A unified multi-functional on-board EV charger for power-quality control in household networks

Applied Energy 215 (2018) 186–201

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

A unified multi-functional on-board EV charger for power-quality control in household networks

T

⁎

Seyedfoad Taghizadeha, M.J. Hossaina, , Junwei Lub, Wayne Waterb a b

School of Engineering, Macquarie University, NSW 2109, Australia Department of Electrical Engineering, Centre of QMNC, Griffth University, QLD 4215, Australia

H I G H L I G H T S a feasible unified control system for a multifunctional on-board EV charger. • Designing EV charger can operate in V2G/G2V mode as its main function. • The EV charger can simultaneously perform three ancillary functions of a STATCOM and an APF. • The on the EV battery is reduced using a two-leg buck-boost DC/DC converter. • Stress • Simulation and experimental results verify the efficacy of the proposed system.

A R T I C L E I N F O

A B S T R A C T

Keywords: Vehicle-to-grid (V2G) Grid-to-vehicle (G2V) Four-quadrant STATCOM Active power filter (APF) Power quality Low-voltage household networks

This paper presents a feasible and reliable unified control system for a single-phase 4.5 kVA on-board multifunctional electric-vehicle (EV) charger that is connected to a low-voltage household network. Based on the proposed control system, the EV charger can operate as both a single-phase four-quadrant static synchronous compensator (STATCOM) and an active power filter (APF). The proposed EV charger can simultaneously perform four functions: charging/discharging the electric-vehicle’s (EV’s) battery; reactive power compensation; voltage regulation; and, harmonic reduction, which are important concerns of the existing power grid. Accordingly, it can enhance the building s voltage profile, power quality, and reliability, which makes the proposed method a complete solution for low-voltage household networks. The stress on the EV battery is also reduced, which can enhance its lifetime. A stability analysis of the proposed unified control system is provided in this paper. The simulation results, with two loads, static and dynamic, confirm the efficacy and reliability of the proposed system. The performance of the designed unified control system is also validated by experimental results.

1. Introduction In recent years the demand for EVs has been growing significantly in developed countries, including Australia [1]. It is predicted that the number of EVs will grow to be 64% of the vehicles on the road in the United States (U.S.) by 2030, and 45% in Australia by 2030 [2,3]. Due to the increased penetration of EV chargers in household charging stations, some critical concerns have appeared for power systems, such as harmonics and voltage fluctuations [4–7]. Modern power electronic devices, for example personal computers, laptops and smart TV power supplies, have also adversely impacted power quality, not only in a house network but also in power systems. Moreover, the growing numbers of inductive and non-linear loads, like washing machines,

⁎

refrigerators, etc. in a house demand the delivery of more reactive power from the grid than ever before. Therefore, a unified advanced control system is required which can provide an effective solution to improve the power quality and provide the required reactive power for each individual house. Many utilities over the last decade have tried to utilize EV charging stations as an effective solution to operate as both chargers and harmonic eliminators, voltage regulators or reactive power compensators (capacitor bank) [8–10]. Such ancillary functions are provided by integrating a control algorithm with the power circuits of the EV charger. According to recent literature, single-phase EV chargers are designed to provide reactive power while also charging or discharging the EV battery [11–16]. The authors in [17–21] address the harmonic problem caused by non-linear loads in a house, or by the

Corresponding author. E-mail addresses: [email protected] (S. Taghizadeh), [email protected] (M.J. Hossain), j.lu@griffith.edu.au (J. Lu), wayne.water@griffithuni.edu.au (W. Water).

https://doi.org/10.1016/j.apenergy.2018.02.006 Received 11 September 2017; Received in revised form 31 January 2018; Accepted 2 February 2018 0306-2619/ © 2018 Elsevier Ltd. All rights reserved.

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

Nomenclature

Iqref1 Iqref2 Iqhref θ P Pref Pc Sref Q Qref Qref1 Qref2 Qc PFc PFL PFs f fsw CD Cf CB Csnubber Lf Lsys L1 L2 LD Rs Xs u (t ) y (t ) K1 (t ) K2 (t ) x (t ) D ω Pa Ts CF

EV electric vehicle STATCOM static synchronous compensator APF active power filter OSG orthogonal signal generation V2G vehicle to grid PCC point of common coupling PWM pulse width modulation THD total harmonic distortion PF power factor source voltage vs v PCC PCC voltage converter output voltage vc vca grid-side capacitor voltage PWM voltage v PWM VB battery voltage VD DC-link voltage D-axis of the source voltage Vsd Vsq Q-axis of the source voltage peak value of the PCC voltage VPCCd ∗ reference value of VPCCd VPCCd D-axis voltage control signal VCd Q-axis voltage control signal VCq is source current iL load current IB battery current IBripple,rms RMS value of the battery ripple current IBmean mean value of the battery current ic converter output current i ca filter capacitor current iα α-axis of the grid-side current iβ β-axis of the grid-side current ihref reference harmonic generated by harmonic controller Id D-axis of the grid-side current Q-axis of the grid-side current Iq Idref total reference value of Id reference value of Id generated by active controller Idref1 reference value of Id generated by harmonic controller Idhref total reference value of Iq Idref

reference value of Iq generated by reactive controller reference value of Iq generated by voltage controller reference value of Iq generated by harmonic controller angle active power demand total reference active power converter output active power reference apparent power reactive power demand total reference reactive power reference reactive power generated by reactive controller reference reactive power generated by voltage controller converter output reactive power converter-side power factor load-side power factor source-side power factor line frequency switching frequency DC-link capacitor filter capacitor battery-side capacitor snubber capacitor filter inductor total series inductance DC-side inductor 1 DC-side inductor 2 total DC-side inductance feeder resistance feeder reactance input signal of harmonic control algorithm fundamental component of u (t ) estimated amplitude of the sine term of u (t ) estimated amplitude of the cosine term of u (t ) difference between u (t ) and y (t ) duty cycle angular frequency DQ transformation matrix time delay crest factor

Although it tackles the mentioned voltage disturbances, it would not be economical to be installed as a separate unit beside an EV charger in household networks. Moreover, this system works as a dynamic voltage restorer (DVR) to eliminate the voltage fluctuation (such as sag or swell), thus it needs a battery energy storage that also increases the cost. Accordingly, it would be a desired solution to equip the singlephase EV chargers to tackle the voltage fluctuations caused by their own operation and occasionally the installed rooftop PV's operation. As a result, no extra unit such as the one presented in [28,29] is required to be purchased and installed in houses. However, it must be noted that, depending upon the feeder parameters such as the R/X ratio of residential feeders, active control and/ or reactive control is required to maintain the voltage profile in an acceptable region. In [30,31] the impact of the R/X ratio in improving the voltage profile of a residential feeder using PV inverters and droopbased battery storages is studied in detail. This study concludes that the reactive power capability of PV inverters is sufficient to improve the voltage profile of urban areas where the R/X ratio is less than a critical ratio (identified as 4.5–5), whereas in rural areas where the critical R/X ratio is greater than that critical ratio (more resistive feeder), both reactive and active compensations are required. Such an investigation is followed in this paper to design an EV charger which can improve the voltage profile of a residential network for urban areas.

operation of the EV charger itself. As a result, an EV charger can operate as an APF to reduce or filter out the harmonics of the network. In [22] a single-phase EV charger is utilized to tackle the voltage disturbances caused by motor startup or inductive loads by operating as a STATCOM. In [23,24], while a single-phase EV charger works in the V2G or G2V mode, it is designed to support reactive power and/or harmonic reduction. As presented in the above literature, the majority of the designed EV chargers are single-functional and only a few of them are double-functional and, thus, are unable to remove all the three mentioned ancillary functions at the same time. Moreover, a single-phase EV charger which is able to improve the voltage profile of a household network is less focused in the literature. The effect of plug-in electric vehicles (PEVs) on the voltage profile of one phase in a low-voltage residential feeder is investigated in [25,26]. It is concluded that the EVs' operation can adversely affect the voltage profile of the residential feeder, particularly where the EVs are plugged in close to the end of the feeder. Similarly, a rooftop photovoltaic (PV) installation can also cause an unbalance voltage in a household network to more than the standard limit [27]. As a solution, the authors in [28,29] propose a transformerless hybrid series active filter for residential buildings in order to combat the voltage disturbances caused by charging/discharging the PEVs. This system is a separate unit and is independently connected to a household network. 187

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

converter operates as a voltage regulator for the DC-link capacitor CD, as shown in Fig. 1, between the DC/DC and AC/DC converters for a wide range of battery voltages. A non-linear diode rectifier load is used for harmonic injection into the household network, while the load variations create voltage fluctuations in the network. Table 2 shows the circuit parameters of the proposed EV charger. Two separate control systems have been designed for the AC/DC and DC/DC converters, respectively. The first one, known as the unified control system in this paper, is designed for the AC/DC converter, and makes the converter operate as a STATCOM and an APF; the second one is for the interleaved two-leg buck-boost DC/DC converter.

