Active Distribution Networks become a reality in the modern power systems. An Active Distribution Network consists of high penetration of Distributed Energy Resources (DER), including Energy Storage (ES) and Demand Response (DR). It may also include Micro-grids (μG), Plug-in-Vehicles (PEV), Power Electronics (PE), distributed Remedial Action Schemes and Protection (RAS), and other devises and entities controllable by local energy management system and controllers, as well as by advanced Distribution Management Systems (DMS).
These components of the Active Distribution Networks are highly sensitive to emergencies in bulk power systems.
The reactions of these components to the emergency changes of the operating conditions may include disconnection of DERs due to frequency and/or voltage protection (including ride-through functionalities with a wide range of possible settings [1]), separation of microgrids with load-rich or generation-rich balance [2], [3], discharge of ES, activation of demand response, and other unpredictable activities. Changes of real power injections by the DERs may also happen during the emergency due to their dependency on ambient conditions. Some of these reactions may reduce the impacts of the original emergency disturbance, while other may aggravate the situation.
The above-mentioned reactions are different at different times and under different ambient conditions and involve a high degree of uncertainty. This suggests a need in adaptive methodologies of the mitigating means of the power systems. One aspect of such adaptation for load shedding methodologies is addressed in [4].
In this article, we address another aspect of adaptation of Under Frequency Load Shedding (UFLS). This adaptation relates to the uncertainty of possible developments of a major generation deficit in large electrical islands. The uncertainty will significantly increase, when substantial Active Distribution Networks are included in a deficit island.
An electric island of a bulk power system may be created as a result of an intentional or unintentional separation of the power system due to a wide-scale contingency. The sizes of the islands and their composition of loads and generation are often unpredictable. Therefore, one of the assurances of adaptation of the UFLS to the island conditions is that the amount of load connected to the UFLS is sufficient for any possible size and composition of the islands. This may be a challenge, because different loads are not equally available for the UFLS.
Another significant uncertainty amplified by the presence of Active Distribution Networks is the unpredictability of the development of the generation shortage. The development of the same total generation shortage may range from a one-time event to a chain of small events during a prolonged time interval. Even when the frequency is above the restoration level the development of the deficit can continue before the island is reconnected with the bulk power system. The DERs running the Watt-Frequency function will reduce the real power injection, some DERs and microgrids can still disconnect due to voltage swings or other reasons, some uncoordinated load restoration may happen, etc.
The UFLS should properly operate under any of these extreme conditions and any conditions between them.
In the one-event cases, the frequency drops monotonically. In these cases, the reducing frequency activates the groups of the UFLS that are distinguished by their frequency settings. Let us call these groups of UFLS – UFLS-1[5]. The amount of load connected to UFLS-1 should be sufficient to stop the drop of the frequency above a given low frequency limit [6], if the maximum possible deficit happened as one event. The level of the frequency stabilized after the operations of UFLS-1 may be lower than the desired restoration frequency. In this case, a small number of the groups of UFLS with high frequency settings and longer time delay settings bring the frequency to the restoration level. Let us call these groups of UFLS – UFLS-2 [5].
In another extreme case of the development of the generation shortage, when the process consists of a chain of small events, the frequency may not reach the levels that activate many groups of UFLS-1. In these cases, the frequency stalls above the settings of UFLS-1 and should be brought to the restoration level by UFLS-2. It means that the amount of load connected to a greater number of of UFLS-2 groups should be sufficient to mitigate significant portion of the total generation deficit. Hence, to accommodate both conditions of the development of the deficit, the total amount of the load connected to the UFLS should be larger than the maximum possible deficit. There may be not enough loads available for the UFLS to meet this requirement.Â
In order to minimize the needed amount of load connected to the UFLS, a combined design of the UFLS was introduced [6]. In this design, the groups of UFLS have the settings of both UFLS-1 and UFLS-2. Let us call this design UFLS-Comb. In this case, if the frequency does not reach the settings of UFLS-1, the same load can be shed later under the settings of UFLS-2. The total load connected to the UFLS should be sufficient to mitigate the maximum available deficit regardless how the deficit is developed. The advantage of this design in comparison to the separate UFLS-1 and UFLS-2 is the smaller amount of load needed to be connected to UFLS.
