Under-frequency load shedding in isolated multi-microgrids

https://doi.org/10.1016/j.segan.2021.100494Get rights and content

Highlights

  • An exclusive under-frequency load shedding scheme is proposed for multi-microgrids.

  • The UFLS scheme provides a fair distribution of load shedding among microgrids.

  • The location of removable loads is determined by solving the optimization problem.

  • A mixed integer non-linear programming optimization problem is proposed.

  • A new index is introduced in order to distribute load shedding among MGs.

Abstract

In this paper, a new under-frequency load shedding (UFLS) scheme is proposed for multi-microgrid (MMG) application. MMG, consists of multiple micro-grids, perhaps with different owners, connected together. Multiple owners make load shedding policies more complex in MMG compared to the policies employed in micro-grids and large power systems. Accordingly, an appropriate UFLS scheme is required, in order to ensure the frequency stability of the MMG, while considering its economic aspects. Therefore, in this paper, an optimization problem has been introduced which minimizes the cost imposed on the owners for mandatory load shedding alongside the cost of MMGs operation after the load shedding procedure. By solving the proposed optimization problem, the optimal locations of loads to be shed are determined. Two optimization algorithmsgenetic algorithm (GA) and exchange market algorithm (EMA) – and GAMS software have been used for solving this optimization problem. A new index is introduced in this paper in order to distribute load shedding among MGs based on their total power generation and load demand. Simulation results confirm the efficiency of the proposed methodology.

Introduction

Ensuring frequency stability is one of the most critical responsibilities of power system protection. In order to maintain the frequency close to the nominal value, different control systems are employed such as governors and Automatic generation controls (AGCs). If the control system is not able to return the frequency to the nominal value in a timely manner, the frequency protection system will be enabled to prevent frequency instability. Under-frequency load shedding (UFLS) is the key protection scheme against frequency instability which is implemented by UFLS relays. UFLS methods can be classified into three main categories [1]:

(1) Conventional method

(2) Adaptive methods

(3) Computational intelligence-based Techniques

In [2], the conventional UFLS method is fully explained. This method, regardless of the extent of the disturbance, disconnects the pre-specified amount of loads in each step which brings about non-optimal load shedding. However, adaptive and intelligent UFLS methods prevent unnecessary load shedding and, therefore, lead to an optimal approach.

Generally, adaptive and computational intelligence-based UFLS methods estimate the amount of disturbance and shed the loads depending on the extent of the imbalance. In [3], [4], [5], [6], [7], [8], [9], adaptive UFLS methods are proposed for the conventional power system. In [3], an UFLS scheme is proposed in which the loads with lower VQ margin and voltage profile are placed on top of the priority list of load shedding. In [4], frequency trajectory is forecasted by using frequency second derivate and the amount of removable loads are determined by continuous forecasting of the minimum frequency value. In [5], for determining the location of loads to be shed, an index is defined for each feeder. In [6], the UFLS and under voltage load shedding schemes are implemented in a coordinated manner. In [7], frequency and time to reach the threshold value are predicted, using curve fitting method. The load priority table is used to determine the location of loads to be shed. In [8], the loads are ranked based on Outage Sensitivity Index (OSI). The load shedding focuses on the loads with higher OSI, i.e., higher impact on the critical elements prone to outage. In [9], a three-stage adaptive UFLS scheme is proposed. In which according to the event type and lookup table, predetermined loads are shed. In [10], instantaneous voltage deviation of load buses is used to determine the frequency relays threshold. The more severe the voltage drop at load buses, the higher the frequency thresholds of the relays. In [11], the value of the frequency threshold of the relays and the time delay between successive steps are determined according to the voltage drop of the load buses.

In [12], [13], [14], [15], [16], intelligent methods are used to determine the amount and location of removable loads in conventional power systems. Genetic algorithm (GA) has been employed in [17], [18] with different objectives, to either minimize the amount of loads shedding, circumvent line congestion, and improve voltage profile [17], or to minimize loads shedding while maximizing lowest swing frequency [18]. In [14], an ANN has been trained for optimal load shedding (OLS) with generator outputs, load demand, total spinning reserve and rate of change of frequency (RoCoF) as inputs. In another study [15], ANN is used to shed loads in five steps, the same inputs are considered here, and the output of the ANN is the OLS amount in each step. In [16], a UFLS scheme based on the neuro-evolution algorithm is proposed. ANN is trained using OLS with overload, frequency, RoCoF.

