Voltage Quality Enhancement of Grid-Integrated PV System Using Battery-Based Dynamic Voltage Restorer

: The advancement of power electronic-based sensitive loads drives the power utilities’ devotion to power quality issues. The voltage disturbance could be happening due to fault conditions, switching of loads, energizing of transformers, or integration of highly intermittent energy sources such as PV systems. This research work attempts to enhance the voltage fluctuation of a sensitive load connected to a grid-integrated PV system using a battery-based dynamic voltage restorer (DVR). The proposed battery energy storage-based DVR has two separate controlling stages that are implemented at the DC–DC buck/boost converter of the battery and voltage source converter (VSC) system. Charging and discharging of the battery is operated based on the state-of-charge (SOC) value of the battery and the measured root mean square (RMS) voltage at the point of common coupling (PCC). The VSC of the DVR detection and reference generation control is done appropriately. In the detection control of the VSC, a combination of RMS and dq0 measurement techniques is used, whereas in the reference generation control, pre-fault strategy is implemented to restore both phase jump and magnitude distortions. Symmetrical and asymmetrical voltage sags scenarios are considered and the compensation demonstration is carried out using power system computer-aided design (PSCAD/EMTDC) software.


Introduction
Exploitation of nonrenewable energy sources threatens our planet with extreme exposure to global warming and climate change. Due to this, for the past few years and nowadays, leaders and policy makers overlay utilization of renewable energy sources (RES) with wind and solar photovoltaic (PV) power standing out in particular [1]. Based on the past trends, PV power is one of the fastest-growing technologies which gets vital emphasis and attracts both industries and customers in view of energy shortages and environmental concerns [1][2][3][4]. Especially, it has immensely evolved in the building energy sector for which it is predominantly relevant and this clearly appears through distributed grid-connected systems. The green energy policies and technological advancements play a vital role in helping PV expansion with progressive cost reduction. Expansion of grid-connected PV systems in various energy sectors is expected to continue in the next decade [5].
Even though RES are environmentally friendly and abundant in nature, their output power is not stable due to their source intermittent characteristics. The global trend of grid-connected PV power plant installation indicates a fast pace. The installation is expected to grow more by developing and applying additional regulations and standards to minimize the potential adverse impacts of large There are different topologies and control strategies of DVR in consideration with performance and cost aspects [25]. The topologies of the DVR could be categorized as with energy storage or without energy storage based on the energy storage perspective [29]. Considering the DVR connection and the energy storage aspect, the topology can be categorized into four types, such as without energy storage supply side connected, without energy storage load side connected, selfcharging (capacitor supported), and with external energy storage [20,25,29]. The detection, reference generation, voltage, and current control and modulation aspects of the DVR different controlling schemes are presented in [16,25]. Detection and reference generation are the main parts of the DVR control system [31]. Voltage sag/swell detection is done by measuring the grid voltage and analyzed so that based on sag/swell detection method, the voltage disturbance can be recognized [32]. There are different detection methods such as RMS measurement, dq0 component measurement, wavelet transform, phasor parameter estimation using Kalman filters [25,32]. Among these, dq0 component measurement-based synchronous rotating frame (SRF) control technique is a more popular technique due to its prompt response and simple implementation [25,33]. Generation of reference signal of series-injected voltage will be determined based on the type of energy storage device, its active power supporting capability, and sensitivity of load to voltage disturbances [15,25,32]. Based on this, four basic voltage compensation methods are commonly used such as pre-fault, in-phase, energy optimized, and hybrid compensation methods [25,32]. The only possible way to mitigate the phase jump is to restore the load voltage to the pre-fault value which is addressed by only pre-fault compensation [15,25,28,32].
