Occurrence of faults is inevitable in electrical power systems and the faults result in service outage in some regions. Power outages affect commerce, industry and everyday life of the general population. By speeding up the restoration process, the cost of blackouts can be reduced directly. Once the fault has occurred, it is desirable to limit the effects of the faults to the faulty section only and to restore the service of the un-faulty sections in the minimum possible time to ensure the continuity of service to the customers. This is one of the objectives of Smart Grid Technology .
Initially distribution restoration was done by centralized architecture based systems; the major disadvantage of this architecture was the duration of restoration time .Then, the restoration time was greatly reduced by the decentralized architecture systems with multi agents as in  and genetic algorithm as in . However, we can further reduce the restoration time to a negligible extent by utilizing reclosing dead time and making use of the modern reliable, efficient and fast communication technologies .
Automation in electrical power systems including distribution automation has been an active research area for electrical engineers in the past few decades . Different communication protocols have been proposed for distribution automation [5, 7].
By making use of the advanced digital communication technology, this task can be achieved in an efficient way. Studies based on agent technology and artificial intelligence are very common in power systems these days [9, 10]. Most of the algorithms proposed in these papers start doing restoration after the completion of recloser’s reclosings, which causes an extra delay in the restoration process especially if the recloser has more delayed settings.
This paper proposes a communication based algorithm for reducing the restoration time. The proposed algorithm starts doing the restoration during reclosing. Restoration during reclosing is a unique idea which is proposed in this paper and is explained in the next section
2. Service Restoration Time and Restoration during Reclosing.
A recloser is a circuit breaker equipped with a mechanism that can automatically close and open the breaker for a preset number of times (usually three or four) after it has been opened due to a fault. The opening and closing action of the recloser is known as a reclosing operation, and the time after the recloser has completed its reclosings until the restoration of service is known as service outage time. A recloser has some fast operations (usually one to three) with a reclosing dead time (time during which the recloser remains open) of two seconds and some delayed operations (usually one or two) having a reclosing dead time of fifteen seconds.
Studies have shown that 80% of temporary faults are cleared during the 1st reclosing operation and 10% during the 2nd fast reclosing operation, so after two fast operations the probability of the fault being temporary is only 0.1. It is unnecessary to wait for the recloser to complete its reclosings.
The proposed algorithm exploits this nature of faults (Fig. 1) and starts doing restoration just after the recloser has completed its two fast reclosing operations. The upstream of the fault will still be cleared by the recloser while the downstream will be restored by the proposed algorithm. The process of doing restoration during the reclosing operations is given the name “Restoration during Reclosing”. The proposed algorithm also caters to those cases where the temporary faults are cleared during the recloser delayed reclosing operations by driving the system back to the original configuration.
Fig. 1.Restoration during reclosing
3. Restoration during Reclosing
3.1 Current state of restoration and problems
Initially, the distribution restoration was done by physically sending the crew to the faulty area, and that resulted in an increased restoration time depending on the efficiency of the working crew. Then, centralized architecture algorithms were proposed where each agent sends information to a central agent and the central agent makes decisions and commands other agents on what to accomplish for optimal restoration. It also takes a lot of time because all the actions are performed in series (one after another) and the central agent makes a restoration plan after the occurrence of a fault. A new algorithm was proposed  in which the central agent makes a restoration plan before the occurrence of a fault and informs each agent beforehand. The actions are performed in-parallel to overcome the shortcomings of the previous algorithms. The restoration time was reduced up to 48 seconds, but it only starts restoration after the responsible recloser has lock-opened the fault. This usually results in an increased restoration time especially if recloser has more delayed settings. (i.e. 2F2D, 1F3D, or 4D).
3.2 Proposed idea
In order to decrease the restoration time, this paper presents a new concept of restoration (Restoration during reclosing), where the restoration is started just after the time allotted for two fast trips. A three link six section network as shown in Fig. 2 has been used for explanation of the proposed algorithm. The Feeder (F1) is primarily feeding the shown six sections. If a fault occurs then the outage region can be fed by any combination of the other three Feeders (F2, F3 and F4). The other feeders are also having their own primary feeding regions and each feeder have maximum extra capacity of feeding two sections. Working details of the algorithm are as follows.
