An improved low‐voltage ride‐through (LVRT) strategy for PV‐based grid connected inverter using instantaneous power theory

Correspondence Soubhik Bagchi, Department of Electrical Engineering, Budge Budge Institute of Technology, Kolkata, India. Email: soubhik.bagchi91@gmail.com Abstract This paper presents a low-voltage ride-through technique for large-scale grid tied photovoltaic converters using instantaneous power theory. The control strategy, based on instantaneous power theory, can directly calculate the active and reactive component of currents using measured grid voltage and currents and generate inverter switching pulses based on the formulated reference current values and thus helping to improve the dynamic response when voltage sag takes place. The dynamic response of the proposed model has been compared with both proportional-integral and fuzzy current controllers to judge their suitability. The proposed strategy can provide both active and reactive power support dynamically during grid side fault. The proposed active reactive current control based technique shows better dynamic response compared to existing techniques. The proposed method is tested through appropriate simulation on a practical system to show the effectiveness of the proposed control method.


INTRODUCTION
Renewable energy based DG systems are becoming increasingly popular for electric power generation in the recent past. Among all, solar photovoltaic (PV) and wind turbines have currently become the strongest pillar for electric power generation as a replacement of conventional methods. Thus, interconnection between large-scale PV plants and electric power grid via voltage source inverters has been widely employed to meet the load power demand. Disconnection or standalone operation of these plants during faults or voltage sag situation is always not possible from the point of view of smooth operation, stability and reliability of the power system. In view of the abovementioned difficulties, several countries such as Italy, Germany, Denmark, Spain and US-FERC have revised their grid codes for grid-connected PV system. These grid codes ensure the availability, quality and reliability of grid-connected PV system during faults or voltage sag [1]. By keeping in mind the standard grid codes of each country, the capability of reactive power control must be incorporated in a large scale PV system to supply the reactive power demand during faults and voltage sags [2]. Power quality, fault management and current limitation are the major aspects for grid-connected voltage source converter during any kind of disturbance of the system. Active power reduction method, peak current limiting method and formation of three current references have been implemented during symmetrical and asymmetrical grid fault to overcome over voltage, over current and also to permit reliable operation [3,4]. Reactive power injection for voltage support during grid faults or voltage sag is essential as evident from the V-I characteristics wherein reactive current is supplied to the system by contributing at least 2% of rated current as per each 1% deviation of voltage provided the voltage sags exceeds 10% of line to line voltage [5]. Several difficulties of low-voltage ride-through (LVRT) operation for current source inverter have been investigated and improvised topologies such as modified maximum power point tracking (MPPT), addition of buck chopper have been applied to control dc-link current which can suppress grid voltage drop [6]. Grid interfacing and inverter control are two major aspects for grid-connected PV system. Generally, inverter and grid are interfaced via a phase-locked loop which is operated in relatively low bandwidth but such practice causes delay to detect the fault. Nonlinear phase-locked loop based on the complex-coefficient filter with adaptive controller gain has been proposed to accelerate the capability of filtering and also to enhance the dynamic performance of phase-locked loop [7]. Integration of high penetrated PV systems is becoming more decentralised and vulnerable day by day. For these reasons, the controllers of PV systems have to be more controllable, intelligent and cost effective. A flexible controller based on PQ theory has been developed to verify its effectiveness and flexibility [8]. Feed forward compensation for reactive current injection into dc-link and active power control to provide power balance between the both sides of converter have been discussed to provide grid support and enhancement of LVRT capability [9]. PV controllers have been optimally tuned by the application of slap swarm algorithm (SSA) to ensure the enhancement of LVRT in order to the percentage undershoots or overshoots, settling time and steady state error of voltage response. A proportional-integral (PI) based open fractional voltage control and cascade voltage control has been implemented here to design the controller where the fitness function is optimised by using SSA [10]. An alternative control strategy based on synchronously reference frame phased-locked loop (SRF-PLL) has been implemented and verified to show efficient control of the inverter for grid-connected solar photovoltaic system [11]. To continue the operation of grid-tied renewable