Produção Nacional
Revisado por pares
Centralised secondary control for islanded microgrids
2020; Institution of Engineering and Technology; Volume: 14; Issue: 9 Linguagem: Inglês
10.1049/iet-rpg.2019.0731
ISSN1752-1424
AutoresBruno de Nadai Nascimento, Antônio Carlos Zambroni de Souza, Diogo Marujo, Jonattan E. Sarmiento, Cristian A. Alvez, Francisco Portelinha, João Guilherme de C. Costa,
Tópico(s)Smart Grid Energy Management
ResumoIET Renewable Power GenerationVolume 14, Issue 9 p. 1502-1511 Research ArticleFree Access Centralised secondary control for islanded microgrids Bruno de Nadai Nascimento, Corresponding Author Bruno de Nadai Nascimento brunonascimento@utfpr.edu.br orcid.org/0000-0001-5134-0919 Electrical Engineering Department, Federal University of Technology, UTFPR, Paraná, Brazil Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorAntonio Carlos Zambroni de Souza, Antonio Carlos Zambroni de Souza Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorDiogo Marujo, Diogo Marujo Electrical Engineering Department, Federal University of Technology, UTFPR, Paraná, BrazilSearch for more papers by this authorJonattan Emanuel Sarmiento, Jonattan Emanuel Sarmiento orcid.org/0000-0001-6367-0587 Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorCristian Adolfo Alvez, Cristian Adolfo Alvez orcid.org/0000-0002-8488-6085 Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorFrancisco Martins Portelinha Jr, Francisco Martins Portelinha Jr Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorJoão Guilherme de Carvalho Costa, João Guilherme de Carvalho Costa Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this author Bruno de Nadai Nascimento, Corresponding Author Bruno de Nadai Nascimento brunonascimento@utfpr.edu.br orcid.org/0000-0001-5134-0919 Electrical Engineering Department, Federal University of Technology, UTFPR, Paraná, Brazil Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorAntonio Carlos Zambroni de Souza, Antonio Carlos Zambroni de Souza Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorDiogo Marujo, Diogo Marujo Electrical Engineering Department, Federal University of Technology, UTFPR, Paraná, BrazilSearch for more papers by this authorJonattan Emanuel Sarmiento, Jonattan Emanuel Sarmiento orcid.org/0000-0001-6367-0587 Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorCristian Adolfo Alvez, Cristian Adolfo Alvez orcid.org/0000-0002-8488-6085 Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorFrancisco Martins Portelinha Jr, Francisco Martins Portelinha Jr Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this authorJoão Guilherme de Carvalho Costa, João Guilherme de Carvalho Costa Electrical Systems Institute, Federal University of Itajubá, UNIFEI, Minas Gerais, BrazilSearch for more papers by this author First published: 07 May 2020 https://doi.org/10.1049/iet-rpg.2019.0731Citations: 9AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract Electric power systems have undergone substantial changes in their operation. The higher penetration of renewable resources, demand response capability, and generators operating via droop control at the distribution level are the main features resulting in the microgrid concept. Microgrids must operate connected or islanded from the main grid, ensuring reliability and quality in the supply of energy in both operating scenarios. In this sense, the secondary control becomes essential in the system's resilience, since it is responsible for restoring the frequency and voltage within acceptable values. This study proposes a unified frequency and voltage secondary controls for microgrids operating in islanded mode. For this sake, a modification in the load flow algorithm considering a Jacobian matrix takes place, enabling a sensitivity analysis to give the adjustments in the set point of generators. The help of the Levenberg–Marquardt method improves the convergence in the modified load flow. All generators are continuously considered in this process, regarding their capabilities and relative control sensitivities concerning the operation point restoration. The proposed methodology is validated in a modified IEEE-37 node test feeder, showing the efficacy of the centralised secondary control under different scenarios of renewable generation penetration and load levels. Nomenclature Symbols frequency of the system V voltages at all buses, except the angular reference one voltage at the angular reference bus references values of frequency and terminal voltage at bus k active and reactive droop angular coefficient at bus k active and reactive powers generated at bus k active and reactive mismatches of all buses active and reactive mismatches of the system, i.e. the difference between the amount of load, including the losses, and the generation active and reactive load in each bus total reactive and active losses in the system calculated during the convergence active and reactive load calculated in each bus, calculated during the convergence angle of the voltage at bus k acceptable limits to voltage and frequency, respectively N number of buses number of generation units connected through power electronic converter in voltage source inverter mode element of the admittance matrix correspondent to admittance between buses k and n lowest voltage reference voltages update necessary range to the violated bus until the acceptable limit reference frequency update set of equations from reduced Jacobian matrix regarding reactive power damping factor to Levenberg–Marquardt method (10−4) Jacobian matrix Eye matrix the total amount of renewable generation renewable generation penetration increment in system's loading line's resistance