Portal de Periódicos da CAPES

Design and tests of a super real‐time simulation‐based power system real‐time decision‐making emergency control system

2020; Institution of Engineering and Technology; Volume: 14; Issue: 9 Linguagem: Inglês

10.1049/iet-gtd.2018.6812

1751-8695

Tannan Xiao, Yu Zou, Yanhui Xia, Weilin Tong, Yifan Gao, Jianquan Wang,

Real-time simulation and control systems

IET Generation, Transmission & DistributionVolume 14, Issue 9 p. 1714-1725 Research ArticleFree Access Design and tests of a super real-time simulation-based power system real-time decision-making emergency control system Tannan Xiao, Tannan Xiao College of Electrical Engineering, Zhejiang University, 38# Zheda Road, Hangzhou, People's Republic of ChinaSearch for more papers by this authorYu Zou, Yu Zou Guodian Nanjing Automation Co., Ltd., 39# Shuige Road, Nanjing, People's Republic of ChinaSearch for more papers by this authorYanhui Xia, Yanhui Xia Guodian Nanjing Automation Co., Ltd., 39# Shuige Road, Nanjing, People's Republic of ChinaSearch for more papers by this authorWeilin Tong, Weilin Tong State Grid Wuxi Power Supply Company, 12# Liangxi Road, Wuxi, People's Republic of ChinaSearch for more papers by this authorYifan Gao, Yifan Gao East China Electric Power Design Institute, 99# Henan Middle Road, Shanghai, People's Republic of ChinaSearch for more papers by this authorJianquan Wang, Corresponding Author Jianquan Wang wangjq@zju.edu.cn orcid.org/0000-0003-0019-6878 College of Electrical Engineering, Zhejiang University, 38# Zheda Road, Hangzhou, People's Republic of ChinaSearch for more papers by this author Tannan Xiao, Tannan Xiao College of Electrical Engineering, Zhejiang University, 38# Zheda Road, Hangzhou, People's Republic of ChinaSearch for more papers by this authorYu Zou, Yu Zou Guodian Nanjing Automation Co., Ltd., 39# Shuige Road, Nanjing, People's Republic of ChinaSearch for more papers by this authorYanhui Xia, Yanhui Xia Guodian Nanjing Automation Co., Ltd., 39# Shuige Road, Nanjing, People's Republic of ChinaSearch for more papers by this authorWeilin Tong, Weilin Tong State Grid Wuxi Power Supply Company, 12# Liangxi Road, Wuxi, People's Republic of ChinaSearch for more papers by this authorYifan Gao, Yifan Gao East China Electric Power Design Institute, 99# Henan Middle Road, Shanghai, People's Republic of ChinaSearch for more papers by this authorJianquan Wang, Corresponding Author Jianquan Wang wangjq@zju.edu.cn orcid.org/0000-0003-0019-6878 College of Electrical Engineering, Zhejiang University, 38# Zheda Road, Hangzhou, People's Republic of ChinaSearch for more papers by this author First published: 20 March 2020 https://doi.org/10.1049/iet-gtd.2018.6812Citations: 1AboutSectionsPDF 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 In many China's enterprise-owned power grids (EPGs), the frequently changing operation state and the poor reliability of some outdated facilities may lead to inaccuracy or even malfunctions of the traditional pre-decision-making emergency control systems, which have already caused several blackouts in EPGs. To solve this problem, a real-time decision-making emergency control system based on the detailed time-domain transient stability simulation is designed. Based on the trajectory characteristics of the stability restoration process of EPGs, the stability restoration criteria are proposed to significantly shorten the simulation time needed to judge the elimination of a certain stability problem. Fast decision-making algorithms for frequency instability, rotor angle instability and branch overload with relatively low computational cost are introduced. Control strategies can be determined within a specific time after faults are detected. The corresponding countermeasures are directly executed when the strategy is obtained. The system configuration is designed to be capable of dealing with complicated and unanticipated faults, such as cascading outages and unexpected grid splitting. The test results of a 31-node practical EPG show that the proposed algorithms and the system configuration are practical and efficient. This work is a meaningful exploration of the real-time decision-making emergency control system. 