss and EMF level evaluation were not addressed.

In this paper, a WPT system model using a lumped circuit model is introduced. The relationship between the relative permeability ($\mu$) and coupling coefficient (k) and that between the core amount and core loss are shown and discussed. Three case studies using shaped bars, a pot core, and a slotted core, for circular coils, are investigated using finite element message (FEM) software ANSYS Maxwell.16 (Commercial software). An EMF was intensively evaluated with a view to reducing core loss in high-power transfer systems and mitigating human exposure to the EMF. Motivated by the investigation results obtained from these three cases, a new optimum core structure is proposed as a trade-off between core loss and core weight. The proposed core reduced core loss by about 33% as a result of reducing EMF by about 44%, while the weight increased by only 15% compared to that of the system using shaped bars as the core. As a result, the proposed core is highly efficient and has a small and comfortable weight. The EMF evaluation methodology in this paper is general and can be applied to such research areas as high frequency transformers [21] used in wind farms, to ensure whether or not the emission from such devices is compliant with allowed exposure limits. The simulation software (ANSYS Maxwell and MATLAB∖SIMULINK) and hardware prototyping are used to validate the proposed optimum core structure. The design was able to transfer 3.5 kW through a 20 cm air gap with 96% coil–coil efficiency.

2. The WPT System for Charging EVs

The inductive coupling power transfer (ICPT) system at resonance can transfer high power over several millimeters with high efficiency, but its efficiency decreases when the distance increases. Moreover, it is very sensitive to lateral and angular misalignment [22].

For EV charging purposes, the ICPT system consists of many stages as shown in Figure 1. The charging station side works as follows: AC utility power is converted to DC using a rectifier with a power factor correction, and a high-frequency (HF) inverter is then used as a source for the transmitter coil and compensation capacitor. The AC current in the Tx coil generates an alternating magnetic flux. The on-board side works as follows: The receiver coil is coupled with an alternating magnetic flux generated by the transmitter. As a result, the current is induced by induction therein, and a rectifier is then used to convert the AC-induced current to DC to charge the battery [23].

Figure 2 shows the equivalent circuit model using lumped parameter model when using a series–series (SS) compensation topology [24].

In the figure, $I_1,R_1,C_1$, and $L_1$ are the primary side current, resistance, capacitance, and inductance, respectively, and $I_2,R_2,C_2$, and $L_2$ are the corresponding parameters on the secondary side, M is the mutual inductance between the primary and secondary coils, and $V_{in},R_s$, and $R_L$ are the HF power source, the power source internal resistance, and the load resistance, respectively. The output power $P_{out}$ is calculated using uncompensated power $S_u$ (VA) and secondary side quality factor $Q_2$. However, $S_u$ can be analyzed using the open circuit voltage $V_{oc}$ and the short circuit current $I_{sc}$ [25] as follows:

$$\begin{array}{lll}{P}_{out}& =& S_u{Q}_2=V_{oc}{I}_{sc}{Q}_2\\ & =& j\omega M{I}_1{\displaystyle \frac{M}{L_2}}{I}_1{Q}_2\end{array}$$

(1)

where $\omega$ is the angular frequency.

When capacitors are added to the primary and secondary coils, the system resonates, and the power transferred to the load will increase [26]. The output power is rewritten as

$$\begin{array}{lll}{P}_{out}& =& \omega M{I}_1{\displaystyle \frac{M}{L_2}}{I}_1{Q}_2\\ & =& \omega {\displaystyle \frac{M^2}{L_2}}{I}_1^2{Q}_2\end{array}.$$

(2)

To increase the power transferred, it necessitates the increase in part $({\displaystyle \frac{M^2}{L_2}})$ through the design of the coil shape and the core structure. $\omega$ is limited by the power electronic components, and $Q_2$, according to [27], is not more than 10. Moreover, increasing $I_1$ will increase the mass of copper needed for the coil and thus increases the cost and weight. Mutual inductance is increased proportionally when the coupling coefficient is increased as in Equation (3), so enhancing the mutual inductance using an optimal core and coil structure can increase the coupling coefficient and hence improve system performance.