Fig. 2.Initial workings of the algorithm
Before the occurrence of a fault, the central agent will download network information and a restoration plan for each agent as shown in Fig. 2 (a). When a fault occurs, some of the protective devices will detect the fault current. When a protective device detects a fault current, it will query about the fault from its upstream and downstream devices. Therefore every protective device involved in that particular fault will have the fault information of its upstream device, downstream device and itself. Based on this information, every device will decide its own role. One of the protective devices will become the Downstream Responsible Device (DSRD) and one will become the Upstream Responsible Device(USRD) while other protective devices will wait for commands from the responsible devices as shown in Fig. 2(b).
In Fig. 2(b), the fault is between Auto Switch (AS1) and Recloser (R2). After the exchange of fault information, R2 will become the responsible device for downstream restoration and AS1 will become the responsible device for upstream restoration.
Once the device is identified as DSRD, it will wait for the time allotted for two fast trips of a recloser which is always less than or equal to 3 seconds. Almost 90% of the temporary faults are cleared during the first two fast trips. Even after two fast trips, the probability of a temporary fault is decreased further.
After two fast trips, the device identified as DSRD will open itself and start the downstream restoration based on the restoration plan provided by the central agent beforehand.
All of the downstream will be restored before the recloser has completed its reclosings. The recloser will declare the fault as a permanent fault after completing its reclosings and will inform all of the protective devices. When the USRD will receive this message, it will open itself and ask the Recloser to close. In this way, all of the upstream will also be restored.
In Fig. 3 (a) R2, DSRD in this case, will open itself after 3 seconds, ask Auto Switch (AS3) to open, and Auto Switch (AS4) & Auto Switch (AS6) to close. This way, the downstream will be restored before the completion of the recloser’s reclosings (Restoration during Reclosing). Just after completing the reclosings, Recloser (R1) will send a permanent fault message to all of the protection devices. After receiving this message, AS1, USRD in this case, will open itself and ask R1 to close. The upstream will then be restored immediately as shown in Fig. 2 (b). The redistributed load is shown in Fig. 2. (c).
Fig. 3.Working of the algorithm in permanent faults
As mentioned earlier, 90% of temporary faults are cleared during first two fast trips, but there still is a 10% As mentioned earlier, 90% of temporary faults are cleared during first two fast trips, but there still is a 10%
When a fault occurs, it is unknown if it is either a temporary or a permanent fault so the initial process for the identification of DSRD and USRD is the same. After the fault is cleared and before the recloser completes its reclosings, the recloser will declare the fault a temporary fault and inform the whole system. If a fault is cleared during the delay trips, it will be declared a temporary fault. When the other protective devices will receive the message, they will undo their previous actions, such as some devices may have been opened or closed by the DSRD for restoration, and the system will go back to its original configuration. The complete process for temporary faults is shown in Fig. 4.
Fig. 4.Workings of the algorithm in temporary faults
3.3 Average restoration time
The time after the recloser completes its reclosing operations until the restoration of service is defined as the restoration time. As mentioned earlier, the algorithm divides the system into upstream and downstream making the restoration time different for upstream and downstream.
Where 3: the time equal to two fast trips of the responsible recloser in seconds. Trds: the downstream restoration time in seconds. Tcd: the communication delay in seconds, which is usually less than or equal to 0.5 seconds. Ti: the interrupting time (time required by CB for opening or closing). It is usually 3 cycles. Tro: the recloser operation time. It is the sum of the recloser dead times, Trip interval times (Time during which recloser remains closed), and Interrupting times, so the Trds depends upon the settings of a responsible recloser. Trus: the upstream restoration time in seconds. Trt: the temporary fault restoration time in seconds.
The system analyzed in this paper has two reclosers, R1 which is 2F2D and R2 which is 2F1D. By considering extreme cases, Tro for R2 will be 4.75 seconds and Trds will be zero if Tcd is less than or equal to 0.75 seconds. If we assume Tcd as 0.75 seconds, Trds will be zero, Trus will be 1.6 seconds and Trt will be 0.8 seconds.
The system analyzed in this paper is a 3-link 6-section system, so faults may occur at six possible points and each fault may be temporary or permanent. By considering both faults at all six possible places and considering 80% of the faults as temporary and 20% as permanent, the average restoration time comes out to be 0.8 seconds.