Rate of frequency drop in the MGs is higher than conventional power systems, due to their low inertia constant. Then, the UFLS schemes in MGs must be faster than conventional power system. References [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], proposed adaptive UFLS schemes for implementation in MGs. In [19], consumers are ranked based on the willingness to pay index and rate of change of frequency (RoCoF) is calculated for each load. UFLS will continue until, the sum of RoCoFs of the loads are greater than the system RoCoF. In [20], [21], UFLS is performed stepwise. In each step, the amount of imbalance and inertia constant of the system are updated to avoid excessive load shedding. In [22], a decentralized UFLS scheme is proposed based on the nearest neighboring consensus algorithm (NNCA). The proposed scheme uses a two-level method. In the first layer, the amount of disturbance is estimated and in the second one, the determined loads are shed. Similarly in [23], the UFLS scheme is implemented as a two-layer method. In [24], to determine the locations of removable loads, an optimization problem has been proposed. The objective is to minimize the cost of load shedding. In [25], a UFLS scheme combining the fixed and random priority of loads is presented. Any combination of loads which has the least difference with the amount of imbalance is selected as removable loads. In [26], a stepwise load shedding takes place at locations where the highest voltage drop and frequency variation are experienced. In [27], using distributed state estimators, the UFLS sheds the determined amount of load commensurate to the disturbance in four steps. In [28], Loads are divided into three groups, vital, semi-vital and non-vital loads. The non-vital loads are placed in the top priority of the load shedding. In [29], a mixed integer linear problem (MILP) is defined in order to minimize the amount of load shedding, considering the uncertainties of the system parameters such as inertia time constant, load damping and generation deficiency. GA is used for solving the proposed MILP. In [30], a security-based load shedding optimization problem is presented in order to minimize the expected load shedding amount.

From 2013 to 2019, about 1669 microgrids (MGs) were installed in the United States, of which 546 out of 1669 were installed in 2019 (32%) [31]. This indicates that the electricity industry is moving towards greater use of MGs. Also, many articles have been published about MGs. MGs compared to the conventional distribution system, are able to improve the reliability of the distribution system, reduce greenhouse gases, and also lower the cost of supplying electricity to customers [32]. So it is expected that in the near future, the traditional distribution system will be replaced by MGs. With the increase in the number of MGs and their connection to each other, a new concept called multi-microgrids (MMG) has emerged. UFLS is one of the essential protection schemes in each network and the proposed methods for MGs are not necessarily suitable for multi-microgrid (MMG). So, a new UFLS scheme for implementation in MMG is proposed in this paper. A MMG consists of several MGs that each of them may have different owners. Load shedding in MMG is in conflict with the interests of the owners of the MGs. As a result, none of the owners of the MGs are willing to remove their loads. In this situation, the protection system should work in a way that UFLS does not create discontent among owners of MGs. Therefore, considering the economic aspects in the UFLS methods of MMGs is an important issue. Determining the location of loads to be shed in MMG is the main innovation of this article. In this paper, a UFLS method is introduced in which, the removable loads are selected by solving a mixed integer nonlinear optimization problem (MINLP) in order to minimize the cost of load shedding as well as the operation cost after load shedding. Since all the MG owners are reluctant to reduce their loads, it is essential to provide a load shedding strategy that not only meets technical requirements of the grid, but also serves the economic interests of MG owners. A new index is defined for loads based on each MG’s total generation and load demand in order to fairly distribute load shedding among different MGs.

The remainder of the paper is organized as follows. In Section 2, the problem formulation is proposed. Section 3, describes the proposed optimization methods. In Section 4, the assumptions and simulation results are presented.