$$M=K\sqrt{L_1{L}_2}.$$

(3)

Since the system is operating at the resonance frequency, the current in primary side can be obtained as [22]

$$I_1={\displaystyle \frac{(R_L+R_2)V_{\mathit{in}}}{(R_s+R_1)(R_L+R_2)+{\left(2\pi fM\right)}^2}}.$$

(4)

Then the input power can be calculated as

$$P_{in}={\displaystyle \frac{(R_L+R_2)V_{\mathit{in}}^2}{(R_s+R_1)(R_L+R_2)+{\left(2\pi fM\right)}^2}}.$$

(5)

The magnitude of the secondary current can be obtained as

$$\mid I_2\mid ={\displaystyle \frac{{\left(2\pi fM\right)}^2{V}_{\mathit{in}}^2}{(R_s+R_1)(R_L+R_2)+{\left(2\pi fM\right)}^2}}.$$

(6)

The output power can be rewritten as

$$P_{out}=\mid I_2{\mid }^2{R}_L={\displaystyle \frac{{\left(2\pi fM\right)}^2{V}_{\mathit{in}}^2}{{\{(R_s+R_1)(R_L+R_2)+{\left(2\pi fM\right)}^2\}}^2}}.$$

(7)

Since the power transfer efficiency $\eta$ is a ratio of the power transferred to load $P_{out}$ to the input power $P_{in}$ as in Equation (4) [28]:

$$\eta =P_{out}/P_{in}={\displaystyle \frac{R_L}{R_L+R_1{\{(R_2+R_L)/\left(2\pi fM\right)\}}^2+R_2}}$$

(8)

where k is the coupling coefficient factor.

3. Coil and Core Design Analysis

Since the coil geometry and the core material permeability as well as the core thickness have an obvious impact on the flux path and the flux distribution inside the air gap, these factors are investigated in this section.

3.1. Coil Geometry

In this part, one-turn coreless coils for circular and square planar coils are analyzed using FEM software. The analysis is conducted by considering the two coil types having the same length (127.23 cm), 85 kHz, 20 cm air gap, and the cross-sectional wire area of 0.25 cm². Supposing the same currents and phases in both primary and secondary coils, and a design is symmetric for primary and secondary sides for both coil types. The coupling coefficient k vs. air gap curves for the two coil types are compared as depicted in Figure 3. Moreover, the misalignment test for the two types is plotted in Figure 3. From Figure 3a,b, the coupling coefficient k (for a range of air gaps from 10 to 30 cm) and alignment tolerance (for offset changes along the X-axis from 0 to 20 cm) are higher for a circular coil compared to its square counterpart. Therefore, the circular geometry is considered hereinafter.

3.2. Core Material Permeability and Core Loss Evaluation

Ferromagnetic material is used as a magnetic core in electrical machines as it has good magnetization characteristics below certain limits (saturation and temperature). When magnetic field intensity H (A/m) is applied to this material, it tends to gather magnetic field B (T) in its interior and the dipole moments tend to align as in Figure 4a. Furthermore, the magnetic flux $\Phi$(Wb/m²) inside the material is greater than outside $B={\mu }_{0}H$ for free space and $B=\mu H$ for inside ferromagnetic material when $\mu \gg {\mu }_0$, as shown in Figure 4b [29,30]. Although the high-permeability magnetic core can reduce the reluctance path and orient the magnetic field [31], it will introduce new factors such as core loss and increase the mass of the charger [32].

