Keywords

1 Introduction

Global consumption pattern has seen continual annual growth over the last two decades and is expected that the demand may even double in coming 10 years [1]. A shift towards renewable based low carbon technologies in power generation has been observed. Key factors such as technological advancement in semiconductors, Solar cells and wind turbines, decarbonisation of power sector, efficient power electronics converters and advanced control strategies allows accelerated deployment of renewable energy sources. Renewable based Microgrids ensures less transmission losses and can even improve the power quality and voltage profile of the Grid [2]. However, increasing penetration of renewable based energy system can pose certain challenges to the power quality, stability and safety of the Grid [3].

Renewable based generation systems or Distributed generations (DGs) as shown in Fig. 1 can either operate in grid connected mode or isolate itself to provide power to local loads. However, events related to unintentional islanding of DGs may occur which needs prompt attention. These events leads to instability, transient overvoltage, frequency deviation, malfunction of protection system that degrades the power quality of the system. Certain standards are adopted widely to address unintentional islanding issues for grid connection of DG units. Some standards are listed in Table 1 that provides technical requirement for Anti islanding capability of DG units before integration in the power system network.

Fig. 1
figure 1

Power islanding condition

Full size image
Table 1 Technical requirements and guidelines defined by various for monitoring islanding events in grid connected DGs [4, 5]
Full size table

For detection of the islanding phenomena different methods are shown in Fig. 2 remote islanding and local islanding methods. Again local islanding methods are categorised as passive, active and hybrid methods [6]. A detailed assessment and comparison of various Islanding Detection (ID) methods along with benefits and limitations are presented for monitoring islanding based activities.

Fig. 2
figure 2

Taxonomy of islanding detection methods

Full size image

  • Considering the future outlook of the electric power industry, wide scale penetration of renewable technologies such as solar, wind, fuel cell etc. in the form of distributed generation requires efficient converters, adoption of advance control strategies and latest technological advancements.

  • Detection of islanding condition is also an important issue to deal with.

  • Due to challenges related to the power quality of the system, progress in the field of islanding detection have gained momentum in recent years.

  • However, very few publications have discussed about the fundamental problem, comparison between contemporary and advance detection techniques, classification of IDMs, performance indices for assessing IDMs, advantages and disadvantages of each IDMs in terms of Non detection zones (NDZ), detection time and power quality. Hence, a comprehensive analysis on various aspects of islanding detection is worth to review and compile.

The work is segregated into following sections. Section 2 deals with challenges related to unintentional islanding and its effect on the power system. Section 3 discusses about various performance indices for selecting a better IDM. Section 4 broadly discusses about different types of local, remote, active and Hybrid islanding detection methods. Section 5 compares different types of IDM based on NDZ, detection time and power quality.

2 Certain Challenges with Islanding

Unintentional islanding into the system leads a big concern since integrated power distribution networks was implemented. Such integrated system together with inverter or non-inverter-based DG systems, different load conditions, and compounded control strategy give rise to many challenge to power system network reliability [7]. There are numerous IDMs presented in research paper, which uses different tolls for effective capacity building, tools name as complex computing and signal analysis [8]. Though, still some problems remain that need to be examined in the field of islanding detection, example as performance of IDMs for concurrent operations, interaction between different DG systems which are fitted with AI devices, when islanding occurs. Tackling mentioned challenges and highlighting the responsiveness of the AI devices, IDMs are analytically analysed in this review. If an island of the power system is created unintentionally, it causes trouble and serious harms. Some of the important harms are discussed as.

  • The voltage and frequency may get affected and not remain in their appropriate standard level.

  • Line staff safety may be vulnerable by DGs feeding into a system after opening and tagging of primary sources.

  • Instant reclosing may cause phase difference. This causes huge torques mechanically and huge difference in current (I) /voltage (V), might shatter the physical systems [9]. Sharp rise leads to devastation of the appliances also.

