**3. Case Study**

An IED-to-IED GOOSE message exchange within a substation is required for fast bus tripping in the case of bus faults and the interlocking of bus-IED in the case of feeder/line faults, the protection scheme traditionally known as the reverse interlocking scheme. The IED-to-IED GOOSE messages can also be used in the case of breaker failure to trip the adjacent breaker(s). This can be done by sending a trip command message to adjacent breakers from a protection IED with a built-in breaker failure function or from a dedicated IED performing only the breaker failure function. The transfer trip may also be required between two substations. The transfer time requirement of 10 ms was set in IEC 61850 for fast trip messages (releases and status changes) between substations (transfer time class TT5) and 3 ms for fast trips and blocking messages between IEDs within a substation (transfer time class TT6) [33,34]. The transfer time requirement also varies with respect to the specific protection function. The transfer times required for various protection functions are given Table 1.

Although very strict time requirements have been demanded in IEC 61850 for type 1A fast trip messages, in this study, an average transfer time of 10–20 ms was considered for the one-way GOOSE message to cover the limitations (the limited failures of LAN within a maximum allowed transfer time of 18 ms), safety margins (errors in the time-stamp accuracy) and redundant GOOSE messages for communication between substations, as explained in [35]. IEC 61850-90-1 [36] recommends a maximum time delay of 5 to 10 ms on the communication channel depending on the voltage level [37]. However, in order to meet the requirements of security, reliability and dependability according to the IEC 60834 standard, the communication system should meet the 3-ms transfer time requirement for 99.9999 percent of the time and should have a delay no longer than 18 ms for the remainder of the time [38]. A fixed transfer time of 20 ms is thus used for both IED-to-IED communication within a substation and IED-to-IED communication between di fferent substations in this study to cover even the worst time delay of 18 ms for the type 1A messages. In practical cases, generally, the transfer time for communication between IEDs at di fferent substations is longer than the transfer time within the same substation. Measuring the one-way communication latency by a round-trip time between two remote substations was discussed in [39]. The selected one-way transfer time of 20 ms for producing results corresponded to the fast messages of type 1B (the ideal case) with performance class P2/P3 (transfer time class TT4) [33,35], and it covers the worst-case delay of 18 ms of type 1A fast messages according to the IEC 60834 mentioned above. The considered transfer time was also within the range of practical observed time delays in the light-weight implementation of the IEC 61850 standard-based GOOSE messages done in [40]. Although GOOSE messages apply the heartbeat messages and an IED will issue a so-called burst of GOOSE messages right after the detection of the fault, for the final trip decision, the IED necessarily needs to know the updated status at the downstream IED after the fault to ensure proper coordination. The selected 20-ms communication delay between IEDs ensures that even the type 1A GOOSE messages with the maximum delay of 18 ms are also subscribed by IED for the trip decision. Another reason of selecting a 20-ms delay between IEDs is the potential requirement to use GOOSE messages to transfer analog data between IEDs for trip decision criterion like vectors of measured values (RMS values), which need to be transmitted only once per cycle of 50-Hz frequency. It means a new analog measurement data is required to be transmitted in just every 20 ms [37]. In the earlier publication [33], it was mentioned that there were two types of GSE (generic substation event) messages: GSSE (generic substation state event) message and GOOSE message. The GSSE message is the old binary-only message type. All the modern systems use the more flexible GOOSE message, which transfers both binary and analog data. Both GSSE and GOOSE can coexist but are not compatible with each other. The proposed protection algorithm in this paper not only uses the binary data but, also, uses the analog data (RMS magnitude of currents) for the trip decision.



