Clustering and supervisory voltage control in power systems. (July 2019)

• Record Type: Journal Article
• Authors: Baranwal, Mayank; Salapaka, Srinivasa
• Is Part Of: International journal of electrical power & energy systems. Volume 109(2019)
• Page Start: 641
• Page End: 651
• Publication Date: 2019-07
• Subjects: Graph clustering -- Power network -- Electrical distance -- Supervisory control -- Islanding; Electrical engineering -- Periodicals; Electric power systems -- Periodicals; Électrotechnique -- Périodiques; Réseaux électriques (Énergie) -- Périodiques; Electric power systems; Electrical engineering; Periodicals; 621.3
• DOI: 10.1016/j.ijepes.2019.02.025
• Languages: English
• ISSNs: 0142-0615

Abstract

Highlights: A new notion of electrical similarity between two buses is developed to consider both topology and physics of power networks. Scalable entropy-based clustering algorithm successfully partitions power network into loosely coupled zones. Decoupling enables design of a decentralized rule-based supervisory control scheme that relies only on "local" measurements. Combined approach of clustering & control is tested on IEEE test cases for scenarios – overloading, underloading, islanding. Our method outperforms existing methods both in terms of finding decoupled clusters & number of corrective control actions. Abstract: In this paper, the problem of decomposing a large interconnected power network into smaller loosely-coupled zones in order to facilitate easy and flexible management of power transmission systems is addressed. This decomposition enables secondary level voltage control at regional levels and controlled islanding, which can be used to prevent the spreading of large-area blackouts. An electrical power transmission system is viewed as a fully-connected, weighted directed graph, where nodes and edge-weights of the graph represent buses and quantifications of electrical similarity between any two buses, respectively. Unlike impedance or admittance based similarity measures which are largely restrictive and do not account for topology of power networks, the electrical similarity between any two buses in this work is considered in terms of their influence over the remainder of the network. In particular, the electrical similarity between two buses is quantified in terms of the respective voltage fluctuations over all the buses in the network as a result of reactive power perturbations at these buses. Moreover, quantification of electrical influence does not have significant bearing on the computational complexity since it is computed using jacobians obtained as byproducts of solving power flow equations. The resulting directed graph is then clustered into prespecified number of zones that are weakly coupled electrically using a graph-theoretic clustering algorithm. A rule-based decentralized control strategy is proposed for effective management of bus voltages in the weakly coupled zones that are obtained as a result of the clustering process. The proposed approach is then tested on IEEE test systems for applications such as supervisory voltage control and islanding, and results in excellent identification of mutually decoupled sub-networks within a large power network.