3 Indices / Parameters of Islanding Detection Methods

The objective of the islanding detection techniques is to monitor certain system parameters such as rate of change of frequency, Voltage, current, \(\Delta {\text{P}}\), \(\Delta {\text{Q}}\) and accordingly determine any possible occurrence of grid disconnection based on the threshold value of the parameters. The salient characteristics of an effective and reliable IDM are listed below:

  • Minimum NDZ

  • Minimum detection time

  • High Power quality (lower value of THD)

  • Error detection ratio close to 1 i.e. minimum events of false detection.

Hence, IDMs can be assessed on the basis of four relevant performance indices namely Non detection Zone (NDZ). Detection time (\(\Delta {\text{t}}\)), power quality and error detection ratio (Ed).

3.1 Non Detection Zone (NDZ)

NDZ became the primary cause for failing of IDMs. NDZ methods is build up by observing of V, I and f deviation that lead to power mismatch. Power mismatch between output of DG and load utilization when distributed generator operates in islanding condition are due to fluctuation of V and f at a point of common coupling (PCC). When \(\Delta {\text{P}}\) and \(\Delta {\text{Q}}\) are almost to zero, at that time V and f fluctuation are not enough to detect the islanding. The range of power mismatch (\(\Delta {\text{P}}\), \(\Delta {\text{Q}}\)), which can’t cause V and f crossing the threshold set value to detect islanding is known non detection zone (NDZ) [6, 10]. NDZ in power mismatch space is shown by [11].

$$\left( {\frac{V}{{V_{{{\text{max}}}} }}} \right)^{2} - 1 \le \frac{\Delta P}{P} \le \left( {\frac{V}{{V_{{{\text{min}}}} }}} \right)^{2} - 1$$
(1)
$$Q_{f} \left( {1 - \left( {\frac{f}{{f_{{{\text{min}}}} }}} \right)^{2} } \right) \le \frac{\Delta Q}{P} \le Q_{f} \left( {1 - \left( {\frac{f}{{f_{{{\text{max}}}} }}} \right)^{2} } \right)$$
(2)

where [\({\text{V}}_{{{\text{min}}}} ,{\text{V}}_{{{\text{max}}}}\)] is the allowable voltage range, \({\text{f}}_{{{\text{min}}}}\) is minimum frequency and \({\text{f}}_{{{\text{max}}}}\) is maximum frequency, V and P are rated voltage and active power, and \({\text{Q}}_{{\text{f}}}\) is the quality factor. NDZ region is shown in Fig. 3. NDZ stands “non-detection zone” the range of conditions where a real grid failure will be filtered out.

Fig. 3
figure 3

Power mismatch space or NDZ region

Full size image

NDZ is a very useful feature used for evaluating and examining the effectiveness of the islanding detection technique. Figure 3, shows \(\Delta {\text{P}}\) versus \(\Delta {\text{Q}}\) plane and this power imbalance happen due to V and f variations. Out of the shaded region AI device sense the islanding condition and thus disconnect the DG but this would not be possible in case of power difference came within power mismatch space and thus it became threat for the safety unit.

3.2 Detection Time

Time interval between the disconnecting of the main grid and detecting islanding b IDMs that is expressed as

$$\Delta t = t_{IdM} - t_{trip}$$
(3)

where \(\Delta t\) is the detection time, \(t_{IdM}\) is the instant at which islanding detected and \(t_{trip}\) is the instant at which circuit breaker of the main grid get trip and isolate the DGs.

3.3 Power Quality

For the power quality, several aspects have to be taken in consideration like voltage, current and frequency stability as well as continuous supply of power and there waveform. Due to all these concerns it doesn’t have the particular definition and thus it is also known as service quality, voltage and current quality but judging the all parameters it can defined as the calculation, analysis, and betterment of the line parameters to maintain a reference waveform.

3.4 Error Detection Ratio

It is related to the false detection of islanding by IDMs whenever utility grid and micro grid (MG) are interconnected. Error measurement caused primarily by disturbance, which results in measured units exceeding the set points [12]. It is described as

$$E_{d} = \frac{{N_{error\,detected} }}{{N_{error\,detected } + N_{correct\,detected} }}$$
(4)

where \(E_{d}\) is error detection ratio, \(N_{error\,detected}\) is count of time false detection and \(N_{correct\,detected}\) is count of time correct detection.