system, it is necessary to detect the voltage sag during any type of fault condition. Maximum point tracking and zero point tracking algorithm have been proposed to detect voltage variation with accuracy along with filtering out the low-order voltage harmonics [12]. Besides the incremental impedance method has been used based on existing PQ measure point for finding out the location of voltage sag [13]. Concerning the stability of single stage grid-connected PV system; modified incremental conductance MPPT method has been implemented by reducing the step length of output power reference in the system. This method improves the stability of the system as well as enhanced the steady state accuracy of the method [14]. Filter design is one of the major concern criteria to restrict harmonic injection to the grid. The electromagnetic decoupling model of transformer integrated filtering system has been introduced in [15] for this purpose to improve the power quality with low transformer loss. Novel controllable inductive power filtering method based on magnetic potential balance theory combining with the hybrid active filtering technology to improve the power quality with better efficiency of industrial dc supply system has been demonstrated [16,17]. In [18], the estimation of real-time filter output and key parameters identification is shown for grid supporting inverters where the extended Kalman filter helps to derive the parameters by using discrete state-space representation of dynamic systems.
Several countries are having their own grid codes to mitigate the reactive power demand and maintain the hassle-free LVRT operation of grid-connected PV system during fault or voltage sag situations. Several control strategies such as droop control, robust control, d-q control, PQ control, etc. have been implemented and verified with PI and fuzzy logic controllers (FLCs) for LVRT operation [19][20][21][22][23]. A concept with a mathematical explanation on instantaneous power theory (IPT) has been described including several examples in [29]. An algorithm is shown in [30] to calculate current reference based on dq-frame for three-level neutral-point clamped inverters (3L-NPC) resulting better ride-through capability during voltage sag situations. A control scheme based on positive and negative sequence with lower voltage stress on inverter devices and dc-link capacitor minimizing the AC side current harmonics is described in [31]. An improved control strategy with decoupled reference grid current for the grid current controller to accelerate the dynamic response of the grid-connected inverter is shown in [32]. Detailed design with fuzzy controller including the procedure to form fuzzy rules has been discussed for the application of PV inverter [33]. A current control strategy incorporating FLC has been carried out for grid-connected PV system to control the inverter [34]. Fuzzy logic based MPPT algorithm along with PI current regulator is proposed in [35] to track maximum power point during rapid change of atmosphere or during fast transient. Comparison of PI and fuzzy controller during zero power flow condition between grid and local network with a PV system is shown in [36]. Recurrent wavelet fuzzy neural network (RWFNN) is proposed to replace traditional PI controller which helps to track the controlling performance of the active and reactive powers under grid faults for the weak grid conditions [37]. However, all the above-mentioned literature emphasise on the controlling operation for LVRT through several topologies where different types of inverter (i.e. 3L-NPC, four leg inverter, etc.) have been used along with suitable filters. Besides, all the inverter control topologies also focused on the reduction of PV inverter current and dc-link over voltage by means of reactive power injection. But no published paper so far deals with the dynamic response improvement from the fault inception to fault clearance. Moreover appropriate reactive power injection principle under LVRT operation has not been adequately discussed in the existing literature.
The proposed method utilises IPT for appropriate injection of reactive power under the LVRT operation of the gridconnected PV system. The proposed method reduces the size of the filter requirement which also has a cost impact. Moreover, the proposed method shows better dynamic performance compared to the existing technology. The novel control strategy combining with PI-IPT and fuzzy-IPT has been developed and implemented on a practical system which results in the faster response of the system to clear the faults. This paper has been organized as: 1. In Sections 1 and 2, a background review has been provided followed by grid code and LVRT requirement as well as control strategy and PQ theory. 2. In Section 3, the proposed grid-connected PV system, the details of PV modelling incorporating with MPPT algorithm have been discussed. 3. In Sections 4 and 5, the mathematical modelling of proposed PI and fuzzy logic based control strategy using IPT have been formulated and implemented to show the enhancement of low-voltage ride-through as soon as voltage sag takes place. 4. In Section 6 and 7, the subsequent results, the summery report, comparison and conclusion have been discussed.