Another issue that is less focused on in the literature is the batteryside stress. Reviewing the above literature, it appears that the proposed single-phase EV chargers in [13,17,19,20] include only one singlephase DC/AC inverter, from which the battery would suffer from the double-frequency (2-f) ripple of the DC-link voltage which is the natural byproduct of single-phase AC/DC rectification [32]. In [11,16,18,21,23,24] this problem is solved by adopting a one-leg buckboost DC/DC converter between the battery and the AC/DC converter. However, the charging and discharging operation of the battery-end inductor in the structure of the DC/DC converter itself generates current ripple which causes stress on the battery and can reduce its lifetime in the long term. In this paper, a multifunctional single-phase EV charger with a new unified control system is presented which can support the main functions of V2G/G2V and ancillary functions such as reactive power support, harmonic reduction and voltage regulation simultaneously as shown in Table 1. The superior performance of the proposed EV charger compared with those of the contemporary literature are as follows: Firstly, the EV charger can address all the three ancillary functions at the same time. Secondly, the stress on the EV battery is reduced by using an interleaved two-leg buck-boost DC/DC converter. The proposed system can provide the three ancillary functions without adding any additional components to the EV charger, or being reliant on the EV battery. The voltage regulation feature of the proposed EV charger is designed for urban areas where the reactive capability of the EVcharger's inverter is sufficient, and no energy is demanded from the EV battery for this purpose. However, for rural areas where active power support is also needed, the voltage regulation function of the proposed EV charger might not be entirely effective and could only partly improve the voltage profile of the network. This paper is organized as follows. Section 2 presents the system configuration. The proposed unified control system is presented in Section 3, including the reactive power compensation, voltage regulation, harmonic detection and elimination, and active-power control functions. Section 4 analyzes the stability of the proposed control system, followed by details of the DC/DC converter and its control system in Section 5. Section 6 includes the simulation results and Section 7 shows the experimental results that verify the proposed idea. A summary of this research is given in Section 8.

3. Proposed unified control system for the AC/DC converter Fig. 1 shows the proposed unified control system. As mentioned, the proposed EV charger can operate in four quadrants of a STATCOM. This means that the charger is able to operate in inductive and capacitive modes while it is charging or discharging (bidirectional) the EV battery. The designed system can support the reactive power demanded by inductive and non-linear loads as well as the voltage regulation at the PCC, demonstrated by the phasor diagram in Fig. 2. When the system operates without reactive power support, there is a lagging angle between the load current (iL ) and the voltage at the PCC (v PCC ) as presented in Fig. 2(a). The PCC voltage itself has a phase deviation with the supply voltage (vs) due to the losses caused by the line impedances (Rs , Xs ). Accordingly, v PCC is always less than vs . Such a voltage drop and the phase difference between supply current (i s ) and v PCC could be removed by injecting a leading current (i c ) supplied by the EV charger as shown in Fig. 2(b). Based on the proposed idea of the unified control system presented in this paper, as soon as i s is synchronized with v PCC , performed by injecting a leading current, the compensating harmonic current could be injected so that the harmonics of i s could be reduced to meet the IEEE 1547 standard. As a result, the three ancillary functions of voltage regulation, reactive power support, and current harmonic elimination can be achieved simultaneously. The designed control system is a current controller that operates based on the DQ transformation technique. The DQ transformation matrix is expressed as:

sinωt − cosωt ⎤ ⎡i α ⎤ Pa = ⎡ ⎣ cosωt sinωt ⎦ ⎢ ⎦ ⎣ iβ ⎥

2. System configuration of the EV charger

(1)

where i α and iβ are the α and β stationary reference frames of the input signal, with a 90° phase shift from each other. In a single- phase system, an orthogonal-signal generation (OSG) block is needed to generate iβ . Fig. 3 shows the equivalent circuit of the designed EV charger. The mathematical model of the grid-connected AC/DC converter and the LC filter is:

Fig. 1 illustrates the overall configuration of the designed EV charger including the power circuit and the proposed unified control system. The power circuit includes an H-bridge single-phase AC/DC converter and an interleaved two-leg buck-boost DC/DC converter. These two converters operate as a 4.5 kVA EV charger in a bidirectional mode to charge and discharge a 400 V EV battery. A 230 V AC network is modeled as a household network, including linear and non-linear loads. The AC/DC converter, which is connected in parallel to the AC network, operates as a single-phase STATCOM and an APF to combat line voltage disturbances, reduce the harmonic currents, and supply the reactive power demanded by inductive and non-linear loads. Bipolar modulation is employed in the design of the AC/DC converter, which is presented in [11]. In order to comply with the IEEE 1547 standard for the AC-side harmonics generated by switching operation, an inductorcapacitor (LC) harmonic filter is used at the front end of the AC/DC converter for reducing the harmonics to less than 5% [3]. An interleaved two-leg buck-boost DC/DC converter, because of its advantages, is used in the design of the proposed EV charger [33,34]. The two inductors of the battery side (L1 and L2 ) reduce the current ripple, which flows through the battery (by 100% at 50% duty cycle). Such a ripple cancellation reduces the stress on the battery, thus increasing its lifetime. Moreover, a smaller capacitor, CB , in parallel with the EV battery, is needed due to the inherent ripple cancellation, which reduces the cost of the system. The interleaved two-leg buck-boost DC/DC

d

⎧ Lsys dt i c = vs−vc ⎪ d ⎨C dt vca = i ca ⎪i c = i ca + i s ⎩

(2)

where Lsys is the total series inductance, vc is the output voltage of the converter, vs is the source voltage, vca is the capacitor voltage, and i ca Table 1 Features of the proposed EV charger compared to the contemporary literature.

188

Considered issues

[11–16]

[17–21]

[22]

[23,24]

Proposed method

Charging/discharging Reactive power support Harmonic reduction Voltage regulation Lower stress on the battery Addressing functions simultaneously

√ √ x x √ –

√ x √ x √ –

√ x x √ √ –

√ √ √ x √ x

√ √ √ √ √ √

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

Duty ratio controller

VD VB

Rsa,b,c Lsa,b,c

Transformer 11 kV/ 230 V

House H Ho use Dynamic Load

11 kV

Equation (33)

D

Static Load

PWM(±1) S5,6 S7,8

Idref1 Pref

Vsd

vpwm Reactive controller

Iqhref Iqref2

vPCC iL

VCd

Equations (8, 9, 10)

dq VCq

Iqref1

Voltage controller

PI1

Non-linear Load

Idref1 Idhref

Lsys

Id dq

Lsys

Iq PI2

Vsq

Iqhref

+++

Qref2

Equation (11)

S1,4

+ ++

+-

* VPCCd

PI3

dq

is vPCC iL

SPWM(±1)

Idhref OSG

VPCCd

S2,3

Harmonic controller

ihref

S4

S3

OSG

-+

Fig. 4

Equations (17, 18)

S8

1 kW 1.5 kVAR

Cf

+ +

iL

VB Qref IB

Lf

CD

+

Active controller

CB

VD

LC filter

S2

-

VB

L2 S7

S1

-

EV Battery 400 V

S6

++

S5 L1

2 kW

Iqref1 Iqref2

PLL

vPCC

Fig. 1. Designed single-phase bidirectional EV charger and the proposed unified control system.

Table 2 Circuit parameters of the proposed EV charger. Item

Parameter

Value

Source voltage Line frequency Switching frequency

vs f fsw Lf Cf Csnubber CD CB L Rs + jXs L1 L2

230 V 50 Hz 13 kHz

Filter inductor Filter capacitor Snubber capacitor DC-link capacitor (high-voltage side) Battery-side capacitor (low-voltage side) Load Feeder impedance DC-side inductor 1 DC-side inductor 2

2.8 mH 5 μF 0.04 μF 1 mF 10 μF 4 kVA 0.23 + j0.71 Ω 1.4 mH 1.4 mH

Fig. 3. Equivalent circuit of EV charger connected to grid.

From Eq. (3) the derivative terms of Id and Iq can be expressed as:

dId V V = −ωIq + sd − cd dt Lsys Lsys

dIq dt

= ωId +

Vsq Lsys

−

(4)

Vcq Lsys

(5)

To obtain the steady-state condition, the obtained derivative terms are set to zero. Following that, Vcd and Vcq are calculated as:

Fig. 2. Phasor diagrams. (a) Without the proposed unified control system, (b) with the proposed unified control system.