In this paper, the application of the BES-based DVR to enhance the voltage fluctuation of a gridconnected PV power system is investigated in which the DVR is connected in series between the grid distribution line and the sensitive load. In this study, the proposed BES-based DVR operational time is considered to be high. This is due to the high intermittent scenario of the PV system in addition to any grid fault conditions which all cause voltage fluctuation. This makes the proposed DVR more economical. The output voltage of the DVR is added to the grid distribution voltage instantaneously to be kept a constant voltage at the load during under grid fault condition, during over and under voltage conditions of the PV system, as well as in normal condition. Hence, the grid voltage disturbances have no direct effect on the sensitive load. Therefore, the proposed method is able to eliminate the effect of grid voltage disturbance on sensitive loads. The control algorithm of the DVR system is proposed to overcome voltage sag and swell conditions at the point of common coupling (PCC) so that the sensitive load feeds by a constant voltage. A combination of RMS and dq0 component measurement methods is used for the detection and pre-fault compensation used for reference generation. In addition, a control scheme for the control of the PV system, in which the PV system is integrated with the low voltage (LV) grid side, is designed. The validity of the proposed control algorithm is demonstrated by using the PSCAD simulation. The overall system configuration is presented in Section 2 and the control and compensation strategy is illustrated in Section 3. In Sections 4 and 5, the simulation and results discussion and the conclusion are presented, respectively.

System Configuration
Electrical power is generated by a PV system through the conversion of sunlight irradiation using the PV cells which are semiconductor diodes. As the sunlight reaches the PV cell surface, charge carriers are generated from the cells and the electric current produced flows through the short circuit [34,35]. The common PV cell model is presented in Figure 1a which comprises a diode connected to a series and parallel resistance, Rp and Rs, respectively [35]. The fast depletion of conventional energy resources, global warming and environmental pollution, technological advancements, and a gradual cost minimization have made RES, such as PV and wind, preferable sources of electric power generation [30,36,37]. The power generated by a PV source is DC and it requires a DC-AC conversion system for grid-integrated operation. DC-DC boost converter, DC-AC converter or VSC, filter, and grid coupling transformer are the main components of the grid-integrated PV system [30] which is illustrated in the proposed scheme in Figures 2 and 5. The PV array is operated at a maximum power point tracking (MPPT) and the converter is synchronized to the grid. As shown in Figure 1b, the maximum output power Pmax of the PV system is described in Equation (1) which is found by the product of maximum current Im and maximum voltage V . However, it is rather convenient to use open circuit voltage Voc and short circuit current Isc to estimate the maximum power of cells. To do this, the fill factor (FF) is used which is the fraction of the maximum power that is generated by PV cells to the product of Voc and Isc as expressed in Equation (2) [34]. Voc and Isc will be calculated using Equations (3) and (4) so that the output power of the PV cell Ppv at any time would be found by using Equation (5) where part of the parameters are obtained mainly from the manufacturer data sheets [5].
where V * , β , and T are the V of the PV cells under standard test conditions, the PV cells' temperature coefficient referent to V , and the cell temperature, respectively.
G and G are the incident solar irradiance and the incident solar irradiance under standard test conditions, respectively. I * and β are the PV cells' I under standard test conditions and the temperature coefficient of PV cells referent to I , respectively.
where P and γ are the PV cells' maximum power under standard test conditions and maximum power temperature coefficient, respectively.
where T , G , T , , and T are the environmental temperature, incident solar irradiance at nominal operating cell temperature, the nominal operating cell temperature provided by the manufacturer, and nominal operating cell temperature, respectively.
As it is illustrated in Figure 2, the grid-connected PV power system supplies power for a sensitive load and to serve this load with a stable power supply, BES-based DVR is connected in series with the system. In case of fault conditions and PV intermittency, the proposed DVR will operate to protect the load from being damaged and phase-out. As it is shown in Figure 5, the output of a DC-DC boost converter is connected to the DC link capacitor of the VSC. The output voltage generated by the DC-DC boost converter is to be kept along with the average DC input voltage. As described in Figure 2, the PCC bus is the bus where PV power generation and LV grid are connected, whereas PCC is the point where the BES-based DVR is connected to the output terminals of the PCC bus and the sensitive load through its coupling or injection transformer. The main parameters of the grid-connected PV system and the load are presented in Table 1.  