Alternating current microgrid protection method utilizing photovoltaic low-voltage ride-through characteristics

The increasingly popular inverter distributed generation in microgrids is leading to changes in system fault characteristics. The fault behaviors of inverter distributed generation are closely related to the control mode. Here, a photovoltaic power supply in constant power mode enters a low-voltage ride-through state when there is a fault in the microgrid. The output current phase in the ride-through state is analyzed, and a local protection method based on the phase difference of the feeder positive-sequence current is proposed. The method is used to avoid interpreting voltage surges as faults, and to detect and locate faults rapidly and accurately. Electromagnetic transient simulation software is used to demonstrate the efficacy of the proposed protection scheme.


Introduction
Microgrids can connect multiple distributed energy resources and supply highly reliable power through a distribution network. Microgrid generators are close to the load; thus, power can be maintained when faults occur. Microgrids can operate in grid-connected and autonomous modes [1,2].
Microgrid power will be interrupted if the network is improperly protected against short-circuit faults that can occur in power lines or distributed energy resources. Different short-circuit fault current levels are applied in gridconnected and autonomous microgrid modes. These differences make it difficult, and occasionally impossible, to configure and coordinate protection devices. It is challenging to locate faults in microgrids that incorporate generation units and complex operating scenarios; some microgrid features (e.g., large phase imbalances) can introduce further challenges. Investment in microgrid protection is limited [3,4].
The integration of distributed energy resources and microgrids has led to grid policies that require renewable energy sources to remain connected to the grid during faults. This B Liuming Jing lmjing@ncut.edu.cn 1 Department of Electrical Engineering, North China University of Technology, Beijing, China is commonly referred to as fault ride-through and is defined using a stepwise/linear voltage-time-after-fault curve [5].
According to some grid codes, photovoltaic (PV) reactive power output is necessary to support the line voltage. An increase in reactive current changes the magnitude and phase of the PV output current. Accordingly, the microgrid line current changes, which can cause directional elements of the microgrid to incorrectly diagnose the fault direction. Therefore, microgrid fault characteristics should be analyzed for PV low-voltage ride-through, and the fault direction identification approach should be studied [6].
Some studies have analyzed PV low-voltage ride-through characteristics. For example, solar inverters act as a positivesequence source and have insignificant negative and zerosequence fault current [7]. Some solar inverters respond to faults in accordance with a preprogrammed control mode. For example, a unity power factor control mode will behave differently from a reactive support control mode.
Typically, microgrid protection includes overcurrent and distance relays, as well as current differential, traveling wave, and data mining methods.
A microgrid overcurrent protection strategy comprising inverter-based distributed generation in islanded operation has been proposed [8]. However, the protection method must be used to set a threshold appropriate to the context, which is difficult and expensive to determine.
Voltage compensation presumably can be used to enhance the reliability and accuracy of distance protection [9]. However, this approach requires many detection modules and complex compensation methods.
A new differential bus protection scheme based on the instantaneous energy coefficient has been proposed [10]. The energy coefficient is calculated using mode separation, but this method is expensive and requires excellent communication conditions.
A traveling wave-based protection scheme that uses current wavefronts as a fault detection mechanism has been developed [11]. However, the line lengths between microgrid nodes are short; therefore, it is difficult to detect the initial wavefront.
An intelligent fault detection method has been used to preprocess current samples at either end of the feed [12]. However, the protection process necessary for this approach is complex, and the overall judgment method must be improved.
Adaptive protection in microgrids and the composition of adaptive protection have been analyzed [13]. However, the complex structure, high cost, and communication requirements are prohibitive.
The fault point is located using a detection method based on the phase differences between the positive-sequence fault component of the bus voltage and the positive-sequence fault components of the currents in the feeders [14]. However, the voltage transformer was required in the bus.
In previous current-only method, the pre-fault and postfault phase angle change was used for fault section location. However, the low-voltage ride-through characteristics of PV were not considered.
In summary, methods to protect against PV low-voltage ride-through typically require voltage transformers, which increases the protection system cost [15]. Therefore, we propose a microgrid line protection method that uses only current information.
The application of voltage information as a reference quantity for the detection of fault direction is a common practice in transmission lines. Nevertheless, this approach is not applicable in microgrid due to the absence of potential transformers. Therefore, it is necessary to use the current-only polarity comparison for microgrids.