In this paper, the conventional GOOSE (tunneled-GOOSE) messages in layer 2 (horizontal communication) with an Ethernet link are considered, because the short distances (a few km) between substations are considered. However, for the longer distances between substations where an Ethernet link is not possible, the routable-GOOSE (R-GOOSE) messages in layer 3 (vertical communication) for wide-area or system protection applications could be used, especially with wireless communication technologies using synchrophasors in compliance with IEC TR 61850-90-5. Some applications of R-GOOSE were reported in [41]. The normal predefined fixed GOOSE message transfer delay of 40 ms (2 cycles of 50-Hz power system) was assumed previously for the adaptive protection of the microgrid

using communication over a WiMAX network in [42], and the actual latency observed was within 35 ms with no data packet loss. However, due to packet loss and, consequently, retransmissions, the overall delay could further increase, thus limiting the application of WiMAX (R-GOOSE) to comparatively slower control and protection actions like status updates and protection settings during scheduled maintenance and load management. The adaptive protection methodology presented in this paper is concerned with primary and backup protections of the microgrid during faults in predefined operational modes: grid-connected or islanded modes. With this regard, a communication-dependent coordination methodology is proposed in Section 4 for the cases when the fault happens between two defined IEDs in grid-connected and islanded modes. The proposed methodology is very generic in nature and can be implemented in any protection IED.

The schematic diagram of a radial AC microgrid under study is shown in Figure 2. The considered AC microgrid consists of one MV bus of 20 kV (Substation Bus-2) and one LV bus of 0.4 kV (Substation Bus-3). A load of 2 MW at Substation Bus-2 is supplied by a wind turbine generator (WTG) of 2-MVA capacity, whereas a load of 0.4 MW at Substation Bus-3 is supplied by a photovoltaic (PV) generator of 0.4 MVA. The MV bus (Substation Bus-2) of the AC microgrid is connected with the LV bus (Substation Bus-3) of the AC microgrid through a 1-km-long, 20-kV cable line and 0.5-MVA, 20/0.4-kV transformer. The AC microgrid is connected with the main grid through a 2-km-long, 20-kV overhead line and an intermediate 20-kV Substation Bus-1. The WTG is connected to Substation Bus-2 through a 0.2-km-long 20-kV cable and a 2-MVA, 0.69/20-kV transformer (inside the WTG model). A 2-km overhead line between Substation Bus-1 and Substation Bus-2 is protected by two circuit breakers, CB1 and CB2, with the related protection IEDs. The protection IED1 is considered to be a nonadaptive IED due to its direct connection with the main grid, whereas the protection IED2 is considered as an adaptive IED. In the grid-connected mode, IED2 operates with settings that enable the tripping of CB2 in the case of bus fault F8 at Substation Bus-2 and facilitates the transfer trips of CB2 after receiving the CB1 open-state signal in the case of short-circuit fault F1. However, if IED2 fails to receive a CB1 open-state signal in the case of short-circuit fault F1 after the opening of CB1 and the AC microgrid already changed to islanded mode with a trip-block signal to all IEDs, this will be the failure of the transfer trip. In this case IED2 can provide a backup operation by opening CB2 with the fault current still coming from the DGs within the AC microgrid. This can be performed by IED2 either with the islanded mode settings or using the current magnitude comparison and the direct transfer trip failure logic, as explained later in Section 4. The IED2 may take quite some time to change its active group settings from the grid-connected mode settings to the islanded mode settings, and this will require DGs to remain online for additional time beyond the standard LVRT curve until the IED2 settings are changed and the tripping of CB2 is executed. However, the backup operation of CB2 by the direct transfer trip failure logic implemented at IED2 could be performed within the standard LVRT curve. In the islanded mode, the IED2 settings are adapted so that the fault F1 is detected when the CB1 is open. The 1-km cable line between Substation Bus-2 and Substation Bus-3 is protected by CB6 and CB7 with the related protection IEDs. The protection IED6 and IED7 are also considered to be adaptive.