GRID CODE AND LVRT REQUIREMENT
In this paper, we are considering Malaysian grid code which is shown in Figure 1. It is seen that PV plant must be connected to the grid in the connection area to avoid power loss and also must not be disconnected up to 150 ms even if the line voltage drops to 0% of its nominal value. Accordingly, voltage must be recovered up to 90% from its pre-fault value within 1.5 s from the occurrence of faults or voltage sag. Moreover, the grid support must be provided by the PV system by injecting reactive power as per the standardization to keep the voltage source inverter in operation [24].

Modelling of grid-connected PV system
The grid-connected PV system configuration is shown in Figure 2. It consists of a PV source, a dc/ac voltage source con-verter along with a step up transformer. The voltage source converter is operated through P & O algorithm to extract the maximum power output from the PV source. A dc-link capacitor is used across the PV output to make smooth the PV output voltage. An R-L filter is connected to the low voltage side of ac grid to minimize the distortion and a delta/star step up transformer is connected to the distribution side. Table 1 represents the value of the parameters used in the proposed system.
The following Figure 3 shows the per phase equivalent circuit of the proposed system consisting with dc/ac voltage source inverter [28]:   From Figure 3, Hence, Grid side active and reactive power are given by where is the power angle, i.e. phase angle between output voltage of the inverter, E∠ and ac grid voltage,V g ∠0, Z ∠ = R eq + jX eq is the output impedance. As the R-L filter, transformer and grid parameters are in series, R eq and X eq are indicated the equivalent resistance and inductance of the proposed system. The inverter parameters have been provided in Table 2.

Modelling of PV system
Quick changing of solar irradiation level has a significant impact on PV output than the temperature which generally changes quite slowly throughout the day. The equivalent circuit of the PV cell is shown in Figure 4 and it has been illustrated through the following equations [25][26][27]: where The parameters which have been applied in the proposed system of PV cell are given in Table 3 In the proposed grid-connected PV system 235 strings have been used along with 16 series module. The maximum current, voltage and dc output power

MODELLING OF PROPOSED CONTROL STRATEGY BASED ON IPT (PQ THEORY)
IPT deals on the basis of instantaneous values of active and reactive power. As per the theory, initially, three-phase grid voltage and current have been transformed to α-β coordinate by using Clark transformation to get the voltage and current vectors, i.e. v α , v β , i α , i β , respectively. Then these vectors are used to calculate instantaneous active and reactive power. The Clark transformation has been shown by equation (11) and (12) [ A voltage vector ⃗ v and current vector ⃗ i are represented as follows [28]: (13) and So, three-phase complex power is as follows: So, instantaneous power, Substituting V a , V b , V c in equation no (11), thus we get In this paper, a novel control strategy has been introduced by extracting active and reactive current based on IPT. The block diagram of this type of inverter control strategy is given in Figure 6.
So, substituting equation (17) into equation (16), we get Multiplying both side of equation (18) by These active and reactive currents have been compared with active (i p * ) and reactive (i q * ) current references and passed through the proportional integral (PI) controller to decompose v p * and v q * . These voltages have been further transformed to three-phase voltage, i.e. pq-αβ-abc and it is sent via PWM signal generator to generate switching pulse of Mathematical model of three-phase dc/ac converter's has been expressed as follows: where R g and L g are the grid side resistance and inductance respectively. V dc is the output voltage of the PV panel. d p and d q are the duty cycles of p-q components of converter switch. Table 4 represents the required proportional and integral gain of current control loop used for designing the control structure of the inverter. The voltage at point of common coupling (PCC) drops during the fault, the inverter must be switched into LVRT operation immediately. As the consequence of fault, the imbalanced power of both PV and grid causes transient in dc side voltage and ac side current. To protect the power electronic devices, the mentioned proposed control strategy based on IPT is applied and it is ensured the sufficient decrement of over voltage and over current which will lead the PV plant stay connected as requirement of LVRT operation.
The proposed control strategy helps to inject appropriate reactive power based on the severity of the fault to support the grid voltage which is essential during LVRT operation. When the magnitude of grid voltage is between 0.9 and 1.1 p.u., the system is operated in normal mode and inverter injects only active current to the grid and consequently the requirement of Two way control as per value reactive current is zero. But as soon as the fault occurs, voltage sag takes place and the amount of reactive current depends on the present grid voltage magnitude while the inverter should inject sufficient reactive current to the grid. By considering the severity of the fault, the ratio of injected reactive current to the nominal current is given below: (26) where V g p and V gn refer to the magnitude of present voltage during fault and the normal grid voltage, respectively. So, the reference value of reactive current and maximum allowable reference active current during fault is defined as: where I n refers to the normal value of inverter rated current. Now, as per Figure 6, the operation mode of the converter is depending upon the values of i p * and i q * which is shown in Table 5.