We analyze existing protection schemes, discuss the limitations of conventional fault protection, and present a protection method that is based on the positive-sequence current phase angle. The phase angle difference between the positive-sequence current of the line in fault and a normal line is obtuse, whereas the phase angle between two normal lines is acute. Finally, we use power system computer-aided design (PSCAD) and electromagnetic transients including direct current (EMTDC) to build a microgrid model with Imax Iq.f Fig. 1 Fault components of a PV microgrid multiple distributed PV generators. The proposed method in this paper using current-only polarity comparison.
The paper is organized as follows. In Sect. 2, we analyze the operating characteristics of PV low-voltage ride-through. In Sect. 3, we analyze the microgrid fault characteristics during the PV low-voltage crossing process, describe the theoretical principles of the protection method, and determine the detection range of the approach. In Sect. 4, we describe the microgrid protection scheme; in Sect. 5, we verify the accuracy of several fault detection methods in various fault conditions. A conclusion is presented in Sect. 6.

PV low-voltage ride-through characteristics
When a microgrid is connected to the grid, active and reactive control is used in conjunction with low-voltage ride-through, as shown in Fig. 1. When a microgrid fault occurs, the PV will reduce the active power and increase the output reactive power to regulate the microgrid voltage. There is a positive-sequence component in the output current on the alternating current side. In this network, the active and reactive-controlled PV is equivalent to a positive-sequence current source [16]. According to the active and reactivecontrolled inverter-interfaced distributed generator (IIDG) equivalent model, the fault current output can be expressed as: where I q.f and I d.f are the reactive and active current generated during the fault, respectively; U d.0 is the d-axis component of the voltage in normal operations; U + d.f is the positive-sequence component of the voltage at the time of the fault; I max is the maximum PV output current during the fault; I amp.f is the fault current amplitude; α is the fault current phase angle; and k is the reactive power compensation coefficient.
According to Eq. (1), a PV connected to the microgrid can be analyzed when a fault occurs using the equivalent model shown in Fig. 1, where Z L is the line equivalent impedance, Zs is the system equivalent positive-sequence impedance, and I is the current fault component.
During a fault, the PV must generate reactive power to suppress the voltage drop; the changes in output voltage and current before and after the fault must be analyzed. Subsequently, the change in bus voltage and current as a function of the PV output are analyzed. The PV output voltage and current phasors before and after the fault are shown in Fig. 2 [17]. Figure 2a shows U d before the fault, where the output active current is I d , U d is the post-fault voltage, θ is the phase lag angle, I 1 and I 2 are the fault currents, and the fault current range is the sector area surrounded by the arc with radius I max . Additionally, I f1 and I f2 are the current fault components. If the voltage drop is large when a distribution network fault occurs (i.e., U d is small), the phase difference between I and U d before the fault is > 90°. If the voltage drop is small (i.e., U d is large), the phase difference between I and U d is < 90°. Thus, the voltage drop magnitude contributes to the phase difference between I and U d . Furthermore, the post-fault voltage has a critical value, as indicated in Fig. 2b. I d and I q are the post-fault active and reactive current, respectively, and the phase angle between I and U d is 90° [18]. As shown in Fig. 2b: and because I = I max , it follows that Regulations state that if a voltage drop is > 10%, then the reactive current must be increased by 2% for every 1% voltage decrease [19]. Therefore, where k ≥ 2. Equation (4) can be substituted into (3) to obtain the relationship between the voltage and critical voltage during normal operation: When the post-fault voltage is less than the critical voltage, the phase angle between I and U d is > 90°. When the postfault voltage is greater than the critical voltage, the phase angle between I and U d is < 90°. Analysis of the PV output fault current characteristics forms the basis of microgrid fault analysis.

Microgrid fault characteristics during PV low-voltage ride-through
Fault characteristics were analyzed by constructing a simple microgrid model [20]. Figure 3a shows a microgrid that comprises four buses, feeders, and a grid-connected PV. Buses  When a fault occurs at point f, the voltage drop of buses E and M is small because the microgrid and distribution network are connected to the utility grid, which can support the necessary voltage. However, if the fault point is very close to bus E, there may also be a large voltage drop between buses E and M.
