Band pass filter and AFVmean‐based unintentional islanding detection
2019; Institution of Engineering and Technology; Volume: 13; Issue: 9 Linguagem: Inglês
10.1049/iet-gtd.2018.5314
ISSN1751-8695
AutoresSuman Murugesan, Venkatakirthiga Murali,
Tópico(s)Islanding Detection in Power Systems
ResumoIET Generation, Transmission & DistributionVolume 13, Issue 9 p. 1489-1498 Research ArticleFree Access Band pass filter and AFVmean-based unintentional islanding detection Suman Murugesan, Suman Murugesan orcid.org/0000-0002-5801-5438 Department of Electrical and Electronics Engineering, National Institute of Technology, Tiruchirappalli, Trichy, IndiaSearch for more papers by this authorVenkatakirthiga Murali, Corresponding Author Venkatakirthiga Murali mvkirthiga@nitt.edu Department of Electrical and Electronics Engineering, National Institute of Technology, Tiruchirappalli, Trichy, IndiaSearch for more papers by this author Suman Murugesan, Suman Murugesan orcid.org/0000-0002-5801-5438 Department of Electrical and Electronics Engineering, National Institute of Technology, Tiruchirappalli, Trichy, IndiaSearch for more papers by this authorVenkatakirthiga Murali, Corresponding Author Venkatakirthiga Murali mvkirthiga@nitt.edu Department of Electrical and Electronics Engineering, National Institute of Technology, Tiruchirappalli, Trichy, IndiaSearch for more papers by this author First published: 10 April 2019 https://doi.org/10.1049/iet-gtd.2018.5314Citations: 3AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract Several international standards have suggested to detect the unintentional island formation to safeguard the working personnel and protect the electrical equipment in the system. An active islanding detection technique, which employs the q-axis current disturbance injection, has been utilised and an analysing technique is proposed in this work based on a combination of band pass filter and mean of absolute frequency variation (AFVmean) to detect and differentiate the island formation from other transient events. The proposed analysing technique (PAT) detects island formation within 180 ms, which is less than the prescribed time of various international standards, applicable to dynamic loading conditions without the need of adaptive threshold and communication, and offers zero non-detection zone (NDZ). The disturbance injection makes significant variation in the frequency of voltage signal on post-islanding condition even for zero-power mismatch condition, therefore, the NDZ is zero. The PAT is corroborated for various islanding and non-islanding events and the simulations are performed in MATLAB/Simulink tool. 1 Introduction Owing to the significant advantages of distributed generators (DGs) namely,, improved reliability, power quality and reduced transmission losses, nowadays they are widely being interfaced/introduced in the distribution systems. Despite numerous advantages, it is not preferable to operate DGs during an unplanned or unintentional islanding [1–5] owing to the risks of loss of control over the frequency and voltage, ineffective grounding in the system, bumbled protection system, unsynchronised reclosing leading to huge transients, severe health hazards to the maintenance personnel etc. According to IEEE standard 1547.4, DGs should be shut down during an unintentional islanding within 2 s. Many researchers have worked upon detecting an island formation during unintentional islanding, which is coined as islanding detection method (IDM). IDMs are broadly classified into remote and local IDMs [6]. IDM which uses communication between DGs and generation station or sub-station falls into the category of Remote IDM. It becomes unsuitable for microgrids of lower capacity due to its high capital cost [6–9]. Local IDMs are further classified into passive and active IDMs [6]. The different parameters (i.e. frequency, THD, phase angle, voltage magnitude etc.) are monitored at the point of common coupling (PCC) and examined to verify whether the MGs are islanded or not, in passive IDMs [6, 10–13]. Novel passive IDMs based on Duffing oscillator and Helmholtz oscillator are proposed in [10, 13]. However, the islanding detection time is increased due to the higher intentional delay [10, 13]. A passive IDM based on reference impedance is implemented in [11]. Guha et al. have utilised voltage ripple as a parameter to detect island formation in [12]. The main advantage of this type of IDM is that the power quality is not degraded, cost is less and easy to implement but the drawback is the existence of non-detection zone (NDZ) the inability of the IDM to detect the island formation on low or zero-power mismatch between the generation and the demand in the islanded region. Active IDMs overcome the shortcomings