1 Introduction The objective of power system emergency control is to maintain the stable operation of the power grid when the power grid is subject to severe transient disturbances. Power grids need to be equipped with a dedicated emergency control system (ECS) to achieve this goal. The ECS can be divided into two categories, namely, the pre-decision-making ECS and the real-time decision-making ECS (RDECS). In the pre-decision-making ECS, a strategy table or a lookup table needs to be determined in advance for faults in the anticipated contingency set. Depending on different ways of generating the strategy table, there are two types of pre-decision-making ECS, namely, the offline pre-decision-making ECS and the online pre-decision-making ECS. The offline pre-decision-making ECS generates the strategy table via numerous offline calculations. In most cases, the effectiveness of the strategies is proven by the detailed time-domain transient stability simulation (TSS). The offline pre-decision-making ECS is well developed and is widely used around the world [1, 2]. The main disadvantage is that the strategy table will remain unchanged when the ECS is operating. If the current operation state or the contingency is not taken into account during the offline pre-decision-making, the strategy table may not work properly, leading to inaccuracy or even mismatch. The online pre-decision-making ECS introduces the telemetering signals and data measured by external systems such as the energy management system and the wide-area measurement system (WAMS) into the offline pre-decision-making ECS so that the strategy table can be updated by online calculations at regular intervals according to the current operation state of the power grid [3–10]. The ECS requires the decision-making process to be fast. Owing to the fact that the detailed TSS is very time-consuming, simplified TSS [3], early-termination techniques [4], direct methods [5], hybrid methods [7, 8, 10] and parallel computing [9] are used for transient stability analysis. Based on the operation state at the time, the strategy table is much more accurate and the probability of inaccuracy and mismatch is obviously reduced. However, the control strategies are still made by pre-decision-making for faults in the anticipated contingency set. When unanticipated faults such as cascading outages and unexpected grid splitting happen there still can be a problem of inaccuracy or mismatch. The RDECS only start decision-making after faults are detected. The anticipated contingency set and the strategy table are no longer needed. Therefore, there is theoretically no problem of inaccuracy or mismatch. The RDECS appears to be the best kind of ECS. However, the implementation of a RDECS is very difficult, though much work has been done in this line of research. Rapid decision-making algorithms are needed. One type of RDECS is built based on the real-time data measured by the WAMS or the multi-agent system before and after the occurrence of a fault [11–17]. The real-time data is usually used for power system dynamic equivalence [11, 17], stability margin calculation [12], trajectory prediction [13, 16] and sensitivity calculation [14, 15]. The decision-making algorithms are generally very quick because TSSs are not required. The effect of the control strategy highly depends on the locations and the number of phasor measurement units or agents. The communication delay between devices affects the decision-making process greatly and the control strategy can sometimes be unreliable. Another type tries to realise the detailed TSS-based RDECS [18, 19]. The algorithms are reliable, but multiple TSSs need to be done during the decision-making process, which is computationally intensive. The computing speed can hardly meet the requirement of real-time decision-making. Sequential controls based on the detailed TSS with a narrow simulation window may lead to excessive control actions [19]. With existing technologies, it is very hard to implement this kind of RDECS in large-scale power grids. In China, energy-consuming enterprises such as metallurgical companies usually establish enterprise-owned power grids (EPGs) to significantly reduce the electricity cost [20]. Currently, most EPGs are equipped with offline pre-decision-making ECSs. However, the frequently changing operating point and unanticipated faults in EPGs may lead to inaccuracy or even mismatch of the strategy table. Inaccurate operations or malfunctions of the offline pre-decision-making ECSs have caused several blackouts in EPGs, resulting in serious economic losses. Many enterprises have expressed the need to develop a RDECS. In this paper, exploratory research on the RDECS has been done with the support of a power equipment corporation and several practical EPGs. A real-time decision-making software (RDS) is programmed using C++ and a detailed TSS-based RDECS is designed. The remainder of this paper is organised as follows. In Section 2, based on the characteristics of EPGs, the modelling of emergency control, stability restoration criteria, and fast decision-making algorithms are presented for EPGs. The hardware and software configuration of RDECS is illustrated in Section 3. Section 4 introduces how the RDECS deals with cascading outages and unexpected gird splitting. A test system is built for a practical 31-node EPG and the test results are shown in Section 5. Some conclusions are drawn in Section 6. 