Simulation Test of a DC Fault Current Limiter for Fault Ride-Through Problem of Low-Voltage DC Distribution

: The low voltage direct current (LVDC) distribution networks are connected with too many kinds of loads and sources, which makes them prone to failure. Due to the small damping value in the DC lines, the fault signal propagates so fast that the impact current with the wave front of millisecond and the transient voltage pose great challenges for fault detection. Even worse, some faults with small currents are di ﬃ cult to detect and the communication is out of sync, resulting in protection misoperation. These problems have severely a ﬀ ected the new energy utilization. In view of this, a DC fault current limiter (FCL) composed of inductance, resistance, and power electronic switch was designed in this paper. The rising speed of fault current can be decreased by the series inductance and the peak value of the fault current can be limited by series impedance, thus in this way the running time can be gained for fault detection and protection. For distributed energy access, by deducing the short circuit fault characteristic expression of LVDC distribution network, the feasibility of FCL was veriﬁed. Based on the structure of the bridge-type alternating current (AC) current limiter, the structure and parameters of the DC FCL were determined according to the fault ride-through target. Then, a low voltage ride-through strategy based on DC FCL was proposed for the bipolar short-circuit fault of LVDC distribution network. Finally, MATLAB / Simulink simulation was used to verify the rationality of the proposed FCL and its ride-through strategy.


Introduction
With the application of large quantities of direct current (DC) equipment and the increasing access and utilization of various distributed energy sources, DC distribution networks have a broad application prospect [1,2]. The DC distribution systems based on two-level voltage source converters (VSC) are suitable for access to distributed generation and DC load with fast and independent active or reactive power control, and they have abilities such as black start, multiterminal network structure expansion and rapid reduction of interference, making them widely applied in middle and low voltage industries [3]. However, the transient characteristics of VSC caused by the fault are accompanied by the rapidly increasing DC fault current, and the poor overload capacity of the power electronic converter affects its stability [4]. With the increase of the installed capacity, North American Electric Reliability Corporation and the workgroup member of Institute of Electrical and Electronic Engineers (IEEE) took full consideration of the high permeability of distributed power supply and required the fault ride-through ability for distributed power supply to support the stable operation of the power grid [5,6]. Therefore, it is necessary to enhance the fault ride-through capability of DC distribution network with distributed power supply. According to the fault position with respect to the VSC converter [16], the faults of DC distribution systems can be roughly divided into the following types: 1) Faults on AC side: VSCs are widely used as AC-DC or DC-AC converters for power conversion. In fact, the AC side fault of gridconnected inverter has been taken into account when designing the protection scheme. 2) Internal failure: internal failure of VSC includes failure of power electronic devices, such as insulated gate bipolar transistor (IGBT), etc., which can protect the system by providing backup converters or redundant devices. 3) DC network fault: DC short circuit fault is the most serious fault of VSCs. In this case, IGBTs may be blocked from self-protection, and the large VSC filter capacitor and low impedance cable may cause overcurrent or undervoltage, so the fault protection and ride-through make use of the FCL installed in the DC network. According to the study in References [17][18][19][20], when there is a short circuit between the positive and negative electrodes of the DC line, a pole-pole fault occurs with low fault impedance. When the positive or negative terminal of a DC line is ground shortcircuited, a pole-ground fault occurs, accompanied by a low impedance fault or a high impedance fault. The location of these faults is either in the DC bus or in one of the DC cables, as F1-F10 in Figure  1. F1-F4 and F5-F8 respectively represent the pole-pole faults and pole-ground faults of the DC cable, while F9-F10 represent the pole-pole faults and pole-ground faults of the DC bus, respectively. Compared with these two faults, although the pole-ground fault is more likely to occur, the current and voltage caused by the ground fault is far less serious than the DC bipolar short-circuit fault. Moreover, when the single-pole bus ground fault occurs, the DC switch can quickly disconnect the fault and restore the load power supply through the nonfault pole bus. Therefore, this paper takes the connection of photovoltaic power system to LVDC network as an example to deduce the analytical expression of the fault characteristic of bipolar short circuit in the system (limited by space, the derivation is described in detail by Shuai [17][18][19][20]; so as to provide a basis for the proposed FCL structure and parameter design. According to the fault position with respect to the VSC converter [16], the faults of DC distribution systems can be roughly divided into the following types: 1) Faults on AC side: VSCs are widely used as AC-DC or DC-AC converters for power conversion. In fact, the AC side fault of grid-connected inverter has been taken into account when designing the protection scheme. 2) Internal failure: internal failure of VSC includes failure of power electronic devices, such as insulated gate bipolar transistor (IGBT), etc., which can protect the system by providing backup converters or redundant devices. 3) DC network fault: DC short circuit fault is the most serious fault of VSCs. In this case, IGBTs may be blocked from self-protection, and the large VSC filter capacitor and low impedance cable may cause overcurrent or undervoltage, so the fault protection and ride-through make use of the FCL installed in the DC network. According to the study in References [17][18][19][20], when there is a short circuit between the positive and negative electrodes of the DC line, a pole-pole fault occurs with low fault impedance. When the positive or negative terminal of a DC line is ground short-circuited, a pole-ground fault occurs, accompanied by a low impedance fault or a high impedance fault. The location of these faults is either in the DC bus or in one of the DC cables, as F1-F10 in Figure 1. F1-F4 and F5-F8 respectively represent the pole-pole faults and pole-ground faults of the DC cable, while F9-F10 represent the pole-pole faults and pole-ground faults of the DC bus, respectively. Compared with these two faults, although the pole-ground fault is more likely to occur, the current and voltage caused by the ground fault is far less serious than the DC bipolar short-circuit fault. Moreover, when the single-pole bus ground fault occurs, the DC switch can quickly disconnect the fault and restore the load power supply through the nonfault pole bus. Therefore, this paper takes the connection of photovoltaic power system to LVDC network as an example to deduce the analytical expression of the fault characteristic of bipolar short circuit in the system (limited by space, the derivation is described in detail by Shuai [17][18][19][20]; so as to provide a basis for the proposed FCL structure and parameter design.