This is highly acceptable and negligible as compared to previous work where it was up to 48 seconds .
3.4 Exceptional case (communication failure)
Though communication is very fast, efficient and reliable these days, there are still some cases where communication may also fail. For example, occurrences of natural disasters or a car striking a pole may result in a breakdown of either electrical and communication wires or both of them.
There are three possible scenarios i.e. only communication failure, only power failure and both power and communication failure. First case (Only communication failure) is out of scope of this paper. A mechanism for reestablishment of the communication service should be defined by the communication algorithm being used. Second case (Only power failure) has been explained in the previous section. This section only deals with the third case, both power and communication failure. When a fault occurs in a power system then it will check the health of the communication links.
There are several possibilities of communication failures, i.e. fault and communication failures in the same section, a fault in one section and a communication failure at another, an entire downstream communication failure, an entire upstream communication failure, or at worst, an entire system communication failure may also occur.
All these possible communication failure cases have been analyzed and eight rules shown in Table 1 have been proposed. If there is a communication failure, each protective device will check the rules and do the necessary action based on the rule which is satisfied for that particular device. The actions will be carried out by the protective device/devices in the sequential order as shown in Table 1, against each rule. The rules are only for communication failures cases followed by power line faults.
Table 1.* Time t1 is equal to the complete operation time of the responsible recloser. * Time t2 is equal to the time of two fast operations of the responsible recloser.
Change of status; if previous state of a protective device is open, it will get closed and vice versa.
Fig. 5 shows an entire downstream communication failure. Before the occurrence of a fault AS2, AS3, AS4, AS6, and R2 will know about the communication error. When a fault occurs between R2 and AS2, Rule 4 will work for AS2, AS3, AS4 and AS6. Therefore, AS2 and AS3, which are closed auto switches, will wait for 22 seconds. This is for the worst case scenario; though, the recloser total operation time will be always less than 22 seconds. If the fault is permanent; however, before 22 seconds, it will be lock-opened by the responsible recloser and check the service. If it still experiences service failure, it will open itself.
Fig. 5.All downstream communication failure
For AS4 and AS6, Rule 4 is also true. Since these are open auto switches, they will wait 23 seconds to avoid overloading and keep the system as an open loop system i.e. opening actions before closing actions. After 23 seconds, if service is still not restored, they will be closed. For R2, Rule 2 is true; after 22 seconds, it will become USRD. After 23 seconds, AS4 and AS6 will be closed, and the downstream will also be restored
In every communication failure case, 100% restoration is not possible. In the worst case scenario, when the fault is next to the Re and the entire communication is down, 60% of the load will be restored. Before the occurrence of a fault, devices know about the communication failure. If there is a communication failure among the protective devices involved in restoration of a particular fault, then the DSRD will follow another restoration plan which is also provided by the Central Agent (CA) beforehand.
4. Case Studies
A three link six section network  has been simulated in SIMULINK (MATLAB) R2010a version. A total of nine protective devices have been developed as agents. The device in the hexagon, which is next to the main feeder, is a relay. The two square shaped devices (2nd and 4th from ReA) are reclosers. The first recloser (2nd from Re) have 2F2D settings, and the second recloser (4th from Re) is set as 2F1D.The devices in the black circles (2nd, 5th and 6th from Re) are closed auto switches. Those in the white circles (gateway to other sources) are open auto switches. Each protective device is uniformly spaced by a distance of 3km, each section has a fixed load of 2MW, and the fault in all the cases is a single phase to ground fault. A restoration plan has been given to every device. When a protective device becomes DSRD, it will follow that plan.
4.1 Permanent fault
A single line to ground fault has been assumed in section 3 which is the primary protection region of recloser 1. Fault duration has been set at 0.05→2.5 in Simulink, which is a permanent fault for recloser 1.
The fault in this case is between AS1 and R2. The following actions have been executed by each device after the occurrence of the fault as shown in Figs. 6, 7 and 8.