The adaptive IED6 primarily protects both the cable line and the 20/0.4-kV transformer from short-circuit fault F2 during both the grid-connected and islanded mode of operations. In the islanded mode, after sensing the fault current at its location, the adaptive IED6 trips CB6 and transfer trips circuit breaker CB7. Additionally, IED6 and IED7 can compare the post-fault magnitude of currents at their locations with a 1.2-p.u. threshold and determine the location and direction of the fault between IED6 and IED7, as explained in the coming section. The adaptive IED7 can also provide backup protection in case of transfer trip failure (like the adaptive IED2 does, as explained earlier) in the case of the fault F2 in the grid-connected mode, in addition to the normal protection against bus fault F4 at Substation Bus-3 in both the grid-connected and islanded modes by direct tripping CB7 and transfer tripping CB9. The IED7 can only provide an "adaptive trip" to CB7 for the transfer trip failure from IED6 in the case of short-circuit fault F2 in the grid-connected mode if su fficient fault current contribution from PV is available beyond the standard LVRT characteristic curve. This is because the IED7 may need quite some time to change its active group settings from the grid-connected mode settings to the islanded mode settings, and PV must remain online until IED7 settings are changed and CB7 tripping is executed. For this purpose, a new LVRT curve proposed later in this paper can be used. The 0.2-km cable connecting WTG with Substation Bus-2 is protected by CB3 with an adaptive IED3. Both MV and LV loads are also provided with adaptive IEDs (IED5 and IED8), which trip CB5 and CB8 adaptively in the case of load-side short-circuit faults F3 and F9 in both grid-connected and islanded modes of the AC microgrid.

**Figure 2.** The radial MV/LV AC microgrid model for adaptive protection study.

The adaptive IEDs with two preplanned setting groups for AC microgrid lines are provided with only under-voltage (UV) local backup protection (Figure 3a) and adaptive IEDs with two preplanned setting groups for loads with both under and over-voltage (UV/OV) backup protection (Figure 3b). The DG protection IEDs (IED4 and IED9) are also considered to be adaptive in order to differentiate between the grid-connected and islanded mode operations. Moreover, DG protection IEDs should not trip instantaneously in the case of all external faults and allow DGs to provide fault current contributions according to predefined standard LVRT characteristics. A multifunctional adaptive IED

for the protection of converter-based DGs is shown in Figure 4, which consists of adaptive OC and anti-islanding protection functions. In practice, DGs may be provided with unit protection and IEDs with several fault protection and anti-islanding protection functions. In this study, the anti-islanding protection functions (passive/active methods) of DG protection IEDs are assumed normally "disabled" if the communication link is continuous and enabled quickly when the communication link is lost. Thus, communication-based loss-of-mains detection with no nondetection zone can be used as a primary means of anti-islanding protection and passive/active methods as backup in the case of communication link failure. All the sensitive protections within the islanded AC microgrid need to be disabled/interlocked during the starting of DGs, motor loads and during the transient period when changing from the grid-connected to islanded mode and vice versa.

**Figure 3.** Adaptive definite time overcurrent (DTOC) relays with two preplanned setting groups: (**a**) for lines with local undervoltage (UV) backup protection and (**b**) for loads with local voltage protection (UV/over-voltage (OV)).

**Figure 4.** A multifunctional adaptive intelligent electronic device (IED) for the protection of converter-based distributed generators (DGs).

## **4. Adaptive Protection Methods and Results**

Although several faults may happen in the presented AC microgrid, only adaptive protection methods and results of three-phase ungrounded short circuit faults with 0.01-Ohm fault resistance at locations F1 and F2 are presented. Moreover, for the sake of simplicity, it is assumed that three-phase fault F2 occurs only in the islanded mode. Nevertheless, the adaptive protection method for fault F2 also considers the protection option in the case of F2 in the grid-connected mode, as explained in the following text. Figure 5 shows the flowchart of communication-based nonadaptive IED1 for protection during fault F1. IED1 provides primary protection for fault F1 and the backup protection with definite

time delay for all other downstream faults using OC relay and UV protection works as backup of the OC relay. The IED1 normally uses the redundant communication link to ge<sup>t</sup> information about downstream faults and use this information for trip decisions. If the fault is downstream, it waits for the CB2 to trip first. On receiving a CB2 failure signal, it trips CB1 and sends a CB1 status "open" GOOSE message (XCBR signal) to all IEDs to change their settings to the islanded mode. Even if CB1 fails, it can transfer the trip incoming breaker CB0 of substation-1 to initiate the islanding. If no communication link is available, IED1 will simply trip CB1 using definite time delays depending on the magnitude of the current. Figure 6 shows the steps for the clearance of fault F1 using GOOSE messages with different transmission delays. In both cases, at step 7, IED2 can be used in an adaptive manner for tripping CB2 to clear F1 completely, if not directly tripped by the CB1 status transfer trip, as mentioned in step 7 of Figure 6. If CB2 is not tripped with the CB1 status transfer trip or the adaptive trip by IED2, then fault F1 will not clear due to fault energization by DGs in the AC microgrid, and DGs will trip after LVRT time is elapsed.