Fault Characteristics of LVDC Distribution Network with Photovoltaic Power System
The equivalent circuit of a pole-pole short-circuit fault for DC distribution network with distributed power supply is shown in Figure 2, from which the short-circuit fault characteristics of the network can be deduced. The photovoltaic unit is connected to the DC bus by a booster converter, where R c1 , L c1 , R c2 and L c2 are the π-type equivalent resistance and inductance of the positive and negative cables. From the VSC and boost converter to the fault position, due to the larger DC bus capacitance C 1 , the grounding capacitance of the cable can be ignored. In fault analysis, considering the fault resistance is far less than the load resistance R load , the load resistance R load can be neglected as well.

Fault Characteristics of LVDC Distribution Network with Photovoltaic Power System
The equivalent circuit of a pole-pole short-circuit fault for DC distribution network with distributed power supply is shown in Figure 2, from which the short-circuit fault characteristics of the network can be deduced. The photovoltaic unit is connected to the DC bus by a booster converter, where Rc1, Lc1, Rc2 and Lc2 are the π-type equivalent resistance and inductance of the positive and negative cables. From the VSC and boost converter to the fault position, due to the larger DC bus capacitance C1, the grounding capacitance of the cable can be ignored. In fault analysis, considering the fault resistance is far less than the load resistance Rload,, the load resistance Rload can be neglected as well.
Photovoltaic Cell AC The fault response process is divided into the following three stages: 1. The equivalent circuit of DC capacitance discharging stage is shown in Figure 3. The fault process is dominated by C1 on the VSC side and Cb of the booster converter. For a bigger C1 than Cb, the discharge time constant τ = RC; when they are approximately equal, the discharge time of C1 will be greater than that of Cb. Assume that failure occurs at the time t0, and the steady-state bus voltage and initial cable current of the two converters on DC side are Applying Kirchhoff Voltage laws (KVL) to the circuit consisted of Rc1, Lc1 and the DC bus capacitance C1 yields: The fault response process is divided into the following three stages: 1.
The equivalent circuit of DC capacitance discharging stage is shown in Figure 3.

Fault Characteristics of LVDC Distribution Network with Photovoltaic Power System
The equivalent circuit of a pole-pole short-circuit fault for DC distribution network with distributed power supply is shown in Figure 2, from which the short-circuit fault characteristics of the network can be deduced. The photovoltaic unit is connected to the DC bus by a booster converter, where Rc1, Lc1, Rc2 and Lc2 are the π-type equivalent resistance and inductance of the positive and negative cables. From the VSC and boost converter to the fault position, due to the larger DC bus capacitance C1, the grounding capacitance of the cable can be ignored. In fault analysis, considering the fault resistance is far less than the load resistance Rload,, the load resistance Rload can be neglected as well. The fault response process is divided into the following three stages: 1. The equivalent circuit of DC capacitance discharging stage is shown in Figure 3. The fault process is dominated by C1 on the VSC side and Cb of the booster converter. For a bigger C1 than Cb, the discharge time constant τ = RC; when they are approximately equal, the discharge time of C1 will be greater than that of Cb. Assume that failure occurs at the time t0, and the steady-state bus voltage and initial cable current of the two converters on DC side are Applying Kirchhoff Voltage laws (KVL) to the circuit consisted of Rc1, Lc1 and the DC bus capacitance C1 yields: The fault process is dominated by C 1 on the VSC side and C b of the booster converter. For a bigger C 1 than C b , the discharge time constant τ = RC; when they are approximately equal, the discharge time of C 1 will be greater than that of C b . Assume that failure occurs at the time t 0 , and the steady-state bus voltage and initial cable current of the two converters on DC side are Applying Kirchhoff Voltage laws (KVL) to the circuit consisted of R c1 , L c1 and the DC bus capacitance C 1 yields: The corresponding solution of (2) is (3). where, The moment i 1 reaches maximum can be calculated by (4).
The moment that voltage of the capacitance C 1 drops to zero is: where, Applying Kirchhoff Current laws (KCL) to the circuit consisted of C pv , L b and D b yields: At this moment, i cb , i cp and i pv are the discharge current of C b and C pv and the photovoltaic cell equivalent current, respectively.
Applying KCL to the circuit consisted of C b , R c2 and L c2 yields: The differential equation can be solved as follows: where, Applying Kirchhoff Voltage laws (KVL) to the circuit consisted of C pv , L b , R c2 , L c2 and ignoring the voltage drop on L b produced by i pv yield: Energies 2020, 13, 1753 6 of 17 The differential equation can be solved as follows: i cp (t) = D 1 cos ω cp t e −α cp t + D 2 sin ω cp t e −α cp t +D 3 cos(ω cb t)e −α cb t + D 4 sin(ω cb t)e −α cb t (10) where, The diode D b of the booster converter will experience a large transient current at this stage which will quickly damage this diode.