Fig.6.Permanent fault simulated case
Fig. 7.A).R1 trip signal B). AS1 open/close signal
Fig. 8.C). R2 open/close signal D). AS3 open/close signal E.)AS4 open/close signal F). AS6 open/close signal
Re, R1, and AS1 have detected the fault, so they will query about the fault from their upper and downstream protective devices. Based to the fault information of the upstream, own, and downstream protective devices, R2 will become DSRD and AS1 will become USRD. R2, which is now DSRD, will wait for 2 fast trips, open itself, and start the downstream restoration R2 will ask AS3 to open and AS4 & AS6 to close. Section 6 will be fed by F3. Section 4 and 5 will be fed by F2 so the entire downstream will be restored before the recloser completes its action. When R1, which is primarily responsible, for this fault completes its action (2F2D), it will declare the fault a permanent fault and inform all of the other protective devices. When AS1, which is USRD, receives this message, it will open itself and ask R1 (the Responsible Recloser) to close. R1 will close and the upstream will also be restored.
4.2 Temporary fault
A single line to ground fault has been simulated in section 4 between R2 and AS2. Duration of the fault has been set at 0.05→0.3, which will be cleared by recloser 2 during a delayed reclosing.
Since the fault in this case is temporary, the following actions have been executed by each of the involved protective devices and are depicted in Fig. 9.
Fig. 9.A). Trip signal of R2 B). Open/close signal of AS2. C) open / close signal of AS4
Re, R1, AS1 and R2 have detected the fault, so these protective devices will ask their upper and downstream devices about the fault. After getting fault information, R2 will become USRD and AS2 will become DSRD. AS2, DSRD in this case, will wait 3 seconds, open itself, and start the downstream restoration. AS2 will ask AS4 to close, and section 5 and 6 will be fed by F3. The fault will be cleared during the recloser delay operation, so the recloser will declare the fault a temporary fault and inform all of the devices. When DSRD and R4, the state being changed by DSRD from opened to closed, receive this message, they will go back to their original state. (DSRD→closed & R4→ Opened) The system will then go back to its original configuration i.e. the whole system will be fed by the main feeder F1.
4.3 Communication failure
A permanent fault has been simulated in section 2 between R1 and AS1 and a communication failure is also between these two devices.
Since there is a communication failure between R1 and AS1, these two devices will check the rules mentioned in Table 1 and follow the necessary course of action as shown in Figs. 10 and 11.
Fig. 10.Simulated case for communication failure
Fig.11.A) R1 Trip signal. B) AS1 open/close signal. C) R2 open/close signal. D) AS3 open/close signal. E) AS4 open/close signal. F) AS5 open/close signal. G) AS6 open/close signal
For AS1, Rule 1 of Table 1 is satisfied so it will wait for 22 seconds and then check the service. Since the fault is permanent, it will find a service failure. It will become DSRD, open itself, and start downstream restoration. AS2, acting as DSRD, will ask R2 and AS3 to open and AS4, AS5 and AS6 to close, so all of the downstream will be restored. For R1, Rule 2 of Table 1 is satisfied; after 22 seconds, it will become USRD.
In this case, since USRD is a recloser (R1), after it completes its after operation service of it will be restored. If it had been an auto switch after 22 seconds, it would have opened itself and asked the responsible recloser to close.
4.4 Advantages of the proposed algorithm
The proposed algorithm is more advantageous for permanent faults, and it has reduced the average restoration time by a negligible extent. The proposed algorithm is a robust algorithm which also caters communication failure cases. A comparison of the proposed algorithm and existing algorithms is shown in Table 2.
Table 2.Restoration time in seconds
The proposed algorithm has reduced the average restoration time to about one second which was 48 seconds in the previous paper . If a temporary fault persists till recloser’s delayed tripping (10% Probability), there will be switching transients equal to the communication time.
A communication based algorithm has been applied to a Smart Grid Distribution Management System to minimize the service restoration time. Different protective devices, i.e. Relay, Reclosers and Auto Switches, have been developed and equipped with the proposed algorithm in Simulink and various cases including different communication failure scenarios that have been analyzed. The algorithm has given the desired results in all cases and proved its reliability and efficiency. The proposed algorithm is an efficient algorithm which also takes into account communication failure cases. It uses the agent technology which has a promising future
Future work on this algorithm could be useful for application in closed looped systems including Distributed Generations to make it a key component for Smart Grid.