1

1 0 ⎤ Vsd ⎤ ⎡ Lsys ⎡ ⎥ −⎢ 1 ⎥ ⎢Vsq ⎥ ⎦ ⎢ 0 Lsys ⎣ ⎦ ⎣

0 ⎤ V ⎥ ⎡ cd ⎤ 1 ⎢ ⎥ Vcq ⎥ ⎦ Lsys ⎣ ⎦

(6)

Vcq = ωLsys Id + Vsq

(7)

Eqs. (6) and (7) are the fundamental equations of the unified currentcontrol algorithm and can be observed in Fig. 1. Using the DQ transformation, the active and reactive components of the pulse-widthmodulation (PWM) signal (output signal of the controller, vc ) are separated and controlled independently. In the proposed control system, the four sub-control algorithms (active controller, voltage controller, reactive controller, and harmonic controller) generate reference signals for the unified current controller. For example, as shown in Fig. 1, the combination of both the output signals of the harmonic controller and the active controller compose the Id reference signal. On the other hand,

and i s are the capacitor current and the source current respectively. Multiplying the DQ transformation matrix (Eq. (1)) by the first equation of Eq. (2), and adding the decoupled terms, results in

⎡ d ⎡ Id ⎤ 0 − ω ⎤ ⎡ Id ⎤ ⎢ Lsys =⎡ + Iq ⎥ ⎣ ω 0 ⎦ ⎢ Iq ⎥ ⎢ 0 dt ⎢ ⎣ ⎦ ⎣ ⎦ ⎣

Vcd = −ωLsys Iq + Vsd

(3) 189

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

any communication infrastructure, while the voltage at the PCC can be monitored easily. The measured voltage at the PCC is compared to the nominal voltage in per unit (pu). The proportional-integral (PI) controller is then employed to make the error signal zero. The controlled output of the PI controller is Qref2 . Accordingly, Iqref2 is calculated by Eq. (11) and is then added to Iqref1, achieved by Eq. (10) as shown in Fig. 1.

the combination of both the output signals of the voltage controller and the reactive power controller generate the Iq reference signal for the unified current controller. Based on this concept, the harmonic detection relies on both the active and the reactive power support. However the required energy to provide this function is negligible [11]. Voltage regulation and reactive power compensation rely only on the reactive part, which consumes the energy of the DC-link capacitor and does not affect the battery. Consequently, the three functions can be supplied even when the EV charger is idle and does not charge or discharge the battery. As mentioned earlier, the main focus of this paper is to introduce a unified control system that is able to perform all of these functions. Furthermore, the paper highlights that, using the proposed strategy, all these functions are addressed at the same time without imposing extra stress on the EV battery, or adding any extra components. In the following section, the sub-control algorithms are explained separately.

2 ⎞ Iqref2 = Qref2 ∗ ⎛ ⎝ VPCCd ⎠ ⎜

⎟

(11)

However, this technique is more suitable for urban areas where the R/X ratio is less than a critical ratio (identified as 4.5–5) [30,31]. The reason is that in rural areas feeders are more resistive, and active control must be combined with reactive control to improve the voltage profile of the feeders. In this paper, the voltage-regulation function of the EV charger is designed for urban areas. Therefore, the reactive control strategy is adopted, and accordingly operating this function does not demand any active energy from the EV battery.

3.1. Reactive power compensation 3.3. Harmonic detection and reduction The proposed EV charger is able to operate in the bidirectional mode, while it can supply the reactive power without relying on the EV battery. The conventional unidirectional chargers are only able to provide reactive power in the charging mode and before the battery is fully charged [35]. They are limited in supplying reactive power to the network when the battery is in a full state of charge (SOC). This problem is solved in bidirectional systems such as the proposed one. The proposed charger is also capable of compensating for reactive power without any detrimental effects on its battery. The required reactive power is provided by the DC-link capacitor and not from the battery. This situation reduces the stress on the battery, thus increasing its lifetime. Moreover, an interleaved two-leg buck-boost DC/DC converter is used to regulate the DC-link voltage. Thus, the control system of the AC/DC converter is not affected by the common 2-f ripple of the DClink voltage which is the byproduct of a single-phase rectification. Hence, a large capacitor is not required to provide the required reactive power with the minimum current and voltage ripple. The reactivepower-control algorithm generates Iqref1, as shown in Fig. 1, based on Eqs. (8)–(10). It measures the PCC voltage and the load current, v PCC and iL ; then, based on these two parameters, the required reactive power can be calculated as follows:

Qref1 =

|v PCC | |iL | ∗ ∗sinφ 2 2

sinφ = ∠v PCC−∠iL

The third function of the proposed EV charger is its capability to detect and reduce the harmonics of the line current so that the EV charger can operate as an active power filter while it performs its other functions. In this paper, an adaptive filter for synchronous detection and extraction of harmonics is employed for the harmonic controller as shown in Fig. 4. The x (t ) signal, which is the difference between the input signal and its fundamental component, includes the harmonics of the input signal. Thus

ihref ≈ x (t ) = iL−y (t )

where ihref can be used as the reference signal utilized for the harmonic reduction. This algorithm operates by measuring the load current iL , thus generating the fundamental signal y (t ) , which is a signal without any harmonics and distortion. The input signal of the algorithm, load current, iL in this case, and its fundamental component y (t ) are:

(8)

Then Iqref1 can be calculated by ⎜

iL = A1 sinωt + B1cosωt

(13)

y (t ) = K1 (t )sinωt + K2 (t )cosωt

(14)

The algorithm employs an adaptive filter which uses the Gradient Descent method to find the amplitudes of the sine and cosine terms, K1 (t ) and K2 (t ) [36]. As a result, these two signals are obtained by:

(9)

2 ⎞ Iqref1 = Qref1 ∗ ⎛ ⎝ VPCCd ⎠

(12)

K1 (t ) = 2μ1sin(ωt ) x (t )

(15)

K2 (t ) = 2μ 2 cos(ωt ) x (t )

(16)

where μ is the algorithm learning rate that adjusts the convergence rate of the algorithm and x (t ) is the difference between the input signal and

⎟

(10)

This signal is then transferred to the unified current controller. 3.2. Voltage regulation

sin( t) The PCC voltage is regulated by supplying reactive power provided by the DC-link capacitor. Therefore, the battery is not involved, which lengthens its lifetime. Based on the literature, the voltage can be controlled using two techniques [22]. In the first method, the reactive power is estimated via global optimization. This method requires the power-flow information of the node, where its voltage needs to be regulated, as well as a communication infrastructure that sends the data for power management. In the second method, the reactive power can be determined and compensated for locally, based on the voltage drop at the PCC. This method detects any changes in the amplitude of the voltage at the PCC and then, based on the difference, the converter circulates leading or lagging VAR into the grid. This technique is more suited to an EV connection in a residential house, as it does not need

K1 x(t)

u(t) K2 cos( t)

Fig. 4. Algorithm of the harmonic controller.

190

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

the house) is needed to perform this task. Available wireless AC current meters with appropriate accuracy and minimum delay are suggested so that the meter can measure the instantaneous value of iL and send it to the controller of the EV charger via a simple communication network. As a result, no additional wiring between the EV charger and the household network is required. However, it should be noted that the wireless current meter is supposed to be able to monitor the waveform of iL . As the delay of communication systems is inevitable, this parameter also needs to be determined and addressed in the control software. The most commonly used controllers (DSP, FPGA, etc.) have serial communication interfaces that can be used for this purpose. For instance, the enhanced controller area network (eCAN), RS485 and RS232 interfaces on the TMS320F28335 device can be used to import data from the wireless current meter [38].

its fundamental component. Based on the calculated K1 (t ) and K2 (t ) , y (t ) from Eq. (14) is obtained. Next, Eq. (12) can be used to obtain ihref . The ihref must be converted to the DQ components, Idhref and Iqhref , before applying to the unified current controller. So Eq. (1) is re-applied and Idhref and Iqhref are added to the active and reactive reference inputs as shown in Fig. 1. Among the different techniques for harmonic detection and compensation, according to the literature, this technique has the advantage of its simplicity and robust structure. The algorithm is also very robust against its internal parameter variations and the external disturbances. The accuracy and stability of the adaptive filter is analyzed and verified in [36]. 3.4. Active power control for charging/discharging the EV battery The active power control generates the active power reference (Pref ) which utilizes charging or discharging of the EV battery in the range of 0–4.5 kW. This range is varied dependent upon both the battery voltage (VB ) and the reactive power demands (Qref1 and Qref2 ). The active power reference is estimated by:

Pref =

2 2 Sref −Qref

4. Stability analysis of the proposed unified control system The stability of the proposed control system is analyzed in this section. Fig. 5, that is used for this purpose, represents the block diagram of the current controller and the plant in a linear time-invariant (LTI) system. The stability analysis of such an LTI control block diagram using an L filter has been performed in [39], verifying that such a system is stable for all values of k p and ki for the two PI controllers. As a symmetrical LC filter is used for the proposed system, the transfer 1 function is used to represent the proposed plant. Since it is LCs 2 + Ls + 1 assumed that the actual voltage and its relative feedforward terms completely cancel each other, they are not shown in this model. The D component (Idhref ) of Ihref generated by Eq. (12) and the reference signal of the charging/discharging controller (Idref1) compose Idref . Similarly, Iqhref , Iqref1 and Iqref2 make Iqref . The signal iβ is generated by the OSG block. The DQ transformation of i α and iβ is estimated by the matrix in Eq. (1). Accordingly, Idq̂ (t ) can be determined by substituting [i α,iβ]T into the DQ transformation as follows:

(17)

where Pref is the determined active power reference, Sref = 4.5 kVA is the apparent power representing the total capacity of the EV charger and Qref = Qref1 + Qref2 is the reference reactive power determined by Eqs. (8) and (11). It should be noted that Qref may contain both sinusoidal (fundamental) and non-sinusoidal (non-fundamental) components due to measuring iL that might be distorted by harmonics. In presence of harmonics, Qref should be Nref which is identified as the non-active power in single-phase non-sinusoidal networks according to the IEEE 1459-2010 standard [37]. Idref1 is determined by:

(± ) Idref1 = Pref / VB

(18)

The positive sign of Idref1 represents the discharging mode and the negative sign the charging mode. When there is no reactive power demand, Qref is zero and the EV battery can be charged or discharged with the full capacity of the EV charger. In this charging mode, when VB is 200 V, the maximum charging current is 22.5 A (level 2 charging according to SAE J1772 standard). Such a charging current can be reduced to 10 A, while VB is increased to 400 V. In another case, if Qref increases in the range of 0–2.5 kVAR demanded by inductive and/or non-linear loads, Pref is reduced subsequently, but not to less than 3.5 kW to meet the total capacity (4.5 kVA) of the EV charger. In this system, the range of active power is set to be between 3.5 kW and 4.5 kW and the range of reactive power is adjusted between 0 and 2.5 kVAR. As mentioned above, the reactive power controller and the harmonic controller require measurement of the load current iL . Such a measurement is not addressed in the infrastructure of the proposed EV charger and an external current meter located at the load side (inside

Id̂ (t ) =

Iq−Iqref Id I I −I + dref + sin(2θ)− d dref cos(2θ) 2 2 2 2

(19)

Iq̂ (t ) =

Iq

Iq−Iqref Id−Idref sin(2θ)− cos(2θ) 2 2

(20)

2

+

Iqref 2

+

The double-frequency terms of Eqs. (19) and (20) converge to zero when Id and Iq approach Idref and Iqref , performed by appropriate tuning

(

1

)

of the PI controllers with the equation k p 1 + k i s . Therefore, eliminating these two terms, Id̂ (t ) and Iq̂ (t ) can be rewritten as:

⎯⎯⎯⎯⎯⎯→ ⎯⎯⎯⎯⎯→ ⎯ ⎯⎯⎯⎯⎯→ 1 ⎯→ Id̂ (t ) − Idref = ( Id − Idref ) 2

(21)

⎯⎯⎯⎯⎯⎯→ ⎯⎯⎯⎯⎯→ ⎯ ⎯⎯⎯⎯⎯→ 1 ⎯→ Iq̂ (t ) − Iqref = ( Iq − Iqref ) 2

(22)

According to Eqs. (21) and (22) the block diagram of Fig. 5 could be

Fig. 5. Block diagram of the controller in the DQ frame and the plant that includes an LC grid-side filter.

191

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

redrawn. Fig. 6 represents the block diagram of the proposed system along with the reference-signal algorithms. As shown, a double control structure is used for Id control and a triple control structure for Iq control. The inner Id control loop is modeled with a transfer function as follows:

GId − Id̂ (s) =

1 LCs 2 + Ls + 1

GIq − Iq̂ (s) =

GIq − loop (s) = GPI2 (s) ∗GIq − Iq̂ (s) ∗

The transfer function is between Id and Id̂ and represents the magnitude of the AC current. Accordingly, the transfer function of the inner current loop, including the delay (Ts ) caused by the sampling time (1∗10−7 s), is represented as:

1−e−sTs sTs

(24)

Gq (s ) =

G Vl − loop (s ) =

The harmonic-detection algorithm that is explained in Section 3.3 operates as a second-order notch filter as follows [36,40]: (26)

G hq − loop (s ) =

(31)

GIq − loop (s) ∗G h (s ) 1 + GIq − loop (s) ∗G h (s )

(32)

The bode plot of G hq − loop (s ) is similar to the bode plot of G hd − loop (s ), as GId − loop (s) and GIq − loop (s) have the same gains. According to the bode plots of the loops, after considering the time delay caused by the sampling time, the Id and Iq control loops have acceptable bandwidths to track the reference signals. The bode plots also confirm the wide stability margins, even with changing the non-linear loads and the realistic low-voltage network parameters. The gain values of the PI controllers are tuned using the Zigler-Nichols method that is explained in [41]. The simulation and experimental results in Sections 6 and 7 verify the robustness of the proposed system in the case of changing the loads. The results verify that the active/reactive power control and voltage regulation are performed in milliseconds, which represents a very fast and reliable response of the system.

GId − loop (s) ∗G h (s ) 1 + GId − loop (s) ∗G h (s )

GIq − loop (s) ∗GPI3 (s) 1 + GIq − loop (s) ∗GPI3 (s)

where the GPI3 (s) controller parameters are k p = 5, k i = 0.05. The bode plot of G Vl − loop (s ) is shown in Fig. 7(d). Similarly, the transfer function of the outer harmonic-detection loop, which is between Iqhref and Iq̂ , is

where ω is the angular frequency and ω b is the bandwidth of the notch filter. In this paper, this algorithm is designed with a bandwidth of 290.0977 rad/s that is tuned to perfectly attenuate a 50 Hz input signal so that the higher order harmonics can be extracted accurately. As mentioned before, such a filter is robust and does not affect the dynamic response of the system. Fig. 7(b) shows the bode plot of Eq. (26) and proves the stability of the harmonic detection algorithm. The transfer function of the outer harmonic-detection loop, which is between Idhref and Id̂ , is given as follows:

G hd − loop (s ) =

(30)

For the outer voltage-control loop, GPI3 (s) is used as the voltage con∗ troller. The transfer function of the loop, which is between VPCCd and Iq̂ , is provided as:

(25)

s 2 + ω2 G h (s ) = 2 s + ω b s + ω2

(29)

GIq − loop (s) 1 + GIq − loop (s)

GId − loop (s) 1 + GId − loop (s)

1−e−sTs sTs

The bode plot of the transfer function is similar to the bode plot of GId − loop (s) , as k p and k i of GPI2 (s) are similar to the gains of GPI1 (s) . Accordingly, the transfer function of the outer reactive-power control, which is between Iqref1 and Iq̂ , is shown as:

where the GPI1 (s) controller parameters are k p = 5, k i = 0.2 . The bode plot of the transfer function is shown in Fig. 7(a). Accordingly, the transfer function of the outer active power-control loop, which is between Idref1 and Id̂ , is given as:

Gp (s ) =

(28)

The transfer function of the inner current-control loop, including the delay caused by the sampling time, is given as:

(23)

GId − loop (s) = GPI1 (s) ∗GId − Id̂ (s) ∗

1 LCs 2 + Ls + 1

(27)

The bode plot of the loop is shown in Fig. 7(c). For the control of Iq , a triple-loop structure is employed. For the inner Iq control loop, the transfer function between Iq and the Iq̂ , is given as:

Fig. 6. Block diagram of the proposed system for bode plot analysis: (a) Id control loop, (b) Iq control loop.

192

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

Fig. 7. Bode plots of the system: (a) inner Id and Iq current loop, (b) inner harmonic detection loop, (c) outer harmonic detection loop, (d) outer voltage control loop.