2 Stability evaluation and control 2.1 Characteristics and requirements of EPGs An EPG is a micro-grid that mainly consists of thermal generators, induction motor loads, static loads etc. In general, an EPG contains <50 nodes, 20 generators and 100 branches (transmission lines or transformers). There is generally no flexible AC transmission system equipment and renewable energy facilities such as wind turbine generators and solar power plants in EPGs. The structure of such a non-utility grid is usually not well-designed and relatively fragile. The geographical area and the network scale of an EPG are generally small, i.e. the electrical distances between components are usually very short, whereas the total load is commonly large and the facilities in the EPG are often heavy-loaded. In EPGs, building a RDECS has practical value and certain feasibility. An EPG may operate in an isolated mode, which means the EPG is islanded, or in an interconnection mode, which means the EPG is connected to a large-scale utility grid. The operation state may change rapidly due to the inadequate planning and scheduling experience of the dispatchers. Considering the cost of facilities, many facilities in these EPGs are outdated and some of them are even obsolete facilities from the large-scale utility grid. As a result, faults happen rather frequently. The three common instability problems of EPGs are frequency instability, rotor angle instability and branch overload. The losses of generators or loads are the faults that are most prone to occur in EPGs. In an EPG, a single generator or a single load usually occupies a large proportion of the grid capacity and the generating reserve capacity is quite insufficient. Therefore, if the EPG is operated in the isolated mode, the large frequency excursions will easily lead to frequency instability when losing a generator or a load. Conversely, the frequency oscillation will be rather small and the EPG will remain with the frequency stability if the EPG is operated in the interconnection mode. When an EPG is operated in the isolated mode, it is almost impossible for the synchronous machines in the grid to lose synchronism, i.e. the grid will keep rotor angle stability, since the electrical distances between elements are generally very small. However, when an EPG is operated in the interconnection mode, faults will cause the angular separation between generators in the EPG and generators in the large-scale power grid to oscillate, which may lead to rotor angle instability. As mentioned before, the facilities in EPGs are often heavy loaded. The power transfer after fault clearance may cause these heavy loaded branches to overload. Especially in the interconnection mode, faults can easily cause the overload of interconnecting branches between the EPG and the large-scale power grid, which is not allowed in the interconnection mode. On the other hand, in most EPGs, the facilities are usually poorly adjustable and the dispatchers commonly lack planning and scheduling experience. As a result, generation rescheduling may not be able to mitigate branch overload in time. Therefore, countermeasures such as generator tripping and load shedding need to be executed to mitigate this stability problem. Besides the three common stability problems mentioned above, complicated faults also happen rather frequently due to the poor reliability of some outdated facilities in EPGs. The RDECS should be capable of dealing with these faults such as cascading outages and unexpected grid splitting. For a certain power grid, the countermeasures need to be executed within a specific time after faults are detected. This specific time is usually an empirical parameter and will vary with the grid. In China, this specific time is often set to 300 ms. The cooperative power equipment corporation and EPGs also require this specific time to be 300 ms. Therefore, in this paper the countermeasures should be executed within 300 ms after faults are detected. The control measures for frequency instability or rotor angle instability should be determined as quickly as possible to prevent further damage to EPGs while the transmission lines and transformers have a certain ability to withstand overload. Obviously, for the frequency instability and the rotor angle instability, the RDECS should meet this requirement. According to the field tests carried out by the cooperative power equipment corporation, the communication delay is generally no more than 20 ms, and the arc extinction time of breakers is generally no more than 100 ms. Therefore, in this paper, there will be up to (300 − 100 − 20 = 180) ms that can be used to calculate the control strategy. As for branch overload, the maximum calculation time can be longer, e.g. the countermeasures should be executed within 500 ms in this paper, i.e. the maximum calculation time is 380 ms. The offline pre-decision-making ECS has already been implemented. The number of stability control devices is fairly small and a high-speed optical communication system has already been established. The cost of building the RDECS should not be too high. The RDECS should be designed and implemented on the basis of the offline pre-decision-making system, making full use of the functions of the existing stability control devices and the high-speed real-time optical communication channels. 