2.
Free-conducting stage of diode: During the discharge process of C 1 , the electrical energy will change to electromagnetic energy and be stored in the DC side inductance. Its drive current of the reverse electromotive force flows through the VSC bridge commutating diodes of each phase. When the capacitor voltage C 1 reaches zero, the equivalent circuit of the stage is demonstrated in Figure 4.
The diode Db of the booster converter will experience a large transient current at this stage which will quickly damage this diode.
2. Free-conducting stage of diode: During the discharge process of C1, the electrical energy will change to electromagnetic energy and be stored in the DC side inductance. Its drive current of the reverse electromotive force flows through the VSC bridge commutating diodes of each phase. When the capacitor voltage C1 reaches zero, the equivalent circuit of the stage is demonstrated in Figure 4. The current and the voltage of C1 are always zero. Under the action of the conduction diode, the cable current i1on the VSC side is one third that of the current flowing through the conduction diode of each phase bridge, i.e., The initial value of i1 is shown in (3). According to (6), in the photovoltaic power system, the transient short circuit current 2 i will flow through the fault point and continue to discharge in the form of oscillation. According to (11), it is an important phase for the free-conducting diode of the VSC, for its current 1 i will mutate largely, which can quickly damage it.
3. Steady state: When the inductance discharge on the DC side ends, the current on the grid side will flow through the diode into the DC side. The equivalent circuit of this state is shown in Figure 5. The current and the voltage of C 1 are always zero. Under the action of the conduction diode, the cable current i 1 on the VSC side is one third that of the current flowing through the conduction diode of each phase bridge, i.e., The initial value of i 1 is shown in (3). According to (6), in the photovoltaic power system, the transient short circuit current i 2 will flow through the fault point and continue to discharge in the form of oscillation. According to (11), it is an important phase for the free-conducting diode of the VSC, for its current i 1 will mutate largely, which can quickly damage it.

3.
Steady state: When the inductance discharge on the DC side ends, the current on the grid side will flow through the diode into the DC side. The equivalent circuit of this state is shown in Figure 5. through the fault point and continue to discharge in the form of oscillation. According to (11), it is an important phase for the free-conducting diode of the VSC, for its current 1 i will mutate largely, which can quickly damage it.
3. Steady state: When the inductance discharge on the DC side ends, the current on the grid side will flow through the diode into the DC side. The equivalent circuit of this state is shown in Figure 5. Figure 5. Equivalent circuits for the short-circuit fault at steady-state stage. At this stage, the short-circuit fault current comes from the AC grid. A three-phase short-circuit current is obtained by the analysis. The phase currents, namely i ga , i gb and i gc flow into the DC side through the corresponding bridge, while the currents on the AC side are under the rectified state without any control.
According to (6), in the photovoltaic power system, the transient short circuit current i 2 is damped oscillation. The three-phase AC power supply and the photovoltaic power unit provide the fault current through the uncontrolled rectifier circuit and the booster circuit respectively. In addition, the amplitude of the fault current in this stage is significantly smaller than that in the previous two stages.

Fault Analysis of LVDC Distribution Network with Photovoltaic Power System
Based on the fault characteristics of the LVDC distribution network connected to the photovoltaic power system, the following information can be obtained.

1.
Peak current and its arrival time According to [17][18][19][20], when a small-resistance short-circuit fault happens, the voltage across the capacitance of the photovoltaic booster converter drops to zero within 0.627 ms, while the time for the voltage of C 1 droping to zero is about 3.1 ms. Since the voltage across the capacitance is close to the rated bus voltage, the fault current i sc will approach the value larger than i 1 and the time for the fault current i 2 to reach the peak is far less than that of i 1 . In addition, i 2 has a higher peak than i 1. In the first stage, the peak value of fault current is the highest and the time is the shortest.
According to (3), the initial amplitude of the fault current is composed of A 1 , A 2 and e −αc1t . Among them, A 1 is the rated current at the steady state before the fault occurs, and it is not a big number. Because the time interval of the moment of failure is very small, the term e −αc1t approaches 1 where α c1 =R c1 /2L c1 , a numerically large value due to the small value of L c1 , and the value of L c1 C 1 is small, resulting in a big value of ω c1 , a big α c1 and ω c1 make a numerically large A 2 , so the amplitude of the fault current is mainly determined by the second coefficient in (3).
According to (4), t i1 (max) is in proportion to 1/ω c1 and y = arctan(x) is a numerical value in the interval of (−π/2, π/2). This is because the big value of ω c1 , t i1 (max) is rather small, which means the time for the fault current to reach the peak is quite short.

