IEC 61850-Based Centralized Protection against Single Line-To-Ground Faults in Ungrounded Distribution Systems

: We developed an International Electrotechnical Commission (IEC) 61850-based centralized protection scheme to prevent single line-to-ground (SLG) faults in the feeders and busbars of ungrounded distribution systems. Each feeder intelligent electronic device (IED) measures its zero-sequence current and voltage signals and periodically transmits zero-sequence phasors to a central IED via a Generic Oriented Object Substation Event message. Using the zero-sequence phasors, the central IED detects SLG faults in feeders and busbars. To achieve centralized protection, angle differences between the zero-sequence currents and voltage phasors are exploited, and their calculation compensates for data desynchronization. The feeder IEDs were implemented using the MMS-EASE Lite library, while the transmitted zero-sequence phasors were calculated based on fault signals simulated by Power System Computer Aided Design / Electro-Magnetic Transient Design and Control (PSCAD/EMTDC). The central IED determined if the SLG fault was in a feeder or busbar by aggregating and analyzing the zero-sequence phasors received from the feeder IEDs. The results conﬁrmed the validity and efﬁciency of our centralized protection scheme. resistances of 5 k Ω . These results demonstrated that the centralized protection scheme is useful for detecting the SLG faults in ungrounded systems when the fault resistance is lower than a pre-deﬁned value.


SLG Faults in an Ungrounded Distribution System
The ungrounded distribution system under study operates radially; four feeders are supplied by a Y/∆ step-down transformer ( Figure 1). A ground potential transformer (GPT) is installed at the 22 kV busbar to measure the zero-sequence voltage. The wyeconnected primary windings of the GPT are solidly grounded by a current-limiting resistor (CLR) connected across the broken delta of the tertiary windings. The CLR provides very high-resistance grounding for ungrounded systems (of a few tens of kilo-ohms); therefore, SLG faults produce zero-sequence currents that are extremely small compared to the phase currents. Thus, it is nearly impossible to calculate zero-sequence currents from phase currents; zero-sequence current transformers (ZCTs) are commonly used to accurately measure zero-sequence currents. Table 1 summarizes the system configuration. Two different SLG faults were considered in this paper: a fault in a feeder and a fault in the 22 kV busbar. It is noted that the central protection proposed in this paper can be easily applied to other ungrounded distribution systems. The system shown in Figure 1 is just an example to explain the characteristics of SLG faults in ungrounded distribution systems.

SLG Fault in a Feeder
The angle difference between the zero-sequence voltage and current phasors is generally used to identify the faulty feeder in an ungrounded system. The zero-sequence angle difference of the mth feeder is given by: where 0 m ∠ I and 0 m ∠ V are the angles of the zero-sequence current and voltage phasors at the mth feeder, respectively. Note that the angle of the zero-sequence voltage phasor should be identical at all feeders, because the GPT supplies the same zero-sequence voltage signal to each feeder IED.

SLG Fault in a Feeder
The angle difference between the zero-sequence voltage and current phasors is generally used to identify the faulty feeder in an ungrounded system. The zero-sequence angle difference of the mth feeder is given by: where ∠ m I 0 and ∠ m V 0 are the angles of the zero-sequence current and voltage phasors at the mth feeder, respectively. Note that the angle of the zero-sequence voltage phasor should be identical at all feeders, because the GPT supplies the same zero-sequence voltage signal to each feeder IED. Figure 2 shows the sequence networks, and their interconnections, in a case when an SLG fault developed in the fourth feeder. The notation is as follows: m Z L0 : Zero-sequence line impedance of the mth feeder; m Z C0 : Zero-sequence line-to-ground capacitive impedance of the mth feeder;  Figure 2 shows the sequence networks, and their interconnections, in a case when an SLG fault developed in the fourth feeder. The notation is as follows:   Where 0 denotes the zero sequence and it is replaced with 1 and 2 to represent the positive and negative sequences, respectively. The zero-sequence impedance of an ungrounded system is very large; therefore, the positive-and negative-sequence impedances can be ignored when considering the SLG fault in a feeder. This simplifies the sequence network, and thus the phasor diagram (Figure 3). Circuit analysis of the simplified sequence network yields the zero-sequence current at the fault point: where the equivalent capacitive impedance of all feeders is Where 0 denotes the zero sequence and it is replaced with 1 and 2 to represent the positive and negative sequences, respectively.
The zero-sequence impedance of an ungrounded system is very large; therefore, the positive-and negative-sequence impedances can be ignored when considering the SLG fault in a feeder. This simplifies the sequence network, and thus the phasor diagram ( Figure 3). Circuit analysis of the simplified sequence network yields the zero-sequence current at the fault point: where the equivalent capacitive impedance of all feeders is 1∼4 Z C0 = 1 Z C0 // 2 Z C0 // 3 Z C0 // 4 Z C0 .
In line with the current divider rule, the zero-sequence current at the relay point of the fourth feeder is: where the equivalent capacitive impedance of the non-faulty feeders is 1∼3 Z C0 = 1 Z C0 // 2 Z C0 // 3 Z C0 . As shown in Figure 3b, if a forward SLG fault develops in front of the relay point, the zero-sequence angle difference becomes between 0 • and 90 • depending on the size of the CLR. If a backward SLG fault develops behind the relay point, the theoretical zero-sequence angle difference becomes −90 • .  In line with the current divider rule, the zero-sequence current at the relay point of the fourth feeder is: where the equivalent capacitive impedance of the non-faulty feeders is As shown in Figure 3b, if a forward SLG fault develops in front of the relay point, the zero-sequence angle difference becomes between 0° and 90° depending on the size of the CLR. If a backward SLG fault develops behind the relay point, the theoretical zero-sequence angle difference becomes −90°.