2.2 Modelling There are many kinds of emergency control measures. In most EPGs, generator tripping and load shedding are the only executable countermeasures. Thus in this paper, the decision-making algorithms are developed with generator tripping and load shedding as the two main countermeasures. For an EPG, which contains n nodes, branches, generators and loads, a combination of countermeasures is expected, as shown below (1)where and are the generator tripping proportion vector and the load shedding proportion vector, respectively, and are the cost coefficient matrices for generator tripping and load shedding, respectively, and and are the electrical active power vectors of generators and loads, respectively. The cost coefficient matrices and are often determined based on the unit generation cost of generators and the importance of generators and loads. These matrices demonstrate the control priorities of generators and loads. It is more inclined to pick generators and loads with a small cost coefficient when generating control strategies. Power system dynamics can be described as a group of differential-algebraic equations (DAEs) shown in (2) and (3) [21]. (2) (3)where is the state vector, whose time derivatives are equal to , is the nodal admittance matrix, is the nodal voltage vector and is the injection current vector. EPGs should maintain frequency stability and rotor angle stability. Branches should not overload. In other words, inequality constraints (4), (5) and (6) should be met. Inequality constraints (4), (5) and (6) are the stability criteria for frequency stability, rotor angle stability and branch overload, respectively. (4) (5) (6) (7) (8)where is the grid frequency under the centre of inertia and complies with (7) and (8), and are the inertia constant and the frequency of generator i, respectively, is the angular velocity under the centre of inertia, and are the lower and upper limit of frequency, respectively, is the maximum angular separation between the rotors of generators, is the maximum allowable rotor angle difference, is the overload duration vector and is the permitted overload duration vector. According to the requirements of the cooperative power equipment corporation and cooperative EPGs, is set to 180 degrees. For branch k, the rated current is and the permitted overload time is . If the load current of the branch k is greater than and this situation continues for longer than , the branch is considered overloaded. In other words, the stability criterion (6) is that no such situation described above occurs during the whole simulation. 2.3 Super real-time detailed TSS Let us define as the integration time step, as the predicted time when the execution of countermeasures completes, as the maximum simulation time and as the actual simulation time. The value of a variable at time t is represented by adding after the variable. In this paper, we let be 0.01 s, let and be 300 ms and 20 s, respectively, after the last fault is detected. Simulations may end early so that . If , the simulation is called a full-process simulation. At each integration step, , , branch load currents, and are calculated. All the constraints must be met during the whole simulation. Otherwise, the EPG is judged unstable. Super real-time detailed TSS is the foundation of RDECS. Although it is not difficult to perform super real-time simulations for EPGs with current technology because of the small scale of EPGs, the CPU time cost by a 20-second simulation is still significant compared with 180 ms. For this reason, the fast alternating solution method [22] with the implicit high precision integration method [23], the approximate minimum degree minimum number of source predecessors ordering algorithm, and the multi-path sparse vector method [24] are applied to solving DAEs. The entire simulation tool including the linear solver is programmed by using C++, which is also used in [23–26]. The 20-second simulations of four power grids are conducted with an integration step of 0.01 s by the RDS. All four power grids are small. The IEEE-39 system and the WEPRI-36 system are typical power grid examples, while the ChiP system and the HuoLH system are practical EPGs covering very small geographical areas. Twenty contingencies for each grid are simulated using the same computer. The detailed information of the computing server can be found in Section 5. The generators are modelled using detailed models including the fifth-order and the sixth-order generator models with governors, exciters and stabilisers. Load models include static load model (Z.I.P) and induction motor. The detailed information of each grid and the average CPU time (ACT) cost