2.
Continuity condition for three fault response According to the fault response analysis in the three stages, the converter is locked after the fault, and the power supply on the AC side and the photovoltaic side is disconnected. In the first stage, the capacitance releases energy storage into the inductance. When the voltage of the capacitance is zero, it stops releasing energy. In the second stage, the current flowing through the inductance follows the VSC side reverse diode. Since there is always the magnetic field energy released by the inductance consumed by the impedance in the discharge circuit, when the energy storage in the inductance is small, the power grid will discharge through the diode until the inductance discharge process on the DC side ends. Thus, the rapid release of capacitor-stored energy is the first condition for the continuity of the three stages.
In the pole-pole short circuit fault, the fault resistance R fault is generally small, thus C(R c + R fault ) 2 /4L c < 1. According to this, the fault response is an underdamped oscillation process. Moreover, the first-stage transient process of the three stages is the most severe (the fault current may reach thousands of amperes in 2 ms). Therefore, it is in the other continuous conditions that the fault resistance meets the condition of underdamped oscillation.
From the perspective of fault protection and fault ride-through, some measures should be taken in the first stage of fault response. On the one hand, the release rate (rise rate of the current and its amplitude) of capacitance-stored energy needs to be reduced; on the other hand, the condition that C(R c + R fault ) 2 /4L c > 1 should be satisfied to ensure the fault current response in the state of overdamping attenuation and protect the sensitive device of the fault branch. At this point, the voltage of the DC bus parallel capacitance will not drop to zero quickly, and the transient fluctuation is within the acceptable range. The latter two stages will also be avoided.

Countermeasures
To realize the fault ride-through, two measures should be taken in the first stage of the short-circuit fault to reduce the damage and break the continuity condition of the fault response. The first is to extend the time when the fault current reaches the peak (delay its rising speed) so as to facilitate the fault detection and protection operation; the second is to limit the amplitude of the fault current (protect the sensitive devices) so as to provide the guarantee for the recovery operation after the fault disappears. Theoretically, it is possible to achieve by increasing the capacitance discharge time constant and the fault circuit resistance.
According to (3), the current of capacitance energy release observes the law of e −t/τ , where τ = L/R. Therefore, the capacitance discharge time constant can be increased by increasing the inductance of the discharge circuit. According to (5), the increase of the inductance of the discharge circuit may also increase the value of 1/ω c1 , resulting in an increment in the time of the C 1 voltage dropping to zero. Due to the design, the FCL is connected in series in the circuit, thus it is unfavorable to add a series of resistance which will increase loss. Besides, it works at the DC steady state, and the appropriate increase of inductance will not affect the stable operation but reduce the coefficient of the second term A 2 . So, with the application of a series of inductance, the capacitance discharge time can be prolonged and the discharge speed can be reduced. In such way, the requirement of delaying the rise speed of the impact current can be met.
According to (3), the decrease of e −αc1t can reduce the amplitude of the fault current rapidly. That is to increase the value that α c1 = R c1 /2L c1 , which can be realized by adding a large resistance in series into the circuit to limit the amplitude of the fault current. When the DC system runs under steady condition, the application of series inductance for current limiting can effectively prolong the rise time of fault current and reduce its rise rate. After that, at the moment the appropriate current is found, the large resistance circuit is connected to limit the amplitude of the fault current. When the fault disappears, the current-limiting circuit is reconnected and the system resumes normal operation. All these functions can be realized by appropriate design of DC FCL.