5. Control system of the interleaved two-leg buck-boost DC/DC converter

the stability margin of the system [32]. In the designed system, as the DC-link voltage regulation is performed by the DC/DC converter, the unified current controller is isolated from such a disturbance. Moreover, a large capacitor is not required, reducing the size and total cost. Using the DC/DC converter, the minimum input current ripple is obtained at the 50% duty cycle, as shown in Fig. 8. The input peak-to-peak current ripple can be obtained as:

An interleaved two-leg buck-boost DC/DC converter is utilized between the EV battery and the single-phase AC/DC converter, as shown in Fig. 1. The DC/DC converter is responsible to regulate the DC-link voltage at 400 V despite the wide range of the battery's voltage. A dutyratio controller monitors the battery voltage and calculates the appropriate duty ratio as:

VD 1 = VB 1−D

ΔIL1,L2 =

(33)

VD−2∗VB ⎧− D, D < 0.5 ∗ fsw ∗L ⎨ ⎩1−D, D > 0.5

(34)

where fsw is the switching frequency and L is the total DC inductance obtained by L1 + L2 . The interleaved two-leg buck-boost DC/DC converter operates with a wide range of battery voltage, between 100 V and 400 V. A lithium-ion battery is a reliable battery type that can be charged or discharged in this voltage range, so that it is suitable for EV application. The current ripple peak-to-peak versus duty ratio is shown in Fig. 8. The EV battery experiences zero ripple at 400, 200, and 0 V (However, a functional battery voltage cannot be 0 V in any operating

where VD is the DC-link voltage, VB is the battery voltage, and D is the estimated duty cycle. In this case, the AC/DC converter is not responsible for regulating the voltage of the DC-link capacitor. As a result, the double-frequency (2-f) ripple of the DC-link voltage caused by the single-phase rectification does not propagate into the unified currentcontrol loop. Conventionally, the 2-f ripple term could be reduced by a large capacitor or lowpass filter, but with high cost, delay or narrowing 193

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

Inductor Current Ripple ( I)

2.5

support and harmonic reduction, at the same time. Moreover, as an advantage of the system, the results show that supporting the ancillary functions does not add extra stress on the EV battery. To do so, firstly, the optimum performance of the proposed control system is verified for voltage regulation and reactive power support. Then, secondly, the harmonic compensation is added with the two existing functions, and the operation of the proposed system is analyzed for all three functions. Importantly, in the two testing conditions, the EV battery is being charged with 1000 W of active power absorbed from the grid as the fourth function of the EV charger. Finally, the proposed system is again analyzed by a 24-h real dynamic load extracted for a residential house in Australia.

Max I

2 1.5

300 V

100 V

1 0.5 0 0

0.2

0.4

0.6

0.8

1

Duty Ratio Fig. 8. Inductor current ripple versus the duty ratio.

6.1. Voltage regulation and reactive power support condition). Such a ripple cancellation occurs because of the 180° phase shift between the inductor currents at these three voltages, so that IL1 and IL2 can cancel each other out completely and, accordingly, no current ripple passes through the EV battery. This technique reduces the stress on the battery, and thus increases its cycle life. It should be noted that the quantitative improvement of the battery lifetime is not investigated in this research at this stage. On the other hand, the maximum ripple of 2 A can be seen at 100 V and 300 V, where the duty cycle is estimated at 75% and 25% respectively. At these voltages, the phase difference between IL1 and IL2 is less than 180° and the two currents cannot cancel each other out 100%. The performance evaluation of the DC/DC converter is evaluated in the following section.

Voltage drop is a common phenomenon in a household network; it is normally caused by load variations and network faults. Fig. 10(a) represents the 0.1 pu voltage reduction caused by adding a 1000 W + j1500 VAR load to the former 1000 W load between 0.1 s and 0.18 s while the EV charger is in the stand-by mode. Fig. 10(b) shows the increase in the amplitude of source current i s caused by the voltage reduction and the load variation. Fig. 10(c) shows that the battery needs to be charged by absorbing 1000 W from the grid. The negative polarity implies the charging mode (G2V operation). Inductive loads also demand 1500 VAR of reactive power as estimated by the Qref1 signal. This signal is calculated by the reactive-power controller (Eqs. (8)–(10)). Fig. 10(d) illustrates the operation of the EV charger between 0.1 s and 0.18 s. As the voltage has dropped by 0.1 pu, the EV charger needs to operate in a capacitive mode to circulate VAR into the line. Based on the network impedance, for this test 500 VAR is needed to compensate for the 0.1 pu voltage drop. In addition, the inductive loads in the house have already needed 1500 VAR. Accordingly, the reactive power controller causes the EV charger to make the current further lead the voltage as shown in Fig. 10(d). As a consequence, the voltage is regulated to 1 pu over the testing time interval. Especially as, sometimes, the line voltage might be increased due to adding capacitive loads, the inductive operation of the EV charger is also simulated, as shown in Fig. 10(e). In this situation, the current leads the voltage to circulate lagging VAR to the network. The supplied reactive power Qc which is required to compensate for the voltage drop (500 VAR) and the reactive power demand (1500 VAR) is shown by 2000 VAR in Fig. 10(f). Pc shows that the EV battery is being charged by 1000 W, while the EV charger operates either in a capacitive or inductive mode by circulating

6. Simulation results of of the multi-functional EV charger controller

0

L2 L1

-4

2.7 A

B

-10

-6 6.00005

(a)

6.0001

i

i

L1

-2

L2

Zero ripple

-4

40 20 0

i

-6 -8

6.00015

Duty cycle = 50%

0

B L2 L1

-5

i 6

i (A),i (A),i (A)

0

L2

L1

-2

Duty cycle = 25%

i

i

B

i (A),i (A),i (A)

The proposed control system for the EV charger is designed, simulated, and its operation is verified by several case studies using the MATLAB/Simulink software. Fig. 9 shows the operation of the buckboost DC/DC converter that was presented in Section 5. Fig. 9(a) shows the response of battery currents IL1, IL2 and IB when the battery voltage is 300 V, the DC-link voltage 400 V, and accordingly the calculated duty cycle is 25%. As can be seen, the battery current IB experiences a 2.74 A ripple which validates the theoretical analysis of the current ripple in Section 5. The 2.74 A ripple is the highest ripple during the total charging region of the battery (0–400 V). It should be noted that the highest ripples of IB (2.74 A), using an interleaved two-leg buck-boost DC/DC converter, is still smaller than the ripples of IB (5.5 A) when a conventional one-leg buck-boost DC/DC converter is used. This comparison is shown in Fig. 9(a). Fig. 9(b) shows zero current ripple of IB when the battery voltage is 200 V, the DC-link voltage 400 V and the estimated duty cycle is 50%. As discussed, at the other voltages, the current ripple of IB is confined between 0 and 2.74 A. This result validates the efficacy of the DC/DC converter in terms of reducing the current stress on the EV battery. The parameters presented in Table 2 are used to design and test the system's performance. Generally, a 4.5 kVA EV charger is connected to a single-phase network that models a 2000 W 1500 VAR residential house. To test the reliability of the system against the harmonics of the current, the load current is distorted manually and the total harmonic distortion (THD) content is increased to 23.2%. To meet the power limitation of the designed EV charger, this THD is chosen, as normally the THD of a house does not go beyond this value (23.2%). The proposed EV charger will also be tested with 35% THD in Section 6.3 and 65% THD in Section 7. Moreover, to create a voltage drop and subsequently verify the operation of the voltage controller (Section 3.2), the test feeder is assumed to be inductive (0.23 + j0.71 Ω). The operation of the proposed system, analyzed in this section, is intended to verify that the unified control system, can address all four functions, charging/discharging operation, voltage regulation and reactive power

-20

B

6

6.00005

Time (s)

6.0001

-40 6.00015

(b) Fig. 9. Battery-side currents during charging mode including inductor currents IL1, IL2 and battery current IB : (a) when battery voltage is 300 V, DC-link voltage 400 V and the calculated duty cycle is 25%, (b) when battery voltage is 200 V, DC-link voltage 400 V and the estimated duty cycle is 50%.

194

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al. 1

v (pu),i (pu)

1

c

v (pu)

0.1pu

-1 0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

-1 0.06

0.22

(a)

0.1

0.12

0.14

0.16

0.18

0.2

0.22

0.16

0.18

0.2

0.22

0.16

0.18

0.2

0.22

(d)

i (A)

c

20

s

0

-40 0.06

0

PCC

-20

0.08

0.1

0.12

0.14

0.16

0.18

0.2

-1 0.06

0.22

1000

P (W),Q (VAR)

(b)

1500 VAR

0.08

0.1

0.12

0.14

(e)

2000

2000 VAR

1000

c

ref1

0

-1000 W

c

ref

0.08

1

v (pu),i (pu)

40

P (W),Q (VAR)

0

PCC

PCC

0

-1000 0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

0

0.06

0.22

-1000 W

-1000 0.08

0.1

0.12

0.14

Time (s)

Time (s)

(c)

(f)

Fig. 10. Operation of the EV charger with three functions (voltage regulation, reactive power support, and charging operation): (a) PCC voltage, (b) source current, (c) active and reactive power reference signals, (d) capacitive operation of the EV charger, (e) inductive operation of the EV charger, (f) active and reactive power supplies.

6.2. Voltage regulation, reactive-power support and harmonic reduction

a lagging or leading 2000 VAR to the line. As presented in Fig. 10(f), it can be concluded that the voltage regulation and reactive power support do not demand any active power from the system, thus the EV battery is not affected.