4 Islanding Detection Methods

Various detection methods are being identified the researchers across the world. Typically, these can be classified into remote islanding detection (RID) and local islanding detection (LID) methods. RID methods are based on measurement and calculation of technical parameters at the central side [13,14,15,16,17,18], while LID methods are based realization at DG side. LID methods can be further divided into passive islanding detection (PID) [19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53], active islanding detection (AID) and hybrid islanding detection (HID) methods as shown in Fig. 2. In this work, Active Islanding [54,55,56,57,58,59,60,61,62,63,64] and Hybrid Islanding techniques [57, 65, 66] are extensively reviewed. Passive detection techniques are discussed in our recent review article [67].

4.1 Active Islanding Detection Methods

In these AID methods, intentionally injecting small perturbation at the output end of the DGs, that injecting signal do some significant changes in the system parameters and if that changes exceed the predefined value then trip signal get activated of the islanding condition and thus DG get disconnected. Remarkably AID methods have advantage as their low or zero NDZ, but major demerit is that these methods deteriorate the power quality. The basic flow chart of AID methods is shown in Fig. 4, and then different AID methods discussed briefly.

Fig. 4
figure 4

Flow chart of AID methods

Full size image

4.1.1 Active Frequency Drift (AFD)

Electrical parameters voltage and current are associated with the magnitude, frequency and phase. AFD uses the frequency/phase related parameters. In this method a small disruption current is injected into the main system, and whenever the islanding phenomena occurs, frequency drift take place by adding a zero time in inverter current and make the inverter current waveform distorted with respect to the original waveform i.e., shown in Fig. 5, however it is not affected during the grid connected system. Distortion meaning is phase difference between inverter output waveform (current) and original waveform of utility voltage at PCC. Due to distortion, inverter detect phase error, and participate in drifting the DG/inverter current to make the phase error zero. To make the drift occurs, DG connected inverter should work at unity power factor, and frequency drift is compared with OUF relay for islanding to be detected. For this a parameter is defined name as chopping fraction (cf) shown in Eq. (5).

$$cf = \frac{{T_{Z} }}{{T_{V} /2}}$$
(5)
Fig. 5
figure 5

Operational waveform of AFD method [54]

Full size image

where, \(T_{Z}\) = zero or dead time and \(T_{V}\) = time period of utility voltage. It is easy in implementation and having small NDZ compare to novel PID methods. Drawbacks are, effective for purely resistive load, as NDZ became large during large values of capacitance (C) and inductance (L), also wouldn’t suitable for parallel inverter systems, and have issue with the power quality [54].

4.1.2 Active Frequency Drift with Positive Feedback (AFDPF)

For supressing drawbacks of AFD (multi-inverter and NDZ issue), a positive feedback is used along with AFD. This feedback amplify the chopping fraction (cf) or dead time and assist the frequency drift at higher rate. Due to increase in the rate of frequency detection, it leads to detect the islanding condition fastly compared to AFD.

$$cf_{K} = cf_{K - 1} + G\left( {\Delta \omega_{k} } \right)$$
(6)

where \(cf_{K}\) and \(cf_{K - 1}\) are \(k{\text{th}}\) chopping fraction and \(\left( {k - 1} \right){\text{th}}\) chopping fraction, \(\Delta \omega_{k} =\) frequency difference = \(\omega_{K - 1} - \omega_{0}\), G is positive gain constant. No matter cf is positive or negative, AFDFP reinforced the frequency drift for any load, but the power quality degradation is still continued due to injected disturbance [55].