3.2.1. Core Material Permeability

Figure 5 shows the relationship between the coupling coefficient and the magnetic core’s relative permeability $\mu$ at 85 kHz and with a 20 cm air gap in the WPT system is simulated for cores having different relative permeabilities. From the curve, we can see that the coupling coefficient is nearly constant for $\mu \ge 3000$. Based on the result obtained, TDK-PC95, which has a permeability about 3300 and saturation magnetic flux density $B_s$ in the range 380–530 mT, was selected for this study.

3.2.2. Core Loss and Core Thickness

Core loss in general consists of two types of losses: eddy current loss and hysteretic loss. The core loss, according to the Steinmetz Formula expressed in Equation (9), depends on the operating frequency f and the magnetic flux density B [10]:

$$P_{cv}=C_m{f}^{\alpha }{B}_{max}^{\beta }(Kw/m^3)$$

(9)

where $C_m,\alpha$, and $\beta$ are coefficients that can be extracted from the manufacturer’s datasheets for the $(B-P_{CV})$ curves for a specific material. Figure 6 shows the Tx and Rx core losses against the core weight when the core thickness changes from 0.05 to 0.5 cm. From Figure 6, we can prove that the lighter the core, the higher the core loss. The significance of this analysis revealed that the light core can reduce production costs and car’s power consumption, especially for the core that mounted on the car. However, the lighter core can decrease system performance due to the higher core loss. From this point of view, the core structure must be a trade-off between core loss and core weight. The figure also shows the optimum core thickness to be 0.3 cm when the Tx and Rx core losses are approximately similar.

4. The Proposed Core Structure

In this section, three types of core shapes, namely pot core (Figure 7b), slotted core (Figure 7c), and shaped bars (Figure 7d) are proposed. The three designs are simulated using FEM software and compared with respect to the magnitude of the magnetic flux density $(B\_Mag.)$ in different places. The Litz wire ($\Phi$3.5 mm, 20 turns) was used for both Tx and Rx coils. The two pads have the same configuration as depicted in Figure 7a when the gap between them is 20 cm. An aluminium plate is used as a shield for the two pads with dimensions as shown in Figure 7a.

Table 1. The simulation parameters.

Parameters	Value
Air gap, d	20 cm
Operating frequency, f	85 kHz
The coil radius	20 cm
Litz wire $\Phi$	3.5 mm
The core radius	25 cm
Number of turns $N_1,N_2$	20, 20
Maximum Tx, Rx diameter	60 cm

shows the simulation parameters when the eddy current solution is used.

5. The Simulation Analysis

In this section, FEM software simulation results are presented and discussed. EMF levels were evaluated for the three proposed cores using the configuration parameters presented in Figure 7a for different evaluation lines. To show how the magnetic flux was distributed inside the air gap, Line 1 is located horizontally in the middle between Tx and Rx pads along the X-axis, as depicted in Figure 8a. The value of $B\_Mag.$ is concentrated at the vertical center of the two coils, while the shaped bars core has its equivalent point slightly higher than the two other cores. Figure 8b shows the magnitude of the magnetic flux density along the Z-axis, and the shaped bars core again has the highest EMF.

As the core loss in the secondary side depends on B, $B\_Mag.$ on the surface of the receiving side on the magnetic core must be evaluated. Figure 8 shows the values of $B\_Mag.$ for Line 3, which is on the Rx core surface from edge-to-edge along the X-axis: the shaped bars also have the highest B value compared to those of other cores. To evaluate the shielding effectiveness, Line 4 is located 20 cm from the shielding surface in the Z-axis direction, and the RMS value of the EMF is tested through Line 4 to ensure that the design complies with the EMF exposure standards. Figure 8d shows the penetrated EMF levels on Line 4 for the three proposed designs. Pot and slotted cores designs are below the exposure limit, while the use of a shaped bars core exceeded the limiting value (27 µT according to the International Commission on Non-Ionizing Radiation Protection (ICNIRP) Guideline (2010) when the frequency is less than 100 kHz) [33].