in passive IDMs by injecting a small disturbance into the system or adding a positive or negative feedback in one or more control parameters of the inverter. This in turn makes significant variations in various parameters (voltage, frequency and THD etc.) on post-islanding and the parameters at PCC are examined to distinguish between islanding and non-islanding conditions [14–22]. The NDZ is reduced or nullified in active IDMs and also the island detection time is lesser than many passive techniques but at the cost of slight/significant degradation in power quality [6]. Many active IDMs destabilise the system on post-islanding, unfortunately [6, 15, 19, 21]. In [15], the variation of voltage phase angle is given as a feedback to converter-based DGs, which ensures noticeable variations in frequency and voltage on post-islanding assisting to detect the island formation. Chen and Li have implemented an active IDM based on reactive power disturbance in [18]. The authors in [19] have proposed an active IDM based on modulation index shift of pulse width modulation control, which results in change in voltage magnitude outside the operating range on post-islanding. Even though, these techniques [15, 18, 19] detect island formation faster and have less/zero NDZ, they destabilise the system on post-islanding conditions. In [14], the authors have suggested an active IDM based on disturbance injection through the q-axis current controller which results in significant variation in frequency on post-islanding condition but did not validate the technique for various non-islanding conditions at PCC. The major advantage of this strategy is that, it does not destabilise the system on post-islanding. The authors in [16, 17, 20] have adopted the technique in [14] and proposed various analysing techniques to differentiate the islanded from the grid connected condition, and validated for various islanding and non-islanding conditions. When the grid frequency exceeds ±0.5 Hz from the nominal, then the proposed technique in [16] mal-identifies the grid connected condition as islanded condition. The islanding detection time is ms for low power mismatch conditions in [17, 20]. At this juncture, an analysing technique is proposed in this work to overcome the earlier shortcomings and also to detect islanding faster than the earlier works which deals with the q-axis current disturbance signal injection for active islanding detection. The disturbance injection through q-axis current controller makes a significant variation in frequency on post-islanding conditions alone, and the proposed analysing technique (PAT) analyses the frequency to accurately differentiate the islanding from non-islanding conditions. The PAT comprises band pass filter (BPF), and mean of absolute frequency variation (AFVmean). The PAT is capable of detecting the island formation within 180 ms, which is less than the islanding detection time of the proposed techniques in [17, 20]. The PAT is capable of differentiating the islanded condition from the grid connected operation accurately. The islanding detection time is found to be faster than the earlier works focused on q-axis current disturbance injection and others as well, and also well within the detection time suggested by various standards. The NDZ is zero for the quality factors specified in various international standards. The PAT is adept of detecting island formation even for dynamic loading conditions without changing the selected threshold. The power quality degradation is very minimal during grid connected mode of operation, since only 1% of the magnitude of the rated d-axis current is injected as disturbance. It is easy to extend the PAT to multiple DGs without any modification, including the threshold fixed in this work. The paper is organised as follows: the introduction in Section 1 is followed by the significance of using active island detection and its control circuitry in Section 2, Section 3 elucidates the PAT continued by endorsement of the efficacy of the PAT in Section 4, and Section 5 concludes the paper. 2 Active islanding detection The test system and the control strategy adopted in this work are described in this section. 