Anti-Islanding Method Development Based on Reactive Power Variation under Grid Support Environments

Featured Application

Grid Connected Inverter.

Abstract

As the proportion of distributed generation (DG), including Photovoltaic (PV) generation, in the power grid system increases, there are dropouts of large-scale distributed power generation sources due to some transient conditions that negatively affect the power grid stability and power quality. Accordingly, the inverter for DG generation is in a transition period, requiring more complex control performance. Anti-islanding function requirements in particular are becoming more strict because sophisticated grid-connection requirements are demanded, such as voltage ride-through, frequency ride-through, rate of change of frequency, and other functions. Thus, highly advanced anti-islanding methods are required to detect the islanding condition quickly and accurately to stop the inverter. This paper presents the improved anti-islanding method based on reactive power variation (RPV) under grid-support environments for single-phase DG inverters. In order to verify the validity of the proposed method, PSIM simulation was conducted. The proposed method meets the requirements of IEEE Std. 1547-2018 and KS C 8564:2021 by preventing islanding within 0.007 s under the newly adopted voltage/frequency trip setting, while the conventional RPV method fails.

1. Introduction

The islanding operation of distributed generation (DG) refers to a phenomenon in which electricity is continuously generated from the DG despite a power outage in the power grid [1]. This islanding operation causes many problems, such as a potential electric shock at the Point of Common Coupling (PCC) that was considered to be inactive, a high inrush current due to reconnection after power recovery, damage to the circuit breaker, an increase in harmonics, and flickering [2,3,4,5]. Therefore, many domestic and foreign standards require the function of the anti-islanding method (AIM) to stop the generation of DGs when the grid is disconnected, with the goal of securing safety [6,7]. As a part of the energy transition, renewable-energy-based DGs, such as Photovoltaic (PV) and wind power, are occupying more and more parts of the power grid worldwide [8,9,10]. For this reason, there is a dropout of large-scale DG sources due to a partial transient state of the power grid, which adversely affects the stability and power quality of the power grid, involving issues such as false tripping of feeders, blinding of protection, and unintentional islanding occurring [11,12,13,14].

In order to respond to this issue, recent power regulations require a function to maintain the connection to the power grid as much as possible, control active power or reactive power, or both in response to variations in voltage and frequency in the power system [15,16,17]. This is called the grid support function of DG resources, and the enactment is being promoted as a grid regulation of each country around the world [18,19,20,21]. However, the grid support function creates a very difficult condition for detecting the islanding phenomenon with the conventional technology. Accordingly, many studies are being conducted to create a new islanding detection technique in response to these challenging problems [22,23,24].

The conventional AIM is divided into a passive method and an active method [1]. The passive method simply monitors the main parameters, such as voltage/frequency/phase of the inverter’s output stage, and when it exceeds the set value, it is determined as islanding operation [25,26,27]. In the passive method, when the generated power and the load power of the DG are balanced, the fluctuation amount of the main parameter is not large, and there is a problem that the islanding cannot be detected. On the other hand, the active method forcibly changes parameters, such as current/frequency/phase/active power/reactive power, and causes a large change in the parameter after islanding operation occurs, so that it is designed to exceed the set value [28,29,30]. As shown in Figure 1, active methods based on the frequency variation include active frequency drift (AFD), active frequency drift with positive feedback (AFDPF), slip mode frequency shift (SMS), and reactive power variation (RPV) methods. The AFD method in particular uses the chopping fraction, cf, as a frequency perturbation parameter, which is defined as the ratio of zero time (t_z) to half of the period of voltage waveform (T₁/2). Its concept is based on varying the frequency of the inverter output current. If the frequency of the inverter voltage is out of the frequency relay set value after islanding occurs, it is determined to be an islanding operation, and the inverter stops. This technique inherently raises the harmonic component of the current, thereby lowering the power quality. It also has a limitation in that the performance varies depending on the value of the resonance index of the RLC local load. Another representative active method is a technique of periodically changing reactive power causing frequency fluctuations due to fluctuations in reactive power when islanding occurs. This method inherently lowers the value of the displacement power factor, which has a problem of lowering the power quality, and also has a limitation in that the performance varies according to the value of the resonance index of the RLC local load.

