1. Introduction

Because of global warming and the exhaustion of fossil fuels, interest in and research of eco-friendly vehicles such as hybrid electric vehicles (HEVs), plug-in hybrid vehicles (PHEVs), fuel cell electric vehicles (FCEV), and electric vehicles (EVs) have increased [1, 2]. Among the eco-friendly vehicles, PHEVs and EVs (xEVs) depend on electrical energy compared with other eco-friendly vehicles. From the electrical energy dependence, research regarding battery chargers and the charging infrastructure for PHEVs and EVs has drawn considerable interest. The battery charger can be distinguished by an on-board charger mounted on the vehicle or an off-board charger installed in the charging station [3].

The charging efficiency performance and the power factor are important factors because the battery charger uses the AC utility grid. Because of the connection with the utility, the charging station and vehicles can be considered as load networks, as shown in Fig. 1. A frequent charging pattern and the high charging capacity have an effect on the utility grid as a large load. Therefore, consideration of the increasing power demand for charging xEVs is an important issue for expanding the electrical infrastructure of xEVs [4-6]. Table 1 summarizes the estimates of the additional required power to charge xEVs according to the increasing penetration rate of vehicles in the Korean market by assuming that the charging power is 16 kW/h for each xEV, and one vehicle of ten vehicles is charged at the same time. From the estimation, 2,776-MW is needed in the short-term, and 20,641-MW is needed in the long-term. These results show that the current electrical infrastructure could not satisfy the additional power for charging xEVs, and the infrastructure should be increased.

Table 1.Estimation of the additional required power according to the increasing penetration rate of xEVs

Fig. 1.Diagram of the charging system for xEVs

However, considerable time and costs are needed to increase the capacity of the conventional power grid, and the space for installation is limited because of the “not in my back yard” (NIMBY) syndrome. Moreover, additional environmental pollution would occur by using nuclear energy and fossil fuels to generate power. As such, alternative solutions for charging xEVs while slightly changing the existing power grid are greatly needed. Photovoltaics (PVs), fuel cells, wind, geothermal heat, etc. have been researched and installed owing to their benefits. Among the renewable energy sources, photovoltaic systems are a better choice for the charging source because of following reasons:

1) The growth rate of PVs has increased steadily over 40%, and the total generated power is forecasted to be 40 TW/h in 2020 [7]. 2) The output of a PV is a DC voltage, i.e., the battery of an EV can be charged directly by using the output of the PV. Thus, the charging efficiency is higher than a conventional charging structure (using an AC utility source). 3) The generation pattern for PVs is similar to the typical parking times of vehicles, i.e., the generated PV power can be stored in the battery during the daytime. Accordingly, the volatile energy of the PV system is reduced and does not disturb the grid status owing to the high transfer of power [8]. 4) PVs can be installed relatively easy in urban areas compared to wind power systems.

Fig. 2(a) shows a conceptual image of an EV parking lot using PV power, and Fig. 2(b) shows the entire block diagram of the power conversion units. From careful observation of this system structure, it is noted that the input source of the charger is at a certain level of the DC source, and it can be accessed from the DC-link source of the PV PCS instead of the AC source via the AC–DC PFC circuit. This means that it is possible to minimize and integrate a power conversion unit for the battery charger using PV power.

Fig. 2.Conventional charging system with a PV generation system

In this paper, a simple and an optimized battery charger based on PV generation system is proposed. The proposed system consists of a PV PCS and a bidirectional DC–DC converter that shares the DC-link of the PV PCS. From the generating power conditions of the PV panel and the power demand of the battery, the proposed system is operated in four modes, and each mode is automatically selected by sensing the DC-link voltage and power command of the battery. In order to deduce the optimized system control algorithm, the configuration and operation mode are analyzed in detail. With the respect of optimal hardware, PV-PCS and charger circuits are integrated for reduction of the number of switches and passive components. And, for control point of view, an optimal algorithm, which can charge the battery using PV energy and utility simultaneously, is developed. The validity of the proposed system and algorithm are verified by experimental results on the basis of a 3.3-kW prototype. Moreover, to examine the feasibility of the proposed system, the simulation results for multi-charger with various load condition are also progressed.