**Figure 5.** Flowchart for communication-based nonadaptive IED1 providing primary protection for fault F1 and remote backup for all downstream faults in the grid-connected mode. UV: Undervoltage protection.

**Figure 6.** Fault F1 clearance time with 10-ms and 20-ms GOOSE message transfer delays (CB2 can trip by transfer trip GOOSE from IED1 or by adaptive IED2 using islanded mode settings).

Figure 7 shows the flowchart for the clearance of fault F2 in both grid-connected and islanded modes by adaptive IED6 by tripping CB6 and sending trip signal XCBR "open" to IED7 for tripping CB7. With CB6 and CB7 open, two separate islands were created within the islanded AC microgrid, one supplied by only PV (LV microgrid) and other supplied by only WTG (MV microgrid). If fault F2 occurs in the grid-connected mode, then only the LV microgrid will be isolated, and the MV microgrid will operate in the grid-connected mode. IED7 will also need the current flow direction in the case of fault F2, since this fault will be energized by both PV and WTG in islanded mode, which will avoid nuisance tripping by IED6 in the case of bus-3 fault or fault F3 at the LV load. IED7 can easily know if the fault is upstream or downstream of its location after receiving "YES fault GOOSE" from IED6 by simply calculating the RMS magnitude of the current at its location. If the magnitude of current at IED7 is ≤1.2 p.u. of the normal set current, the fault is considered to be upstream of IED7, since the fault contribution at IED7 will come from downstream PV only. In this case, IED7 will send "NO fault GOOSE" to IED6. If the magnitude of the current at IED7 is >1.2 p.u. of the normal set current, the fault is considered to be downstream of IED7. In this case, IED7 will send "YES fault GOOSE" to IED6, and IED6 will wait until the next GOOSE from IED7. The red and green colors in Figure 7 differentiate between the grid-connected and islanded mode features. On the failure of CB6, IED6 will trip CB2, CB3 and CB7 to clear F2 in the grid connected mode, whereas CB7 and CB3 will be tripped in the islanded mode to clear fault F2 completely. Hence, CB6 failure during fault F2 in both grid-connected and islanded modes will cause complete power interruptions to MV microgrid loads. Figure 8 shows the steps for the clearance of fault F2 using GOOSE messages with different transmission delays. It should be noted that, in steps 5 and 6 and 7 and 8 of Figures 6 and 8, the time delay for circuit breaker operation is considered 20 ms, which is one cycle of 50-Hz supply. This means high-speed AC circuit breakers operating in one cycle [43] will be required for the implementation of the proposed adaptive OC protection scheme.

**Figure 7.** Flowchart for communication-based adaptive IED6 providing primary protection for fault F2 and remote backup for all downstream faults in both grid-connected and islanded modes.

**Figure 8.** Fault F2 clearance time with 10-ms and 20-ms GOOSE message transfer delays (CB7 can trip by transfer trip GOOSE from IED6 or by adaptive IED7 using islanded mode settings).