4.1.3 Sandia Frequency Shift (SFS)

This method is also an add-on of AFD, using a positive feedback for the frequency of the output voltage at the inverter end, whose chopping fraction is expressed as

$$cf = cf_{0} + K\left( {f_{PCC} - f_{line} } \right)$$
(7)

where, \(cf_{0}\) is chopping fraction without frequency variation, K is accelerating gain, \(f_{PCC}\) is frequency of voltage at PCC and \(f_{line}\) is the frequency of the original line.

During normal operation DG sets try to change the frequency but system stability maintain the same frequency, but during utility grid disconnected mode, there is rise in the frequency attempted at PCC and thus increase the chopping fraction. Whenever the frequency at the inverter end crosses the set value then islanding is detected. SFS method has very small NDZ, as well as detecting efficiency and power quality is improved [56].

4.1.4 Sandia Voltage Shift (SVS)

This method uses the same principle as SFS, just change is that it uses a positive feedback of magnitude of voltage at PCC and changes the current and power. When grid is linked there is no effect shown of this positive feedback, but during open linked voltage is reduced at PCC and thus current and power reduced. For that reduction of voltage a UV relay is used for the protection. Possibility is also of rising of voltage, and hence OUV relay is used, if the PCC voltage reaches OUV standards then islanding is detected [57]. Among all AID methods this method is quite easy in implementation, having small NDZ but facing the problem of power quality issue.

4.1.5 Slip Mode Frequency Shift (SMFS)

Practice is carried out by applying positive feedback to the phase of the PCC voltage. A small perturbation in the phase will cause frequency deviation during the grid disconnected, but a small disturbance in phase during grid connected mode will not lead to frequency deviation, hence phase difference in normal operation is almost zero as unity power factor maintain by the inverter. A SMFS curve Fig. 9, using Eq. (8).

$$\theta_{SMFS} (k) = \theta_{m} \sin \left( {\frac{\pi }{2}\frac{{\left( {f^{{\left( {k - 1} \right)}} - f_{0} } \right)}}{{\left( {f_{m} - f_{0} } \right)}}} \right)$$
(8)

where \(\theta_{m}\) is maximum phase deviation which occur at maximum frequency \(f_{m}\) and \(f^{{\left( {k - 1} \right)}}\) is previous cycle frequency and \(f_{0}\) is rated frequency. Equation (8) is showing the relation between frequency and phase angle. Working principle also validate that slope of the SMFS line is greater than the load line during unstable region. When grid get disconnected, operation will move towards the new stable point (shown by dotted as f1 and f2), during going through unstable to stable a set threshold value of inverter frequency crossed in either direction and thus islanding get detected by OUF relay. Problem with this method is having large NDZ and quality of power issue [58].

4.1.6 Automatic Phase Drift (APS)

This method is extension of SMFS, used to rectify the problem of NDZ of AFD and SMFS methods by utilizing positive feedback to the phase angle of output current of inverter. Hence only starting angle of inverter current alter consistently by using previous frequency of voltage, expression is shown as

$$\theta_{APS}^{\left( k \right)} = \frac{1}{\alpha }\left( {\frac{{f^{{\left( {k - 1} \right)}} - fo}}{fo}} \right)360^\circ + \theta_{0}^{\left( k \right)}$$
(9)

where \(\theta_{APS}^{\left( k \right)}\) is beginning angle of inverter I, \(\alpha\) is constant and \(\theta_{0}^{\left( k \right)}\) is additional phase shift induced each time in the terminal voltage.

$$\theta_{0}^{\left( k \right)} - \theta_{0}^{{\left( {k - 1} \right)}} = \Delta \theta *sgn\left( {\Delta f_{s} } \right)$$
(10)

where \(\Delta \theta\) = constant, \(\Delta f_{s}\) = change in steady state frequency, sgn = sign or signum function i.e., defined as:

$$sgn(\Delta f_{s} ) = \left\{ {\begin{array}{*{20}c} {1\;if\;f_{s} > 0} \\ {0\;if\;f_{s} = 0} \\ { - 1\;if\;f_{s} < 0} \\ \end{array} } \right.$$
(11)

Due to additional phase shift, frequency deviate of the voltage to stabilizing to new operating point, and during islanding this deviation of the frequency leads the OUF standards, thus islanding get detected. Disadvantages of this method are speed is slow and not suited for nonlinear load [59].