Structure of FCL
The FCLs can be divided into superconducting current limiter, solid-state FCL (solid-state current limiter) and electromagnetic FCL [21][22][23]. During normal operation, it works in a negligible impedance Energies 2020, 13, 1753 9 of 17 state; while in the failure mode, the high impedance is connected to inhibit the fault current. These two modes can be switched quickly.
The application object is the LVDC power distribution network with complex structure and various control strategy whose lines are short and damping is small. It is prone to failure and the failure propagation speed is rather fast. In the event of fault, the fault current will rise to a dozen times or even one hundred times that of the rated current within two milliseconds, which seriously threatens the safety operation of sensitive devices and their protection performance. In addition to general current limiting function, four other functions are required. First, the current limiting works at the same time as the fault occurs, and there is no need for detection. If the detection, operation and high-speed DC circuit breaker are added, the reliability of protection and quick action are difficult to meet. Secondly, the current limiting function should be first operated to protect the sensitive devices by suppressing the impact currents caused by rapid disturbances on the one hand, while on the other hand it can provide sufficient time for detection and execution devices (such as fault detection, signal acquisition, signal transmission, information processing, switch and converter control, and DC protection devices) or the others. Thirdly, during the process, the energy storage in the inductance needs to be released to restore the capacity. Fourthly, for the current fluctuation caused by impact load or motor start, FCL is not needed.
Therefore, the configuration principle of FCLs in this paper is as follows: First, in order to avoid current limiting caused by startup, the power electronic switch (antiparallel IGBT, etc.) is closed on startup to short-circuit the FCL and then disconnected after it operates normally. Secondly, at steady state, the FCL is connected to the DC distribution network in series. Since the fault location is uncertain, the position should be located at the connection joint between the parallel capacitance and the line cable on the DC side of the distributed power supply or the connection joint between the DC bus and the line cable, so as to suppress the impact current caused by rapid disturbance and protect the sensitive devices. Thirdly, during the operation of current limiting, when the fault current exceeds two times the rated one, the circuit is switched to the large-resistance-limiting circuit. Fourthly, the energy dissipation circuit should be deployed for the current-limiting inductance to recover its current limiting capability before another failure.
The proposed structure of direct current fault current limiter (DC FCL) is shown in Figure 6.
Energies 2018, 11, x FOR PEER REVIEW 9 of 17 needs to be released to restore the capacity. Fourthly, for the current fluctuation caused by impact load or motor start, FCL is not needed. Therefore, the configuration principle of FCLs in this paper is as follows: First, in order to avoid current limiting caused by startup, the power electronic switch (antiparallel IGBT, etc.) is closed on startup to short-circuit the FCL and then disconnected after it operates normally. Secondly, at steady state, the FCL is connected to the DC distribution network in series. Since the fault location is uncertain, the position should be located at the connection joint between the parallel capacitance and the line cable on the DC side of the distributed power supply or the connection joint between the DC bus and the line cable, so as to suppress the impact current caused by rapid disturbance and protect the sensitive devices. Thirdly, during the operation of current limiting, when the fault current exceeds two times the rated one, the circuit is switched to the large-resistance-limiting circuit. Fourthly, the energy dissipation circuit should be deployed for the current-limiting inductance to recover its current limiting capability before another failure.
The proposed structure of direct current fault current limiter (DC FCL) is shown in Figure 6.  In Figure 6, Lsh is the current-limiting inductance, Rsh is the internal resistance of Lsh , Rrse is the leak resistance, D is the energy release circuit diode, and Lsh is the limiting resistor. Table 1 shows the switch logic functions of DC FCL.  In Figure 6, L sh is the current-limiting inductance, R sh is the internal resistance of L sh , R rse is the leak resistance, D is the energy release circuit diode, and L sh is the limiting resistor. Table 1 shows the switch logic functions of DC FCL. In the table, "1" represents close while "0" represents disconnect.
Function of switch state logic: K1, K2, K3 and K4 are the four switches for the four-state realization of fault ride-through. Before the distribution system starting, K1 is closed to make the FCL short-circuit to avoid the protection and switch action caused by the large current fluctuation at the start of the impact load. K1 is switched off and K2 is closed after the system operates normally. The current-limiting circuit is connected in series to the lines to prevent the overshoot of the impact current in real time. When the fault current reaches 2I N , the K2 is disconnected and K3 is closed, and the amplitude of fault current is limited in the range of 50% I N -I N . During failure, K4 is closed while K2 is disconnected. In the amplitude-limiting process, the current-limiting inductance discharges through the loop of L sh -r sh -R rse -D and thus its current-limiting capacity is recovered.

Parameter Design of FCL
The FCL designed in this paper is used to realize the fault ride-through when the bipolar short circuit occurs in the LVDC distribution lines, while the faults inside the converter and on the AC side are not involved, and normal operation should be restored immediately after the failure. Thus, the parameter design should consider these two operation modes. In the failure mode, the safe operation of the sensitive devices should be ensured and designed according to the analysis above; in the normal operation mode, we should refer to the standards. Due to the fact that it is installed at the outlet of parallel capacitor or the bus, every power outlet should have its own inductance and resistance.

Design of L sh
The design principle is to suppress the instantaneous overshoot of the fault at the initial stage. Due to the short interval between the early failure and failure, the designed current limiting and amplitude limiting interval should be short too, because the current-limiting inductance L sh cannot be too large (when the volume is too large, too much energy storage causes overvoltage, and the current reducing the system response speed is too slow, etc.). Therefore, the effect of current-limiting inductance is to delay the rising speed of impact current and prolong the time when the impact current reaches the peak but not to change the peak value of impact current. Among them, the maximum value of L sh,max should satisfy the overdamped oscillation condition during normal operation, that is, L < (R 2 C/4); the minimum value of L sh,min should be larger than the value that can withhold the fault current calculated by (3) and (6). It should be pointed out that the current should not break down the capacitance at the output terminal on the DC side. The most serious failure situation should be checked: a bipolar metallic short circuit occurred between the FCL connected in series at the parallel capacitor end and the connection point of the cable line, that is, R fault = 0.

Design of R rse
The simplest design principle is to quickly release the energy stored by the current-limiting inductance and restore its current limiting capability through a diode and a small resistor.
The resistance for energy release is related to the recovery time of current-limiting inductance, and the allowed maximum time interval should refer to the time from the access of the amplitude-limiting R sh to the disconnection. In practice, the time should be designed according to the most serious fault and the shortest time interval between two faults.

Fault Protection and Ride-Through Strategy Based on FCL
Based on the proposed DC FCL, the fault protection and ride-through of LVDC system can be realized, and the flowchart is depicted in Figure 7. In case of permanent failure, the differential protection is used as the FCL backup protection and the DC circuit breaker is disconnected at both ends of the line. In Figure 7, ∆t min > t 3 − t 0 , which means a longer duration of the voltage drop of the parallel capacitor than that of the fault when a short circuit occurs at the connection between the line terminal and the FCL. It shows 40% P Nload < P load < 60% P Nload , which means that the measured power on the load is within ±10% of half of the rated load power.  Figure 7. Fault ride-through strategy based on DC FCL.