If the voltage drop between buses E and M is large, the analysis of bus E is similar to the analysis of bus G. Therefore, we focus on the case where the voltage drop at buses E and M is small. As shown in Fig. 3b, where U E is the positive-sequence fault voltage component at bus E, and Z S is inductive; the phase angle between I E1 and U E is < 90°and in the third quadrant. From this analysis, we can extract the phase relationship between the fault voltage and current components for bus E, as shown in Fig. 4b.
As shown in Fig. 3b, I F3 = − I G1 ; thus, the phase relationship between the fault voltage and current components of bus G can be analyzed. I G2 = − I 1 , I G3 = − U G /Z 1 , and I G1 = −( I G2 + I G3 ), which is the positive-sequence fault voltage component. The load impedance is inductive; thus, the phase difference between I G3 and U G is < 90°, which is in the third quadrant. There  The phase difference between I 1 and U G is > 90°; therefore, I 1 is in the third quadrant. As shown in the fault component diagram, the fault components of each bus F feeder are I F3 = − I G1 , I F2 = − U F /Z 3 , and I F1 = −( I F2 + I F3 ). Therefore, the fault vector direction I G1 is known. As shown in Fig. 4c, the phasor diagram of each fault component for bus F can be obtained using this analysis approach.
When a fault occurs at point f, the positive-sequence fault current component at buses E and F reveals that the fault occurred between feeder E2 on bus E and feeder F1 on bus F.
A fault is indicated in the microgrid model shown in Fig. 3a. Four fault points (f 1 , f 2 , f 3 , and f 4 ) were set as shown in Fig. 5a, and b shows the positive-sequence fault additional network considering a failure at point f 4 . Z F is the resistance of the fault point, Z S is the equivalent positive-sequence system impedance, and Z 11 and Z 22 are the equivalent positive impedances of the lines. Z EF , Z BF , and Z FG are the equivalent positive-sequence impedances between the buses indicated by the subscripts.
The fault components on buses E, F, G, and M were analyzed for cases in which faults occurred at f 1 -f 4 . Because failure at point f 1 is discussed above, the analysis is not repeated here. However, analysis of bus G is necessary. When the bus G voltage drop is small, the phase angle between I 1 and U G is < 90°; thus, I 1 is in the fourth quadrant. Figure 6 shows the phasor diagram for the superimposed fault network.
Bus F was analyzed, and the fault current phasor diagram is shown in Fig. 7. According to the additional network I F1 = − I E2 , I F3 = − I G1 , and I F2 = −( I F1 + I F3 ).
When a failure occurs at f 3 , the relationships between the bus E and M positive-sequence fault phasors are similar to the relationships shown in Fig. 4a, b.
Because bus F is supported by the utility power grid, the voltage drop is small and can be obtained from the current fault component diagram where I F1 = − I E2 , I F2 = U F /Z 3 , and I F3 = −( I F1 + I F3 ). Phasor I E3 is similar to the phasor shown in Fig. 7. The positive-sequence impedance of the load is inductive; the phase difference between I F2 and U F is thus < 90°, which is in the third quadrant. The phase relationship between the bus F fault voltage and current components can be obtained as shown in Fig. 8.
The positive-sequence fault additional network for a fault at f 4 is shown in Fig. 5b. In this case, the fault phasor components for buses E, M, and F are similar to the components in Fig. 4a, b and Fig. 8.
Bus G is supported by the utility power grid, the voltage drop is small, and the phase difference between I 1 and U G is < 90°; accordingly, I 1 is in the fourth quadrant. As shown in Fig. 5b, I G2 = − I 1 , I G1 = − I F3 , and I G3 = −( I G1 + I G2 ). Therefore, the fault voltage and current components can be obtained as shown in Fig. 9. When faults occur at different locations, the relative phase angles of all positive-sequence fault current components on all bus branch feeders can be obtained based on the phase relationships between fault components. In normal operation, the phase angle between current components on any two feeders is between 0°and 90°.
When a microgrid fault occurs, the difference between the fault and normal current phase angles is between 90°and 180°.