4.1.7 Variation in Active and Reactive Power

Method is based on the capacity of inverter to produce both active power and reactive power independently. During the utility disconnected mode, DG has to supply active power, and depending upon the load condition, voltage will rise or fall as voltage is related directly to the active power. When the change in the voltage exceeds the set threshold value, then islanding condition is detected by the help of OUV relay.

Similarly reactive power cause frequency variation, and it’s detected by the OUF relay. The detection time is around 0.3–0.75 s, easy in implementation and has small NDZ. Weaknesses are it can’t support multi-inverter as false detection sometime raised, power quality issue continued [60].

4.1.8 Negative Component of Current Injection (NCCI)

This method is involving the use of small injection of negative component of current (less than 3%) to the voltage source inverter (VSI) at the PCC terminal, leads to disturbing the PCC voltage as unbalancing of the voltage occurs during the grid disconnected. In normal mode NCC flow into grid without affecting the PCC voltage. Thus islanding get detected if negative sequence voltage exceeds the threshold value within 60–70 ms. Method is beneficial for any load, zero NDZ and detecting fast compare to positive sequence voltage [61, 62]. This method can also use to detect islanding by negative sequence impedance as it’s the ratio of negative component of voltage and current.

4.1.9 Impedance Measurement (IM)

Ratio of rate of change of voltage (ROCOV) to the rate of change of current (ROCOC) of the inverter gives the impedance. This (\(dv/di\)) technique is used for the detection as during normal operation variation in the voltage is very less and hence impedance is low, but during grid disconnected mode voltage varies significantly by perturbation in the current and thus impedance crosses the threshold value and islanding get detected. Detection time is around 700–900 ms [63]. It’s advantage is that NDZ is low and work effectively in case of single inverter and synchronously connected multi inverter, but all time synchronously would not be possible and thus multi inverter is became demerit for it.

4.1.10 Impedance Detection at Specific Frequency

This technique is exceptional case of harmonic detection method. A perturbation of the special frequency harmonics of the current are injected in the inverter. Due to this abruptly change occurs in the voltage of inverter placed at PCC terminal when the utility is disconnected, but not significant change during utility connected to the DGs. Due to production of harmonic voltage in the system, impedance can be measure and if it crosses more than the set point, then islanding detected. It is not appropriate for the multi inverter system [64].

4.2 Hybrid Islanding Detection Methods

As passive and active methods are suffering from large NDZ and power quality issue respectively. To eradicate both the problems, a HID methods are used, that combine both PID and AID methods. In this case AID methods apply after the use of the PID methods as shown in the flow chart of Fig. 6. Some of the HID methods are discussed in brief as the following.

Fig. 6
figure 6

Flow chart of HID methods

Full size image

4.2.1 Voltage Unbalance (VU) and Frequency Set Point

As VU (passive) is more sensitive to the load fluctuation, so instead of VU/THD (passive), this method uses VU as a passive parameter. Other active parameter should be needed that is completed by positive feedback to the voltage or current that leads to deviation in the frequency by using SFS/SVS. Whenever some changes in the system occur the voltage spike rises suddenly, and from precaution of the false detection due to load switching or transient, a maximum VU is set for their threshold i.e., around 35 times of the VU average as per the Menon and Nehrir proposal [65]. Voltage spike also leads to frequency deviation, if the frequency change in a specified time is greater than the threshold value then islanding condition get detected. Technique have advantages as very small NDZ, very low power quality issue and clearly discriminate islanding and non-islanding conditions.

4.2.2 Voltage Fluctuation Injection

This method consist of ROCOF/ROCOV (passive) and chopping fraction, CF (active) method. As shown in the HID flow chart, if the rate of change of F/V is greater than the large (maximum) setting value of threshold then islanding is confirmed directly, and when the ROCOF/V is greater than the small (minimum) setting value of threshold then immediately AID method get activated [57]. In AID chopping frequency (CF) is checked, and if CF value is also more to their corresponding threshold value then islanding detected, otherwise it would be considered as non-islanding case.