SLG Fault in a 22 kV Busbar
For busbars in transmission systems, differential protection is commonly used. However, in distribution systems, overcurrent protection is considered to be adequate and thus preferred. For an ungrounded distribution system, however, neither differential nor overcurrent protection can be used to prevent busbar SLG faults because the fault current is small. A dedicated busbar protection is nonetheless required; therefore, we focused on IEC 61850-based protection. Figure 4 shows the sequence network and interconnections when an SLG fault develops in a 22 kV busbar.

SLG Fault in a 22 kV Busbar
For busbars in transmission systems, differential protection is commonly used. However, in distribution systems, overcurrent protection is considered to be adequate and thus preferred. For an ungrounded distribution system, however, neither differential nor overcurrent protection can be used to prevent busbar SLG faults because the fault current is small. A dedicated busbar protection is nonetheless required; therefore, we focused on IEC 61850-based protection. Figure 4 shows the sequence network and interconnections when an SLG fault develops in a 22 kV busbar.  Similar to when an SLG faults develops in a feeder, the positive-and negative-sequence impedances can be ignored when considering an SLG fault in a busbar. This simplifies the sequence network and phasor diagram ( Figure 5). Similar to when an SLG faults develops in a feeder, the positive-and negative-sequence impedances can be ignored when considering an SLG fault in a busbar. This simplifies the sequence network and phasor diagram ( Figure 5). Similar to when an SLG faults develops in a feeder, the positive-and negative-sequence impedances can be ignored when considering an SLG fault in a busbar. This simplifies the sequence network and phasor diagram ( Figure 5). Although the topology in Figure 5a differs from that in Figure 3a, the Thevenin equivalent circuits at the fault points, and thus also the zero-sequence currents, are identical. Although the topology in Figure 5a differs from that in Figure 3a, the Thevenin equivalent circuits at the fault points, and thus also the zero-sequence currents, are identical. However, the zero-sequence current at the relay point is not the same as that of Equation (3), although it can be calculated using the current divider rule. For example, the zero-sequence current at the relay point of the fourth feeder is: As shown in Figure 5b, if an SLG fault develops in a busbar, the zero-sequence angle differences become −90 • at the relay points of all feeders; this enables identification of a busbar SLG fault. Figure 6 shows the configuration of the IEC 61850-based centralized protection scheme for ungrounded distribution systems. Each feeder IED measures the zero-sequence current and voltage signals at the relay point. After calculating the zero-sequence phasors, the IED transmits them to a central IED via a GOOSE message. The central IED identifies SLG faults based on the zero-sequence phasors. For accurate analysis, data desynchronization among feeder IEDs must be compensated for. Data desynchronization is caused by discrepancies in measurement times (i.e., when data sources update at different times) and time desynchronization (i.e., timestamp errors) [28]. Angle differences between the zero-sequence voltage phasors of feeder IEDs are used to estimate data desynchronization. This is possible because the GPT supplies the same zero-sequence voltage signal to each feeder IED (see Section 2.1). Therefore, centralized protection is based on the zero-sequence angle difference, instead of the angle itself of the zero-sequence voltage and current phasors. Data desynchronization is compensated for when the centralized protection calculates the zero-sequence angle difference. ferent times) and time desynchronization (i.e., timestamp errors) [28]. Angle differences between the zero-sequence voltage phasors of feeder IEDs are used to estimate data desynchronization. This is possible because the GPT supplies the same zero-sequence voltage signal to each feeder IED (see Section 2.1). Therefore, centralized protection is based on the zero-sequence angle difference, instead of the angle itself of the zero-sequence voltage and current phasors. Data desynchronization is compensated for when the centralized protection calculates the zero-sequence angle difference.