by one simulation are displayed in Table 1. Obviously, the simulation speed of the RDS is rapid. However, even with the above algorithms, only 3–4 full-process simulations can be done within 180 ms. Table 1. ACT (ms) cost by a 20-second detailed TSS Grid n ACT IEEE-39 39 46 10 19 52 WEPRI-36 36 42 8 10 48 ChiP 31 43 11 15 53 HuoLH 59 71 10 26 65 2.4 System workflow Evidently, (1)–(6) formulate an optimisation problem to find a combination of countermeasures with the lowest F that meets the equality and inequality constraints. For this problem, optimisation algorithms such as sequential quadratic programming [27, 28] and the interior point method [29–32] can be used. In many cases, the optimisation algorithm can obtain a solution that is globally optimal or very close to the global optimum. All the constraints can be taken into account at the same time during the optimisation process. However, for strongly non-linear systems such as power systems, it is sometimes difficult to ensure convergence. On the other hand, heuristic algorithms can also be used to solve this problem. Generally speaking, heuristic algorithms do not have the problem of convergence, but the results may deviate far from the global optimum. The primary task of ECS is to generate a combination of countermeasures that can restore stability. Meanwhile, the control cost of this combination is expected to be as small as possible. Obviously, if only 3–4 full-process simulations can be done within 180 ms, it is hardly possible to realise real-time decision-making whether using optimisation algorithms or heuristic algorithms. To this end, the RDS adopts a piecewise and successive control scheme. Only one stability problem is identified at a time. For each stability problem, a stability restoration criterion is proposed and a heuristic decision-making algorithm is adopted based on the characteristics of EPGs. The stability restoration criteria and decision-making algorithms will be demonstrated later in Section 2.5 and Section 2.6, respectively. Fig. 1 shows the flowchart of the strategy calculation process of the RDS. The process can be divided into the stability discrimination stage and the decision-making stage. Fig. 1Open in figure viewerPowerPoint Flowchart of control strategy calculation of the RDS 2.4.1 Stability discrimination stage After receiving the fault information, the detailed TSS is performed and the stability of the EPG is judged according to the stability criteria shown in (4), (5) and (6) at each integration step. If the EPG remains stable during the whole simulation, consider the EPG stable and end the control strategy calculation. If the EPG is unstable, discriminate the stability problem, end the simulation and begin the decision-making process immediately. The RDS should accurately judge the stability condition of the EPG and only one stability problem is determined for the decision-making stage at a time. Cascading outages may occur in EPGs due to the poor reliability of some components in the grid. The losses of multiple components simultaneously or successively may lead to combined stability problems. In this case, the RDS determines the stability problem according to the severity and the coupling of stability problems. Compared with the frequency instability and the rotor angle instability, the priority of branch overload is low, as branches have a certain ability to withstand overload and the calculation time allowed is much longer. Thence in the stability discrimination stage, the decision-making for branch overload is only carried out when a full-process simulation has been performed and the EPG does not lose frequency stability and rotor angle stability. There is a certain coupling between frequency instability and rotor angle instability. Before the maximum rotor angle difference violates the stability criterion, the frequency may have reached the limits since the inertia of the whole EPG is generally small and the amplitude of frequency oscillation is large. Therefore, the RDS also calculates the lowest frequency and the highest frequency at each integration time step. When the frequency stability criterion is violated, the grid is considered to lose frequency stability only if the maximum rotor angle difference and the difference between the lowest frequency and the highest frequency are small. Otherwise, there is still a risk of losing rotor angle stability so that the simulation will continue until the stability problem can be discriminated. 