2. Principles of the Optimized Battery Charger

2.1 Configuration of the proposed battery charger

In order to integrate the common parts of the different systems, an analysis of the operation of each system is required. For the PV PCS, DC power is generated with the maximum power point tracking (MPPT) algorithm of the PV converter, and this power is converted into AC power by the DC–AC inverter in order to transfer the power to the AC utility grid. The power of the PV PCS is unidirectional, and all generated power is transmitted to the grid.

For the battery charger, the PFC circuit converts the AC voltage of the grid to a DC voltage above 350 V, and the DC–DC converter charges the battery by using the DC-link voltage. If the DC-link voltage is ensured, the battery can be charged by the DC–DC converter without the PFC circuit. The power flow of the battery charger is unidirectional and in the reverse direction compared with the PV PCS.

To transmit the generated power to the AC utility grid, the DC link of the PV PCS requires a higher voltage than the grid voltage, and the DC-link voltage of the PV PCS is similar to the DC-link voltage of the battery charger. Therefore, the battery charger, except the PFC circuit, can be connected to the DC link of the PV PCS, which means that the PFC circuit of the battery charger can be eliminated. When the PV power is lower than the charging power or not generated, the system should be supplied with power from the grid with bidirectional operation. In this case, the GC inverter can be operated as a PFC circuit because it is composed of an H-bridge structure. From the bidirectional operation of the GC inverter and the individual operation of the PV and DC–DC converters, the entire system can control the PV power, charging power, and grid current. Thus, two different systems can be integrated into one system.

In order to achieve a high-performance, light-weight, small-volume, and high-efficiency battery charger for xEVs, a non-isolated type DC–DC converter is adopted, which satisfies SAE J1772, the standard for safety of xEVs[13]. According to integrating PV PCS and the battery charger, the optimized charging system is realized as shown in Fig. 3. After the integration of the two systems, the system operation modes should be defined, and the proper control methods should be developed according to the operation modes.

Fig. 3.Configuration of the proposed system

2.2 Analysis of the operation modes

In order to develop the system control algorithm, the system operation modes need to be analyzed according to the magnitudes of the generated power and charging power. The operation modes of the proposed system are categorized into four modes, as shown in Fig. 4, and explained in detail as follows:

1) Mode I: This mode occurs when the charging requirement of the battery does not exist, and the battery charger is not operated. The PV converter is operated with MPPT, and the GC inverter transfers the generated power with DC/AC inversion. This mode is the same operation of a conventional PV PCS. 2) Mode II: In this mode, the battery is connected with the charging system, and the PV generates enough power to supply the battery charging. Thus, surplus power is generated, and this power is transferred by the GC inverter to the grid. 3) Mode III: When the battery requires more power than the generated power of the PV, all PV generation power is transferred to the battery, and the power difference between the battery and the PV is supplied from the grid. In this case, the power flow of the GC inverter is changed; that is, this power conversion unit is operated as the PFC circuit. Thus, the battery is supplied by power from the PVs and grid. 4) Mode IV: This mode is present when the amount of the PV power is inadequate for charging the battery, for instance, the PV modules are in the shade or operate at nighttime. All of the charging power for the xEV batteries is supplied from the grid, and the PV converter does not operate.

Fig. 4.Operation modes of the proposed system

A possible scenario of the proposed system operation throughout the day is presented in Fig. 5, which shows that the amount of PV power varies according to the time of day, and the operation mode is determined by amount of the PV power and the charging requirements. The operation conditions are summarized in Table 2, where PPV is the generated power, and Pbatt is the charging power.

Fig. 5.Power relationship according to the PV generation and charging requirements of xEVs during the day

Table 2.Operation conditions by the operation modes

3. Proposed Control Algorithm

In order to establish the system control algorithm, an analysis of the power variation according to the operation mode is needed.

3.1 Power equation for the control algorithm

According to Fig. 3, all of the power of the proposed system passes through the DC-link capacitor. Thus, the equations for the power flow of each operation mode need to be analyzed with regard to the DC link. The power equations of each power conversion unit are calculated as follows:

where VPV, IPV, Vbatt, Ibatt, Pgrid, Vgrid, and Igrid are the PV output voltage, the PV output current, the battery voltage, the battery current, the grid power, the grid voltage, and the grid current, respectively. The power equation for each power conversion unit is expressed as

The power equation in (4) becomes (5) by substituting (1~3), and (5) is expressed as (6) because VPV and Vbatt can be expressed as the DC-link voltage according to the voltage transfer ratio:

where VDC, k1 and k2 are the DC-link voltage, the voltage transfer ratios of the PV converter (boost converter), and the battery charger (buck converter), respectively.