In response to the regulation of the wider voltage/frequency relay setting value change for grid support function, this paper presents the improved anti-islanding method based on reactive power variation (RPV) for single-phase DG inverters. Compared with the conventional RPV method, the proposed method has the advantages of the ability to detect islanding under the recent wider voltage/frequency relay threshold conditions with the same perturbed reactive power variations.

This paper is organized as follows: The recent regulation review on wider voltage/frequency relay threshold is presented in Section 2. Section 3 presents the proposed method with the operational principle. Then, simulation verification is provided in Section 4. Finally, Section 5 concludes this paper.

2. Standard Synopsis

IEEE Std. 1547-2018 and KS C 8564:2021 are representative technical standards that include the latest grid support functions domestically and internationally. This section compares and analyzes the change in voltage/frequency regulation related to the prevention of islanding operation in these two technical standards and the non-detection zone for islanding operation. The major difference between the islanding detection requirements in IEEE Std. 1547-2018 and KS C 8564:2021 is that the required islanding prevention times are 2 s and 0.5 s, respectively. This difference comes from the technical specifications, such as the reclosing time to fault current of each country’s power grid. In both standards, the grid code for abnormal voltage magnitude and frequency threshold value has changed as the proportion of DG increases in the power grid.

Table 1. DG response to abnormal voltage by the standards.

Standard	OV1		OV2		UV1		UV2
	p.u.	Clearing Time [s]	p.u.	Clearing Time (s)	p.u.	Clearing Time (s)	p.u.	Clearing Time (s)
IEEE Std. 1547-2003	1.1	1	1.2	0.16	0.88	2	0.5	0.16
IEEE Std. 1547-2018	1.1	2	1.2	0.16	0.7	2	0.45	0.16
KS C 8564:2015	1.1	1	1.2	0.16	0.88	2	0.5	0.16
KS C 8564:2021	1.1	1	1.2	0.16	0.7	2	0.5	5

and

Table 2. DG response to abnormal frequency by standards.

Standard	OF1		OF2		UF1		UF2
	Value (Hz)	Clearing Time (s)	Value (Hz)	Clearing Time (s)	Value (Hz)	Clearing Time (s)	Value (Hz)	Clearing Time (s)
IEEE Std. 1547-2003	60.5	0.16	-	-	59.3	0.16	-	-
IEEE Std. 1547-2018	61.2	300	62.0	0.16	58.5	300	56.5	0.16
KS C 8564:2015	60.5	0.16	-	-	59.3	0.16	-	-
KS C 8564:2021	61.5	0.16	-	-	57.5	300	57.0	0.16

show the clearing time of DG according to the abnormal voltage and frequency, respectively. Clearing time is defined as the time between the start of an abnormal condition and the DG ceasing to energize the connected grid. Table 1 shows the required clearing times for Over Voltage (OV), a voltage larger than the nominal voltage level, and Under Voltage (UV), a voltage smaller than the nominal voltage level. In the case of IEEE Std. 1547-2018, compared to the previous version, there was no change in the setting value of OV trip; only the clearing time was partially changed, and the setting value of UV trip was lowered, so the allowable range of grid connected operation was wider. In the case of KS C 8564, the UV trip setting value was found to be wider in the same trend. Table 2 shows the required clearing times for Over Frequency (OF), frequencies greater than the nominal frequency level, and Under Frequency (UF), lower than the nominal frequency level. In the case of IEEE Std. 1547-2018, the OF trip set value was 1.5 Hz wider, and the UF trip set value was wider by 2.8 Hz, compared to IEEE Std. 1547-2003, and the allowable range of grid-connected operation was wider. In the same trend, in the case of KS C 8564-2021, the OF trip set value was 1 Hz wider and the UF trip set value was 2.3 Hz wider compared to KS C 8564-2015, so the allowable range of grid-connected operation was wider. This is because a wider voltage/frequency setting was required to prevent the nuisance tripping of DGs, which occurs due to transient operations, such as the switching of capacitor banks and starting of motors. According to this trend, it becomes more difficult to prevent islanding operations with most inverter-based anti-islanding technologies, based on OV/UV/OF/UF relays.

