Proper Splitting of Interconnected Power Systems

Abstract

Power system islanding is the last defense line to protect power grids from incidence of wide-area blackout. As a wide-area control action, power system splitting is a comprehensive decision making problem that includes different subproblems. This paper introduces a novel approach for separation of the entire power system into several stable islands in different loading levels. The proposed method combines both the dynamic and the static characteristics of interconnected power network and determines the proper splitting schemes. The presented algorithm searches for proper islanding strategy in the boundary of primary determined coherent machines using Krylov subspace method and finds the proper splitting points by transferring some of the buses in one island to another island such that total load shedding is minimized. A spanning tree-based depth first search algorithm is used to find all possible combination of transferred buses. The presented method reduces the huge initial search space of islanding strategy considering dynamic characteristics of integrated power system and reduction of search space to only boundary network. The speed of the proposed algorithm is remarkably high and can be applied for islanding the power system in real-time. The presented algorithm is applied to IEEE 118 BUS test system. Results show the robustness, effectiveness, and capability of the algorithm to determine fast and accurate proper islanding strategy. Time domain simulation of the islanding strategies confirms that all the islands which are specified by the proposed method are stable. 

Segmentation of interconnected power systems considering microgrids and the uncertainty of renewable energy sources

“…In addition, the current challenges and future trends of ICI in smart power systems are discussed. In order to identify the number and location of islands in power networks, various techniques including Ordered Binary Decision Diagram (OBDD) [2,3], Slow Coherency Method [4][5][6][7][8][9], Graph Theory [10][11][12], Spectral Clustering [13,14], and Optimization-based Methods [15][16][17] have been proposed before. This section will review them.…”

“…In the following, the Krylov projection method for the identification and selection of slow modes is described briefly [8]. Clustering based on coherency is done by considering a set of system modes and finding coherent groups of states.…”

Multi‐objective optimal preventive islanding based on stochastic backward elimination strategy considering uncertainties of loads and wind farms

“…In normalized spectral clustering procedure, use of minimum power-flow disruption as the objective function instead of minimum power mismatch improves the transient stability of the resulted islands, reduces the time complexity, and allows a real-time splitting strategy. Yingjun et al 15 used a genetic algorithm combined breadth-first search (BFS) [16][17][18] method to simplify an entire power system into a simple equivalent graph and find feasible partitioning branches. Sun et al 19 formulated an ordered binary decision diagram (OBDD)-based 3-phase approach to generate proper and accurate separating Diagonal eigenvalue matrix of J R ; P m , Output mechanical power of wind turbine in MW; λ i i th , Eigenvalue of J R ; u in , u out , u r , Wind turbine generator cut-in, cut-out and rated speeds in m/sec; ξ i i th , Column of right eigenvector matrix; P r , Wind turbine generator rated power in MW; η i i th , Row of left eigenvector matrix; Z 0 , R , X, Impedance, resistance, and reactance in per unit; β, Weighted factor between 0 and 1; μ PL i ; μ Q L i , Mean of active and reactive power of i th load; N Island , Number of islands; σ PL i ; σ Q L i , Mean and standard deviation of active and reactive powers of i th load; N G (i), Number of generators in i th island; μ p n , Mean value of n th random variable; N L (i), Number of loads in i th island; μ p l ; σ p l , Mean and standard deviation of l th input random variable; N g , Number of generators in pre-contingency system; ξ l , k , Standard location of l th input random variable in k th point; N l , Number of loads in pre-contingency system; λ l , 3 , Skewness of l th input variable; P i G g ð Þ, Active power of g th generator of i th island; M, Number of input random variables Abbreviations: CDF, Cumulative distribution function; MCS, Monte Carlo simulation; OBDD, Ordered binary decision diagram; PDF, Probability distribution function; PEM, Point estimation method; PMEA, Primary maximum expansion area; ST-BFS, Spanning tree based breadth first search; SVSI, Static voltage stability index scenarios and demonstrated its applicability using power system transient simulations software.…”

“…In normalized spectral clustering procedure, use of minimum power‐flow disruption as the objective function instead of minimum power mismatch improves the transient stability of the resulted islands, reduces the time complexity, and allows a real‐time splitting strategy. Yingjun et al used a genetic algorithm combined breadth‐first search (BFS) method to simplify an entire power system into a simple equivalent graph and find feasible partitioning branches. Sun et al formulated an ordered binary decision diagram (OBDD)‐based 3‐phase approach to generate proper and accurate separating scenarios and demonstrated its applicability using power system transient simulations software .…”

A phase‐plane trajectory vector‐based method for real‐time identification of critical machines

“…The coherent generators identified by model‐based methods stay unchanged when subjected to different disturbances under different system conditions, which is apparently improper for CMs identification. The measurement‐based methods, such as fast Fourier transform , empirical mode decomposition with the Hilbert transform , graph theory , support vector clustering , k ‐means clustering , and principal component analysis , can adapt to different disturbances and system conditions. However, they either need excessive computation, enormous training effort, or a wide time window.…”