Assume that bus A has three branch feeders A1, A2, and A3. If A3 fails, the additional network of the positivesequence fault is shown in Fig. 10; the positive-sequence fault currents in each branch are I A1 , I A2 , and I A3 , respectively. In Fig. 10, Z 1 is the equivalent positive-sequence impedance of bus A between the front feeder and load, Z 2 is the equivalent impedance of branch 2, Z 3 is the impedance from the fault point to bus A in branch 3, Z 4 is the impedance from the fault point to the end of branch 3, and Z F is the additional impedance of the power supply at the fault point.
Because the positive-sequence impedance of the line and load is inductive, Z 1 , Z 2 , Z 3 , and Z F are inductive. The phase angle of U F is θ u ; the phase angles of Z 1 to Z F are θ 1 to θ F , respectively. Therefore, the limit: where θ x is the phase angle that corresponds to the addition of vectors Z 1 to Z F . As shown in Fig. 10, the fault current of each branch feeder can be represented by an equivalent positive-sequence voltage and resistance at the fault point, where The following can then be obtained: Because the positive-sequence impedance of the line and the load is inductive, 0 • < θ 1 < 90 • and 0 • < θ 2 < 90 • ; If the absolute values of (9) are transformed to 0 • → 180 • , we obtain: Therefore, the fault can be identified and located using the phase angle difference of each branch feeder. Current information is also used as a highly reliable protection method that can rapidly detect failures with high sensitivity.

Microgrid protection scheme
The microgrid protection system proposed here is a centralized distributed protection scheme. The unit protection module is distributed at the bus node to measure the current in each feeder.
In a conventional unit protection module, each feeder must be equipped with measurement devices. The protection method proposed here only utilizes a voltage transformer installed at the grid-connected bus, which reduces cost. The unit protection module obtains current data from the grid-connected bus and extracts the positive-sequence fault current component. A trigger is transmitted to the unit protection module when a voltage drop occurs at the grid-connected bus, thereby beginning the process to locate the feeder in fault. Communication tie lines can be used to transmit protection information between adjacent modules (Figs. 11, 12).
A fault is identified if a bus voltage drop is > 10% of the original value. When this trigger is met, the process shown in Fig. 13 is initiated.
The phase angles of the feeder fault current components are extracted and translated between -180°and 180°, and the absolute differences between the feeder phase angles are where α is the phase angle difference, I Y is the target line, and Y Yn are the other lines on bus X. A branch is in fault if the difference is between 90°and 180°, and the fault state criterion output is -1. Otherwise, the branch feeder is healthy and the fault state is 1. When the detection process is initiated, all bus branch feeders are simultaneously inspected for faults; a trip signal is triggered if a faulty line is detected. A trip signal is sent to the opposite side. The system will trip if the signal is received simultaneously; otherwise, it will not trip.

Case study
The microgrid model was constructed using PSCAD/EMTDC electromagnetic transient simulation software (with reference to the model diagram shown in Fig. 5). The microgrid secondary voltage was 10 kV, the length of each feeder was 500 m, the positive-sequence resistance of the line was 0.64 km -1 , and the positivesequence inductive reactance of the line was 0.12 km -1 . The zero-sequence resistance of the line was 2.00 km -1 , the zero-sequence inductive reactance of the line was 0.4 km -1 , and the load was 0.4 MW. The test fault was a three-phase ground fault that occurred at 1 s.
The parameters of the microgrid are shown in Table 1.  The current phase angle and reactive output of a distributed generator with fault resistances measuring 0.01 , 0.1 , and 1 are shown in Fig. 14. The alternating current line fault occurred after 1 s, and the reactive current output increased rapidly. The fault ended at 2 s, and the reactive current returned to 0 A. The output current phase angle increased as the distributed generation current increased.
