Abstract: This paper introduces the adverse effects of the islanding effect and its detection criteria, and analyzes the differences between traditional passive and active islanding detection methods. A novel and effective method, the "positive feedback active frequency shifting method," is proposed, and its feasibility is verified through simulation.
Keywords: Solar photovoltaic system islanding detection, detection blind spots
Abstract: This paper introduces the adverse impact of island effect and the testing standards of islanding detection, analyzes the traditional methods, which include passive and active islanding detection. A new and effective approach – Active Frequency with Positive Feedback (AFDPE) is put forward, and is validated by the simulation.
Key words: Photovoltaic system Islanding detection Non-detection Zone (NDZ)
[Chinese Library Classification Number] ???? [Document Identification Code] B Article Number 1561-0330(2012)08-0000-00
1 Introduction
Energy shortages and environmental degradation are increasingly serious global problems. Since entering the 21st century, green and harmonious development has become a human goal. With the continuous increase in human demand for energy, traditional fossil fuels face the danger of depletion. After nearly half a century of effort, solar energy utilization technology and its industry have emerged as a rising star in the energy industry. However, the islanding effect is a common problem in grid-connected photovoltaic power generation systems. The islanding effect refers to the situation where, when the power grid trips due to faults or power outages, individual user-end solar grid-connected power generation systems fail to detect the outage in time and fail to disconnect themselves from the grid, forming a self-sufficient power supply island composed of the solar grid-connected power generation system and surrounding loads. The islanding effect is a phenomenon unique to grid-connected power generation systems and has considerable harmfulness, not only endangering the entire distribution system and user-end equipment, but also, more seriously, threatening the lives of transmission line maintenance personnel. Currently, distributed power sources are growing rapidly, especially solar power generation. When many photovoltaic grid-connected systems simultaneously supply power to the public grid, the probability of the islanding effect increases accordingly. Therefore, solving the problem of isolated islands is particularly important.
2. Detection criteria for the island effect
The internationally recognized grid connection standards for photovoltaic systems, IEEE Std. 2000-929, UL 1741, and the distributed power station grid connection standard IEEE 1547, all require the islanding detection function of grid-connected inverters. Specifically, IEEE Std. 2000-929 stipulates that when the frequency of the point of common is outside the range of 59.3-60.5Hz, the grid-connected inverter should stop supplying power within 6 cycles; the standard for the inverter's power supply stoppage time under abnormal point of common voltage is shown in Table 1.
Table 1 Response under voltage anomalies (IEEE Std.2000-929)
my country's photovoltaic system grid connection technology (GB/T 19939-2005) requires that the allowable frequency deviation after grid connection of a photovoltaic system be ±0.5Hz. When this range is exceeded, over/under frequency protection should operate within 0.2s to disconnect the photovoltaic system from the grid. The corresponding system response time to abnormal voltage detection is shown in Table 2.
Table 2 Response under abnormal voltage conditions (GB/T 19939-2005)
For islanding detection test circuits, RLC resonant loads are commonly used in North America and Europe. When using an RL load to simulate a power grid without power factor correction, as Qf changes from 0 to 2.5, the corresponding power factor changes from 1 to 0.37. IEEE Std.2000-929 uses Qf as 2.5 to simulate the maximum value that may occur in a real load after power factor compensation. This paper still uses a resonant load with Qf of 2.5, and is required to detect islanding and implement protection within 2 seconds.
3. Analysis of Islanding Detection Methods for Photovoltaic Power Generation Systems
3.1 Principle of Island Detection Method
As shown in Figure 1, PV represents the solar panel, Grid represents the power grid, and the local load is the RLC parallel load; PPV and QPV are the active and reactive power provided by the photovoltaic system; Pload and Qload are the active and reactive power consumed by the local load; ΔP and ΔQ are the active and reactive power provided by the grid side; and point a is defined as the common coupling point.
According to the principle of nodal power conservation, if ΔP and ΔQ are not zero, Pload and Qload will change before and after the grid is disconnected (S1 is disconnected). The change in Pload will cause a change in the point-of-common voltage Va, and the change in Qload will cause a change in the common frequency f. If the change in Va or f is large enough, an abnormal grid condition can be detected by voltage or frequency, thus cutting off the electrical connection with the grid. Otherwise, relying solely on voltage or frequency detection cannot detect the abnormal grid condition, and the system continues to operate in grid-connected mode, entering a detection blind zone. Blind zone description is one of the commonly used tools for evaluating the effectiveness of islanding detection. The so-called blind zone refers to the sum of all loads that cause detection failure under a certain islanding detection method.
3.2 Passive Island Detection Method
Passive islanding detection methods primarily determine islanding based on changes in electrical parameters at the common coupling point a. However, due to difficulties in setting the detection threshold, the existence of detection blind spots, and significant susceptibility to load interference, passive detection can only serve as an auxiliary method for islanding detection. Therefore, active detection methods have been proposed.