If the utility grid is in a stable state, the grid voltage is a constant value, which means that the entire system can be controlled by maintaining the DC-link voltage. There are three power conversion units for controlling the DC-link voltage in the proposed system: the PV converter, the GC inverter, and the DC–DC converter. However, while controlling DC-link voltage, the proposed system may be unstable owing to the multiple control requirements for MPPT voltage control, battery voltage control, and grid current control at the same time in a single controller. Therefore, among the three power conversion units, it is important to select a proper power conversion unit for DC-link voltage control and system stability.

3.2 Principle of proposed control algorithm

Based on the aforementioned analysis, it is necessary to determine which power conversion unit controls the DC-link voltage. There are two power conversion units, the PV converter and GC inverter, controlling the DC-link voltage and the basic operation at the same time in the proposed system. The control function of each power conversion unit is summarized in Table 3 when the DC-link voltage is controlled by the PV converter and GC inverter according to the operation mode. In Mode I and Mode II, the PV converter controls the DC-link voltage, and the MPPT algorithm is executed by the GC inverter with gridconnected operation. When the generated power is less than the charging power (Mode III), the PV converter tracks the MPPT voltage of the PV module, and the DC-link voltage is controlled by the GC inverter with gridconnected operation. In this mode, the current of the grid is reversed, which means that the GC inverter is operated as the PFC circuit. When the solar irradiation disappears, the PV converter does not generate any power, and all the power for charging the battery is supplied from the utility grid by the GC inverter. The GC inverter is operated as the PWM converter with power factor control to supply the grid power to the battery, which is same as the PFC circuit operation, and the battery charger always controls the charging voltage of the battery.

Table 3.Control algorithm for each operation mode

The control algorithm should maintain stable operation regardless of any changes in the generation or charging conditions. However, the control algorithms in Table 3 are changed from mode variation, which disturbs the system stability. Moreover, these conditions are varied frequently in a real system in contrast to Fig. 3, which means that the system stability can be much worse. Therefore, the system algorithm needs to be optimized in order to increase the control performance from a transient state. Table 4 summarizes the modified system algorithm for the different modes to avoid a transient status due to mode variation. The PV converter controls the MPP voltage of the PV module, the GC inverter (PFC circuit) controls the DC-link voltage and grid current, and the battery charger controls the battery voltage. Thus, the transient system status can be removed from the mode variation.

From Table 4, the controller of each power conversion can be composed a single one as shown in Fig. 6. A perturbation and observation algorithm is adopted for the MPPT algorithm of the PV converter, and the output voltage of the PV converter is determined by the condition of the PV module and the output impedance of the PV converter. Thus, the GC inverter (PFC circuit) regulates the DC-link voltage by transferring power to the utility grid or by receiving power from the utility grid, which means that the GC inverter operation and PFC circuit operation are determined automatically without an additional algorithm. For the battery charger, the charging methods are applied according to the state of charge (SOC). The constantcurrent–constant-voltage (CC–CV) method can be adopted as the charging method because this method stably charges the battery [14]. In order to control the constant battery current or power, a proper output voltage reference is needed. However, it is difficult to calculate the impedance of the battery. Thus, we obtain the voltage reference by sweeping the output voltage. This is expressed as follows:

Table 4.Modified control algorithm for each operation mode

Fig. 6.Controllers of each power conversion unit

where k3 and I*CC are control coefficients that determine the voltage sweeping steps and a reference of constant current magnitude.

3.3 Battery charger algorithm

A battery charger is needed in order to stably charge the battery. Various topologies can be applied to the battery charger according to the input and output voltages. In this study, a buck converter is adopted for the battery charger because the DC-link voltage is always higher than the battery voltage in the proposed system. In order to reduce the size and inductance of the inductor, the inductor current is controlled in the discontinuous conduction mode (DCM), which eliminates the turn-on switching loss and reverse recovery loss by the operation of a zero-current-switching (ZCS) diode without an extra component and algorithm [15]. In order to reduce current ripple in the DCM, a twophase buck converter with interleaving control is adopted.