The measurement procedure which was introduced by International Standard IEC 62110 [34] is used to evaluate the EMF level. The environment surrounding the charger is tested to ensure whether or not the design is compliant with allowed exposure limits. The EMF level was observed for each of the three designs through a line located vertically at 20 cm from the edges of the Tx and Rx pads instead of by way of the three-point measurement procedure. According to ICNIRP Guidelines, the EMF level for public exposure must be less than 27 µT. Figure 9 shows that the pot and slotted core designs are compliant with this guideline, while the shaped bars core exceeds it.

Referring to Figure 8c, the magnetic flux density on the receiving magnetic core is very high for the shaped bars structure, so, according to Equation (9), the core loss for shaped bars is higher than that for the two other core structures. Figure 10 shows the receiver magnetic core loss for different frequencies, and the shaped bars core has the highest core loss among those tested here.

6. Optimal Core Structure Design

In spite of the shaped bars core having a high core loss and high magnetic flux leakage, it was the lightest weight: since the Rx core is installed directly at the bottom of the car, weight is a critical factor in designing the wireless charger for EV applications. Therefore, the proposed optimum core design based on a shaped bars core is highly recommended. Optimization aimed to reduce the magnetic flux density on the surface of the Rx magnetic core and made it as uniform as possible. When considering the result in Figure 8c, the flux density is high in regions located at the center of the winding coil and decreases significantly towards the inner and outer edges of the core. Therefore, the magnetic core amount needs to be increased at the center of the winding coils to suppress the magnetic flux leakage and hence reduce core loss. The optimal core structure design is a trade-off between core weight and core loss.

The flux density at any point (x,y,z) on the core is a function of many factors as the core thickness ($t_{core}$) and core permeability (${\mu }_r$), core length ($L_{core}$), core width ($S_{core}$), the position of the two coils ($\Delta$), the two coils currents, and the number of turns ($I_1,I_2,N_1,N_2$), the air gap (d), and the total width of the coil ($w_{coil}$) as given in Equation (10):

$$B_{core}(x,y,z)=f(t_{core},L_{core},S_{core},{\mu }_r,\Delta ,I_1,I_2,N_1,N_2,w_{coil},d).$$

(10)

The optimized shaped bars are proposed as in Figure 11a, where $L_1$ and $L_2$ are the bar lengths to be optimized after setting an initial length of 6.5 cm for the primary and secondary coils, respectively. $S_1,S_2$ are the bar widths to be optimized for the primary and secondary sides, respectively, whereas the thickness remained the same at 3 mm. For minimizing the CPU and memory usage issues [35], reducing the computational time is highly recommended. Thereby, the FEM software was used to optimize the lengths and the widths for Rx and Tx sides by sweeping all the lengths from 2 to 16 cm and sweeping those widths from 4 to 8 cm, while the other parameters remained constant. A part of the non-extreme data arising from the sweeps was then compared and is plotted in Figure 11b. Lastly, the optimized parameters were selected as given on

Table 2. Optimized bar lengths and widths for Tx and Rx cores.

Parameters		Optimized Value
$L_1$		16 cm
$L_2$		12 cm
$S_1$		6 cm
$S_2$		8 cm

, and the pad configuration is depicted in Figure 12. Moreover, the magnetic flux density on the Rx core surface (Line 3) for shaped bars and the proposed optimum shaped bars are depicted in Figure 13. The proposed optimum shaped bars successfully reduced the magnetic flux density by about 44%, while the flux became more uniformly distributed.

Figure 14 shows the loss comparison between shaped bars and optimum shaped bars to confirm whether or not the optimization can reduce core loss. From Figure 14, we realized that the optimum design reduced core loss by approximately 30% (at 85 kHz), while the weight was increased by only 15%. Again the EMF (RMS) level was evaluated and compared through the lines located 20 cm vertically from the edges of the charger and 20 cm horizontally from the Rx aluminium sheet as depicted in Figure 15 and Figure 16, respectively. These figures show that the optimum shaped bars core design can reduce the EMF to within allowable limits.