Table 2 shows the normal flow of currents measured at the IED1, IED2, IED6 and IED7 locations during four different DG scenarios in the grid-connected mode. The maximum currents used for the adaptive DTOC settings are also indicated in Table 2. The fault current magnitudes at the concerned IEDs during the short-circuit fault F1 in the grid-connected mode and the short-circuit fault F2 in the islanded mode are shown in Table 3. Table 4 shows the DTOC settings and time grading of the IEDs 1, 2, 6 and 7 in the grid-connected and islanded modes. In this study, only the high-stage setting group (I>>) of Table 4 was used for the detection of the three-phase short circuit faults, and the isolation of the fault (tripping) is totally dependent on the GOOSE message transfer, according to Figures 6 and 8. However, in the case of complete communication failure, the time grading of Table 4 ensures selective operation. Table 5 explains the adaptivity and current magnitude comparison (CMC) requirements at IEDs of Figure 2 for three-phase faults (F1–F9) at different locations in different modes of operation. In the grid-connected mode, the higher setting group like that given in column two of Table 4 will be applied for all IEDs irrespective of the connection and disconnection status of DGs. This higher setting group is denoted by SG GM in Table 5 to represent the active settings in the grid-connected mode. For all IEDs except IED1, IED5 and IED8, a separate local current magnitude comparison function/logic is proposed for the operation in both the grid-connected and islanded modes when DGs are actively participating in load sharing with the connection status "YES". The CMC function in the IEDs will logically work in parallel with the communication-based DTOC protection and would act quickly with the "transfer trip communication failure" signal to trip the local downstream IED during the upstream fault if it does not receive the transfer trip signal (CB status "OPEN") from the upstream IED after a predefined time period (time period between step 6 and 7 in Figure 6). The CMC function will continuously receive the analog value of the current from the local MU and compare it with the current threshold of 1.2 p.u. of the max current; if the current is less than the threshold and the upstream IED also sends a fault detection GOOSE "YES", the fault is assumed to be the upstream fault. Then, if no transfer trip GOOSE is received from the upstream IED within 110 ms of the fault (event 7 in Figure 6), the IED will trip the local circuit breaker to clear the fault completely. In this way, the CMC function implemented in the communication-based adaptive protection (Figure 7) could not only detect the direction of the fault, but it could also act as a backup for direct transfer trip (DTT) communication failure. The proposed CMC function can only work for the feeder with a strong DG source on one side and a comparatively weaker DG source with a predefined fault current contribution on the other side. For example, during the fault F1 in the grid-connected mode when the connection status of both the WTG and PV is "YES", the IED2-IED4, IED6-IED7 and IED9 all will need the CMC function to determine if the fault is at the downstream or the upstream location. However, only IED2 will activate the CMC trip function after the transfer trip GOOSE message from IED1 is found undetected (not received). In the same way, during the fault F2 in the islanded mode of operation when the connection status of both the WTG and PV is "YES", the IED6-IED7 and IED9 need the CMC function to know if the fault is upstream or downstream of these relays. IED6 using the CMC function will detect the fault F2 to be downstream, while IED7 and IED9 will detect the fault to be upstream of their locations. Then, when the transfer trip GOOSE from IED6 is not received by the IED7, it will initiate the backup using CMC logic to trip CB7 in order to clear the fault F2 completely, while IED9 will continue following the LVRT curve. The topmost AND logic in Figure 9a,b presents the backup for DTT failure from IED6 during the fault F2 using the CMC function/logic. It should be noted that, in the islanded mode, when the connection status of both the WTG and PV is "YES"(the scenario of Table 5, column 3), then the WTG of 2 MW acts as a comparatively stronger source than the PV of 0.4 MW. Therefore, the grid-connected mode settings "SG GM" for IED6-IED9 will remain e ffective and unchanged in this islanded mode scenario, and only IED3-IED5 need to change to the islanded mode settings (SG IM). This way, the minimum number of IEDs will need to adapt to the islanded mode settings, and simple logics for the primary and backup protections, for example, at IED7 (Figure 9a), could be used. This will also prevent the OC function of IED7 and IED9 to start pickup for the "out of zone" fault F2 in the islanded mode scenario of Table 5, column 3.


**Table 2.** Normal RMS (Root mean square) currents at considered intelligent electronic device (IED) locations with four different distributed generator (DG) scenarios in grid-connected mode. WTG: wind turbine generators and PV: photovoltaic.

1 The maximum currents used for adaptive overcurrent (OC) settings.

**Table 3.** Fault currents (RMS) at considered IED locations with DG scenario-4 (F1 in grid-connected mode and F2 in islanded mode).


2 Before CB1 tripping (grid-connected mode). 3 After CB1 tripping (used for IED2 adaptive tripping).



4 Definite time overcurrent (DTOC) setting group for the grid-connected mode (SG GM). 5 DTOC setting group for the islanded mode (SG IM). 6 Conventional time-coordination only used in the case of communication failure.

**Table 5.** The adaptivity of SG GM or SG IM settings and CMC 8 requirements at various IEDs (Figure 2) during different faults in grid-connected and islanded mode scenarios with predefined fixed fault current contributions from DGs.