Operation of the Central IED When an SLG Fault Occurs in a Feeder
If an SLG fault develops in a feeder, the central IED operates only when the following conditions are all satisfied for the mth feeder:

Operation of the Central IED When an SLG Fault Occurs in a Feeder
If an SLG fault develops in a feeder, the central IED operates only when the following conditions are all satisfied for the mth feeder: The zero-sequence voltage should be larger than the threshold value V T 0 . Note that the zero-sequence voltage at each feeder should have the same phasor regardless of SLG fault location, because the Thevenin equivalent circuit at the CLR is identical. Below, we deal with an SLG fault in the fourth feeder; SLG faults in other feeders are handled similarly. As shown in Figure 3a, the zero-sequence voltage at the relay point is: when the zero-sequence current at the CLR is: Substitution of (2) into (3) yields: Assuming that the fault resistance ranges up to 5 kΩ, it is possible to determine V T 0 and m I TF 0 using (6) and (8), respectively. The threshold value for the zero-sequence angle difference m θ TF 0 is also determined by (6) and (8). Note that the zero-sequence angle difference is independent of the fault resistance, as indicated by the relationship between 4 V R 0 and 4 I R 0 own in Figure 3b. Although the sum of the zero-sequence currents at the relay points of non-faulty feeders is the same as the capacitive zero-sequence current at the relay point of the faulty feeder, the polarities are opposite. This is useful for determining whether the ZCT polarities are correct. For this purpose, an operator P is defined as follows: where A = A e jθ A and B = B e jθ B . Note that P(A|B) yields the component of A orthogonal to B. Therefore, the following condition should be satisfied if there is an SLG fault in the fourth feeder:

Operation of the Central IED When an SLG Fault Occurs in a Busbar
If the SLG fault is in a busbar, the central IED operates only when the following conditions are satisfied for every feeder: The zero-sequence voltage at each feeder is identical for SLG faults with the same fault resistance; therefore, the V T 0 of (5-1) is used in . As shown in Figure 5a, the zero-sequence voltage at the relay point is: where the zero-sequence current at the CLR is: The zero-sequence current at the relay point of the mth feeder is: Similar to the case of an SLG fault in a feeder, assuming that the fault resistance ranges up to 5 kΩ, it is possible to determine V T 0 and m I TB 0 using (12) and (14), respectively. In addition, the threshold value for the zero-sequence angle difference θ TB 0 can be easily found from (14); this becomes the angle of − m Z C0 .
As shown in Figure 7, m θ TF 0 is between 0 • and 90 • , depending on the size of the CLR, while θ TB 0 is −90 • for all feeders. Figure 8 shows the flowchart for centralized protection against SLG faults in a feeder and busbar. Each feeder IED periodically transmits its zero-sequence current and voltage phasors to the central IED via GOOSE messages. The central IED uses their magnitudes and angle differences to identify an SLG fault in a feeder or busbar. Data desynchronization is compensated for when the zero-sequence angle difference is calculated. ranges up to 5 kΩ, it is possible to determine T 0 V and TB 0 I m using (12) and (14), respectively. In addition, the threshold value for the zero-sequence angle difference TB 0 θ can be easily found from (14); this becomes the angle of  Figure 8 shows the flowchart for centralized protection against SLG faults in a feeder and busbar. Each feeder IED periodically transmits its zero-sequence current and voltage phasors to the central IED via GOOSE messages. The central IED uses their magnitudes and angle differences to identify an SLG fault in a feeder or busbar. Data desynchronization is compensated for when the zero-sequence angle difference is calculated.

Algorithm for Centralized Protection
The presence of an SLG fault is checked for each feeder sequentially. If the operating conditions given in Equation (5) are satisfied for the mth feeder, the central IED determines that an SLG fault develops in the mth feeder and then sends GOOSE messages to trip the circuit breaker of the mth feeder. As shown in Figure 8, the presence of an SLG fault in a busbar is checked independently of this process. If the operating conditions given in (11) are satisfied for every feeder, the central IED determines that an SLG fault develops in the busbar and then sends GOOSE messages to trip the circuit breaker of a main transformer.