A DVR is used to restore the load-side voltage in case of any voltage fluctuation by injecting a voltage of required magnitude and frequency based on the adequate active power capacity. Based on the availability of adequate energy source, the DVR can exchange both active and reactive power at the PCC to compensate the voltage magnitude and phase of sensitive load which is caused by fluctuation. The injected power is a function of the voltage injected and the sensitive load current [17]. As shown in Figures 2 and 7, the DVR has its own BES to have adequate active power capacity and this BES-based DVR has mainly six components such as: battery, buck/boost converter, DC link capacitor (C ), VSC, LCL filter, and injection transformer. The proper integration and control of these components will result in effective operation of the DVR. The battery charges and discharges based on its capacity, which is measured by its state of charge (SOC). Here, the LCL ripple filters which are placed at the terminals of VSC in both PV generation and BES-based DVR are used for absorbing switching ripples which are produced by the VSC. The design of the LCL filter's parameters is mainly based on its capability to eliminate the most low-and high-order harmonics of the output voltage waveform. The resonant frequency (f ) attenuates the harmonics generated by the sinusoidal pulse width modulation (SPWM) of the inverter, which occurs near the switching frequency (f ) which is 5040 Hz. The f is formulated based on Equation (8), which is dependent on the filter capacitance and filter inductance and obeys the condition expressed in Equation (7) [38][39][40]. Based on this, the filter inductance was chosen to be 100 uH and the capacitor value should be chosen so that the resonant frequency is less than half of the switching frequency [39,40]. Accordingly, a filter capacitance of 55 mF was chosen to avoid any harmonics with the f value of around 68 Hz, which is less than half of the f which is 2520 Hz. During normal power generation of the PV system, the generated power is fed to the grid in unity power factor mode of operation. However, if the PV power faced intermittency or fault exists at any point upstream of the PCC, then the BES-based DVR operates to protect the sensitive load by stunning the fluctuation. The DVR restores voltage only downstream from the PCC and the compensation is carried phase by phase. The parameters of the proposed BES-based DVR are described in Table 2.
where f is the grid frequency, which is 60 Hz.   Figure 2 depicts the schematic of the proposed system which consists of PV system, grid system, BES-based DVR, and nonlinear load or sensitive load. As depicted in Figure 2, the proposed DVR has a battery system as its external energy storage, which is one of the DVR topology as shown in Figure  3. The different compensation strategies of the DVR are also presented in Figure 4. Figures 5 and 6 illustrate the detailed block diagram of the PV system and its DC-DC converter controlling block diagram respectively. The proposed BES based DVR detailed block diagram is also illustrated in Figure 7. The variable RLC load is used to comprehend a sensitive load. An extensive simulation work is carried out using PSCAD under possible dynamic circumstances, considering the load as sensitive. The control system mainly consists of four subdivisions: the PV system DC-DC boost converter control, the BES DC-DC buck/boost converter control, the PV system VSC control, and the BES-based DVR VSC control as depicted in Figures 6 and 8-10, respectively.

Control Technique and Strategy
There are different multilevel modulation methods which have been developed for converters and the most commonly used can be classified into three types: (1) carrier-based sinusoidal modulation (SPWM), which includes phase-shifted PWM and phase disposition PWM; (2) space vector modulation (SVPWM); and (3) nearest-level modulation (NLPWM) [41][42][43][44][45]. The SVPWM works with line-to-line voltages which deals simultaneously with all phases, while the other two methods are phase-voltage modulation techniques. The obstacle for applying SVM to multilevel converters is the difficulty caused by the largely increased number of switching states and switching sequences that accompany the higher number of levels. For a large number of converter levels, NLPWM may be preferable, which directly computes the switching states and duty cycles for each phase of the converter by avoiding triangular waves which are used for SPWM [43]. By selecting proper common-mode voltage injections, SPWM can be functionally equivalent to SVPWM and more specifically, for two-level three-phase converters and lower level converter cases, there are no comparative differences with them [43,45]. Due to this, for all control systems, the SPWM technique is applied in this study. A double loop control is implemented for the two-level three-phase inverters for effective dynamic control of the PV system and the BES-based DVR. The inner loop is used to generate the reference voltages by controlling the output current of the inverter and the outer loop is used to generate the reference currents by controlling the DC voltage.
Based on the PV-VSC energy demand, the controlling of the boost converter of a PV system can operate in two modes [41], which are the MPPT and DC voltage control modes as shown in Figure 6. The two control modes can be chosen alternatively by a switch and the main difference between the two modes is the priority of the VSC to supply active or reactive power. The MPPT technique is used to harvest maximum power from the PV array based on Equation (1) when the PV-VSC priority is active power, whereas the DC voltage control mode is used to maintain the DC voltage so that the PV-VSC will control the active power of the DC link and the reactive power is its priority.