7 Current magnitude comparison: IIED ≤ 1.2 p.u. of max possible normal current at a location, then fault is upstream; IIED > 1.2 p.u. of max possible normal current at a location, then fault is downstream. 8 Fault locations (F1 to F9) for which the CMC feature is required at the corresponding IED. Note: Table 5 considers the minimum required settings of the main (primary) protection for the detection of the faults, and the final successful islanded scenarios are considered.

If, during the islanded mode (Table 5, column 3 mode), the settings of IED6, IED7 and IED9 are changed to the islanded mode settings (SG IM), then OC function of all the three IEDs will start picking up not only during the downward fault F5 but, also, during the upward faults F4 and F2. Hence, the protection logic for the detection of the "in-zone" fault direction and location will become more complex with the islanded mode settings (Figure 9b) compared with the protection logic with the grid-connected mode settings (Figure 9a). The bottom AND logic of Figure 9a represents the protection logic of IED7 with the grid-connected mode settings (SG GM) for "in-zone" fault F4, where only "NO fault" detection signals at downward IED8 and IED9 are sufficient to activate the primary and time-delayed backup protection at IED7, in addition to the local IED7 and upward IED6 "fault YES" signals and the CMC function outputs for "downward" fault. The bottom AND logic (black) of Figure 9b represents the protection logic of IED7 with the islanded mode settings (SG IM) for the "in-zone" fault F4, where not only "NO fault" detection at downward IED8 and IED9 are required, but also, the CMC function output at IED9 for "upward" and "downward" faults will also be required to activate the primary and time-delayed backup protection at IED7, respectively, in addition to the local IED7 and upward IED6 "fault YES" signals and the CMC function outputs for the "downward" fault. The protection logics presented in Figure 9 will be necessary in order to quickly detect the fault and isolate the faulty section within 150 ms, keeping the stability of the remaining system intact if the communication system performance is according to the predefined boundaries. In the case of communication failure, the normal communicationless time coordination described in Table 4 will be applied.

(**a**) 

(**b**) 

**Figure 9.** The application of the current magnitude comparison (CMC) function as GOOSE logic at communication-based IED7 to provide backup for the delayed/missing direct transfer trip (DTT) from the IED6 at a remote station bus when: (**a**) SG GM settings are used and (**b**) SG IM settings are used. (1) Backup for DTT from IED6, (2) normal DTT, (3) in-zone primary OC protection of IED7 and (4) time-delayed backup OC protection for IED8 and IED9.

#### *4.1. Results for Fault F1 in Grid-Connected Mode and Transition to Islanded Mode*

For three-phase fault F1, the adaptive OC relay logics are implemented in PSCAD (Power Systems Computer Aided Design) simulation software according to Figure 6 and settings according to Figure 3 and Table 4. The fault starts at 1.2 s and ends at 5 s; this fault duration is small, but it is assumed to be a permanent fault. The fault current magnitude at IED1 and IED2 before and after the fault F1 is shown in Figure 10. It shows that fault current magnitude is enough at IED1 location and can be detected easily with higher OC settings (SG GM). However, the fault current magnitude at IED2 is not enough to detect the fault with its higher settings, since at IED2, the fault contribution mainly comes from DGs, which are set to provide a fault current up to 1.2 p.u. of rated DG current. Hence, to remove fault F1 completely, either CB2 should be remotely direct transfer-tripped by IED1 using "CB1statusOpen" signal communication or IED2 should adapt to lower settings (SG IM) and issue an "adaptive trip" command to CB2 if "CB1statusOpen" is not received at IED2 within the predefined delay. Figure 11 shows the tripping of CB1 at about 1.29 s with a delay of 90 ms after the fault at 1.2 s and CB2 tripping at 1.334 s with a delay of about 134 ms after the fault at 1.2 s, as per method-B steps 6-8 in Figure 6; in this case, CB2 is successfully transfer-tripped. Figure 12 shows the RMS magnitude of the current of DGs before, during and after the fault F1 with the successful CB2 transfer trip method. Figure 13 shows the detection of fault F1 by lower settings of adaptive IED2 in islanded mode (CB1 already open, and transfer trip from IED1 to IED2 failed) and subsequent tripping of CB2 to clear F1 completely. The CB1 is tripped at 1.29 s according to method-B step 6 in Figure 6, and IED1 sends the circuit breaker status "Open" to all IEDs within next 20 ms. The DGs and IEDs except IED2 within the islanded AC microgrid receive the circuit breaker status "Open" from IED1 and change their mode/settings at 1.31 s; all IEDs (except IED2) remain in a "trip block" state until a "CB2 open" or "CB2 breaker failure" signal from IED2 is received. Meanwhile, an adaptive IED2 changes to lower settings due to transfer trip failure; it detects the fault, sends an "adaptive trip" command to CB2 at 1.345 s and CB2 finally trips at 1.37s. On receiving the "CB2 open" signal, all IEDs within the AC microgrid may issue "block release" to their CBs after the terminal voltage of DGs will reach a value >50% of its normal value. This will ensure no IED tripping during transition to the normal islanded mode, because DGs will continue LVRT and fault contribution until 50% terminal voltage is reached after the fault clearance recovery.

