1.3. Pregunta problema:
3.6.2. Analizar:
3.6.2.3. Análisis gesto de control de balón (Tabla de análisis):
The conceptual design can be adapted to accommodate DGs or networks with FCL connection. As a matter of fact, FCL connection is a special case of the system given above. Here, the exact fault current contribution of each DG is known and not calculated.
When FCLs are utilized, the data map in the MCPU can be updated as shown in Table 3.2. The only difference is with the DG parameters where IfaultDG is directly recorded rather than
It is assumed that all DGs are interfaced with an inverter that has FCL ability which limits the DG output current to a predetermined IfaultDGx value. Since fault currents are known exactly
rather than calculated or approximated, the performance of the entire protection system will be more accurate and reliable.
The MCPU continuously monitors the statuses of all DGs in a synchronized manner by communicating with them using a Publish/Subscribe peer-to-peer messaging (similar to IEC 61850‟s GOOSE [161]) system. The MCPU therefore receives status report from all DGs
and records their status as ON/OFF. If a DG disconnects at any time for example, the new fault current levels are re-calculated by considering the fault current contribution (IfaultDGx) of
that particular DG. In this way, the system is ready for any fault condition before any fault, actually, occurs. The operation of the system through communication and calculation is identical with the generic system given above.
TABLE 3.2.DATA MAPS IN THE CENTRAL PROTECTION UNIT
Grid-Connected Islanded
Operating Mode 1 0
Grid Fault
Contribution IfaultGRID Relays Operating Fault
Current
Fault Detection (1 – Y, 0 – N)
Time delay for Selectivity
R1 IR1 0 t1
R2 IR2 0 t2
R3 IR3 0 t3
…
DGs IfaultDGx Status (1-ON, 0-OFF)
DG1 IfaultDG1 1
DG2 IfaultDG2 1
DG3 IfaultDG3 0
Alternatively, the use of FCLs makes it possible to simplify this system further. Since fault contribution of each DG is constant, it may be reported to the MCPU before DG‟s
deployment and the communication line for that might be omitted. Furthermore, the status of a DG can be monitored through the status of its respective relay (e.g. R3 for DG1, R5 for DG2 etc.). For example, if R3 is connected then it means DG1 is in operation and in case of a fault IfaultDG1 should be taken into account and if R3 is open it implies DG1 is OFF and no
fault current is in question. In this case, relay‟s status is sufficient and the communication
lines from the MCPU to DGs can be eliminated altogether. In such a case, fault current contributions of DGs will be stored in the MCPU before deployment and updated in rare cases of modification. In this way, not only the fault currents of DGs will be known precisely but also the burden of installing expensive and complex communication lines to each DG, the dark and light bluelines in Figure 3.6, can be eliminated. This simplification has effects on the MCPU, as the area marked with dashed frames will be eliminated if DG communication is omitted. Thus, the data map in MCPU will include relay parameters and static DG parameters which only require local storage. Worthy to note, though, this simplification does not include intermittent DGs such as Wind turbines or PVs. These DGs may be connected to the grid but may not generate any power due to intermittent nature of their sources, i.e. the wind and the sun. On the other hand, diesel generators, fuel cells and other DGs that have constant power output can be incorporated to this simplified protection scheme.
Under these new conditions, for a particular relay, the operating fault current (the current level that causes the relay to trip) is calculated as in (3.2):
(3.2)
where m is the total number of DGs in the microgrid, IfaultGRID, IfaultDGi and StatusDGi are as described above.
The MCPU uses a simple algorithm shown in Figure 3.12. The status of R1 has been given a special consideration since it imposes the operating mode and grid fault contribution.
Figure 3.12. Protection Algorithm employed in MCPU for FCLs.
This flow chart is similar to the one given in Figure 3.7 with a fundamental difference. Unlike the algorithm in Figure 3.7, this flow chart only follows the connections of relays. This is thanks to the unique feature of FCL interfaced DGs where the whole monitoring can be realized over relays. In this manner, management and communication processes have been simplified. Here, the operating condition of the microgrid is checked over R1 and the fault contribution of the grid is taken into account. Then, the relays for DG connections are checked to update their fault current contributions. Finally, the new operating current conditions are updated in the relays. In case of a fault in the network, relays operate independently and follow selective protection procedures, as discussed above.
3.6. Conclusion
The studies on microgrids will aid in the development of secure, reliable, and stable real-life networks with greater penetrations of RE sources. This chapter has presented a new protection scheme for microgrids with high DG penetration. An MCPU is utilized to monitor all entities inside the microgrid and new operating conditions are calculated for every change experienced in the microgrid. The developed system considers two events as crucial changes: microgrid‟s connection/disconnection (i.e. islanding) from the grid and a DG‟s
connection/disconnection from the microgrid. The MCPU is used to perform fault current contribution calculations by considering the rated currents and fault contribution coefficients `k` of DGs. After being updated with these operating conditions, relay make decisions locally depending on the local measurements. This makes the system more robust and reliable since continuous communication is not needed in the decision making process.
The system is very versatile and can be developed to suit more specific purposes or meet different purposes. It can be extended to suit much larger and more complex microgrids. Furthermore, it can be adapted to accommodate different grid components such as FCLs. Limited fault current helps anticipate fault conditions and protection system is programmed accordingly. Communication lines and MCPU can be simplified thanks to the constant and predetermined fault current contribution of DGs. Since the exact fault current contribution is known, this protection system is more reliable than the systems which predict or approximate fault conditions. DGs with FCL can be deployed merely by creating an additional record in MCPU. Relative fault current and the relay which is responsible for its connection are recorded.