Simulation Model
In order to verify the feasibility of the proposed fault ride-through strategy, the DC FCL was connected to the LVDC distribution network containing photovoltaic power generation system, as shown in Figure 8. The corresponding simulation model was built in MATLAB/Simulink, and the simulation was conducted before and after the occurrence of bipolar short-circuits in the middle of the 1km distribution line. The detailed parameters of the simulation model are listed in Table 2. In the figure, ILsh1and ILsh2 are the currents in the current-limiting inductances; Idc1and Idc2 are the DC line currents on the VSC side and boost converter side, respectively; Iload and Ishort are the currents

Simulation Model
In order to verify the feasibility of the proposed fault ride-through strategy, the DC FCL was connected to the LVDC distribution network containing photovoltaic power generation system, as shown in Figure 8. The corresponding simulation model was built in MATLAB/Simulink, and the simulation was conducted before and after the occurrence of bipolar short-circuits in the middle of the 1km distribution line. The detailed parameters of the simulation model are listed in Table 2.
In the figure, I Lsh1 and I Lsh2 are the currents in the current-limiting inductances; I dc1 and I dc2 are the DC line currents on the VSC side and boost converter side, respectively; I load and I short are the currents in the load resistance and short-circuit resistance; U dc1 and U dc2 are the bus voltage on the C 1 side and C b side, respectively. Energies 2018, 11, x FOR PEER REVIEW 12 of 17 Figure 8. Equivalent circuit of LVDC distribution network connected by DC FCL.
In the simulation, the bipolar short circuit fault is divided into two categories: the type of large current (Idc ≥ 2IN) and the type of small current (Idc < 2IN), as shown in Figures 9-10 and Figures 11-12. The whole simulation process lasted for four seconds, in which 0s-1s was the starting process of the distribution network, and after that, the system went into normal operation. During the interval of 1s to 2s, the system operated normally. At 2s, the bipolar short circuit occurred. During 2s-4s, the fault proceeded. After 4s, the limiter circuit was switched back to the current-limiting circuit and stable operation was resumed if the fault disappeared; otherwise, for permanent fault, the DC circuit breaker of the fault branch should be disconnected for fault alarm.

Bipolar Short Circuit Fault of Large Current Type
In the simulation of large-current fault, the bipolar fault resistance is 0.01 Ω, smaller than the load resistance (8.3 Ω). Because of the similarity between the VSC DC side wave and the boost converter DC side wave, the simulation of bipolar short-circuit simulation of large current is given in this paper with the time interval of 0s-4s, as shown in Figure 9. In Figure 9a, the voltage Udc1 across the capacitance C1 is drawn; in Figure 9b, the circuit current Idc1 on the VSC side is given. In Figure  9c, the waveform of the current ILsh1 on the current-limiting inductance in the FCL1 is given. Figure  10 is the VSC side simulation curve before and after the failure.   In the simulation, the bipolar short circuit fault is divided into two categories: the type of large current (I dc ≥ 2I N ) and the type of small current (I dc < 2I N ), as shown in Figures 9 and 10 and Figures 11 and 12. The whole simulation process lasted for four seconds, in which 0 s-1 s was the starting process of the distribution network, and after that, the system went into normal operation. During the interval of 1 s to 2 s, the system operated normally. At 2 s, the bipolar short circuit occurred. During 2 s-4 s, the fault proceeded. After 4 s, the limiter circuit was switched back to the current-limiting circuit and stable operation was resumed if the fault disappeared; otherwise, for permanent fault, the DC circuit breaker of the fault branch should be disconnected for fault alarm.
Energies 2018, 11, x FOR PEER REVIEW 12 of 17 Figure 8. Equivalent circuit of LVDC distribution network connected by DC FCL.
In the simulation, the bipolar short circuit fault is divided into two categories: the type of large current (Idc ≥ 2IN) and the type of small current (Idc < 2IN), as shown in Figures 9-10 and Figures 11-12. The whole simulation process lasted for four seconds, in which 0s-1s was the starting process of the distribution network, and after that, the system went into normal operation. During the interval of 1s to 2s, the system operated normally. At 2s, the bipolar short circuit occurred. During 2s-4s, the fault proceeded. After 4s, the limiter circuit was switched back to the current-limiting circuit and stable operation was resumed if the fault disappeared; otherwise, for permanent fault, the DC circuit breaker of the fault branch should be disconnected for fault alarm.