**Figure 10.** RMS magnitude of the current before, during and after fault F1 at: (**a**) IED1 and (**b**) IED2 (CB2 transfer trip).

**Figure 11.** Operating time and status of breakers before, during and after the clearance of fault F1: (**a**) CB1 (main protection trip) and (**b**) CB2 (direct transfer trip). CB Status: 0 = False (NO trip/Close) and 1 = True (YES trip/Open).

**Figure 12.** RMS magnitude of current per phase of DGs before, during and after fault F1 (CB2 transfer trip): (**a**) LV side of the PV system and (**b**) LV side of WTG. Note: I\_lim is the preset limit of the fault current contribution of DGs.

**Figure 13.** Adaptive IED2 trip after the unsuccessful direct transfer trip from IED1: (**a**) RMS magnitude of the current at adaptive IED2 and relay operating time and (**b**) CB2 status and operating time during an IED2 adaptive trip.

Figure 14 shows the RMS magnitude of the current of DGs before, during and after the fault F1 with an adaptive IED2 trip. The results in Figure 13 show that it takes 20 ms more than the required 150 ms time for clearance of the fault with adaptive IED2 settings. This is because the DG output current takes an extra 10 ms to stabilize and reach the threshold setting of adaptive IED2. Moreover, both IED2 and CB2 take 5 ms more than the set time of 20 ms for fault detection and tripping. Figure 13 also shows that it will take 15 ms extra for DGs to restore to normal operations (normal currents) in the islanded AC microgrid after the removal of fault F2 completely at 1.37 s. These types of extra delays highlight the demand of faster communication with less than 20 ms one-way transfer delay or extension of the initial time after the fault inception in the LVRT characteristic curve for effective adaptive protection in order to maintain the supply in islanded mode. Otherwise, it will cause the complete loss of DGs within the AC microgrid during fault F1 due to the tripping by anti-islanding (e.g., UV) protection at 150 ms after the fault. This, of course, will decrease the reliability of supply for AC microgrid loads due to extra time for the black start of DGs in the islanded mode, then removal of fault F1 by adaptive IED2 and, finally, the restoration of the normal load. In the presented cases by using a 20-ms delay of one-way GOOSE transfer for the complete removal of fault F1, the transfer trip method is faster than using adaptive IED2 tripping in the islanded mode: a direct transfer trip takes about 150 ms after the fault, and the second backup procedure (adaptive tripping) takes about 190 ms after the fault. However, if the delay of 10 ms for one-way GOOSE transfer is used (Method-A in Figure 6), then even the backup adaptive tripping after the failure of the direct transfer trip could be accomplished within 150 ms after the fault, and the present LVRT curve of the DGs will remain valid.

**Figure 14.** RMS magnitude of the current per phase of DGs before, during and after fault F1 (adaptive IED2 trip): (**a**) LV side of the PV system and (**b**) LV side of the WTG. Note: At 1.29–1.3 s, CB1 trips (Figure 10a), and at 1.37 s, CB2 adaptively trips (Figure 12b).