I

The aim of this case study is to analyze the performance of the proposed system when the load current is distorted, and the PCC voltage is not stable. In Fig. 11, as well as the same voltage reduction by 0.1 pu (Fig. 11(a)) and the need for 1500 VAR of reactive power, the

II

I

II

40

0.1 pu

i (A)

0

20 0

href

PCC

v (pu)

1

-1

0.05

0.1

0.15

0.2

0.25

-20 -40

0.3

0.05

0.1

0.15

(a)

1

(d)

THD (%)

PCC

v (pu)

35%

0

-1

0.05

0.1

0.15

0.2

0.25

0.3

0.25

0.3

0.25

0.3

23.2%

25%

IEEE 1547

15% 5% 0

0.05

0.1

3.55% 0.15

0.2

(e)

(b) 0

P (W)

50

c

0

s

i (A)

0.2

-50

-500

-1000 W

-1000 -1500

0.05

0.1

0.15

0.2

0.25

0.05

0.3

Time (s) (c)

0.1

0.15

0.2

0.25

0.3

Time (s) (f)

Fig. 11. Operation of the EV charger with four functions (voltage regulation, reactive power support, harmonic reduction and charging operation) during the time intervals I and II: (a) PCC voltage before voltage regulation, (b) PCC voltage after voltage regulation, (c) source current, (d) harmonic-detection reference signal, (e) THD of the source current, (f) active power absorbed for charging the battery.

195

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

load current is distorted with the THD content of 23.2%. As shown, this test is performed during two different time intervals I and II. During the time interval I, the system supports voltage regulation and reactive power support while it is charging the EV battery by 1000 W. As a result, the PCC voltage v PCC is regulated to 1 pu in Fig. 11(b) but i s is still distorted by the harmonics. The distorted current and the percentage of the THD during time interval I are shown in Fig. 11(c) and (e) respectively. During the time interval II of the test, the harmonic-reduction function of the unified controller is activated. As a result, the distorted current waveform in Fig. 11(c) reshapes to a non-distorted waveform. This is because of injecting the ihref , supplied by the proposed EV charger (Fig. 11(d)). Consequently, as Fig. 11(e) shows, the THD of i s is reduced to 3.55% which complies with the IEEE 1547 standard. This test validates supporting simultaneously the three ancillary functions while the EV battery is being charged. As Fig. 11(f) shows, activating the harmonic controller slightly increases the charging power Pc . However, as the results confirm, such an inevitable transient disappears quickly and does not have an adverse effect on the system's performance. Fig. 12(a) shows the harmonic spectrum of i s during time interval I, and Fig. 12(b) presents the harmonic spectrum of i s during time interval II. Table 3 summarizes the network parameters before and after the EV charger's operation in this test.

Table 3 Comparison of the network parameters with a static load before and after the EV charger operation. Item

Before

After

Improvement

Load voltage Current harmonics Reactive power supplied from power station Reactive power supplied from EV charger Source-side power factor EV charger's output power factor

0.9 pu 23.2% THD 1500 VAR

1 pu 3.55% THD 0 VAR

10% 85% 100%

0 VAR

2000 VAR

NA

0.89 1

0.98 0.45

9.1% NA

section, the discharging mode is studied. Accordingly, while the EV battery has enough charge, it can act as battery energy storage for a house. As Fig. 13(a) shows, the demanded active power (P) and the power (Pc ) produced by the EV charger are approximately matched over the test period, showing that the designed system can exploit the energy of the battery to cover the load power demand. The negligible difference between P and Pc in the time interval II is caused by the harmonic reduction process that forces the EV charger to either inject or absorb the small amount of active power needed for eliminating the harmonics. However, such a process does not disturb the system's normal operation, as the results show. Although, in reality, an EV battery might not be able to supply the required active power of a house for 24 h a day, to show the reliable dynamic response of the proposed system we were obliged to assume that the battery is adequately charged during that time period. Therefore, for this test, an infinite DC source is used as an EV battery. Fig. 13(b) illustrates the variation of the demanded reactive power. The inductive load demands up to 2000 VAR of reactive power (Q ) between 6 pm and midnight which must be provided by the grid. As can be seen, the EV charger can provide reactive power (Qc ) more than the need of the inductive loads as shown in Fig. 13(b) to regulate the voltage drop as demonstrated in Fig. 13(d). Such an excessive reactive power supply, that is determined by the PI3 controller in Fig. 1, maintains the line voltage at 1 pu. To show the dynamic

6.3. Analyzing the proposed system with a real dynamic load The main objective of this section is to verify the performance of the proposed EV charger when it is connected to a real house that includes a dynamic load. Same time intervals as mentioned in the previous case study are used in this test. The dynamic load, that demands active and reactive powers, was recorded for 24 h a day from a residential house, located in Australia. The load data have been provided by Energex Company, the main power distributor in Queensland, Australia. A nonlinear diode-rectifier load is used at the connection point of the load in order to increase the THD content of the line current to 35%. The proposed system is tested with this condition and the network parameters are recorded subsequently. In the previous section, the designed system was tested in the charging mode of the EV battery and, in this

Mag (% of Fundamental)

Fundamental (50Hz), THD= 23.26% 20 15 10 5 0 0

100

200

300

400

500

600

700

800

900

1000

(a)

Mag (% of Fundamental)

Fundamental (50Hz), THD= 3.55% 20 15 10 5 0 0

100

200

300

400

500

600

700

800

900

Freuqency (rad/s)

(b) Fig. 12. Harmonic spectrum of the source current i s : (a) during time interval I, (b) during time interval II.

196

1000

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

I

I

II P

v (pu)

c

2000 1000

04:48 AM

02:24 PM

09:36 AM

s

c

1000 0 12:00 AM

04:48 AM

09:36 AM

02:24 PM

L

Q 07:12 PM

1.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2

v

PCC

04:48 AM

02:24 PM

(d)

(After)

07:12 PM

12:00 AM

07:12 PM

12:00 AM

s

PF

L

PF

c

04:48 AM

09:36 AM

02:24 PM

Time (h)

50%

THD(%)

09:36 AM

PF

(b)

(e)

Before

45%

(Before)

0.9

12:00 AM

12:00 AM

PCC

1

0.8

c

Q

2000

v

1.1

12:00 AM

12:00 AM

PF , PF , PF

c

07:12 PM

(a)

3000

Q(W), Q (W)

PCC

P

0 -1000 12:00 AM

II

1.2

c

P(W), P (W)

3000

35% 25%

After

15% 5% 0

12:00 AM

04:48 AM

09:36 AM

02:24 PM

07:12 PM

12:00 AM

Time (h) (c)

Fig. 13. Operation of the proposed system connected to the real dynamic load: (a) active power demanded and injected to network, (b) reactive power demanded and injected to network, (c) THD of the source current, (d) PCC peak voltage, (e) power factor.

tested when there is 65% THD on the load current generated by an RLC non-linear load. In this case, the EV charger is supposed to operate as a single-phase APF. Fig. 15 shows that both the load-current (iL ) and the source-current (i s ) are in phase with the grid voltage (v PCC ), which means that no reactive power is demanded by the load. As shown, the EV charger perfectly reduces the THD of i s to 2.5% which complies with the IEEE 1547 standard. In this paper, the current is in the same phase position with voltage during the V2G operation. The crest factor (CF) of i s , which is defined as the ratio of the peak value to the RMS value of a current waveform, calculated by Eq. (35), is also decreased from 1.7 to 1.41.

response of the proposed system, the harmonic-reduction function is set to operate during time interval II. Fig. 13(c) shows that the THD of the source-side current is reduced from 35% to around 4%; that meets the IEEE 1547 standard. Finally, Fig. 13(e) shows the power factors at the source side, the load side and the converter side. Before 2:00 pm, when the voltage has not substantially dropped, only reactive power is injected to respond to the demand of the inductive load. As a result, the source-side power factor PFs is almost unity. After this time, both the excessive reactive power shown in Fig. 13(b) and the small reactivepower deviation caused by the harmonic-reduction function reduce the source-side power factor gradually by around 0.1. Consequently, the results in Fig. 13 verify that the proposed system could supply the active power, reactive power, and voltage regulation while the THD is kept below 5%. Table 4 shows the transitions of the network parameters before and after the EV charger's operation in this test. As mentioned, the EV charger is designed to provide 4.5 kVA of apparent power. If we consider the highest produced active and reactive powers at 9:00 pm, 2.8 kW at shown in Fig. 13(a) and 2.5 kVAR in Fig. 13(b), using S = P 2 + Q 2 , 37.53 kVA of apparent power (S ) is transferred to the grid, which is still less than the nominal capacity (4.5 kVA) of the designed EV charger.