and reactance, respectively Operators real part of a complex number imaginary part of a complex number transpose of a matrix phasor representation conjugate of a complex number dot to dot multiplication 1 Introduction Among many changes and challenges associated with the evolution of electric power systems, microgrids (MGs) have emerged as a potential solution to ally reliability and quality at the power supply. Owing to many advantages, such as renewable resources integration, electric plugin vehicles connected at the distribution level and demand response capabilities, this issue has continuously received more attention in the current literature [1-3]. A MG is commonly defined as a group of interconnected loads and sources in low and medium voltage that can operate as a single entity from the central system, enabling the operation in the connected or islanded modes. For this sake, there is a robust and efficient data processing centre integrated with a telecommunication infrastructure to interconnect and autonomously operate all agents [4, 5]. When disconnected from the main grid, the MG central controller (MGCC) plays an essential role in islanded and autonomous operation. A defined control hierarchy in the MG, composed of three layers [5-7] is presented in the literature to standardise the operation mode. It may be summarised as Primary control: characterised by the emulated response of generators, i.e. the frequency and voltage ranges in accordance to demand. Commonly, the droop control is used in this level to emulate the traditional synchronous machines response [4, 8, 9]. Secondary control: responsible for frequency and voltage restoration into acceptable limits. The smooth switching between islanded and grid-connected operation is also performed in this layer. Tertiary control: acts usually in grid-connected mode. In this layer, optimum power flow is performed considering the renewable resources, economic issues, and losses minimisation. Generally, only the primary control is treated as a local approach, and the other control layers, as central procedures implemented inside the MGCC. Hence, the primary control acts decentralised by the demand profile while MGCC performs secondary and tertiary controls to change the operation point based on a centralised decision [4, 5]. The telecommunication system plays a crucial role to provide secure and robust data exchange between MGCC and the inverter-based distribution generators (DGs) in the secondary control operation [6, 10]. Portelinha et al. [11] highlight the telecommunications' infrastructure importance inside MGs under secondary control operation, showing the necessity to consider it in the system's operation, due to its high-power consumption when operating in emergency scenarios. The performance and efficiency of a wireless communication infrastructure applied to active distribution networks are demonstrated in [12]. That work drafts the possibility to maximise the reliability, bandwidth, and coverage area of telecommunication systems with minimum power consumption under secondary control performance, essential to a centralised approach when the MGCC defines the setpoints and send to the generators by a telecommunication infrastructure. In islanded operation, MGs still present challenges related mainly to the secondary control, as exposed in [13-17]. Since the frequency is a global variable, decentralised approach via a local frequency controller can be done easier than voltage control [17]. Additionally, the authors of [18, 19] consider only frequency regulation in islanded MGs when a dynamic approach with small signal stability is taken into account through proportional–integral–derivative (PID) controllers that set new values of the reference frequency. A decentralised approach of secondary control to voltage and frequency restoration can be found in [20-22]. These references cover local voltage restoration by inverter-based MGs, however, only when the terminal voltage of the inverter is considered. The system's frequency is treated in the same way. Therefore, a centralised approach to voltage restore may be an alternative. Since the voltage is a local variable depending upon load and generation features, local control may affect the limits of the generator, mainly when a controlled bus is located far from the generators able to restore its voltage. In [23], a voltage secondary control scheme was proposed to large power systems. This reference presents a methodology based on fuzzy logic to meet the voltage limits while preserving the load margin. Similarly, Nascimento et al. [24] applied fuzzy logic to a secondary control in MGs by considering that only the most sensitive bus can restore the voltage range, the main gap in this previous reference is the increment in the reactive generation only in one bus, which can increase the system's losses resulting in generators' capability violation. In the same way, the authors of [25, 26] proposed some alternatives for the management resources problems to maximise the power supply in MGs operating in islanded conditions. A load shedding scheme to under-frequency and under-voltage corrective actions implemented inside the load flow algorithm is proposed in [27]. Furthermore, the latter addresses the necessity of a central agent able to make decisions about MG's operation, including generators' setpoints. The power flow calculation, besides being a simulation tool in MGs, also plays an essential role in the static analysis point of view. Nevertheless, some aspects of this kind of system must be taken into account when compared to traditional large power ones. In [9], the droop formulation is