2.1 Test system description The solar-based renewable power is evacuated and supplied to the grid through a dc–dc converter, link, the three-phase inverter, the filter inductance () with a resistance of () and circuit breakers C.B.1 and C.B.2. The single line diagram of test system configuration is shown in Fig. 1. The parallel Resistor, Inductor and Capacitor (RLC) load is connected at the PCC, which is tuned to the resonant frequency. The current disturbance is injected into the grid through q-axis current controller which is explained in the subsequent sections. Fig. 1Open in figure viewerPowerPoint Single line diagram of test system configuration 2.2 Control strategy for active islanding detection The real and reactive power outputs of the inverter are controlled using the conventional d-axis and q-axis current controllers, respectively, [14, 16] and are represented as follows: (1) (2)where, P and Q– the real and reactive powers, respectively, – d-axis current and – q-axis current. The current output of the conventional d and q axes current controllers is expressed as follows: (3)where and – d-axis and q-axis current references. In order to operate the inverter in unity power factor, is set as zero. For active islanding detection, a current disturbance signal is injected into the PCC through the q-axis current controller. So, the expression in (3) is reframed as follows: (4)where – q-axis current disturbance injection. (5)where – 1% rated d-axis current of DG at rated voltage and power, – angular frequency of the current disturbance signal injected into the grid. The 'a' phase voltage at PCC is written as follows: (6)where and represent the sum and difference of the fundamental frequency and disturbance frequency, respectively. (7) (8) (9)The derivation of the above equations is given in Appendix, and the other phase equations are also written similarly. The overall control scheme for active islanding detection is shown in Fig. 2. Fig. 2Open in figure viewerPowerPoint Overall control scheme with disturbance injection During the grid connected mode of operation, the injected disturbance current signal flows into the grid due to the low impedance path offered by it, and hence does not make any significant variation in any parameter. On the other hand, the current disturbance injection is imposed to flow into the load, which results in significant variation in frequency at PCC in an islanded mode of operation. The frequency of the voltage signal is being examined continuously to differentiate the non-islanding conditions from the islanded conditions accurately. 2.3 Significance of active islanding detection This section shows the necessity of active IDM. The investigations are done based on two cases, namely, (i) without (ii) with . The microgrid is separated from the utility grid at 1 s. The frequency variations subject to islanding are presented in Fig. 3 for the above mentioned two cases for a perfectly matched load and generation condition. Fig. 3Open in figure viewerPowerPoint Importance of active islanding detection It is clear from Fig. 3 that for an exactly power matched situation, the variation in frequency is insignificant without . It is significant to note that in this condition, the passive IDMs fail to detect island formation in contrary to the active IDMs. Hence, it is preferred to use active IDMs for island detection in spite of a slight degradation in power quality. It is found that lower the frequency of the disturbance signal, higher is the PCC frequency variation on post-islanding. On the contrary, the detection time reduces with the increase in [16]. So, in this work is chosen as 20 Hz at 1% of rated d-axis current of DG at rated voltage and power, in such a way to get minimum detection time with higher magnitude of frequency variations on post-islanding [16]. 3 Proposed analysing technique The q-axis current disturbance injection through the inverter results in a significant variation in frequency at the PCC on post-islanding. However, it is obligatory for an analysing technique to detect island formation more rapidly and accurately distinguish the islanding from non-islanding (transients occurred due to various reasons) conditions. The analysing technique is used to differentiate, whether the frequency variation has occurred due to the islanded condition (frequency variation due to disturbance injection) or due to the transients. The obtained frequency is allowed to pass through the BPF and the mean of absolute frequency variation (AFVmean) is computed and used for the purpose of islanding detection. 3.1 Band pass filter If the grid frequency exceeds ±0.5 Hz, then the IDM in [16], mal-identifies the grid connected operation as islanded operation. This drawback is eliminated by an attempt of using a BPF in this work. The frequency at PCC is extracted from the three phase instantaneous voltage and allowed to pass through the BPF. This filter attenuates the frequency, which is outside a certain range (the range is decided by the bandwidth of the BPF) (10)where s = Laplace operator, = damping ratio = 0.707, = , = 20 Hz (frequency of the disturbance signal which is injected through the current controller). Bandwidth of the BPF is given by the expression . A BPF rejects both low- and high-frequency transients. So, a transient signal or the grid connected operation at