Equations (1) and (2) are quantitative analyses of the non-detection zones of islanding operations according to the voltage/frequency relay threshold values and are summarized in

Table 3. Non-detection zone analysis by standards.

Standard	Non Detection Zone as $\frac{\Delta \mathit{P}}{\mathit{P}}$	Non Detection Zone as $\frac{\Delta \mathit{Q}}{\mathit{P}}$
IEEE Std. 1547-2003	$-30.6\%\le \frac{\Delta P}{P}\le 300\%$	$-2.4\%\le \frac{\Delta Q}{P}\le 1.6\%$
IEEE Std.1547-2018	$-30.6\%\le \frac{\Delta P}{P}\le 394\%$	$-12.8\%\le \frac{\Delta Q}{P}\le 6.3\%$
KS C 8564:2015	$-30.6\%\le \frac{\Delta P}{P}\le 300\%$	$-2.4\%\le \frac{\Delta Q}{P}\le 1.6\%$
KS C 8564:2021	$-30.6\%\le \frac{\Delta P}{P}\le 300\%$	$-10.8\%\le \frac{\Delta Q}{P}\le 4.8\%$

. By adopting the threshold values like V_max, V_min, f_max, and f_min in Table 2 into Equations (1) and (2), quantitative non-detection zones are generated, as shown in Table 3. Since the amount of change in effective power for islanding operation detection must be varied up to 393.8% compared to the rated power, the detection technique by effective power fluctuation is not widely used as a type of active method.

$$\left[{\left(\frac{V}{V_{max}}\right)}^2-1\right]<\frac{\Delta P}{P_{inv}}<\left[{\left(\frac{V}{V_{min}}\right)}^2-1\right]$$

(1)

$$Q_f\left[1-{\left(\frac{f}{f_{min}}\right)}^2\right]<\frac{\Delta Q}{P_{inv}}<Q_f\left[1-{\left(\frac{f}{f_{max}}\right)}^2\right]$$

(2)

Active techniques based on reactive power fluctuation techniques are widely used. In IEEE Std. 1547, they increase from a maximum of 2.4% to 12.8%, and in KS C 8564, they increase from a maximum of 2.4% to 10.8% to enable the detection of islanding operations based on OF/UF. As can be seen from this point, it can be confirmed that the fluctuation range of reactive power for islanding operation detection based on frequency relay is about 5 times larger.

With this amount of fluctuation, the size fluctuation of reactive power may cause the deterioration of power quality. Therefore, in this paper, based on a small level of reactive power fluctuation, the authors propose an improved technique that can detect islanding operation without reaching the threshold value of the frequency relay.

3. The Proposed Method

As analyzed in the previous section, the amount of change in reactive power for detecting islanding operation by the standards should be increased by up to 13% of the rated power of the solar unit inverter. In this paper, the anti-islanding method with small reactive power variations (RPVs) in a grid-supported environment is presented by proposing an islanding detection indicator without depending on OF/UF thresholds. The magnitude of the injected reactive power is proposed as a change in the phase angle defined in the basic concept of power factor in Equation (3). As shown in Equation (4), the phase angle deviation is designed to use the positive feedback according to the frequency deviation for two consecutive periods. While the grid is connected to DG, the frequency fluctuation for two consecutive cycles is not large and has a very small value due to the frequency sensing error. Therefore, only a very small change in reactive power occurs before islanding occurs. However, when islanding occurs, a small amount of reactive power change causes a corresponding frequency change, which needs to be designed to exceed the threshold value of the frequency relays. As introduced in the previous section, since the latest technical standards require a wider set value of the frequency relay, the size of the reactive power that needs to be changed to detect islanding operation is relatively large, which is a problem. Therefore, in this paper, an islanding detection indicator that can detect islanding operation for even minute-long fluctuations in reactive power is presented as the correlation parameters in Equation (5).