2.4.2 Decision-making stage In the decision-making stage, the decision-making process is carried out for one specific stability problem. If admissible controls are insufficient to restore stability, issue the control commands corresponding to the insufficient control strategy, send a warning, and end the control strategy calculation. These countermeasures cannot eliminate, but rather mitigate the stability problem. If the countermeasures can be determined within a specific time, send the control commands corresponding to the control strategy to the stability control devices for execution, and return to the stability discrimination stage to check whether there are other stability problems. These iterations of stages continue until the EPG restores stability or the admissible countermeasures become insufficient to restore stability. Each time the control commands are sent out, the admissible control set is refreshed. In the stability discrimination stage, the computational burden lies in the detailed TSS. In the decision-making stage, no matter which decision-making algorithm is adopted, multiple detailed TSSs are often required. The computational cost needed for the decision-making algorithm itself is relatively low. It can be seen that most of the computational burden required for control strategy calculation lies in the detailed TSSs. Paradoxically, even for small-scale power grids with fewer than 50 nodes, only 3–4 full-process simulations can be performed in 180 ms. Therefore, the main challenge is to improve simulation speed so that the RDS can perform as many simulations as possible within a specific time. Meanwhile, the decision-making algorithms adopted by the RDS should have small computational cost and requires a low number of TSSs. 2.5 Stability restoration criteria Parallel computing can achieve enormous speedup in the simulations of large-scale power grids [25, 26]. However, when it comes to small-scale power grids like EPGs, the parallel overhead is fairly considerable compared with the computation time, leading to a very low concurrency. The efficiency of applying parallel TSS in these small-scale power grids is generally very limited or even none in some cases. Therefore, serial simulations are carried out in the RDS. Since it is very hard to improve the speed of serial simulation, another intuitive idea is to shorten the simulation time required for stability judgment, i.e. shorten . The stability criteria are set up according to the specific conditions and engineering requirements of the EPG. For a certain power grid, the stability criteria are often clearly given. In the stability discrimination stage, the simulation ends and the EPG is judged unstable once any stability criterion is violated. Only when the stability criteria are satisfied during the whole simulation is the EPG judged stable. At this time, the simulation speed is not affected by the stability criteria. In other words, the simulation time required in the stability discrimination stage cannot be shortened. In the decision-making stage, one particular stability problem has been discriminated. Simulations are carried out to determine whether the control strategies can eliminate the stability problem. Judging the elimination of a certain stability problem by meeting the stability criterion during the whole simulation can make sure that the stability is restored. However, the decision-making process will also be very time-consuming since a full-process simulation is needed whenever a control strategy that can restore stability is determined. During the decision-making process, many sufficient strategies with relatively large control cost may be obtained. As mentioned earlier, only a very limited number of full-process simulations can be done in 180 ms even for small-scale power grids like EPGs. Therefore, the stability restoration criteria (SR criteria) are proposed to shorten the simulation time needed to judge the stability of EPGs in the decision-making stage. More specifically, the can be shortened when any sufficient control strategy is tested and the stability is restored. During the simulations carried out in the decision-making stage, the time derivatives of , , and branch load currents are calculated at each integration step to analyse the trajectory trend. Based on the trajectory characteristics, a criterion that determines whether the stability has been restored after applying a certain control strategy is selected for each stability problem. 2.5.1 Frequency Inequality constraint (4) is the frequency stability criterion. After the control measures are executed, the frequency curve will exhibit an oscillating attenuation if the frequency stability is restored. The amplitude and period of frequency oscillation are affected by many factors. For the ease of explanation, the following analysis is illustrated using the low-frequency instability as an example. Fig. 2 shows three typical frequency curves obtained by simulations with when the countermeasures are executed and the frequency stability is restored. Fig. 2Open in figure viewerPowerPoint Typical frequency curves of EPGs In Fig. 2, the characteristic SW corresponds to the short-period and weakly damped frequency oscillation characteristic. The frequency cannot reach a steady state at the end of the simulation. The characteristic SS corresponds to the sh