Figure 17 shows the electromagnetic field distributions on the receiving aluminum plates for shaped bars and optimum shaped bars. The electromagnetic field is consternated at the outer edges of the bars for both designs, but the optimum shaped bar design has a slightly low EMF level.

7. Experimental Set-Up for Series–Series Compensation Topology

In general, there are four basic compensation topologies: series–series (SS), where the two capacitors are connected in series for primary and secondary sides, series–parallel (SP), where the primary side capacitor is connected in series and the other side is in parallel, parallel–series (PS), where the primary side capacitor is connected in parallel and the other side is in series, and parallel–parallel (PP) where the primary and secondary side capacitors are connected in parallel [36]. In this research, we selected the SS topology because the primary side capacitor is not dependent on the coupling factor and the load, which is suitable for EV charging applications [36,37].

To validate the WPT system using optimum shaped bars, the prototype was developed as shown in Figure 18. The power transferred is 3.5 kW, over a 20 cm air gap and 10 $\Omega$ resistive load for fully aligned pads using 85 kHz as a fixed resonant frequency.

Experimental Results

Matching the phase angle between the secondary voltage and the secondary current can improve the power transfer efficiency [38]. To prove the secondary side capacitor is resonating with the secondary inductance, the inductances and resistances of the two coils were measured using an LCR meter 3522-50 LCR HiTESTER (HioKi E.E. Corporation, Nagano Prefecture, Japan). Following the schematic diagram provided in Figure 19a, and using MATLAB/SIMULINK, the load voltage and current waveforms for AC resistive load were plotted, as in Figure 19b, where the current and voltage signals are in phase. Figure 19c presents the primary and the load currents waveforms, where the output current is lagging by 90°. From the results, the coil-to-coil power transfer efficiency is about 96%.

To verify the simulation results, Magnetic Flux Density B is measured using a magnetic field probe EHP-50G (Narda, Segrate, Italy), when the power transfer is fixed to 3.5 kW. The measurements were taken vertically above the center of the secondary aluminium plate (as illustrated inside the figure) and compared to the simulation result as depicted in Figure 20a. Furthermore, the magnetic flux density is measured horizontally from the edge of the Rx pad (as illustrated inside the figure) and compared to the simulation result as in Figure 20b. From Figure 20a,b, it may be seen that the measured and simulated results were in good agreement.

8. Conclusions

Coil geometry and optimum core designs for WPT systems used in EV charging were presented. Three core structure including pot, slotted, and shaped bars cores are proposed and compared using FEM software ANSYS Maxwell. From the simulation analyses, it was deduced that the shaped bars core had the highest core loss, but it was the lightest of the tested designs. From these observations, the optimum core is recommended to be based on the shaped bars, and the design aims to provide a trade-off between core loss and core weight. The proposed optimum core structure successfully reduced the magnetic flux density on the Rx core by about 44%, while the flux was more uniformly distributed. Consequently, the core loss was reduced by about 30% compared to that when using shaped bars. A practical WPT system for transferring 3.5 kW through a 20 cm air gap at an operating frequency of 85 kHz was conducted to validate the simulation results. The EMF environment surrounding the charger was evaluated, and the results show that the system satisfies the permitted EMF guidelines (ICNIRP). The simulation and measurement used to validate the system are in good agreement. The system obtained about 96% full alignment coil-to-coil efficiency at a 20 cm air gap. Since, the iron core is far from its saturation point, the proposed core structure is suitable for commercialized high-power EV chargers. However, in a high-power transfer system, the current stress on the coils and switches must be carefully considered, as well as the shielding effectiveness, and the mechanical durability of the proposed core needs further evaluation. The EMF evaluation methodology in this study can be applied for evaluating the environment surrounding the other high-frequency devices. Future work can concentrate on designing the optimum shield when the power transfer is increased.