a different frequency from the rated frequency is not mal-identified as islanded condition unlike in [16]. By the way, the limitation in [16], is eliminated in this work, because of BPF. 3.2 Mean of absolute frequency variation The mean of the absolute frequency variation of the output signal of the BPF is obtained using the following expression: (11)where T– time period of the disturbance signal (s), t– instantaneous time (s), FV – frequency variation A suitable combination of BPF and AFVmean, accurately differentiates the islanding and non-islanding conditions. The test system is operated for various operating conditions and a premeditated time delay of 90 ms is chosen such that the PAT does not mal-identify the transients as islanded conditions. 3.3 Trip signal generation The three-phase voltage at PCC is measured and the frequency is extracted from the voltage signal. The voltage signal is allowed to pass through the BPF which attenuates the frequency signals that are outside the defined bandwidth. The AFVmean is calculated using (11) and it is compared with the threshold to generate the trip signal. Whenever AFVmean exceeds the threshold for 90 ms (continuously), it indicates that the system is islanded. So, the trip signal is set as HIGH, i.e. 'Logical 1'. The time period 90 ms is the intentional delay, decided from the investigations by simulation of various operating conditions. Within 90 ms, normally the transients arising due to the switching conditions and grid frequency fluctuations fall below the threshold, and thus are differentiated from the islanded conditions. The block diagram for the trip signal generation is shown in Fig. 4. Fig. 4Open in figure viewerPowerPoint Block diagram of trip signal generation The test system is simulated for various quality factors and the peak deviations are noted. Using the trendline option in Microsoft Excel, peak deviations with respect to various quality factors are plotted. The equation for frequency peak trend for varying quality factors is obtained as follows: (12)The quality factor is defined as the ratio of amount of energy stored, to the energy dissipated by the load's reactive and resistive elements, respectively, and it is expressed as follows: (13)The AFVmean for the various is obtained through simulations for zero-power mismatches on post-islanding conditions and are shown in Table 1. It is found that the peak deviation in frequency of the voltage signal for zero-power mismatch condition for any power rating at any particular quality factor is almost the same. The quality factor of the load is varied by varying inductance and capacitance value for a fixed resistance value. The threshold is set to 95% of AFVmean for safety reasons. The reason for injecting the disturbance signal of 1% at 20 Hz is explained in [16], and hence the same frequency is adopted in this work. Table 1. AFVmean values for various quality factors Power rating, MW R, L, mH C, mF Peak deviation, Hz AFV mean, Hz Threshold (95% of AFV mean Hz) 0.5 1 0.16 1.019 9.947 0.1293 0.0823 0.0782 1 1 0.16 0.509 19.894 0.1007 0.0641 0.0609 1.8 1 0.16 0.283 35.810 0.0757 0.0482 0.0458 2.5 1 0.16 0.204 49.736 0.0617 0.0393 0.0373 In real-time scenario, the power demand varies continuously so as the quality factor. It is obvious from Table 1 that the threshold is different for various quality factors. In order to avoid measuring the quality factor of the loads in real time and adaptive threshold for different quality factors, a common threshold is selected. According to the international standards [1–5], the quality factor of the load should be <2.5. In this work, the value corresponding to the load of quality factor 2.5 is fixed as a threshold, because it is clear from Table 1 that the frequency variation on post-islanding condition is less for the load of , when compared to loads of . If threshold pertaining to some other is chosen as fixed threshold, it is riskier for the PAT to mis-identify the islanded mode as grid-connected mode for the loads pertaining to higher . The overall proposed algorithm is depicted as flowchart and is shown in Fig. 5. The motive of the trip signal is to show whether the system is islanded or not. It can be used to trip the DGs according to various standards or shift the DG control from PQ to V/f. If multiple DGs are available, only one DG is operated in V/f control and the rest continue to operate in PQ [23]. Fig. 5Open in figure viewerPowerPoint Flowchart of the proposed analysing technique 4 Results and discussions A good IDM should be accurate enough to distinguish the islanded conditions from the grid connected conditions without de-stabilising the system on post-islanding. Such an IDM is attempted in this work. The PAT is verified for the various operating