Bipolar Short Circuit Fault of Large Current Type
In the simulation of large-current fault, the bipolar fault resistance is 0.01 Ω, smaller than the load resistance (8.3 Ω). Because of the similarity between the VSC DC side wave and the boost converter DC side wave, the simulation of bipolar short-circuit simulation of large current is given in this paper with the time interval of 0s-4s, as shown in Figure 9. In Figure 9a, the voltage Udc1 across the capacitance C1 is drawn; in Figure 9b, the circuit current Idc1 on the VSC side is given. In Figure  9c, the waveform of the current ILsh1 on the current-limiting inductance in the FCL1 is given. Figure  10 is the VSC side simulation curve before and after the failure.  As shown in Figure 9, the system started at the interval of 0-1s. In Figure 9a, the voltage Udc1 across the capacitance CC1 gradually rose to the rated voltage of 500 V; in Figure 9b, the current waveform of Idc1 demonstrated a fluctuation on the VSC side during the start-up process. In Figure  9c, because K1 in the FCL1 was switched off and the current-limiter was short-circuited, ILsh1 was zero.
At 1s, K1 was disconnected while K2 was closed, and the current-limiter began to work. During 1s to 2s, the system operated normally. Due to the current-limiting inductance, the DC bus voltage and DC line current entered steady state after a short fluctuation. Udc1 in Figure 9a reached 500 V while Idc1 in Figure 9b reached 60 A; ILsh1 in Figure 9c gradually increased from 0 to the rated current of 60 A. A short circuit fault occurred in the distribution line at 2s. The fault process of 2s-2.05s is shown in Figure 10. As analyzed in Section 3.2.1, the current-limiting inductance extends the discharge time of the capacitor and the rise speed of the impact current. In Figure 10a, Udc1 of the capacitor C1 did not decrease to 0 V rapidly. At 2.017 s, the line current Idc1 rose to about 170 A and reached the threshold of amplitude-limiting, that is, Idc ≥ 2IN. At this moment, K2 was disconnected and K3 was closed. The amplitude-limiting circuit was connected to prevent the capacitance from discharging rapidly. In this case, Udc1 in Figure 10a only decreased to 200 V and restored to normal operation at 500 V in 5ms. The DC side line current Idc1 in Figure 10b restored to normal operation at 50 A in 4 ms. The K4 was closed while K3 was closed, and the energy leakage circuit was connected, which caused the current flowing through the current-limiting inductance ILsh1 to decrease from 170 A to 0 A in 0.34 s in Figure 10c and thus the current-limiting capacity was restored.

Bipolar Short Circuit Fault of Small Current Type
In the small current fault simulation, the resistance for a bipolar short circuit fault is 16 Ω, 2 times larger than the load resistance (8.3 Ω). The DC simulation curves on the VSC side and boost converter side are demonstrated in Figure 11 and Figure 12, respectively.

Simulation curve on VSC side
As shown in Figure 11, the system started at 0s-1s and operated normally during 1s-2s. At this time, K1, K3 and K4 were disconnected while K2 was closed, and the current-limiting circuit of FCL1 was connected. At 2s, a small current short circuit fault occurred in the distribution line. Due to the variation range of the line current which did not meet the large current condition, K2 was still closed and the FCL1 worked in the current-limiting state. In Figure 8, the capacitance C1 discharged through the short-circuit resistance, resulting in a decrease of Udc1to 400V in Figure 11a. The short-circuit

Simulation Curve on Boost Converter Side
As shown in Figure 12, the system started at 0s-1s and operated normally during 1s-2s. At this time, K1, K3 and K4 were disconnected while K2 was closed, and the current-limiting circuit of FCL2 was connected. At 2 s, a small current short circuit fault occurred in the distribution line. Due to the variation range of the line current which did not meet the large current condition, K2 was still closed and the FCL2 worked in the current-limiting state. In Figure 8, the capacitance Cb discharged through the short-circuit resistance, resulting in a decrease of Udc2 in Figure 12a to 400 V. The short-circuit resistance shunt caused a decrease of booster converter side line current Idc2 from −60 A to −50A in Figure 12b (follow the specified direction in Figure 9).
In Figure 12c, the load current dropped from the rated value of 60 A to 48 A. Moreover, the close of K2 made the line current the same as the one in the current-limiting inductance in FCL2, so the current ILsh2 in Figure 12d which flowed through the current-limiting inductance became 50 A. At this point, because the circuit breaker trip conditions were not reached, the fault current remained stable, which conformed to the condition that Iload < 10%IN.

Bipolar Short Circuit Fault of Large Current Type
In the simulation of large-current fault, the bipolar fault resistance is 0.01 Ω, smaller than the load resistance (8.3 Ω). Because of the similarity between the VSC DC side wave and the boost converter DC side wave, the simulation of bipolar short-circuit simulation of large current is given in this paper with the time interval of 0 s-4 s, as shown in Figure 9. In Figure 9a, the voltage U dc1 across the capacitance C 1 is drawn; in Figure 9b, the circuit current I dc1 on the VSC side is given. In Figure 9c, the waveform of the current I Lsh1 on the current-limiting inductance in the FCL1 is given. Figure 10 is the VSC side simulation curve before and after the failure.
As shown in Figure 9, the system started at the interval of 0-1 s. In Figure 9a, the voltage U dc1 across the capacitance CC 1 gradually rose to the rated voltage of 500 V; in Figure 9b, the current waveform of I dc1 demonstrated a fluctuation on the VSC side during the start-up process. In Figure 9c, because K1 in the FCL1 was switched off and the current-limiter was short-circuited, I Lsh1 was zero. To sum up, the FCL and fault ride-through strategy proposed in this paper can realize both the fault ride-through of large currents and fault ride-through of small currents.