added into the power flow equations to voltage stability analysis, while the authors of [8, 28-32] considered the frequency and voltage magnitude of the swing bus as state variables. Incorporating the secondary control into the power flow equations constitutes the main motivation of this study. Sarmiento et al. [33] exposed a complex-valued Newton–Raphson formulation to radial systems when the system operates in grid-connected mode. Based on these references in the context of active distribution networks, incorporating the secondary control into the power flow equations constitutes the main motivation of this study. Some studies are focusing on secondary control in transmission systems [23, 34]. In this case, the idea is to control the voltage level in a specific area of interest. For this, some generators are called to contribute by generating a share of reactive power. However, the centralised secondary control proposed in this study to islanded MGs consists of adjusting the reference level in each dispatchable unit, so all generators connected in the system controls the voltage level. Both frequency and voltage must be considered in the operation point restoration so, the concept of unified can be used in the presented context. As exposed in the cited papers, generally the secondary control consists of the use of PID controllers in the sharing of active and reactive powers. In this way, the frequency and inverters terminal voltages may be restored. However, the possibility to control all load voltages of the system by simultaneous participation of the DGs still may be considered a challenge in the literature. So, this study presents a proposal to fill this gap. Table 1 summarises a comparison among the presented proposal of secondary control and some papers stated in the literature, highlighting the novelty of this study in the context of secondary control. Table 1. Comparison of the proposed secondary control Work Frequency restoration Voltage restoration Unified control Approach Precision [17] ✓ Ø Ø decentralised/PID guarantee [18, 19] ✓ Ø Ø decentralised/PID guarantee [20-22] ✓ terminal voltage ✓ decentralised/PID guarantee [23] Ø system's voltage profile Ø centralised/fuzzy logic no guarantee [24] ✓ critical bus ✓ decentralised/centralised/fuzzy logic no guarantee [34] Ø system's voltage profile Ø centralised/Impedance dompensation no guarantee presented approach ✓ system's voltage profile ✓ centralised/load flow/sensitive analysis guarantee The main contribution of this study is a proposal of a unified secondary control for balanced MGs using a load flow algorithm to give both the state and the reference values of generators. The potential contributions are highlighted as follows: Proposal of a methodology to control the voltage level of critical buses applying secondary control. In this scenario, all dispatchable units simultaneous act in the voltage regulation. Controls the voltage setpoint of the dispatchable sources, so their voltage level is adjusted while the frequency also takes place at this step aiming the system's operation point maintenance within acceptable limits. This is what characterises a unified perform of the presented proposal. It incorporates voltage and frequency reference values in the Levenberg–Marquardt method to improve the convergence process. This may be used in predictive actions to preview the secondary control operating points during an islanding operation. It centralises the set-points of generators defined by the MGCC and sends them through a telecommunications infrastructure. Moreover, the control strategy proposed here controls both the voltage setpoint in the generators and the load buses of interest in a centralised approach, which defines the setpoint through sensitive analysis from all generators. These contributions are obtained by introducing a novel Newton–Raphson approach for MGs with the help of secondary control ideas inside the problem. The whole analysis is made only in a static way, i.e. by the state of the system in an operation point. The remainder of this work is organised as follows: Section 2 presents the overview in MGs islanded operation. The power flow and the proposed secondary control approach are comprised in Section 3. Finally, the main results are discussed in Section 4, showing the efficacy of the methodology. Section 5 concludes the work. 2 Overview of MGs operation An essential element for a MG survival is the MGCC, generally located at the primary substation. The management resources and agents by the MGCC must ensure the power supply following regulated limits for as long as possible. The MGCC centrally controls the MG, so the power flow algorithm implemented in the MGCC became an essential tool. Moreover, all setpoints are defined and sent to the local controller of agents, divided into a load controller and a micro-source controller, through the communication infrastructure. According to Peças Lopes et al. [4], the amount of data exchanged between MGCC and controllers is small. This dataset includes mainly set points to local controllers and systems information, such as terminal voltages, reactive and active power levels. The primary sources are usually not directly connected to MGs because of the DC technology. The interface between the MG and the sources is done through a power electronic converter (PEC) that can operate in two modes described as [4, 5, 9] Current source inverter (CSI): in this mode, the PEC injects defined values of active and reactive power, in the same way as a current source. The latter is done by extracting the maximum power of the non-dispatchable source, e.g. photovoltaic (PV) panels. In grid-connected