$$\mathrm{Power} \mathrm{factor}=\frac{1}{\sqrt{1+TH{D}_i^2}}\cos \left({\theta }_{RPV}\right)=\frac{1}{\sqrt{1+TH{D}_i^2}}\cos \left\{{\tan }^{-1}\left(\frac{Q_{RPV}}{P_{inv}}\right)\right\}$$

(3)

$${\theta }_{RPV}\left[k\right]=k_p\times \left(freq\left[k-n\right]-freq\left[k-n-1\right]\right)$$

(4)

$$\begin{gathered} C_P\left[k\right]={\displaystyle \sum }_{n=0}^{n=N-1}{k}_C\left({\theta }_{RPV}\left[k-n\right]-{\theta }_{RPV}\left[k-n-1\right]\right)\hspace{1em}\hspace{1em}\hspace{1em} \\ \hspace{1em}\hspace{1em}\hspace{1em}\times \left(\left(freq\left[k-n\right]-freq\left[k-n-1\right]\right)\right) \end{gathered}$$

(5)

where Q_RPV is the injected reactive power, P_inv is the inverter output effective power, k_p is the positive feedback gain, freq[k] is the measured frequency at the time instant k, C_p is the correlation parameter, and k_c is the scaling factor.

Before islanding occurs, the frequency of the inverter voltage is maintained robustly by the power grid, so the frequency is determined regardless of the reactive power fluctuations by the proposed method. Thus, the proposed correlation parameter has a value close to 0, even considering the sensing error when the inverter is connected to the grid. On the other hand, after islanding occurs, the frequency changes in the same direction by the variation of the proposed reactive power, which increases the value of the proposed correlation parameter. When the calculated correlation parameter reaches the designed threshold, it is determined as an islanding operation, and the inverter stops. The author has previously proposed an islanding detection technique using a correlation index based on an AFD method rather than reactive power variations. The previously proposed islanding operation detection method using correlation based on the AFD method has a problem of inherently generating harmonic components while the grid frequency changes [22]. However, the proposed method in this paper has an advantage in power quality as it has a structure in which the harmonic component does not change, but the size of the reactive power component changes. In this paper, the main design parameters are the threshold value of C_p to determine islanding and the magnitude of RPV. The magnitude of the reactive power variation is with positive feedback, and the threshold value of the correlation parameter is designed through a trial-and-error by simulation.

4. Discussion

To verify the feasibility of the proposed method, a 3 kW class grid-connected DG inverter system was implemented using PSIM software, as shown in Figure 2. The main circuit design parameters are shown in

Table 4. Electrical specification of the simulation circuit.

Parameters	Value
Nominal DG inverter effective power, Pinv	3 [kW]
Nominal grid voltage, Vgrid	220 [V]
Nominal grid frequency, fgrid	60 [Hz]
Number of phases	Single
Allowable frequency range	56.5 [Hz] ≤ f ≤ 62 [Hz]
Threshold value for Cp	6 (when N = 4, kc = 100,000, kp = 0.0056)
Local load R, L, C	PR = 3 kW, QL = 3 kVar, QC = 3 kVar

.

First, the power quality on harmonics was compared and analyzed for the proposed method and the conventional AFD method. Figure 3 shows how the waveform changes as cf, which is the control target of the AFD technique when the grid is connected, and

Table 5. Power quality comparison based on total harmonic distortion (THDi) under the AFD and RPV anti-islanding implementation.