MODELLING OF RULE-BASED FLC
Conventional controllers are based on mathematical modelling whereas FLC is based on the concept of interface engine. FLC controller is used as an intelligent controller which avails the concept of fuzzy set theory, IF-THEN rules, linguistic variables and decision making. The proposed controller which is shown in Figure 7 begins by retrieving the latest values of active and reactive current [i p (t )  and i q (t )] and compare with the reference values of active and reactive current [i * p (t ) and i * q (t )] respectively which generates the error, i.e. Δi q (t ) and Δi q (t ) Each FLC contains two input variables (error and change of error) and one output variable. The two input variables have five triangular membership functions, i.e. negative big (NB), negative medium (NM), zero (ZE), positive medium (PM) and positive big (PB) with the range of -1 to +1 whereas output variable has three triangular membership functions, i.e. zero (ZE), positive medium (PM) and positive big (PB) with the range of 0 to +1. The set of knowledge base control rules are obtained as per the behaviour of the proposed system and it is given in Table 6.

6
SIMULATION RESULTS

Maximum power point tracking
The most commonly used MPPT algorithm, i.e. perturbation and observation (P & O) algorithm has been used in the proposed system to extract maximum power from the PV module. This algorithm is very simple in use by measuring voltage and current of PV panel and also it is worked efficiently during the slow changes of irradiation. Figure 8 indicates the block diagram of P & O application technique to track the maximum power point.

Normal operating condition
During normal operating condition the inverter works smoothly but as soon as the fault occurs and the voltage at PCC drops significantly then the inverter must be work in LVRT mode. The graphical representation of PV side and grid side voltage, current and power are shown in Figures 9 and 10.

LVRT operation
At LVRT mode, the system becomes imbalance and transient occurs in the dc side voltage and ac side current. Here, the proposed PI and fuzzy-based novel control strategy using IPT ensures that the PV power plant can stay connected during a fault condition and both the over voltage and over current will die out by keep in mind the Malaysian grid code. When the amplitude of grid voltage is between 90% and 110% of rated voltage as per standard grid code, the system is running in normal operating mode. But when the grid voltage deviates from its limit, the system switches to LVRT operating mode. During faults or voltage sag, it is important to set the reference of reactive current such that the inverter can inject sufficient active power to recover the voltage as per LVRT requirement. For the proposed control technique, the reactive current is set at zero. As per LVRT requirement, the graphical representations of PV side voltage and power, grid side active and reactive power and the analysis of inverter control during a symmetrical and asymmetrical fault in the grid side have been shown here.