mode, all generators can operate in CSI mode. Voltage source inverter (VSI): PECs in VSI mode operate as voltage sources, i.e. controlling the voltages and frequency magnitudes according to the power demand, which is defined by the droop control. In this case, generally, this PEC connects only dispatchable generators or energy storage. The traditional droop control equations are shown in (1) and (2). The frequency and voltage vary linearly with active and reactive power demands, respectively [4, 5, 9, 27, 28] (1) (2) This study considers the traditional coupling, i.e. active power with frequency and reactive power with the voltage. The main idea behind this consideration is to share the active power variations in the system's operation point among the generators. However, the P/V and Q/f coupling can take place when the virtual impedance in the inverters are not considered as shown in [8]. In this context, the droop method is the most common approach to represent the primary control. The main idea behind it is to reproduce the behaviour of traditional synchronous generators, which increase active power by the reduction of the frequency. Similarly, reactive power increases vary as a function of the voltage level [14, 17]. Although the primary control is efficient and does not require any communication between the agents, it results in a steady-state frequency and voltage deviation. In this sense, the secondary control acts aiming to restore the operation point by changing the values of PECs' set points. This is equivalent to a vertical displacement of the droop equations, by changing their linear coefficients, i.e. the values of reference from and to and , respectively. Fig. 1 depicts this kind of control that moves the operation point from A to B. In the figure, the continuous and dashed lines represent the primary and secondary steps, respectively. Fig. 1Open in figure viewerPowerPoint Primary and secondary control performances 3 Proposed algorithm This section presents the main proposal of this study, i.e. a modified load flow algorithm aiming to determine voltage and frequency reference values to control the operating point of the system. Current literature is the base to implement a load flow algorithm for islanded MGs, which is presented in Section 3.1. Section 3.2 details the complementation in the load flow, aiming the adjustments to secondary control. 3.1 Power flow formulation Power flow algorithms are widely used in planning and operation analysis. In MGs, they are usually processed inside the MGCC to determine the state variables, usually carried out by Newton–Raphson's method (NRM). However, some aspects concerning MGs may be highlighted when compared to traditional systems [8, 9, 29] The frequency is not kept constant. The impedance and the load vary under the frequency. There is no swing bus. The voltage level in generation's buses may vary. Generally, there are no PV buses, since the generators in VSI mode cannot ensure the constant voltage. Based on these previous assumptions, only two types of buses are used in power flow algorithms for MGs [8] Voltage and frequency (VF) bus: represents the generators operating with PECs in VSI mode. In this way, the frequency and terminal voltage are dependent on the power load. Active and reactive powers (PQ) bus: only active and reactive powers are known. Generally, this type of bus consists of load buses, or eventually, buses with non-dispatchable sources modelled as negative loads. The model proposed in [35] is used in this study to represent PV sources in the MG. The traditional NRM must be adapted to become useful in MG's studies since these differences are crucial in the problem structure. Considering that all buses' voltages and frequency are state variables, the NRM can be written as exposed in (3). This methodology was previously proposed in [8] (3) The set of equations is formed by (4) (5) (6) (7) In (6) and (7), the losses are determined by (8) (9) In such a way, and tend to vary by the voltage and frequency ranges. So, the active and reactive power generated can be rewritten from (1) and (2), resulting in (10) (11) In (10) and (11), the subscribed k represents the kth amount generators connected through PECs in VSI mode. Linearising (4)–(7) concerning the state variables yields [8] (12) The convergence of NRM results in the state of the MG for a specific operating scenario, which represents the performance of primary control, i.e. the droop response of the generators [8]. The increase of load results in frequency and voltage reduction, as shown in Fig. 1, which may result in an infeasible operating condition from the technical and economic point of view. This process may bring the voltage level and the frequency to unpractical values. For this sake, the unified secondary frequency and voltage control method proposed in this work is presented below. 