The diode in the buck converter can be replaced with a MOSFET to reduce the conduction losses in the semiconductor owing to synchronous rectifier operation. For the DCM, the inductor current is negative after the zero inductor current point DA in Fig. 7, which means a circulation current is generated because the switch maintains the turn-on state [16]. Therefore, the RMS current of another phase is increased to maintain the output current. Because of the increased inductor current, the conduction and iron losses of the inductor are generated. Thus, the MOSFET is turned off when the inductor current becomes negative in order to block the negative current.

Fig. 7.Profile of the inductor current according to the switching pattern operating in the DCM

4. Experimental Verification of the Proposed System

In order to verify the proposed topology and control algorithm, experiments have been performed with the 3.3-kW prototype shown in Fig. 8 with the system parameters in Table 5.

Fig. 8.Prototype of 3.3-kW PV BESS hardware

Table 5.Experiment parameters

4.1 Experimental results for the battery charger

The experimental results for the circulation current of the battery charger are shown in Fig. 9. In Fig. 9(a), the inductor current becomes negative because of the synchronous rectifier switching method. Therefore, the inductor current flows into the MOSFET (not the body diode), which generates a circulation current and increased the peak current of the inductor. Fig. 9(b) shows the results for an improved waveform in which the negative current is blocked by turning off the MOSFET. Because of the elimination of the negative current, the peak current of the inductor is reduced from 2.83 A to 2.62 A, which decreases the conduction losses of each component and the switching losses of the MOSFET.

Fig. 9.Experimental waveforms of battery charger

4.2 Experimental results of mode I

In order to validate the proposed control algorithm and hardware, experimental tests for each mode have been carried out by using the 3.3-kW hardware, and a PV simulator is used to emulate a PV panel.

The requirement of charging power does not exist in Mode I. Thus, the PV converter and GC inverter transfer generated power to the grid, and the battery charger does not operate. Fig. 10 shows the experimental waveforms of the PV converter and GC inverter in Mode I. The output voltage of the PV module is controlled to 270 V with the MPPT algorithm, and the GC inverter controls the DC-link voltage to 380 V with grid-connected operation. The PV module generates maximum power, and the grid current is controlled at high power factor.

Fig. 10.Experimental waveforms in Mode I

4.3 Experimental results of mode II

Mode II is started when a charging requirement occurs in Mode I or when the magnitude of the generated power is less than the magnitude of the charging power in Mode III. The charging current is a constant value from the CC algorithm regardless of the magnitude of the generated power, and only the transferred power to the grid is varied. Fig. 11 shows the experimental waveforms in Mode II when the generated power is varied from 1.95-kW to 1.5-kW. The charging power is maintained constantly. When the generated power is 1.95-kW [Fig. 11(b)], the difference in power between the generated power and the charging power is 1.1-kW (including losses) and is transferred to grid. Fig. 11(c) shows the decreased grid current in order to maintain the charging power for the same operating mode.

Fig. 11.Experimental waveforms in Mode II

4.4 Experimental results of mode III & IV

In Mode III, insufficient charging energy is supplied from the grid because the generated power decreases. The experimental results in Mode III are shown in Fig. 12(a)-(b). The grid current is reversed to supply energy to the battery. Because PV power (0.15-kW) exists, the grid power is smaller (0.57-kW) than the charging power (0.7-kW). When the PV does not generate power, all of the charging power is supplied from grid, and the system operates in Mode IV. Fig. 12(b) shows the waveforms of Mode IV. The grid current is increased to supply energy to the battery without PV power.

Fig. 12.Experimental waveforms in Mode III and Mode IV

4.5 Verification of the entire control algorithm and hardware

The overall experimental results are shown in Fig. 13. In the initial state (Mode I), battery power requirements do not exist, and all of the generated power from the PVs is transferred to the grid. When the battery is connected to the charger, Mode I changes to Mode II, and the battery charger is started in the CC charging mode. As the charged power increases, the surplus power decreases (grid current). When the battery voltage reaches the CV voltage (250 V), the charging algorithm changes to the CV mode. In the CV mode, the charging power decreases as the charging current decreases. Therefore, the grid current increases because the intensity of radiation is maintained constantly. When the solar irradiation decreases gradually, the grid current also decreases. Mode II changes to Mode III because the magnitude of the generated power is reduced below the magnitude of the charging power. When the solar irradiation disappears, Mode III changes to Mode IV, and all of the charging power is supplied from the grid. Regardless of the operating mode, the DC-link voltage is controlled constantly at 380 V. As shown in Fig. 13, a dynamic response in each power conversion unit does not occur with the proposed control algorithm. This is because the proposed algorithm automatically selects the operation modes and individually controls each power unit.