4.2.3 SFS and Q-f (ROCOF) Technique

This method is based on the integration of the Sandia frequency shift, SFS (active) and ROCOF (passive) technique [66]. For large power variation, islanding confirmed directly by ROCOF method only, but if the variation is in between the large and small then active method (SFS) will come into the action and by perturbation the islanding can be detected. Similarly many HID methods (like ROCOV and P, ROCOF and IM, ROCOF and frequency injection) can be formed by implementing the integrated passive and active IDMs.

5 Comparison of IDMs

Key technical factors such as (NDZ), detection time and power quality were considered for comparing RID, PID, AID and HID methods. A fair comparison of widely adopted IDMs has been presented in in Table 2. Each method has its own pros and cons which are listed in Table 3. Selection of IDMs are dependent on number of factors such as system power rating, network loading conditions, cost effectiveness, protection devices and integration guidelines and allowable limits. In terms of cost, PID has advantages over the others. RID requires high initial and maintenance costs.

Table 2 Technical comparison of different IDMs based on three key performance metrics
Full size table
Table 3 Merits and Demerits of Different Islanding Detection Methods
Full size table

6 System Architecture of a Grid Connected Solar Based DG

Figure 7 shows a Solar photovoltaic based DG connected to the 110 kV Utility grid. A DC-DC boost converter stabilizes the output of the PV arrays at a constant DC voltage of 500 V. A 3 level DC/AC power converter is used to process the power to the local loads and the main grid. CB1 and CB2 are the breakers for 20 and 110 kV bus respectively. Two local loads are connected to 20 kV feeder. Five different islanding detections methods are modelled to provide a trip signal during islanding events. They are Over/Under Voltage IDM, Over/Under Current IDM, Over/Under Frequency IDM, modified ROCOF IDM and Vector Shift IDM.

Fig. 7
figure 7

Grid connected solar photovoltaic based DG under study

Full size image

7 Results and Discussion

Five different IDMs are modelled and tested on a grid connected Solar photovoltaic based Distributed Generation system in MATLAB/SIMULINK environment. The PV inverter is connected to a 20 kV Bus through a 3-Φ 100 kVA ∆/Y transformer. Circuit breaker CB1 connects the two local loads to the 20 kV Bus. The DG units and local loads are connected to 110 kV, 2500 MVA, 50 Hz Utility Grid through a circuit breaker CB2. Islanding event is achieved by opening CB2 at t = 0.1 s. Unintentional islanding usually pose power quality issues when the local generation capacity is adequately not enough to supply the local consumer demands. Hence, two loading scenarios have been considered for the simulation study. Trip Signal is generated on successful detection of islanding event in two cases described below:

  • Local generating capacity is either equal to the local load requirement

  • Local generating capacity is less than the local load requirement

In this study, Over/Under voltage, Over/Under current, Over/Under Frequency, modified ROCOF and vector shift methods are examined.

7.1 Case 1: Local Load is Equal to Local Generation

This test is performed at the 20 kV feeder where local loads are connected. Load 1 and Load 2 demands 50 kW each.

For the performing the islanding test, circuit breaker (CB2) of the main grid is opened at the instant of 0.1 s and reclose at instant of 0.25 s. Within the time interval of CB2 open condition (islanding condition) many parameters are changed, and if any parameters are crossed the corresponding threshold value then trip signal is generated for the removal of DG.

As seen in Fig. 8, when the islanding events occurred, suddenly voltage and current parameters are affected. In Fig. 9 change of frequency and modified ROCOF are observed. RMS Value of Phase voltages can be observed in Fig. 10. Trip signal is generated when the selected parameters crosses the set threshold values. Islanding is detected by OUV, OUC, OUF, m-ROCOF and VS within 10 m, 15 ms, 65 ms, 25 ms and 25 ms respectively as shown in Fig. 11. During reclosing of the load at 0.25 s, a transient is seen which is suppressed within few cycles.