3.2 Secondary control approach In this study, all generators can help in the secondary control process. Hence, in an infeasible operation point, the MGCC defines the new frequency reference, which is the same for all generators, and the voltage references, considering the influence for each generator following the lowest voltage bus. Thus, this bus is the pilot one in the MG for voltage control purposes. This is an important feature regarding this control. Commonly, secondary control is implemented by impedance compensation [34] or the pilot bus concept. However, they are meant to transmissions systems, driving one to select adequately the generators to play reactive power redispatch. In this study, because of the MG dimension, adjusting the inverters is enough to emulate the pilot bus control. Fig. 2 shows the MG's control topology, where the dashed line indicates the data communication flow while the continuous one indicates the energy flow. Moreover, and are the generation dispatch before and after the secondary control performance, respectively. Fig. 2Open in figure viewerPowerPoint MG topology under the proposed secondary control approach For the secondary control approach, the NRM assumes a new formulation. The convergence process is maintained; however, the method considers the dependency of frequency and voltage reference values during the convergence process. This idea represents the performance of secondary control that acts to restore the operation point, maintaining the frequency and the voltages within predefined limits. Thus, the problem assumes the form shown in (13) and (14), where and take place as variables of the problem (13) (14) where is the set of power flow equations, extracted from the Jacobian matrix in relation only to reactive power terms. For this sake, the reduction used in [23] was considered. Briefly, when a MG operates under infeasible conditions, i.e. under-frequency and under-voltages, during the power flow convergence, the reference values of voltage and frequency are defined in accordance to the current operating point, aiming the simultaneous restoration within limits. Moreover, the new setpoints are results of the MGs load flow convergence. In this condition, the frequency is set in the frequency limit , and, from this, the reference frequency () can be treated as a state variable, as shown in (15) and (16) (15) (16) where (17) The remainder of partial derivatives in (12) and (16) can be found and better explained in [8], where this methodology has been initially proposed. As a way to determine the new reference voltages values, a sensitivity analysis was performed about the bus with the lowest voltage level. Hence, the step voltage necessary for secondary control is given by the difference between the limit voltage and the voltage at this bus , as described in (18) (18) The sensitivity analysis can be evaluated similarly, as described in [23, 36]. Here, the update of the reference voltage in each PEC in VSI mode () is estimated following their influences in , as exposed in (14). Thus, from (18), the sensitivity can be done by the partial derivate, as shown in (19) (19) In (19), the terms and are the reference voltages upstaged necessary to increase in the violated voltage bus, respectively. Therefore, can be written as (20). Hence, the updates in voltages reference are expressed by the terms of violated voltage bus and partial derivate from the Jacobian matrix (20) The terms and in (18) are calculated by the partial derivate of concerning , the control variable; and , the state variable, as shown in (21) and (22). is obtained from indicating the sensibility of all generators in the violated voltage bus. (21) (22) In this way, the updated values in and are done during the power flow until the convergence and complete operation point restoration within the established limits. This process is demonstrated by (23) and (24), where t indicates the tth power flow iteration (23) (24) The generators' capabilities are considered to make the secondary control approach more realistic due to operational purposes. This feature limits the participation of each generator during the secondary control performance, if necessary. Finally, the values of and are updated in the reference setpoint of the generators to the next iteration. The generators are dispatched following the new amounts of references and operation point, i.e. as in (10) and (11). This process is done until the complete convergence. Additionally, the convergence process may be compromised because of the weak coupling in active power/frequency and reactive power/voltage, a direct consequence of low X/R ratio, i.e. (25) This feature of distribution networks may drive the Jacobian matrix to the singularity, demanding a high computational effort to inverse it or in the worst case, the convergence of NRM in the traditional form is no longer guarantee. To overcome this problem, the Levenberg–Marquardt method [37] or the axis rotation [38] may be used. Hence, the update of the state variable of combined secondary and primary controls is made through the Levenberg–Marquardt method as written in (26) (26) Additionally, Fig. 3 shows the flowchart of the algorithm, evidencing the primary and secondary approaches. Fig. 3Open in figure viewerPowerPoint Proposed methodology's flowchart The step-by-step of the MG power flow with secondary control is described below. Set frequency and voltage references. Dispatch generators through (8) and (9). Solve the load flow to give all voltages and frequency of the system. This is done simultaneously with the dispatch of all generators. If the system presents under frequency or voltage, the secondary control is executed by updating the voltages' and frequency's references. After the convergence, the limits are verified, setting the generator power within the thresholds if necessary, and solving the load flow again. 4 Results and discussions A modified IEEE-37 node test feeder is used to validate the proposed methodology. Some considerations are done to turn this system into an islanded balanced one [39]. Additionally, the imbalance among the phases generally does not exceed critical values and the voltage control is made through the positive sequence value [8, 27, 28, 39], which assures the accuracy in a representation of an equivalent balanced system. Brazilian electrical procedures, for example, do not allow distribution networks under more than
Referência(s)