DISCUSSION
The proposed system has been simulated under the normal operating condition and the plots are provided in Figure 10. It is seen that the inverter is operating smoothly during the normal operating condition and the output voltage of 796.4 V power of 1504 kW (approximate) from PV power plant as well as grid parameters, i.e. grid voltage of 33 kV and grid power of 1 MW are also maintaining normally. The maximum possible active current is injected during normal operation whereas reactive current injection becomes zero.
But these values are deviating from its normal operation values during fault condition or when voltage sag takes place. At this moment, the inverter tries to disconnect from the PV to grid operation but the proposed novel control strategies are helping to switch the inverter into LVRT mode and stay connected to the grid. A brief observation and a comparative analysis on the performance of the PI and fuzzy-based proposed controller has been discussed here by considering the subsequent results which are produced as graphical representation between Figures 11 and 14 during a symmetrical and asymmetrical fault condition. From Figures 11 and 12, it is observed that three phase fault (L-L-L fault: 0.45 s-0.6s) and line to line (L-L fault: 1.3 s-1.45 s) occurs at the grid side and continues during 0.15 s (based on Malaysian LVRT requirement) which causes a huge voltage drop in the grid side. Hence dc-link voltage leads PV plant to operate towards the maximum point which is equivalent to open-circuit voltage of PV plant. At this moment, huge active and reactive current support is required to the grid. The proposed control strategy helps to inject the required reactive current to the grid and overcomes the situation. In case of three-phase fault, the entire process takes long time to recover the PV plants parameters, i.e. voltage, current and power at its pre-fault values whereas during line to line fault, the entire process takes comparatively less time to get back the pre-fault situation. As per the consideration of performance based on dynamic response PI-based controller takes 570 ms and 200 ms whereas fuzzy based controller takes 310 ms and 190 ms during three-phase fault and line to line fault respectively to recover the situation which leads the inverter to stay connected.
From Figures 13 and 14, it is observed that double line to ground fault (L-L-G fault: 0.45 s-0.6 s) and single line to line (L-L fault: 1.3 s-1.45 s) occurs at the grid side and continues during 0.15 s (based on Malaysian LVRT requirement) which causes to different scenario based on the implementation of PI and FLC. By implementing PI controller, it is seen that the total extracted PV power reaches to zero during both the cases whereas the implementation of FLC leads the decrement of the same PV power up to the certain limit. In both situations, the inverter needs proper active and reactive current support and simultaneously the grid requires sufficient reactive power support to enhance the grid voltage towards normalizing. In this context, it is seen from Figure 14(b) that the necessary reactive current support has been given up to 0.65 s and 1.5 s respectively during the mentioned fault situation. After clearing the fault, the scenario goes toward normal operation where active current injection reaches its maximum limit and reactive current injection again becomes zero. As per consideration, the performance based on dynamic response PI-based controller takes 200 ms in both the cases whereas fuzzy based controller takes 200 ms and 250 ms during double line to ground fault and a single line to ground fault respectively to recover the situation.

Comparison
A comparative analysis has been shown in Table 7 where the analysis has been done based on the response of the system during fault inception to fault clearance. The comparative survey reflects how fast the proposed method is responding to clear the fault than the other published method. The analysis procedure of the proposed method has already been explained in the discussion section and the same procedure has been followed for the other literature survey to frame the comparative structure.

CONCLUSION
The proposed control strategy can efficiently handle during grid fault condition, e.g. voltage dip and over current condition. The proposed control strategy has been effectively verified through various case studies and it has been observed that the inverter connectivity can be uninterrupted during fault condition. The inverter gives continuous support to the grid by maintaining the active and reactive power and current. It has been noticed that the LVRT requirements do not violate the procedure of controlling the inverter and also helps the system to operate smoothly by maintaining the grid code. Further, the comparative study shows the effectiveness of the proposed control strategy based on IPT than the conventional strategy. By analyzing and comparing the subsequent results, both the controller shows effective performance during different cases but it is concluded that FLC is more effective than others for smooth operation during large or severe disturbances.