conditions which include islanded conditions for various quality factors of zero-power mismatch condition and various power mismatching conditions, and non-islanding conditions such as load switching, grid frequency variation, unbalanced loading condition and open-conductor faults. The parameters of the test system are depicted in Table 2. Table 2. Test system parameters Parameter Value Parameter Value 50 Hz — load and DG power rating 1 MW 0.4 kV — 2.5 0.8 kV load component R 160 mΩ 7.9 kHz — L 204 μH 1 mΩ — C 49.74 mF 410 μH — 20 Hz 4.1 Islanding events 4.1.1 Zero power mismatch condition The PAT is substantiated for the zero-power mismatch condition (i.e. exactly power matched situation), since it is the most critical operating condition when compared to the power mismatched conditions of considerable differences between generation and demand. The unintentional island formations are mimicked to happen at 1 s by opening C. B. 1 (Fig. 1). The responses of zero-power mismatch operating condition before and after islanding are shown in Fig. 6. The value of the load of 2.5 is chosen as a common threshold for all the and the reason for the same is elucidated in Section 3.3. Fig. 6Open in figure viewerPowerPoint Responses of PAT for zero-power mismatch condition The plots in Fig. 6 show that the frequency varies on post-islanding according to the disturbance injection. The islanding detection times of all loads of are always less for the loads of . The island formation is identified within 180 ms for the exactly power matched situation for all the loads of mentioned in the international standards. The detection time of various for zero and various power mismatch condition is given in Table 3. Table 3. Detection time of different power mismatch conditions Quality factor () Detection time, ms for power mismatch conditions 0% 2.5% 5% 10% 15% 0.5 139.7 137.7 133.9 122.6 119.6 1.0 144.7 142.8 140.5 136.2 127.4 1.8 153.7 152.6 147.7 147.3 145.2 2.5 174.3 173.5 169.6 149.3 146.8 The damping ratio in the BPF is chosen arbitrarily and it is found that the increase in damping ratio results in decrease in islanding detection time by few ms and decrease in damping ratio results in increase in islanding detection time. 4.1.2 Power mismatch conditions The PAT is tested for several power mismatch conditions for different and the results are presented in Table 3. From Table 3, it is apparent that the injection of current disturbance signal in q-axis followed by the PAT is capable of detecting island formation even for 0% power mismatch condition. So, the NDZ is zero for the PAT for the reasonable quality factors (<2.5) mentioned in the standards and literature. The [17, 20] use q-axis disturbance injection for islanding detection, the PAT is faster than both the works and also more accurate than the [16]. It is explicit from Table 4 that the detection time of PAT for various quality factors is very much less than the various standards, depicted in Table 5. Table 4. Comparison of islanding detection time with literature Reference Detection time, ms Reference Detection time, ms [13] 0.96 <454 [11] 2.5 <350 [12] 1 200 [19] 2.5 204 [10] 1.57 <600 proposed technique 0.5 200 — 1 <150 [20] 1.8 200 — 1.8 <160 — — — — 2.5 <180 Table 5. Island detection time of various standards Standard Quality factor () Islanding detection time, t, ms IEEE 1547 [5] 1 t < 2000 IEC 62116 [3] 1 t < 2000 Korean standard 1 t < 500 UL 1741 [4] ⩽1.8 t < 2000 VDE 0126-1-1 [2] 2 t < 200 IEEE 929-2000 [1] 2.5 t < 2000 4.2 Applicability to multiple DGs environment The effectiveness of the PAT for multiple DG environments is verified in this section and the test system adopted for the same is shown in Fig. 7. The test system parameters of the multi-DG system are tabulated in Table 6. Fig. 7Open in figure viewerPowerPoint Test system for multiple DGs Table 6. Test system parameters Parameter Value Parameter Value DG 1 400 kW — 0.4 kV DG 2 300 kW — 0.8 kV DG 3 200 kW — 2.5 DG 4 100 kW load component R 160 mΩ 7.9 kHz — L 204 μH 50 Hz — C 49.74 mF 20 Hz load rating 1 MW — Four DGs are connected to the utility grid at the same PCC. The PAT is incorporated with all the DGs and each DG injects 1% of its d-axis current rating as disturbance. The references of disturbance injection for all the DGs are pre-defined and time synchronised (offline). If it is not, due to the unsynchronised disturbance injection, overall current disturbance magnitude would be reduced and subsequently the peak frequency variation on post-islanding would be reduced. This leads to unidentified island formation. In this section, the proposed technique is tested for the most critical islanded condition, i.e. zero-power mismatch condition (perfectly matched power condition between DGs and load) in a multiple DG