Test Envirionment
To determine the efficiency of the centralized protection scheme, the ungrounded distribution system shown in Figure 1 was modeled using PSCAD/EMTDC. As shown in Figure 9, the system had four feeders, each supplying a 5 MVA load (pf 0.98, delta connection). The distribution line was 20 km in total length. The presence of an SLG fault is checked for each feeder sequentially. If the operating conditions given in Equation (5) are satisfied for the mth feeder, the central IED determines that an SLG fault develops in the mth feeder and then sends GOOSE messages to trip the circuit breaker of the mth feeder. As shown in Figure 8, the presence of an SLG fault in a busbar is checked independently of this process. If the operating conditions given in (11) are satisfied for every feeder, the central IED determines that an SLG fault develops in the busbar and then sends GOOSE messages to trip the circuit breaker of a main transformer.

Test Environment
To determine the efficiency of the centralized protection scheme, the ungrounded distribution system shown in Figure 1 was modeled using PSCAD/EMTDC. As shown in Figure 9, the system had four feeders, each supplying a 5 MVA load (pf 0.98, delta connection). The distribution line was 20 km in total length.

Test Envirionment
To determine the efficiency of the centralized protection scheme, the ungrounded distribution system shown in Figure 1 was modeled using PSCAD/EMTDC. As shown in Figure 9, the system had four feeders, each supplying a 5 MVA load (pf 0.98, delta connection). The distribution line was 20 km in total length. IEC 61850-based feeder IEDs were implemented using MMS-Ease Lite Library 6.2000.2v, a commercial development tool from SISCO. As shown in Figure 10, the simulated fault signals were imported by the feeder IEDs in advance, and each feeder IED then IEC 61850-based feeder IEDs were implemented using MMS-Ease Lite Library 6.2000.2v, a commercial development tool from SISCO. As shown in Figure 10, the simulated fault signals were imported by the feeder IEDs in advance, and each feeder IED then transmitted zero-sequence phasors to the central IED every 100 ms via GOOSE messages. IEC 61850 packets, including the GOOSE messages, were monitored using IEDScout, a commercial testing tool from OMICRON. This environment was able to test the centralized protection because it is not sensitive to data desynchronization. transmitted zero-sequence phasors to the central IED every 100 ms via GOOSE messages. IEC 61850 packets, including the GOOSE messages, were monitored using IEDScout, a commercial testing tool from OMICRON. This environment was able to test the centralized protection because it is not sensitive to data desynchronization.

SLG Faults in the Fourth Feeder
At a fault resistance of 5 kΩ, the threshold values T 0 V and 4 TF 0 I were set to 155.5 V and 1.555 mA using (6) and (8), respectively. The threshold value for the zero-sequence angle difference, 4 TF 0 θ , was set to 62.14° using both (6) and (8). In the case studies, six SLG faults (resistances of 0, 1, 2, 3, 4, and 5 kΩ) were considered. Figure 11 shows the test results for an SLG fault in the fourth feeder with a fault resistance of 0 kΩ.

SLG Faults in the Fourth Feeder
At a fault resistance of 5 kΩ, the threshold values V T 0 and m I TF 0 were set to 155.5 V and 1.555 mA using (6) and (8), respectively. The threshold value for the zero-sequence angle difference, m θ TF 0 , was set to 62.14 • using both (6) and (8). In the case studies, six SLG faults (resistances of 0, 1, 2, 3, 4, and 5 kΩ) were considered. Figure 11 shows the test results for an SLG fault in the fourth feeder with a fault resistance of 0 kΩ.