Specifically, the system is designed to be generic and may serve for plug-and-play purposes. Special attention must be paid to the communication protocol utilized for the communication of data within the system and a standard packet-type for the relay and distributed generator communication can be developed. If each DG unit is equipped with a simple communication device to report its status, rated current and fault contribution, then a universal concept of DG connection can be achieved. This may prove useful for the implementation of plug and play concept in microgrids.
This flexible yet strong conceptual design has been reinforced with various aspects of this research. The automated microgrid structure detection and its impact on selectivity will be discussed in Chapter 4. The specific details of the relay parameter assignments, i.e. fault current and time delay, will be discussed in Chapter 5. Chapter 6 presents the standard communication framework developed in accordance with IEC 61850 to universalize this conceptual design. Chapter 7 provides the results of the simulation works undertaken to investigate behavior of microgrids under different conditions. The conceptual design presented in this chapter has been developed based on these behaviors. Finally, Chapter 8 extends this design further, and shows how this system can be used to support other protection systems where they are insufficient. The eventual system has the ability to address varying challenges of new generation microgrids.
Chapter 4
Object Oriented Modeling of Microgrids and Automated
Structure Detection
Publications pertaining to this chapter:
1) Taha Selim Ustun, Cagil Ozansoy, Aladin Zayegh, "Implementation of Dijkstra’s Algorithm in a Dynamic Microgrid for Relay Hierarchy Detection," in the Proceedings of Smart Grid Communications (SmartGridComm), 2011 IEEE International Conference on , vol., no., pp.481-486, 17-20 Oct. 2011.
2) Taha Selim Ustun, Cagil Ozansoy, Aladin Zayegh, "Using Object Oriented Data Structures for
Dynamic Communication and Control in Electrical Networks," in the Proceedings of the Twenty-fifth Canadian Conference on Electrical and Computer Engineering (CCECE), Canada, 29April – 2 May 2012.
4.1. Introduction
Microgrids have very dynamic structures and automated control and protection of microgrids would only be possible when the microgrid structure is continuously monitored. This chapter details the ideas and concepts developed to enable such a process. Object Oriented (OO) models have first been developed that allow information on various elements of a microgrid systems (power system networks in general) to be defined. The use of the graph theory concept for modeling the structure of the microgrid is then discussed and finally the use of a famous Computer Science algorithm for relay hierarchy detection is explained.
It is known that microgrids have dynamic structures which change very often. The following may be counted among the reasons for the changes in the microgrid structure [146]:
New DG or load deployments Islanding of the system Fault conditions
Reconfiguration of the structure for reasons such as maintenance
This dynamic behavior of microgrids is a major protection challenge since the conventional selectivity methods assume a fixed network structure and a predetermined relay hierarchy [108]. Whenever restructuring occurs, the selective levels assigned prior to that become erroneous. For a proper operation, the selective levels of relays should follow the changing conditions of the network. Whenever a structure variation takes place in a microgrid, new relay hierarchy should be extracted and corresponding settings including time delays should be determined before updating these settings within the relays via communication lines [109, 110]. It is essential that this process is performed automatically so that the processing time is minimal and there is no interruption in power delivery. This requires an algorithm which will determine the current structure of the system and yield the relay hierarchy at all branches of the network.
There are some studies presented in the literature which emphasize on the importance of such an adaptive selective operation such as in [162] and [73]. However, the prior discusses the issue qualitatively without any technical details whereas the latter implements an algorithm which includes a look-up table. The work presented in [162] touches upon the challenging nature of relay hierarchy and selectivity concepts in microgrids. A communication-assisted selectivity approach is explained in concept. However, the technical details of this approach,
its implementation in microgrids with different set of components are not discussed. On the other hand, the research presented in [73] provides a robust system for microgrid protection. However, it uses a look-up table which lists all the cases that may exist in the microgrid under consideration. This is a large set-back because it requires the knowledge of all possible microgrid configurations beforehand, plus human input for the preparation of this look-up table and finally it requires that the microgrid should always match one of the predetermined structures. Moreover, any kind of a new deployment, which is very common to microgrids, requires that the whole selectivity table should be re-written.
In order to address all of the above-mentioned challenges, the developed protection system shall be equipped with an automated operation capability that determines the structure of the system and yields the relay hierarch. This chapter details the work carried out as part of this research to address this development need. Firstly, the novel representation of power networks with OO models is explained. Then, the challenging nature of microgrid structure variations is given. OO models have thereby been developed to handle the dynamic and changing nature of microgrids. Accordingly, a microgrid system is modeled based on graph theory where the components are represented as nodes. Later, Dijkstra‟s algorithm, which is
famous for shortest-path calculation purposes, is run over the microgrid to determine the relay hierarchy at any point in time. In this manner, regardless of the dynamic changes occurring in the system, the hierarchy of the network components can be extracted. The implemented algorithm not only ensures proper selective operation under fault conditions but also facilitates the introduction of new connections and new devices to the system. Since the relay hierarchy is detected automatically, even with new connections, this algorithm serves for plug-and-play concepts in electrical networks.