AFD	cf	0%	1%	2%	3%	4%	5%
	THDi	3.18	3.19	3.53	4.10%	4.79%	5.54
RPV	${\theta }_{RPV}$	0%	1%	2%	3%	4%	5%
	THDi	3.18%	3.19%	3.20%	3.21%	3.22%	3.23%

shows its Total Harmonic Distortion (THDi) of the current. When the cf value is 0% and there is no disturbance at all, the base THDi is 3.18%, and when the cf value is 5%, the THDi is 5.54%. As cf increases, THDi increases, as the distortion in the fundamental sinusoidal wave increases. Figure 4 shows how the waveform changes as the phase, which is the control target of the proposed method, and Table 5 shows its THDi for the proposed method. It can be seen that the base THDi is 3.18% when the phase disturbance, which is the change in reactive power, is 0%, and the THDi is 3.23% when the phase disturbance is 5%. In other words, it can be seen that even when the phase disturbance increases, there is little change in the THDi because there is no distortion of the fundamental sinusoidal wave component.

As suggested in IEEE Std. 1547-2018 and KS C 8564-2021, the anti-islanding performance evaluation is conducted under the worst test condition, in which the power flows to the grid (∆P, ∆Q) are almost zero since the balance between the DG generation power and the load is made.

Figure 5 shows the inverter output voltage (V_inv), current (I_inv), frequency (Freq), Correlation parameter (C_p), and phase variation (θ_RPV) for the grid-connected DG inverter when islanding occurs at 0.4 s. In Mode A1 before islanding occurs, the difference in measured frequency between two consecutive periods is very small, up to 0.007, and the phase angle for the proposed reactive power variation is also very small, close to about 0.004, as shown by Equation (4). Therefore, at this time, it can be confirmed that the THDi in Mode A1 is also maintained at the level of 3.18%, as discussed in the previous section. In addition, in Mode A1, the correlation between the frequency difference and the reactive power variation is very low, and the C_p value stays at the maximum level of 0.08. In Mode A2, after islanding occurs, the frequency fluctuations by the proposed method and the corresponding fluctuations in C_p values are shown. By the reactive power variation technique with positive feedback gain K_p of 0.0056, it can be confirmed that the inverter frequency in the steady state stays around 60.55 Hz. Additionally, it can be seen that C_p in Mode A2 has a maximum value of about 7.45. For these frequency fluctuations, it can be confirmed that islanding cannot be detected under the condition of the frequency relay set value in IEEE Std. 1547-2018 and KS C 8564:2021.

Figure 6 shows the waveform that prevents islanding operations by using the proposed technique at about 0.007 s after islanding occurs. Similar to Mode A1, Mode B1 in Figure 6 is the section before islanding operation occurs, and islanding occurs at 0.4 s. When the value of C_p reaches the threshold value of 6 in Mode B2, islanding operation is determined, and in Mode B3, the DG inverter stops. As stated, in IEEE Std. 1547-2018, islanding should be prevented within 2 s under prescribed load conditions, and in KS C 8564, islanding should be prevented within 0.5 s. As described above, the proposed method minimizes the occurrence of harmonics to THD 3.18% in the grid-connected condition and detects the islanding operation in 0.007 s in the worst condition to meet the required technical standards for IEEE Std. 1547-2018 and KS C 8564:2021.

5. Conclusions

This paper presents the improved anti-islanding method based on reactive power variation (RPV) under grid-support environments for single-phase DG inverters. The proposed method meets the requirements of IEEE Std. 1547-2018 and KS C 8564:2021 by preventing islanding within 0.007 s under the newly adopted voltage/frequency trip setting, while the conventional RPV method fails. Since the wider frequency trip range was adopted for grid support environments, the amount of the conventional reactive power variation should be higher. Based on the proposed method, the amount of reactive power for anti-islanding could be minimized by using the proposed correlation parameter. In order to verify the validity of the proposed method, PSIM simulation was conducted. The proposed method could be applied to the reactive power compensation together with an anti-islanding function based on the appropriate reactive power supply profile.