SLG Faults in the Fourth Feeder
At a fault resistance of 5 kΩ, the threshold values T 0 V and 4 TF 0 I were set to 155.5 V and 1.555 mA using (6) and (8), respectively. The threshold value for the zero-sequence angle difference, 4 TF 0 θ , was set to 62.14° using both (6) and (8). In the case studies, six SLG faults (resistances of 0, 1, 2, 3, 4, and 5 kΩ) were considered. Figure 11 shows the test results for an SLG fault in the fourth feeder with a fault resistance of 0 kΩ.  All of the conditions in (5) were satisfied; the centralized protection scheme detected the fault and the central IED transmitted GOOSE messages that tripped the fourth feeder ( Figure 12). Table 2 summarizes the test results according to the fault resistance. Up to a resistance of 4 kΩ, all of the conditions in (5) were satisfied for the fourth feeder. However, for a fault resistance of 5 kΩ, the measured values were the same or slightly lower than the threshold values, and the centralized protection scheme did not detect the fault. Thus, the fault resistance must be lower than a predefined value. All of the conditions in (5) were satisfied; the centralized protection scheme detected the fault and the central IED transmitted GOOSE messages that tripped the fourth feeder ( Figure 12). Table 2 summarizes the test results according to the fault resistance. Up to a resistance of 4 kΩ, all of the conditions in (5) were satisfied for the fourth feeder. However, for a fault resistance of 5 kΩ, the measured values were the same or slightly lower than the threshold values, and the centralized protection scheme did not detect the fault. Thus, the fault resistance must be lower than a predefined value.
189.6 1.899 62.15 0.002   As mentioned in Section 3.2, θ TB 0 was set to −90 • for every feeder. At a fault resistance of 5 kΩ, the threshold value V T 0 was set to 155.5 V using (12). The threshold values 1 I TB 0 , 2 I TB 0 , 3 I TB 0 , and 4 I TB 0 were set to 0.573, 0.344, 0.458, and 0.917 mA, respectively, using Equation (14). Figure 13 shows the test results for an SLG fault in a busbar with a fault resistance of 3 kΩ. All of conditions in (11) were satisfied for every feeder, and the centralized protection scheme thus detected the SLG fault and transmitted GOOSE messages that tripped the main transformer ( Figure 14).    Table 3 summarizes the test results for busbar SLG faults with resistances 1, 3, and 5 kΩ. All of the conditions in (11) were satisfied for every feeder when the resistances were 1 and 3 kΩ. For a fault resistance of 5 kΩ, the measured values were the same or slightly lower than the threshold values, and the centralized protection scheme did not detect the fault. To reiterate, centralized protection can detect SLG busbar faults when the fault resistance is lower than a predefined value.
(c) (d)   Table 3 summarizes the test results for busbar SLG faults with resistances 1, 3, and 5 kΩ. All of the conditions in (11) were satisfied for every feeder when the resistances were 1 and 3 kΩ. For a fault resistance of 5 kΩ, the measured values were the same or slightly lower than the threshold values, and the centralized protection scheme did not detect the fault. To reiterate, centralized protection can detect SLG busbar faults when the fault resistance is lower than a predefined value.

Conclusions
We proposed a centralized protection scheme against SLG faults in ungrounded distribution systems associated with centralized environments, such as digital substations, wherein data desynchronization occurs among IEC 61850-based IEDs. The proposed scheme detects against the SLG fault in each feeder and checks whether the ZCT polarities are correct. Particularly, in order to cope with the absence of protection against the SLG fault in a busbar, the proposed scheme provides a dedicated busbar protection with the help of centralized environments. Each feeder IED measures its zero-sequence current and voltage signals and periodically transmits zero-sequence phasors to a central IED via GOOSE messages. To detect SLG faults, the scheme analyzes the angle differences between, and magnitudes of, the zero-sequence current and voltage phasors. The zerosequence voltage at each feeder IED should be identical; therefore, data desynchronization is compensated for when the zero-sequence angle difference is calculated.
The centralized protection was tested using IEC 61850-based IEDs and fault signals simulated by PSCAD/EMTDC. The ungrounded distribution system under study was modeled using PSCAD/EMTDC and then various cases were simulated considering fault location and fault resistance. IEC 61850-based IEDs were implemented using MMS-EASE Lite library 6.2000.2v, and the fault signals simulated by PSCAD/EMTDC were imported to the feeder IEDs. Each feeder IED transmitted its zero-sequence current and voltage phasors to the central IED every 100 ms via GOOSE messages. The central IED aggregated the zero-sequence phasors and transmitted GOOSE messages with trip signals when SLG faults were detected. The system detected SLG faults up to a fault resistance of 5 kΩ; six fault resistances were tested (0, 1, 2, 3, 4, and 5 kΩ). The scheme did not detect faults with resistances of 5 kΩ. These results demonstrated that the centralized protection scheme is useful for detecting the SLG faults in ungrounded systems when the fault resistance is lower than a pre-defined value.

Conflicts of Interest:
The authors declare no conflict of interest.