Fig. 13.Experimental waveforms of overall operation

Fig. 14(a) shows the efficiency of the PV PCS and the battery charger of the proposed system. Operation of the PV PCS means the system is operating in Mode I, and the PV converter and GC inverter are operated. Therefore, the efficiency can be calculated by multiplying the efficiencies of the PV converter and the GC inverter. The efficiency of the battery charger is measured when the battery voltage is 250 V. A high charging efficiency greater than 96% is achieved because of a non-isolated type, a single charging stage, and the ZCS characteristic of the DCM operation.

A conventional charging system based on PV power systems is supplied with charging power via the PV PCS and the battery charger. For many power conversion processes, it is difficult to achieve a high charging efficiency. However, the charging power of the proposed system is supplied by two power conversion units, which are the PV converter and the DC–DC converter. Thus, a high charging efficiency can be achieved. Fig. 14(b) shows the charging efficiency of the conventional system and the proposed system using the PV power. To calculate the charging efficiency of the conventional PV charging system, the efficiency of the developed 3.3-kW charger [17], which consists of a PFC circuit and series resonant converter (SRC), is used. The charging efficiency of the conventional structure, which consists of the PV PCS and the battery charger, is less than 88%. The proposed charging system, however, achieves a higher efficiency than the conventional system by approximately 7% to 15%. From the experimental results, the proposed system can reduce not only the system size and cost but also the system losses.

Fig. 14.Comparison of system efficiency

5. Feasibility Study of the Proposed System

The proposed system can be applied to a parking lot of buildings, supermarkets, and etc., where several batteries can be connected simultaneously with different charging energy requirement. Therefore, in order to verify the feasibility of the proposed system, a multi-load system is considered, as shown in Fig. 15.

Fig. 15.Block diagram of proposed system expansion

In this paper, a simulation platform consists of three battery chargers linked to DC-link in parallel, and single PV PCS. A simulation scenario considers charging and discharging operation of chargers; that is, one of xEVs batteries is discharged, and other two batteries are charged.

Fig. 16(a) shows entire simulation results including the grid voltage, the grid current, the DC-link voltage, and a current of the battery1.

Fig. 16(b) shows the expanded waveforms of sections ①-③ in Fig. 16(a). Section ① is the MPPT process of the PV converter, and all of the generated power is transferred to the grid. From the discharging operation (0.8-kW) of the battery1, the grid current is increased at section ②. The charging operation (1.0-kW) of the battery2 and the battery3 is started at section ③. Thus, the grid current is decreased at section ①.

Fig. 16(c) shows the expanded waveforms of ④-⑤ sections, and it is started with end of the charging operation of battery2. The grid current is varied according to the variation of solar irradiation. With end of the battery3 charring, section ⑤ is started, and the grid current is increased because of the constant solar irradiation. Lastly, the grid current become to zero because the discharging of the battery1 is ended, and the solar irradiation is disappeared.

Fig. 16.Simulation of system expansion

From the simulation results, it is noted that the proposed charging system is successfully operated regardless of the various system operation conditions such as the solar irradiation variation and the charger operations with various load requirements.

6. Conclusion

The amount of electrical infrastructure should be increased to supply power to the batteries of xEVs, which have been increasing gradually. We considered a PV system as a solution for the charging source and proposed a battery charger integrated with PV systems to improve its performance and to reduce the system size and volume. The proposed system is operated in a small transient state because each power conversion unit is controlled individually by maintaining the DC-link voltage and by using single controller of each unit. Moreover, DCM operation was applied to the battery charger to improve the charging efficiency. The proposed system and algorithm were verified by experiments based on 3.3-kW hardware, and the efficiency of the charging from the PV increased in comparison with the conventional charging structure. Moreover, the feasibility study of the proposed system by expanding charging systems were progressed and verified. Therefore, it is expected that the proposed system and algorithm can serve as an effective method for supplying power to the batteries of xEVs owing to its advantages.