3.3 Active Island Detection Method
The idea behind active detection is to disturb the voltage, current, and other electrical quantities output by the inverter, monitor the circuit response, and thus determine whether the inverter is connected to the power grid.
(1) By adding perturbation
Islanding can be determined by monitoring the impedance changes in the inverter output circuit after introducing a disturbance. (EU standard EN50330-1 stipulates that an impedance change of 1 ohm is considered islanding, and the power grid must be disconnected within 5 seconds.)
• Power Disturbance Method: This method involves disturbing the output current of the inverter and detecting the change in the common point voltage to determine if a grid connection exists. When a single unit is connected to the grid, the dead zone is very small. However, when multiple units are connected to the grid, the asynchronous disturbances dilute the effect, affecting the detection accuracy. Therefore, this method is only suitable for small-scale single-unit systems.
• Specific signal injection method: This method requires complex signal processing techniques and is suitable for situations with limited signal bandwidth. Since the injected signal is mostly subharmonic, excessively large signals will degrade the performance of equipment such as transformers. A dilution effect also exists when multiple systems are connected in parallel.
(2) Output voltage positive feedback
Applying positive feedback to the effective value of the inverter's output current causes the point-of-common voltage to quickly deviate from the normal range after the grid is disconnected, thus detecting islanding. While this method does not affect power quality, it results in power loss, a relatively long disturbance period, and a slow islanding detection time, making it unsuitable for rapid response.
(3) Active frequency or phase shift
The inverter applies positive feedback perturbation to the frequency or phase of the output current. After the grid is disconnected, the common point frequency is quickly pushed away from the normal range. This method has advantages such as minimal impact on power quality and ease of implementation, and is equally effective in multi-unit grid-connected systems.
4. Active frequency-based island detection
The preceding text compared and analyzed the advantages and disadvantages of various active detection methods. Among them, the impedance detection method suffers from a dilution effect when multiple inverters are operating in parallel. The remaining methods mainly focus on the continuous deviation of the inverter's output electrical parameters due to disturbances, eventually triggering the corresponding protection. Refer to the expression for the inverter output current:
(1)
The quantities that can be subjected to perturbations are: current amplitude, current frequency, and current phase. These are used to generate amplitude-based perturbations, with the Sandia Voltage Shift (SVS) algorithm as a representative example. Frequency-based perturbations are represented by the Auto Frequency Drift (AFD) algorithm. While the SVS method incurs energy loss, the AFD method effectively detects islanding with minimal interference to output power quality and remains effective even in multi-machine operation, thus becoming the focus of this paper.
4.1 Principle of Active Frequency Shift Islanding Detection
The active frequency offset method samples the frequency at the common node a and shifts the frequency of the inverter's output current, causing a disturbance in the load-side voltage frequency, as shown in Figure 2. The output current frequency is adjusted to be slightly higher than the voltage frequency until the voltage zero-crossing point arrives, at which point the current begins its next half-wave. When the mains power is interrupted, the frequency of the common point voltage deviates from its original value due to the influence of the current frequency; if this deviation exceeds the normal range, islanding can be detected.
In Figure 2, Va is the point-of-common voltage, which is the grid voltage during grid-connected operation. Tv is the corresponding period, i is the inverter output current, tz is the current cutoff time, and i1 is the fundamental component of i. cf = 2tz/Tv is defined as the cutoff coefficient. Through Fourier analysis, the phase difference between i1 and i is ωtz/2 (radians), which is defined as the active frequency shift angle, i.e.:
As shown in Figure 3, when the load's resonant frequency f0 is less than the grid frequency, the load is capacitive when connected to the grid; when the load's resonant frequency f0 is greater than the grid frequency, the load is inductive when connected to the grid. Grid-connected inverters are generally controlled to unity power factor. When the grid is normal, the frequency of the point-of-common voltage is clamped by the grid voltage. When the grid loses voltage, an equivalent model is shown in Figure 4, from which it can be seen that the frequency (or phase) signal undergoes two parts: the PLL and the AFD algorithm.
The system uses a phase-locked loop (PLL) to make the output current track the phase of the common point voltage. Taking the phase detection point of the PLL as the cycle start time, the following relationship can be derived from Figures 2 and 4:
Where [m] represents the current quantity, and [m-1] represents the quantity of the previous cycle. The input of the phase-locked loop is the phase difference B between Va and i. From equation (4), if B is greater than 0, i lags Va, which will increase the output current frequency and prevent the frequency from rising further; if B is less than 0, i leads Va, and the system will decrease the output frequency and prevent the frequency from falling further; if B=0, then
The system will maintain the same output frequency as last week, that is, reach a stable operating state. When the frequency exceeds the limit, the island will be detected; otherwise, it will enter the detection blind zone. Equation (5) is the basis for phase angle judgment.