Section snippets

Proposed methodology

In this paper, a new centralized UFLS scheme is proposed for isolated multi-microgrids (MMGs). As mentioned before, MMGs consist of several interconnected microgrids (MGs) with possibly different owners. Shedding loads from each MG will be unfavorable for the owners of the MG. However, in times of disturbances, it is necessary to shed some amount of load to maintain the frequency of isolated MMG within the permissible range. Therefore, a policy is required to establish a trade-off between MMG

Optimization methods

The proposed optimization problem for determining the location of removable loads is a mixed integer nonlinear programming (MINLP) problem. Solving such problems is very complicated. In this paper, Metaheuristic algorithms, genetic algorithm (GA) and exchange market algorithm (EMA), are used to solve the problem. The proposed optimization problem is also modeled in GAMS software.

Simulation studies

In this section, first, test system data and modeling assumptions are introduced. Then, simulation results for different case studies are presented in order to demonstrate the effectiveness of the proposed UFLS scheme.

Conclusion

In this paper, a centralized UFLS scheme is proposed for an isolated multi-microgrid (MMG). Since different MGs do not necessarily have the same owner, implementation of the UFLS method in an isolated MMG is a significant challenge, which is to determine the location of the removable loads with the least economic costs while maintaining system stability. Load shedding will cause dissatisfaction of the consumers which will also result in the dissatisfaction of the owners of the MGs. In this

CRediT authorship contribution statement

Seyed Mohammad Sajjadi Kalajahi: Conception and design of study, Acquisition of data, Analysis and/or interpretation of data, Writing - original draft - review & editing. Heresh Seyedi: Conception and design of study, Analysis and/or interpretation of data, Writing - review & editing. Kazem Zare: Acquisition of data, Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgment

This project is supported by a research grant of the University of Tabriz, Iran (grant number 789). All authors approved the version of the manuscript to be published.

References (40)

  • ElmoreW.A.

    Protective relaying theory and applicantions

    Prot. Relaying Theory Appl.

    (2003)
  • SeyediH. et al.

    New centralised adaptive load-shedding algorithms to mitigate power system blackouts

    IET Gener. Transm. Distrib.

    (2009)
  • RudezU. et al.

    A novel approach to underfrequency load shedding

    Electr. Power Syst. Res.

    (2011)
  • BaiD. et al.

    Under frequency load shedding scheme based on information sharing technology

    (2013)
  • MollahK. et al.

    Coordinated strategy for under-voltage and under-frequency load shedding

  • RudezU. et al.

    WAMS-Based underfrequency load shedding with short-term frequency prediction

    IEEE Trans. Power Deliv.

    (2016)
  • HoseinzadehB. et al.

    Centralized coordination of emergency control and protection system using online outage sensitivity index

    Electr. Power Syst. Res.

    (2018)
  • ShekariT. et al.

    An adaptive wide-area load shedding scheme incorporating power system real-time limitations

    IEEE Syst. J.

    (2018)
  • HoseinzadehB. et al.

    Adaptive tuning of frequency thresholds using voltage drop data in decentralized load shedding

    IEEE Trans. Power Syst.

    (2015)
  • HoseinzadehB. et al.

    Decentralized coordination of load shedding and plant protection considering high share of RESs

    IEEE Trans. Power Syst.

    (2016)
  • Cited by (15)

    • A community energy management system for smart microgrids

      2022, Electric Power Systems Research
      Citation Excerpt :

      Neither of these studies consider user satisfaction as a result of load shedding to decrease costs and increase component lifetimes. Mohammad et al. [12] explore managing user dissatisfaction and maintaining system stability by determining which loads to shed in an isolated multi micro-grid infrastructure so that shedding imposes the least cost to the system. They employ Genetic Algorithms (GA), Exponential Moving Averages (EMA) and General Algebraic Modeling Systems (GAMS) to solve the optimisation problem and compare the results with a conventional under-frequency load shedding (UFLS) scheme, which brings about non-optimal load shedding.

    • A new optimal under-voltage load shedding scheme for voltage collapse prevention in a multi-microgrid system

      2022, Electric Power Systems Research
      Citation Excerpt :

      In this section, the proposed method is applied to the simulated model of 55-bus MMG system. The MMG system is adopted from [16, 40, 41] and modified. The schematic of the test system is illustrated in Fig. 2.

    View all citing articles on Scopus
    1

    Co-authors.

    View full text