environment. The four DGs (1 MW) with equally rated load of are isolated from the grid at 0.6 s. The responses are plotted in Fig. 8. Fig. 8Open in figure viewerPowerPoint Responses of PAT for multiple DG system It is pronounced from Fig. 8 that the PAT is proficient enough to detect the island formation in multiple DG environment without the need of any modification in the circuit. The islanding detection time is 172.4 ms and it is almost the same as the single DG isolation. The island detection time of each DG is same for all the four DGs, since all the DGs are connected at the same PCC. Hence, only the result of one DG among four DGs is shown in Fig. 8 4.3 Analysis on NDZ The NDZ is an operating zone where frequency and voltage variations are lesser on islanded conditions with lower or zero-power mismatches between generation and demand [10], where most of the IDMs fail to identify the island formed. The zero-power mismatch is the worst scenario to be tested, since the voltage and frequency variations are insignificant. The PAT is verified for such an operating condition and the island detection time is 90 ms. So, this transient condition is considered as non-islanded condition by the PAT. The analysing technique proposed in [16], mal-identifies the grid connected operation as islanded condition if the grid frequency operates outside ±0.5 Hz. The range of operating frequency of several standards in Table 5 is , leading to the technique in [16] mal-detecting the grid connected as islanded condition if the grid frequency is between . This drawback has been eliminated in this work. The responses of PAT for grid frequency fluctuations are shown in Fig. 9. Fig. 9Open in figure viewerPowerPoint Responses of PAT for grid frequency fluctuations Even though, the measured value in Fig. 9c exceeds threshold after the fluctuations at 0.6, 0.9 and 1.2 s, it does not sustain for the intentional delay of 90 ms long. So, the trip signal remains LOW. It is clear from Fig. 9, that even if the grid frequency exceeds ±0.5 Hz, the PAT effectively discriminates the grid connected mode from the islanded condition. It is evident from Fig. 9, that PAT does not mal-identify the grid frequency variation as island formation. 4.4.2 Uncertainties in the solar irradiation The solar irradiation is more prone to fluctuations in real time and hence it becomes inevitable to validate the PAT for sudden variations in solar irradiation. In order to reflect the real-time scenario of the sudden and gradual variations, the solar irradiation is varied drastically from 1000 to 200 W/m2 at 0.7 s (to represent drastic variation in solar irradiation) and linearly increased from 200 to 900 W/m2 between 1 and 1.3 s (to represent the linear variation in solar irradiation). The responses of the PAT are depicted in Fig. 10. Fig. 10Open in figure viewerPowerPoint Responses of PAT for uncertainties in solar irradiation It is clear from Fig. 10 that the measured value does not exceed the threshold even for the abrupt change in solar irradiation. The PAT does not mal-identify the transients due to the variations in solar irradiation as islanded condition. Hence, it is evinced that the PAT is adroit enough to distinguish the islanded conditions from the uncertainties due to solar insolation. 4.4.3 Load switching The distribution system is frequently subject to load switching conditions. So, it is mandatory to validate the PAT for such varying load switching conditions. In this section, the PAT is inspected for various load switching conditions. The different types of load (resistive, inductive and capacitive) are switched IN and switched OUT at various instants, and their switching timings are depicted in Table 7. The corresponding results are plotted in Fig. 11. Table 7. Switching timings of different types of load Type of load Switching IN, s Switching OUT, s resistive 0.7 0.9 inductive 1.1 1.3 capacitive 1.5 1.7 Fig. 11Open in figure viewerPowerPoint Responses of PAT for various load switching Even though, measured value exceeds threshold in Fig. 11c after 1.5 s, it does not sustain for 90 ms. So, the trip signal is not set as HIGH. It is clear from Fig. 11, that AFVmean does not even exceed the threshold due to the resistive and inductive load switching. Despite the measured value exceeds the threshold for capacitive load switching, it falls below the threshold within the intentional delay of 90 ms. Hence, the load switching conditions are distinguished as non-islanded conditions. It is evident from Fig. 11 that the PAT exactly discriminates the non-islanding load switching conditions from the islanding events. 