CF =

Peak current RMS current

(35)

In another test in Fig. 16, the power factor (PF) is set to 0.8 leading and the load demands reactive power of 200 VAR. In this test, the reactivepower support and harmonic-reduction function of the proposed EV charger are tested simultaneously. Accordingly, the EV charger is assigned to work as a single-phase APF and a single-phase STATCOM. As can be seen, after the operation of the EV charger, the THD of i s is reduced considerably (from THD 65% to 3.1%) while it also regulates the PF of the source side to be almost unity. The CF of i s is again regulated from 1.7 to 1.4. Comparing Figs. 15 and 16 verifies that the reactive power support does not change

7. Experimental results of the multi-functional EV charger controller In this section, the performance of the proposed system is verified with experimental results. The prototype of a single-phase bidirectional EV charger is shown in Fig. 14. A four-leg SEMISTACK – IGBT is used to implement the single-phase DC/AC inverter and the interleaved two-leg buck-boost DC/DC converter. A DC power supply is used to represent an EV battery so that only the V2G mode is tested in this experiment. A DSP TMSF28335 controller is programmed in C language in Code Composer Studio V6.02 and used to monitor, process and generate the PWM signals. The performance of the proposed EV charger, designed with parameters as given in Table 2, is experimentally validated by various tests. In order to test the efficacy of the proposed unified control system, first the harmonic-reduction function of the proposed EV charger is

Table 4 Comparison of the network parameters with a dynamic load before and after the EV charger operation.

197

Item

Before

After

Load voltage Current harmonics Active power supplied from power station Active power supplied from EV charger Reactive power supplied from power station Reactive power supplied from EV charger Source-side power factor Load-side power factor EV charger's output power factor

1–0.95 pu 35% THD 0–2800 W 0 0–2000 VAR 0 VAR 0.97 0.97 1

1 pu 4% THD 0 0–2800 W −500 to 0 VAR 0–2500 VAR 1–0.9 0.97 1–0.3

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

DC/AC inverter and Current AC-side and &voltage DC-side DC/DC converter sensors Inductors

Single-phase load

simultaneously providing the three ancillary functions including voltage regulation, reactive power support and harmonics reduction. It should be noted that at the operating point, as shown in Fig. 19(a), the DC-link voltage VD is perfectly regulated at 220 V without exhibiting any transient. As mentioned in Section 5, this task is performed by the interleaved two-leg buck-boost DC/DC converter that regulates VD regardless of varying battery voltage (VB ). In another test, in Fig. 20, such a voltage regulation is revalidated while VB is changing from 150 V to 220 V and vice versa. As can be seen, the interleaved twoleg buck-boost DC/DC converter continuously changes the duty cycle of the switching signals, and as a result VD is maintained at 220 V during the test period. Fig. 21 shows the battery current, IB , when load changes. As shown during the test, P , which is the active power demand, increases from 116 W to 233 W, and then decreases to 116 W while Q , that is the reactive power demand, rises from 0 to 155 VAR. Accordingly, IB , the battery current, changes during each time interval to provide the required energy for the load. IBripple,rms , that represents the RMS value of the ripples of the battery current, is measured at 70 mA. The maximum current ripple allowed for Li-ion batteries and/or lead-acid batteries is between 5% and 10% of the mean value of the charging current [11,42]. In this test when P = 233 W and Q = 155 VAR, IBmean = 1.5 A, I thus the current ripple is calculated as Bripple,rms = 4.5% which is in an IBmean acceptable range. It should be noted that while the load demand increases, IBmean increases subsequently, whereas IBripple,rms changes slightly. This guarantees that the ripples which impose stress on the battery will be diminished at the higher load.

DSP controller F28335

DC power supply Step-down transformer Fig. 14. Experimental setup.

the amplitude of i s , which means that this function does not demand any energy from the EV battery. In the previous tests, the EV charger does not inject/absorb active power to/from the grid. In the third test in Fig. 17, the EV charger starts injecting active power to the grid (V2G operation) while it supplies reactive power. Fig. 17(a) shows the STATCOM operation of the EV charger in inductive mode and Fig. 17(b) shows the operation in capacitive mode. As can be seen, the DC-link voltage VD and the battery voltage VB are regulated during both inductive and capacitive modes, verifying the steady-state performance of the designed controller. Fig. 18 shows the dynamic response of the proposed system during load switching from 20 W to 200 W. As shown, the system is quite stable, as no harmful transient occurs on VD, VB and the grid-side current i c validating the appropriate dynamic response of the system. Finally, in the last test the three ancillary functions that are proposed in this paper are validated simultaneously. In this test, as Fig. 19(a) shows, after the standby region, v PCC experiences a voltage drop of 10 V (∼0.1p.u.) due to manually increasing the load demand from 0 to 860 W. At the same time, applying non-linear loads increases the harmonics of iL as well as deviating the power factor from unity. Such a transition is shown in Fig. 19(b) before operating the EV charger. At the operating point, the EV charger starts operating in both APF and STATCOM mode. As Fig. 19(a) shows, after the operation, the 10 V voltage drop is removed, the harmonics are reduced (shown in Fig. 19(c)) as well as the PF is approached close to unity. To summarize, this test verifies that the proposed EV charger is capable of

8. Conclusion A unified control system for an EV on-board charger is designed and implemented which can significantly improve the power quality in a low-voltage household network. The proposed EV on-board charger can operate as a multifunctional four-quadrant STATCOM and an APF while it is charging or discharging the EV battery. The main distinction of the designed EV charger is that it can simultaneously provide the three ancillary functions, namely voltage regulation, reactive power compensation, and harmonics reduction, while operating in the V2G/G2V mode. Despite an extra current sensor that is needed to measure the load current, this design does not require additional components nor demand any extra energy from the EV battery. The stress on the EV battery is also reduced by using an interleaved two-leg buck-boost DC/ DC converter that could be effective to increase the battery's lifetime. The robustness of the proposed system is analyzed with different types of load under several operating conditions. A stability analysis of

Fig. 15. Experimental results – APF performance: grid voltage v PCC (Ch.1, 80 V/div), source current i s (Ch.3 8 A/div), load current iL (Ch.4 8 A/div) and converter's output current i c (Ch.2 4 A/div). Load demand: P = 485 W, Q = 0 VAR, CF = 1.7 and PF = 1. EV charger supply: Pc ≈ 0 W, Qc ≈ 0 VAR.

198

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

Fig. 16. Experimental results – APF and STATCOM performance: grid-voltage v PCC (Ch.1, 80 V/div), source-current i s (Ch.3 8 A/div), load-current iL (Ch.4 8 A/div), and converter's output current i c (Ch.2 4 A/div). Load demand: P = 485 W, Q = 200 VAR, CF = 1.7 and PF = 0.8 leading. EV charger supply: Pc ≈ 0 W, Qc = 200 VAR, PF = 0.2 lagging.

significant and important step in improving the low-voltage distribution system’s performance using distributed energy resources.

the proposed control system is performed to show the reliability of the system. Finally, the simulation in MATLAB Simulink and experimental results validate the feasibility of the proposed EV charger. Using the proposed system, each EV on-board charger will be able to contribute to solving the network’s power quality problems. This ability would be a

Fig. 17. Experimental results – V2G and STATCOM performance: (a) inductive mode Pc = 165 W, Qc = 150 VAR lagging, (b) capacitive mode: Pc = 165 W, Qc = 150 VAR leading. Grid voltage v PCC (Ch.1, 70 V/div), inverter-current i c (Ch.2 4 A/div), DC-link voltage VD (Ch.3 80 V/div) and battery voltage VB (Ch.4 80 V/div).

199

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

Fig. 18. Experimental results-transient test during load switching from Pc = 20 W to Pc = 200 W: Grid voltage v PCC (Ch.1, 70 V/div), inverter-current i c (Ch.2 1.5 A/div), DC-link voltage VD (Ch.3 80 V/div) and battery voltage VB (Ch.4 80 V/div).

Fig. 20. Experimental results – DC-link voltage regulation versus battery voltage. Voltage across the top switch in half bridge 1 (Ch.1, 100 V/div), voltage across the top switch in half bridge 2 (Ch.2, 100 V/div), VD (Ch.3 100 V/div), VB (Ch.4, 100 V/div).

Fig. 21. Experimental results-battery current during load switching. Grid voltage v PCC (Ch.1, 100 V/div), inverter-current i c (Ch.2 5 A/div), and battery current IB (Ch.4 2 A/ div).

Acknowledgement The authors of this paper forward their sincerest thanks to ENERGEX, the electric power distribution company owned by the Queensland government, Australia, for providing the load data of the residential house model for this research study. References

Fig. 19. Experimental results – STATCOM and APF performance for reactive power support, voltage regulation and harmonics reduction: v PCC (Ch.1, 60 V/div), i s (Ch.2 3 A/ div), VD (Ch.3 80 V/div) and VPCCd (Ch.4, 80 V/div).