Conclusion
To deal with the fault ride-through problem of LVDC distribution network, based on the collected electrical quantities and control methods, we designed an FCL which is connected in series at the ends of the distribution lines. The conclusions are drawn as follows: 1. The feasibility of the FCL connecting to the network was discussed in detail and the structure of the d FCL was given in detail; the principle of parameter design was discussed and presented. 2. To verify our proposed FCL, under the most serious conditions, a model was built and simulated. From the simulated results, we can see that FCL can reduce the fault current impact and limit fault current amplitude, which earns some time for fault detection, control, protection and ride-through action. After fault disappears, it will switch back to the current-limiting state and restore the normal operation. If the fault still exists, through cooperation with DCCB, the differential protection should be used as the backup protection of FCL. The identification of short-circuit fault of small current can be realized by monitoring the DC power at the load end.
In the future, we will focus on the design of the DC FCL of multiterminal LVDC system connected by distributed power supply and on developing an intelligent FCL with multiple functions such as detection, operation and protection for practical application.  At 1s, K1 was disconnected while K2 was closed, and the current-limiter began to work. During 1 s to 2 s, the system operated normally. Due to the current-limiting inductance, the DC bus voltage and DC line current entered steady state after a short fluctuation. U dc1 in Figure 9a reached 500 V while I dc1 in Figure 9b reached 60 A; I Lsh1 in Figure 9c gradually increased from 0 to the rated current of 60 A. A short circuit fault occurred in the distribution line at 2 s. The fault process of 2 s-2.05 s is shown in Figure 10. As analyzed in Section 3.2, the current-limiting inductance extends the discharge time of the capacitor and the rise speed of the impact current. In Figure 10a, U dc1 of the capacitor C 1 did not decrease to 0 V rapidly. At 2.017 s, the line current I dc1 rose to about 170 A and reached the threshold of amplitude-limiting, that is, I dc ≥ 2I N . At this moment, K2 was disconnected and K3 was closed. The amplitude-limiting circuit was connected to prevent the capacitance from discharging rapidly. In this case, U dc1 in Figure 10a only decreased to 200 V and restored to normal operation at 500 V in 5ms. The DC side line current I dc1 in Figure 10b restored to normal operation at 50 A in 4 ms. The K4 was closed while K3 was closed, and the energy leakage circuit was connected, which caused the current flowing through the current-limiting inductance I Lsh1 to decrease from 170 A to 0 A in 0.34 s in Figure 10c and thus the current-limiting capacity was restored.

Bipolar Short Circuit Fault of Small Current Type
In the small current fault simulation, the resistance for a bipolar short circuit fault is 16 Ω, 2 times larger than the load resistance (8.3 Ω). The DC simulation curves on the VSC side and boost converter side are demonstrated in Figures 11 and 12, respectively.

1.
Simulation curve on VSC side As shown in Figure 11, the system started at 0s-1s and operated normally during 1 s-2 s. At this time, K1, K3 and K4 were disconnected while K2 was closed, and the current-limiting circuit of FCL1 was connected. At 2 s, a small current short circuit fault occurred in the distribution line. Due to the variation range of the line current which did not meet the large current condition, K2 was still closed Energies 2020, 13, 1753 15 of 17 and the FCL1 worked in the current-limiting state. In Figure 8, the capacitance C 1 discharged through the short-circuit resistance, resulting in a decrease of U dc1 to 400V in Figure 11a. The short-circuit resistance caused the load equivalent resistance to decrease, resulting in an increase of DC side line current I dc1 to 73 A in Figure 11b; moreover, the close of K2 made the line current the same as the one in the current-limiting inductance in FCL1, so the current I Lsh1 in Figure 11c which flowed through the current-limiting inductance was 73 A, too. At this point, for the circuit breaker trip, conditions were not reached and the fault current remained stable.

2.
Simulation Curve on Boost Converter Side As shown in Figure 12, the system started at 0s-1s and operated normally during 1 s-2 s. At this time, K1, K3 and K4 were disconnected while K2 was closed, and the current-limiting circuit of FCL2 was connected. At 2 s, a small current short circuit fault occurred in the distribution line. Due to the variation range of the line current which did not meet the large current condition, K2 was still closed and the FCL2 worked in the current-limiting state. In Figure 8, the capacitance C b discharged through the short-circuit resistance, resulting in a decrease of U dc2 in Figure 12a to 400 V. The short-circuit resistance shunt caused a decrease of booster converter side line current I dc2 from −60 A to −50A in Figure 12b (follow the specified direction in Figure 9).
In Figure 12c, the load current dropped from the rated value of 60 A to 48 A. Moreover, the close of K2 made the line current the same as the one in the current-limiting inductance in FCL2, so the current I Lsh2 in Figure 12d which flowed through the current-limiting inductance became 50 A. At this point, because the circuit breaker trip conditions were not reached, the fault current remained stable, which conformed to the condition that I load < 10%I N .
To sum up, the FCL and fault ride-through strategy proposed in this paper can realize both the fault ride-through of large currents and fault ride-through of small currents.

Conclusions
To deal with the fault ride-through problem of LVDC distribution network, based on the collected electrical quantities and control methods, we designed an FCL which is connected in series at the ends of the distribution lines. The conclusions are drawn as follows: 1.
The feasibility of the FCL connecting to the network was discussed in detail and the structure of the d FCL was given in detail; the principle of parameter design was discussed and presented.

2.
To verify our proposed FCL, under the most serious conditions, a model was built and simulated. From the simulated results, we can see that FCL can reduce the fault current impact and limit fault current amplitude, which earns some time for fault detection, control, protection and ride-through action. After fault disappears, it will switch back to the current-limiting state and restore the normal operation. If the fault still exists, through cooperation with DCCB, the differential protection should be used as the backup protection of FCL. The identification of short-circuit fault of small current can be realized by monitoring the DC power at the load end.
In the future, we will focus on the design of the DC FCL of multiterminal LVDC system connected by distributed power supply and on developing an intelligent FCL with multiple functions such as detection, operation and protection for practical application.