Fig. 8
figure 8

Variation of three phase voltage, current and active power in case local load equals PV generation

Full size image
Fig. 9
figure 9

Variation of frequency (50 Hz), and ROCOF in case local load equals PV generation

Full size image
Fig. 10
figure 10

Variation of Variation in RMS value of voltage Va, Vb and Vc in case local load equals PV generation

Full size image
Fig. 11
figure 11

Islanding detection (trip signal) in case local load equals PV generation

Full size image

7.2 Case 2: Local Load Exceeds Local Generation

This test is also performed at the 20 kV feeder where local load are connected as per the specification: load 1 requires 2 MW and load 2 requires 30 MW and 2 MVAR (lagging) respectively. For performing the islanding test CB2of the main grid is opened for some time i.e., from 0.1 to 0.25 s. Within this period, electrical parameters crosses the set threshold value then islanding get realized. In Fig. 12, during islanding events, sudden change of voltage and current magnitude is seen, and the change crosses the threshold value, hence islanding get detected. The detection time of the OUV is around 7 ms, while detection time of over/under current is around 12 ms, i.e., the fast detection corresponding to the previous case discussed. Variation in active power is also shown in Fig. 12, i.e., sudden dip of power during islanding event. PV based DG is unable to supply the load resulting in load shedding.

Fig. 12
figure 12

Variation of three phase voltage, current and active power in case local load exceeds PV generation

Full size image

In Fig. 13 change of frequency and modified ROCOF are seen and islanding is detected with the corresponding detection time as 54 ms and 65 ms respectively. RMS value of Phase voltages can be observed in Fig. 14. A vector shift is also detected having the detection time is equal to 64 ms. Trip signals are generated on successful islanding detection using OUV, OUC, OUF, m-ROCOF and VS as shown in Fig. 15.

Fig. 13
figure 13

Variation of frequency (50 Hz), and modified ROCOF in case local load exceeds PV generation

Full size image
Fig. 14
figure 14

Variation in RMS value of voltage Va, Vb and Vc in case local load exceeds PV generation

Full size image
Fig. 15
figure 15

Islanding detection (trip signal) in case local load exceeds PV generation

Full size image

Out of these five, three methods OUV, over/under current and OUF are based on the magnitude comparison of voltage, current and frequency. It is observed that the fast detection is done by OUV and followed by over/under current but problem with these are having large false detection. So, further methods modified ROCOF and voltage vector shift are observed and found that, it eliminate the false detection condition to somehow. Hence, accuracy is improved in both methods as two passive parameters are checked there and hence detection time increased. Detection time and non-islanding conditions are discussed in Table 4.

Table 4  Islanding detection time for each IDM
Full size table

8 Conclusion

In this paper, active islanding detection techniques have been described for the utility connected micro grid system. The IDMs comparison is based on standards listed in Table 1, mainly IEEE 1547. Remote methods requires noise free, efficient and faster communication infrastructure, and hence the cost is high but is most reliable, as NDZs are minimized and prevents degradation of power quality. An active method has very small NDZ but degrade the power quality and that is why hybrid methods come into the picture to eradicate the high NDZ of passive and power quality issue of active, by combining the both methods (passive and active), detection time became prolonged. Generally passive detection time is low compared to both active (for seeing the response after perturbation, needed somehow more time) and hybrid. After being easy installation and lesser cost of PID methods, still it is not chosen because of their limitation with multi inverter systems and NDZ during balance islanding. In most of the cases AID are being used, but future trend will be hybrid and RID as both giving the good result over islanding and non-islanding, and capable of use of multi inverter systems with good reliability. Hence, this literature is providing a good concept over islanding detection, and it will be helpful for the future research. A grid connected architecture of a Solar based Distributed generation system is examined under islanding events using five IDMs. Modified ROCOF and voltage vector shift are observed to be more effective than conventional under/over voltage, current and frequency detections as they eliminates false detection conditions.