[1] Chau K, Wong Y. Overview of power management in hybrid electric vehicles. Energy Convers Manage 2002;43:1953–68. [2] Pinto J, Monteiro V, Goncalves H, Afonso JL. Onboard reconfigurable battery charger for electric vehicles with traction-to-auxiliary mode. IEEE Trans Veh Technol 2014;63:1104–16. [3] Taghizadeh S, Hossain M, Lu J. Bidirectional isolated vehice to grid (V2G) system: an optimized implementation and approach. 2015 IEEE PES Asia-Pacific on power

200

Applied Energy 215 (2018) 186–201

S. Taghizadeh et al.

2016;65:1007–20. [24] Boynuegri A, Uzunoglu M, Erdinç O, Gokalp E. A new perspective in grid connection of electric vehicles: Different operating modes for elimination of energy quality problems. Appl Energy 2014;132:435–51. [25] Shahnia F, Ghosh A, Ledwich G, Zare F. Predicting voltage unbalance impacts of plug-in electric vehicles penetration in residential low-voltage distribution networks. Electr Power Compon Syst 2013;41:1594–616. [26] Shahnia F, Ghosh A, Ledwich G, Zare F. Voltage unbalance sensitivity analysis of plug-in electric vehicles in distribution networks. 2011 21st Australasian Universities on power engineering conference (AUPEC). 2011. p. 1–6. [27] Shahnia F, Majumder R, Ghosh A, Ledwich G, Zare F. Voltage imbalance analysis in residential low voltage distribution networks with rooftop PVs. Electr Power Syst Res 2011;81:1805–14. [28] Javadi A, Al-Haddad K. A single-phase active device for power quality improvement of electrified transportation. IEEE Trans Ind Electron 2015;62:3033–41. [29] Javadi A, Ndtoungou A, Blanchette HF, Al-Haddad K. Power quality device for future household systems with fast electric vehicle charging station. 2015 IEEE on vehicle power and propulsion conference (VPPC). 2015. p. 1–6. [30] Kabir M, Mishra Y, Ledwich G, Xu Z, Bansal R. Improving voltage profile of residential distribution systems using rooftop PVs and Battery Energy Storage systems. Appl Energy 2014;134:290–300. [31] Kabir M, Mishra Y, Ledwich G, Dong ZY, Wong KP. Coordinated control of gridconnected photovoltaic reactive power and battery energy storage systems to improve the voltage profile of a residential distribution feeder. IEEE Trans Ind Inform 2014;10:967–77. [32] Khajehoddin SA, Karimi-Ghartemani M, Jain PK, Bakhshai A. DC-bus design and control for a single-phase grid-connected renewable converter with a small energy storage component. IEEE Trans Power Electron 2013;28:3245–54. [33] O’Loughlin M. An interleaving PFC pre-regulator for high-power converters. Texas Instrum 2006:1–14. [34] Wang H, Dusmez S, Khaligh A. Design and analysis of a full-bridge LLC-based PEV charger optimized for wide battery voltage range. IEEE Trans Veh Technol 2014;63:1603–13. [35] Khaligh A, Dusmez S. Comprehensive topological analysis of conductive and inductive charging solutions for plug-in electric vehicles. IEEE Trans Veh Technol 2012;61:3475–89. [36] Karimi H, Karimi-Ghartemani M, Iravani MR, Bakhshai AR. An adaptive filter for synchronous extraction of harmonics and distortions. IEEE Trans Power Deliv 2003;18:1350–6. [37] Mansour A, Chengning Z, Nasry H. Measurement of power components in balanced and unbalanced three-phase systems under nonsinusoidal operating conditions by using IEEE standard 1459–2010 and Fourier analysis. 2013 international conference on technological advances in electrical, electronics and computer engineering (TAEECE). 2013. p. 166–71. [38] Instruments T. TMS320F28335 digital signal controllers data manual, ed; 2007. [39] Ebrahimi M, Khajehoddin SA, Karimi-Ghartemani M. Fast and robust single-phase current controller for smart inverter applications. IEEE Trans Power Electron 2016;31:3968–76. [40] Luo S, Hou Z. An adaptive detecting method for harmonic and reactive currents. IEEE Trans Ind Electron 1995;42:85–9. [41] Hossain M, Mahmud M, Pota HR, Mithulananthan N. Design of non-interacting controllers for PV systems in distribution networks. IEEE Trans Power Syst 2014;29:2763–74. [42] Nguyen V-S, Tran V-L, Choi W, Kim D-W. Analysis of the output ripple of the DC-DC boost charger for Li-ion batteries. J Power Electron 2014;14:135–42.

and energy engineering conference (APPEEC). 2015. p. 1–5. [4] Haidar AM, Muttaqi KM, Sutanto D. Technical challenges for electric power industries due to grid-integrated electric vehicles in low voltage distributions: a review. Energy Convers Manage 2014;86:689–700. [5] Staats P, Grady W, Arapostathis A, Thallam R. A statistical analysis of the effect of electric vehicle battery charging on distribution system harmonic voltages. IEEE Trans Power Deliv 1998;13:640–6. [6] Lee J-Y, Chae H-J. 6.6-kW onboard charger design using DCM PFC converter with harmonic modulation technique and two-stage DC/DC converter. IEEE Trans Ind Electron 2014;61:1243–52. [7] Rahman MS, Hossain M, Lu J. Coordinated control of three-phase AC and DC type EV–ESSs for efficient hybrid microgrid operations. Energy Convers Manage 2016;122:488–503. [8] Bizon N. Energy efficiency of multiport power converters used in plug-in/V2G fuel cell vehicles. Appl Energy 2012;96:431–43. [9] Bizon N. Energy efficiency for the multiport power converters architectures of series and parallel hybrid power source type used in plug-in/V2G fuel cell vehicles. Appl Energy 2013;102:726–34. [10] Rubino L, Capasso C, Veneri O. Review on plug-in electric vehicle charging architectures integrated with distributed energy sources for sustainable mobility. Appl Energy 2017. [11] Kisacikoglu MC, Kesler M, Tolbert LM. Single-phase on-board bidirectional PEV charger for V2G reactive power operation. IEEE Trans Smart Grid 2015;6:767–75. [12] Gorla NBY, Ali K, Lin CC, Panda SK. Improved utilization of grid connected voltage source converters in smart grid through local VAR compensation. IECON 2015-41st annual conference of the IEEE on industrial electronics society. 2015. p. 002532–7. [13] Silvestre C, Sousa DM, Roque A. Reactive power compensation using on board stored energy in electric vehicles. IECON 2012-38th annual conference on IEEE industrial electronics society. 2012. p. 5227–32. [14] Kisacikoglu MC, Ozpineci B, Tolbert LM. EV/PHEV bidirectional charger assessment for V2G reactive power operation. IEEE Trans Power Electron 2013;28:5717–27. [15] Manbachi M, Farhangi H, Palizban A, Arzanpour S. A novel Volt-VAR Optimization engine for smart distribution networks utilizing Vehicle to Grid dispatch. Int J Electr Power Energy Syst 2016;74:238–51. [16] Monteiro V, Pinto J, Exposto B, Gonçalves H, Ferreira JC, Couto C, et al. Assessment of a battery charger for electric vehicles with reactive power control. IECON 201238th annual conference on IEEE industrial electronics society. 2012. p. 5142–7. [17] Jia-zheng L, Yun Z, Bo L. Application of parallel active filter for the electric vehicle AC charge spots. 2012 Asia-Pacific on power and energy engineering conference (APPEEC). 2012. p. 1–4. [18] Rodrigues MC, Souza I, Ferreira AA, Barbosa PG, Braga HA. Integrated bidirectional single-phase vehicle-to-grid interface with active power filter capability. 2013 Brazilian on power electronics conference (COBEP). 2013. p. 993–1000. [19] Fathabadi H. High benefits approach for electrical energy conversion in electric vehicles from DC to PWM-AC without any generated harmonic. Energy Convers Manage 2014;77:324–33. [20] Islam F, Pota H. Virtual active filters for HVDC networks using V2G technology. Int J Electr Power Energy Syst 2014;54:399–407. [21] Rodrigues MC, Souza ID, Ferreira AA, Barbosa PG, Braga HA. Simultaneous active power filter and G2V (or V2G) operation of EV on-board power electronics. IECON 2013-39th annual conference of the IEEE on industrial electronics society. 2013. p. 4684–9. [22] Falahi M, Chou H-M, Ehsani M, Xie L, Butler-Purry KL. Potential power quality benefits of electric vehicles. IEEE Trans Sustain Energy 2013;4:1016–23. [23] Monteiro V, Pinto J, Afonso JL. Operation modes for the electric vehicle in smart grids and smart homes: present and proposed modes. IEEE Trans Veh Technol

201