Three phase fault was considered in the simulation. Figure 1 summarizes the simulation of a fault that occurred at f 1 . The phase angle of feeder F 2 was opposite to the phase angles of feeders F 1 and F 3 . The fault lines were I E2 , I F1 , and I M1 ; the healthy lines were I E1 , I E3 , I F2 , I F3 , I M2 , and I M3 . The phase angle between the faulty and healthy lines was between 90°and 180°; the phase angle between the healthy lines was in the range of 0°to 90° (  Fig. 15). Figure 16 summarizes the simulation of a fault that occurred at f 2 . The phase angle of feeder F 2 was opposite to the phase angles of feeders F 1 and F 3 . The fault lines were I E2 , I F2 , I G1 , and I M1 ; the healthy lines were I E1 , I E3 , I F1 , I F3 , I M2 , I M3 , I G2 , and I G3 . The phase angle between the faulty and healthy lines was between 90°a nd 180°; the phase angle between the healthy lines was in the range of 0°to 90°.  Figure 17 summarizes the simulation of a fault that occurred at f 3 . The phase angle of feeder F 3 was opposite to the phase angles of feeders F 2 and F 1 . The fault lines were I E2 , I F3 , and I M1 ; the healthy lines were I E1 , I E3 , I F2 , I F1 , I M2 , and I M3 . The phase angle between the faulty and healthy lines was between 90°and 180°; the phase angle between the healthy lines was in the range of 0°to 90°. Figure 18 summarizes the simulation of a fault that occurred at f 4 . The phase angle of feeder G 3 was opposite to the phase angles of feeders G 2 and G 1 . The fault lines were I E2 , I F3 , I G3 , and I M1 ; the healthy lines were I E1 , I E3 , I F2 , I F1 , I M2 , I M3 , I G1 , and I G2 . The phase angle between the faulty and healthy lines was between 90°a nd 180°; the phase angle between the healthy lines was in the range of 0°to 90°. Table 2 shows the results of simulated feeder faults. The experimental results show that the proposed algorithm can successfully identify the internal fault and locate the appropriate section when the microgrid is connected to the grid.
The single-phase ground fault and two-phase short-circuit fault were analyzed when the fault resistance at f 4 was set to  Fig. 19 Simplified model of microgrid when a fault occurs at f 4 0.1 . Differences between the positive-sequence currents of each bus F feeder were measured to demonstrate the efficacy of the proposed process. When a fault occurs at f 4 , the faulty line is I F3 , whereas the healthy lines are I F1 and I F2 . This fault condition is illustrated in Fig. 5b, and a simplified model of the microgrid is shown in Fig. 19.
The phase relationships and angles of the bus F fault components when f 4 is a single-phase ground fault are shown in Fig. 20. Figure 20 shows that the phase angle between I F3 , and I F1 and I F2 is between 90°and 180°when a single-phase grounding fault occurs. The phase angle between the two healthy lines is in the range of 0°to 90°. The phase angle relationship is shown in Fig. 21.
The phase relationships and angles of the bus F fault components when f 4 is a two-phase short-circuit fault are shown in Fig. 22. Figure 22 shows that the phase angle between I F3 , and I F1 and I F2 is between 90°and 180°when a two-phase short-circuit fault occurs. The phase angle between the two  Fig. 23. Figures 21 and 23 demonstrate that single-phase grounding faults and two-phase short-circuit faults result in phase angle differences that satisfy the fault judgment ranges proposed here. Therefore, the proposed fault discrimination is appropriate for various fault types.
Results have already been compared with existing schemes. The traditional fault location method is using zerosequence voltage and zero-sequence current phase angle comparison. (c) The current-only polarity comparison-based protection method for microgrids uses local measurement information and is suitable for the PV low-voltage ride-through condition (Fig. 24).
When the fault occurs, the intelligent electronic device (IED) is needed to control the circuit breaker to clear the  Fig. 17 as an example, and the fault occurred at 3.5 s. When the fault occurs, the current waveforms at each branch feeder of bus F are shown in Fig. 25.
The positive-sequence current is measured at each feeder. The angle difference between different branches is calculated, respectively. The positive-sequence angle difference is shown in Fig. 26.
The IED devices detect that angle difference meets the fault operation condition and sends a trip signal to the circuit breaker. After the circuit breaker action delay of 10 ms, it

Conclusion
Fault current phase characteristics during PV low-voltage ride-through were analyzed and a local protection method based on current data was proposed. The method utilizes the phase angle difference between the positive-sequence current of healthy feeders and the faulty feeder (i.e., feeder near the faulty end). The phase angle between two healthy feeders is acute, whereas the phase angle between a healthy and faulty feeder is obtuse, which allows faults to be located rapidly and accurately. Finally, a PV simulation model that included a low-voltage ride-through was constructed using PSCAD/EMTDC, and an internal microgrid fault was simulated and verified. The simulation results demonstrated the effectiveness of the fault characteristic analysis process, as well as the feasibility of the proposed protection scheme.