To avoid significant over shedding of load by the UFLS, the UFLS is divided in many groups. When there are many groups of UFLS-2, the time settings of the different groups should differ by a significant time interval to avoid over shedding when the frequency increases. This may lead to longer restoration times and to violations of the low frequency limit curve. This curve elevates with time, providing a “triangle” frequency-time safety zone [6]. The UFLS-Comb does not guarantee a “triangle” shape of frequency restoration. The frequency, in this case, may go beyond the “triangle” limit curve. Â
In order to guarantee the “triangle” shape of the frequency restoration by the UFLS, an UFLS design with frequency settings changing in time was introduced [7], [8]. In this case, the initial settings are similar to the settings of UFLS-1. When the frequency drops to a given level that indicates that there is a significant generation deficit, the settings start rising in time toward the dropping frequency. The change of system frequency is typically, exponential. Therefore, it makes sense to change the settings of the UFLS also in an exponential manner. The frequency settings should rise in time slightly slower than the rise of the system frequency, when the system frequency can reach the restoration level, to avoid the over shedding. If the system frequency is not going to reach the restoration level, the settings of the UFLS meet the frequency below the level, and additional load is shed. Let us call this design UFLS-Exp.
The above-discussed designs of the UFLS, provided that sufficient load is connected to the UFLS, adapt to any development of the generation deficit without involvement of central control. However, the performances of these three designs are different.
The conventional design of UFLS requires an excessive amount of connected load to be able of mitigating both the one-event and cascading development of the generation deficit.
The combined design of UFLS needs less of connected load, but may either prolong the restoration, or increase the over shedding.
The design with changing settings has the potential for shorter restoration times, smaller over shedding and fewer violations of the established frequency limits.
With the impending high penetration of active distribution networks, it is worth considering an upgrade of the conventional UFLS to more adaptive designs.Â
More details are available in [10].
References
Recommended Settings for Voltage and Frequency Ride-Through of Distributed Energy Resources, EPRI’s White Paper prepared by Jens C. Boemer, May 2015. Available: http://www.epri.com/abstracts/Pages/ProductAbstract.aspx?ProductId=000000003002006203
Use Case #1: Information Support for Coordination of EPS and Microgrid Load Shedding. Available: Â http://smartgrid.epri.com/Repository/Repository.aspx/Â and https://members.sgip.org/kws/groups/sgip-pap24wg/documents?folder_id=0
Nokhum Markushevich, Information Exchange between Advanced Microgrids and Electric Power Systems. Available:Â https://www.scribd.com/document/376567099/Information-Exchange-between-A...
N. Markushevich, Automatic Load Shedding in Active Distribution Networks. Available: http://www.energycentral.com/c/iu/automatic-load-shedding-active-distrib...
G. D. Boutin, N. S. Markushevich, M.G. Portnoy, et al, Automatic Frequency Load Shedding in USSR Power Systems, , CIGRE, Paris, 1972
Standard PRC-006-2— Automatic Underfrequency Load Shedding. Available: http://www.nerc.com/_layouts/PrintStandard.aspx?standardnumber=PRC-006-2&title=Automatic%20Underfrequency%20Load%20Shedding&jurisdiction=United%20StatesÂ
Nokhum Markushevich, An Under-frequency Relay with Time Delay, USSR Certificate of Invention # 201504, 1964;, Available: http://www1.fips.ru/fips_servl/fips_servlet?DB=RUPAT&rn=3118&DocNumber=201504&TypeFile=html
N. S. Markushevich, Under Frequency Load Shedding in Power Systems (from Experience of the Latvian Power System), Moscow, 1975 (Russian)
R. S. Rabinovich, Under Frequency Load Shedding in Power Systems, Moscow, 1989 (Russian)
Nokhum Markushevich, Under Frequency Load Shedding in Power Systems with Active Distribution Networks. Available: https://www.scribd.com/document/376565395/Under-Frequency-Load-Shedding-in-Power-Systems-with-Active-Distribution-Networks