4.2 Defects and Improvement Methods of Active Frequency Shift Islanding Detection
To prevent the frequency from reaching a steady state before exceeding the over-limit protection, a large cf value is required to obtain a large AFD. However, this will severely distort the current waveform, and the impact of cf on power quality cannot be ignored. Due to the relationship between cf and current THD, when the current THD reaches the grid connection standard specified in IEEE Std1547-2003, the absolute value of cf cannot exceed 0.05. Therefore, it is only suitable for islanding detection of non-capacitive loads with low quality factors. Thus, to reduce the impact on power quality during grid connection, to quickly detect islanding when disconnected from the grid, and to have a small detection blind zone, the AFD method introduces positive frequency feedback, which is the Active Frequency Offset Method with Positive Feedback (AFDPE).
In this improved method, we introduce a frequency positive feedback coefficient m, an initial cutoff coefficient cf0, and a grid frequency fg, then:
4.2 Analysis of Blind Zones in Active Frequency Shift Island Detection
There are currently several methods for describing the dead zone of islanding detection. For the active frequency shifting method with a defined expression, the success of islanding detection depends on the quality factor and resonant frequency of the local load. Therefore, the dead zone can be described by the spatial method. This spatial method can cover the load of all RLC combinations and can intuitively reflect the quality factor and resonant frequency of the load, directly corresponding to the standard specified in IEEE std 929-2000. The above description of the dead zone space is based on the phase angle criterion shown in equation (3). The process of spatially describing the dead zone (where is the phase shift angle of the inverter, and is the system frequency when islanding is formed):
From Figure 5, we can analyze that:
(1) The curves 1 and 2 correspond to m=0, which is the blind zone of the AFD method. The size of cf only changes the vertical position of the blind zone.
4.3 Simulation Analysis of Active Frequency Shift Islanding Detection
The islanding detection performance of a 2kW single-phase photovoltaic grid-connected power generation system was simulated using Matlab/Simulink. The grid voltage was set to 220V/50Hz, the frequency protection action threshold was set to 50±0.5Hz, the active power of the load was set to 2kW, the quality factor was 2.5, and the resonant frequency was f0=50Hz.
As shown in Figures 6 and 7 (where Qf=2.55, m=0.07, cf0=0.02; Waveform 1: Voltage 200V/div; Waveform 2: Current 10A/div; Waveform 3: Frequency), the islanding detection meets the requirement of detecting islands within 2 seconds as specified in the international standard IEEE Std.2000-929 and the national standard GB/T 19939-2005. To speed up the detection, cf0 can be appropriately increased.
5. Conclusion
This paper analyzes the principle of islanding occurrence and provides a detailed analysis and demonstration of the frequency-shifted islanding detection method. Through blind zone analysis based on the spatial method and Matlab/Simulink simulation analysis, it is concluded that the AFD method is only suitable for islanding detection of non-capacitive loads, while the AFDPF method has no restrictions on load characteristics and can further improve detection efficiency by introducing positive feedback. Therefore, this improved method has many advantages such as high detection efficiency, fast detection speed, and no restrictions on load. This active detection method will become an important method and development direction for photovoltaic islanding detection.
About the Author
Cheng Peng: Born in 1983, with a bachelor's degree, major research direction: photovoltaic and wind power converter control system.
References
[1] Zhao Jing, Zhao Zhengming, Zhou Dejia. Current Status and Development of Solar Photovoltaic Power Generation Technology [J]. Electrical Application, 2007, 26(10).
[2]JM Carrasco, JT Bialasiewicz, RCP Guisado, et al. Power-Electronic Systems for the Grid Integration of Renewable Energy Sources: a survey. IEEE Trans Industrial Electronics, 2006,53(8):1009-1012
[3]Mike Meinhardt, Gunter Cramer. Past. Present and Future of grid connected Photovoltaic-and Hybrid-Power Systems. IEEE, 2000:1283-1288
[4] Technical requirements for grid connection of photovoltaic systems. National Standard of the People's Republic of China GB/T 19939-2005
[5] Zheng Shicheng, Ding Ming, Su Jianhui, et al. Simulation and experimental study on photovoltaic power generation system and its islanding effect [J]. Journal of System Simulation, 2005, 17(12): 3085-3088.
[6] Yang Haizhu, Jin Xinmin. Anti-islanding control of photovoltaic grid-connected inverter based on positive feedback frequency offset [J]. Acta Energiae Solaris Sinica, 2005, 26(3).
[7] Liu Furong, Kang Yong, Duan Shanxu, et al. Parameter optimization of active frequency shift islanding detection method [J]. Proceedings of the CSEE, 2008, 28(1):95-99.