4.4.4 Open-conductor faults (series faults) Albeit, the open-conductor faults are uncommon, it is essential to validate the PAT for such operating conditions. Since, due to the disturbance injection there is a notable change in frequency during open-conductor faults and that should not be identified as an islanded condition by the PAT. Hence, the PAT is verified for the open-conductor faults in this section. Two phases in C. B. 1 are opened at 0.7 s to create open conductor fault and closed at 1.2 s. Figs. 12a–c show the results for the double conductor open fault. Fig. 12Open in figure viewerPowerPoint Responses of PAT for open conductor fault It is ascertained from Figs. 12a to c that during the open-conductor faults, the measured value does not exceed the threshold, since, one phase remains grid connected. The disturbance injection through that respective phase flows into the grid, due to the low impedance path offered by it. It is proved that the PAT is skilful enough to differentiate the non-islanding conditions alike open-conductor faults and islanded conditions. 4.4.5 Short-circuit faults (shunt faults) The PAT is tested for the short-circuit faults (shunt faults) in this section. It is considered that the various faults occur for 100 ms long. The ground resistance considered for the ground faults is 1 μΩ. The minimum fault resistances for various short-circuit faults for a fault duration of 100 ms are chosen, such that frequency does not exceed the maximum and minimum operating limits of IEEE 1547 standard and the same are tabulated in Table 8. The various short-circuit faults are introduced into the systems at 0.5 s and removed at 0.6 s. The responses of parameters for various types of faults are shown in Fig. 13. Table 8. Minimum resistances for various types of short-circuit faults Type of fault Minimum fault resistance, mΩ Fault duration, ms LLLG 30.5 100 LLL 30.5 100 LLG 21.4 100 LL 16.0 100 LG 10.0 100 Fig. 13Open in figure viewerPowerPoint Responses of PAT for short-circuit faults Even though, the measured value exceeds threshold during the introduction and removal of faults, it falls below the threshold within the intentional time delay of 90 ms. Hence, it is clear that the PAT is capable of identifying the fault condition as non-islanding condition for all the depicted minimum resistance values for the respective types of faults. However, the PAT could not distinguish the fault conditions as non-islanded conditions, if the fault occurs with a fault resistance less than the minimum fault resistance as stated in Table 8, because voltage at fault point and frequency exceeds the operating limit. 4.5 Influence on power quality The influence on total harmonic distortion (THD) due to the current disturbance injection on q-axis current controller is given emphasis in this subdivision. The THD of the inverter output current with and without disturbance is plotted in Fig. 14. Fig. 14Open in figure viewerPowerPoint Comparison of the THD of the current components with and without disturbance It is shown in Fig. 14 that even though the current disturbance is being injected, the THD obtained is very well within the IEEE standard [24]. The THD of the current with disturbance injection is slightly more than the THD of the current without disturbance. Even though the power quality is slightly degraded, the island formation is detected faster. The transients are discarded accurately by the PAT, i.e. the non-islanded conditions are accurately distinguished from the islanded conditions. 5 Conclusion In this work, an active islanding detection methodology based on disturbance injection through q-axis controller is utilised and an analysing technique is proposed to distinguish the islanding from the non-islanding conditions. The proposed analysing technique comprises BPF and AFVmean. It is capable of detecting the unintentional island formation faster than the earlier works based on injection of disturbance signal through q-axis current controller. The PAT is tested for different quality factors represented in the standards for perfectly matched power condition and the results are the evidences to prove that the PAT is adept to detect island formation within 180 ms for the loads of quality factor <2.5. The PAT distinguishes the grid frequency fluctuations, load switching events, uncertainties in the solar irradiation, and open and short-circuit faults from the islanding conditions accurately. Further, the PAT is being extended to multiple DGs connected at various bus locations. Appendix 7 The current output of the conventional d and q axes current controller is expressed as follows: (14)For active islanding detection, a current disturbance is injected into the PCC through the q-axis current controller. So, the above expression is reframed as follows: (15) (16)In order to operate the inverter in unity power factor, is set as zero (17)Hence, (15) is rewritten as follows: (18)The abc components of the respective dq 0 components are calculated based on the following equation: (19)The current equation of 'a' phase is (20)Substituting the value of in the above equation and rewriting (21) and are the sum and difference of the fundamental frequency and disturbance frequency. Multiplying impedance with their respective frequencies, the 'a' phase voltage at PCC is given as follows: (22) 6 References 1 IEEE Std 929-2000: ' IEEE recommended practice for utility interface of photovoltaic (PV) systems', 2000 2 DIN-VDE Std. 0126-1-1: ' Automatic disconnection device between a generator and the low-voltage grid', 2005 3 IEC 62116: ' Testing procedure of islanding prevention measures for grid connected photovoltaic power generation systems', 2008 4 UL 1741: ' UL standard for safety for inverters, converters, controllers and interconnection system equipment for Use with distributed energy resources', 2010 5 IEEE Std 1547.4: ' IEEE guide for design, operation, and integration of distributed resource island systems with electric power systems', 2011 6Khamis, A., Shareef, H., Bizkevelci, E., et al.: 'A review of islanding detection techniques for renewable distributed generation systems', Renew. Sustain. Energy Rev., 2013, 28, pp. 483– 493 7Sun, R., Centeno, V.A.: 'Wide area system islanding contingency detecting and warning scheme', IEEE Trans. Power Syst., 2014, 29, (6), pp. 2581– 2589 8Bayrak, G.: 'A remote islanding detection and control strategy for photovoltaic-based distributed generation systems', Energy Convers. Manag., 2015, 96, pp. 228– 241 9Bayrak, G., Kabalci, E.: 'Implementation of a new remote islanding detection method for wind – solar hybrid power plants', Renew. Sustain. Energy Rev., 2016, 58, pp. 1– 15 10Vahedi, H., Gharehpetian, G.B., Karrari, M.: 'Application of duffing oscillators for passive islanding detection of inverter based distributed generation units', IEEE Trans. Power Deliv., 2012, 27, (4), pp. 1973– 1983 11Liu, N., Diduch, C., Chang, L., et al.: 'A reference impedance-based passive islanding detection method for inverter-based distributed generation system', IEEE J. Emerg. Sel. Top. Power Electron., 2015, 3, (4), pp. 1205– 1217 12Guha, B., Haddad, R. J., Kalaani, Y.: 'Voltage ripple-based passive islanding detection technique for grid-connected photovoltaic inverters', IEEE Power Energy Technol. Syst. J., 2016, 3, (4), pp. 143– 154 13Bakhshi, M., Noroozian, R. B., Gharehpetian, G.: 'Novel islanding detection method for multiple DGs based on forced Helmholtz oscillator', IEEE Trans. Smart Grid, 2018, 9, (6), pp. 6448– 6460 14Gonzalez, G.H., Iravani, R.: 'Current injection for active islanding detection of electronically-interfaced distributed resources', IEEE Trans. Power Deliv., 2006, 21, (3), pp. 1698– 1705 15Pourbabak, H., Kazemi, A.: 'A new technique for islanding detection using voltage phase angle of inverter-based DGs', Int. J. Electr. Power Energy Syst., 2014, 57, pp. 198– 205 16Gupta, P., Bhatia, R.S., Jain, D.K.: 'Average absolute frequency deviation value based active islanding detection technique', IEEE Trans. Smart Grid, 2015, 6, (1), pp. 26– 35 17Emadi, A., Afrakhte, H.: 'A reference current perturbation method for islanding detection of a multi-inverter system', Electr. Power Syst. Res., 2016, 132, pp. 47– 55 18Chen, X., Li, Y.: 'An islanding detection method for inverter-based distributed generators based on the reactive power disturbance', IEEE Trans. Power Electron., 2016, 31, (5), pp. 3559– 3574 19Liu, S., Zhuang, S., Xu, Q., et al.: 'Improved voltage shift islanding detection method for multi-inverter grid-connected photovoltaic systems', IET Gener. Transm. Distrib., 2016, 10, (13), pp. 3163– 3169 20Gupta, P., Bhatia, R.S., Jain, D.K.: 'Active ROCOF relay for islanding detection', IEEE Trans. Power Deliv., 2017, 32, (1), pp. 420– 429 21Murugesan, S., Murali, V., Daniel, S.A.: 'Hybrid analyzing technique for active islanding detection based on d-axis current injection', IEEE Syst. J., 2018, 12, (4), pp. 3608– 3617 22Bei, T.-Z.: 'Accurate active islanding detection method for grid-tied inverters in distributed generation', IET Renew. Power Gener., 2017, 11, (13), pp. 1633– 1639 23Eid, B. M., Rahim, N. A., Selvaraj, J., et al.: 'Control methods and objectives for electronically coupled distributed energy resources in microgrids: a review', IEEE Syst. J., 2016, 10, (2), pp. 446– 458 24 IEEE Std 519-2014: ' IEEE recommended practice and requirements for harmonic control in electric power systems – redline'. (Revision of IEEE Std 519-1992) – Redline, 2014, pp. 1-213 Citing Literature Volume13, Issue9May 2019Pages 1489-1498 This article also appears in:2021 Winners of the IET Premium (Best Paper) Awards FiguresReferencesRelatedInformation
Referência(s)