The PV system VSC controlling is done using vector-based SRF technique so that reference voltages are generated using instantaneous values of active and reactive voltages and currents. The V component of the three-phase grid voltage at the terminal of the PV system is formulated using park transformation as shown in Equation (9) and the reference voltage is obtained using the inverse park transformation as described in Equation (10). Finally, the reference grid voltage V is compared with the triangular carrier signal using SPWM block [12] as shown in Figure 9.
where ∅ denotes the transformation angle and represents the grid voltage. As illustrated in Figure 9, the controlling of PV-VSC is implemented using the SRF transformation. The three-phase voltage and current at the terminal of PV-VSC transform to their respective dq0 component. The voltage vector V is affiliated with the direct-axis and V is with the quadrature axis. The DC link capacitor voltage V is regulated by the power balance at the DC link capacitor by controlling direct axis component of current , whereas the quadrature axis current is used to regulate the reactive power. The MPPT implemented with the PV panels ensures maximum available PV power to flow into the DC link and later to the grid [42]. As shown in Figure  9, the difference of the reference reactive power Q and actual reactive power Q is processed through the proportional integral (PI) controller and gives the quadrature reference current I . To regulate the quadrature current I , the difference of I and actual I is processed through another PI. On the other hand, the reference direct current I is produced either from the difference of the reference DC link voltage V _ and measured DC link voltage V or the difference of the actual PV power P and reference PV power P and processed through a PI controller using MPPT or V selection switch. Through these processes, the reference direct component voltage V * and quadrature component voltage V * are processed so that, using the inverse park transformation, the three phase reference voltages V _ * , V _ * and V _ * will be formulated. These reference voltages are compared with carrier signal using the SPWM technique to generate pulses for each respective insulated-gate bipolar transistor (IGBT) switches G − G . Finally, the high-frequency component and DC offset voltages at the terminal of PV-VSC is filtered by LCL low pass filter and integrated into the grid.
Disturbance detection and reference generation are the main parts of the DVR control system. There are different detection methods such as RMS measurement, dq0 component measurement, wavelet transform, and phasor parameter estimation using Kalman filters [25,32]. Among these, vector-based SRF dq0 component measurement control technique is a more popular technique due to its prompt response and simple implementation [25,33]. Generation of reference voltage will be determined based on the type of energy storage device, its active power supporting capability, and voltage disturbances sensitivity of the load [15,25,32]. Based on the reference generation, the DVR voltage compensation strategies can be classified as pre-fault, in-phase, energy optimized, and a hybrid compensation (combination of pre-fault and in-phase compensation) [17,25,32] as depicted in Figure 4. The pre-fault compensation restores the voltage to its pre-fault value both in magnitude and phase angle, whereas phase compensation restores voltage only in phase with the voltage during fault which means it only restores the magnitude to pre-fault value but not the phase angle. The energy optimized compensation tries to reduce the real power injection so it may restore phase angles only partially. The hybrid compensation has the features of both pre-fault and in-phase compensation which operates for some time in pre-fault compensation and other times shifts to in-phase compensation. For sensitive loads, restoration of both magnitude and phase angle is necessary, so pre-fault compensation is used in this study due to its capability to restore both.
In the proposed control scheme of the BES-based DVR, two control strategies are used where one is for the buck/boost converter and the other is for the VSC as depicted in Figures 8 and 10, respectively. For the buck/boost converter controlling, the RMS voltage is measured on the upstream side of the PCC and fed as input to permit the battery to charge and discharge based on its SOC value as shown in Figures 7 and 8. The boost switch (G ) and buck switch (G ) are used to discharge and charge the battery, respectively. Based on the reference power (P ) and rated battery power (P ), the reference charging and discharging signals are generated using separate PI controllers. In the other case, using the RMS measured voltage (V ) and SOC value of the battery, the SOC permits blocking and deblocking of G and G switches. To protect the battery from overcharging and discharging, charging of the battery is permitted if and only if the SOC value is less than 100% and discharging of the battery is permitted if and only if the SOC value is in the interval of 5-100%. Figure 10 shows the proposed control block diagram of the VSC of the DVR which developed using RMS voltage measurement and dq0 component measurement detection techniques and prefault compensation strategy. Here, V (V , V , V ) and I (I , I , I ) are the voltage at the upstream side of PCC and the current flowing at the terminal of the DVR VSC, respectively. Based on the park transformation, the three-phase abc voltage and current components change to dq0 components and the phase angle is taken from phase locked loop (PLL) which is used to lock the phase angle of the system during normal condition before the fault. I is used to regulate V and I is used to regulate the reactive power. To regulate V , the difference between V _ref and measured V is processed by the PI controller and generates I _ . Then after, I and I _ difference is processed through another PI controller to regulate I . On the other hand, by detecting the sag/swell condition using V value, its output value Blk/Deblk is multiplied with _ and the difference with I is processed through the PI controller to regulate component. Finally, the reference direct axis (V * ) and quadrature axis (V * ) voltages are processed so that using the inverse park transformation, the reference voltages (V _ , V _ , V _ ) are processed. These reference voltages are compared with the triangular carrier signal using the SPWM block to generate pulses for each respective IGBT switches G − G . The terminal voltage of VSC will be transmitted to the LCL filter to eliminate harmonics and DC offsets and finally the DVR will inject this appropriate voltage to the sensitive load though the injection transformer.

Simulation Results and Discussions
A fault was applied at the PCC bus by varying the fault resistance to form different sag depths for both symmetrical and asymmetrical scenarios. Figures 11 and 12 illustrate a single line to ground (LG) fault with different fault resistance which is 2 ohms and 4 ohms, respectively. Due to this, the sag depth and the DVR compensation voltage are higher in Figure 11 compared to Figure 12. Even though the fault is LG, due to the delta-wye connection of the distribution transformer and severity of the fault, the impact of the fault affects the other phase also and makes double phase sag with different sag depth. Figure 11a shows the load voltage during LG fault at phase a produces a deep sag for the particular phase a, and a slight sag for phase c. The phase a sag is more than 33% and in phase c, it is around 20%. Since the voltage variation is more than 10% from the nominal value, based on the control techniques applied, the BES-based DVR operates and compensates the voltage sag as it is seen in Figure 11b,c. In Figure 12a, LG fault is applied at phase a with higher fault resistance value compared to Figure 11 and due to this, the fault affects only the faulted line.
As described in Figure 13a, all the three-phase load voltages are sagged more than 40% from the nominal value due to a three-phase ground fault (LLLG) at the PCC bus with a fault resistance of 2.5 ohms. By operating the BES-based DVR, the sagged voltage is compensated as shown in Figure 13b,c. On the other hand, the LLLG fault happened at the PCC bus with fault resistance of 4 ohms and a more than 30% symmetrical voltage sag happened as shown in Figure 14a. By using the BES-based DVR, it is restored to normal as presented in Figure 14b,c. During PV system intermittence situations also, voltage fluctuation will happen at the PCC bus so that the sensitive load will face PQ problems. To overcome this problem, the proposed BES-based DVR will operate in the same way as the fault cases simulated above. Due to this, the probability of operation of the DVR will be high during night time and cloud conditions since the PV system will not work with its nominal voltage. This makes the proposed DVR more economical as its operation increases.

Conclusions
The advancement of power electronic-based devices increases the issue of PQ due to its sensitivity to power fluctuation. When sensitive loads get power from grid-integrated PV/wind power systems, their vulnerability will increase due to the intermittent nature of PV/wind power sources. In this study, the sensitive load gets power from the grid-integrated PV system. The supply voltage for the load could be fluctuated either due to faults in the upstream of the load or due to the source intermittency of the PV system. In this study, a fault is applied at the PCC bus to form voltage fluctuation at the load. The proposed BES-based DVR operates when voltage deviation is greater than 10% of nominal voltage. The PV system is integrated with the grid system with a proper controlling of its boost converter and VSC. Based on the priority of active power or reactive power support of the system, these converters can be controlled by the MPPT mode or DC-link mode. The BES-based DVR operates by proper controlling of buck/boost converter of the battery and VSC of the DVR. The detection and reference generation are the two main controlling techniques of the proposed DVR where a combination of RMS and dq0 measurement is used for detection, and pre-fault compensation is used for reference generation control due to its capability of phase jump and magnitude restoration. The effectiveness of the proposed BES-based DVR is demonstrated for both symmetrical and asymmetrical sag conditions with different sag depths. The proposed system also could be very useful for microgrid systems which provide power to sensitive loads.