Constraint Screening for Security Analysis of Power Networks

• Published in: IEEE transactions on power systems Vol. 32; no. 3; pp. 1828 - 1838
• Main Authors: Madani, Ramtin, Lavaei, Javad, Baldick, Ross
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.05.2017; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Computational modeling; constraint screening; Emergency power systems; Generators; Load modeling; Mathematical model; Model reduction; Power systems; Security; Security-constrained optimal power flow; Uncertainty; Unit commitment; Upper bounds
• ISSN: 0885-8950; 1558-0679
• DOI: 10.1109/TPWRS.2016.2611589

Consider a general security-constrained unit commitment (SCUC) problem for an arbitrary power network. This problem includes discrete variables corresponding to commitment parameters as well as demand and generation constraints, among others. Aside from its nonconvexity, SCUC is a large-scale problem for real-world systems due to the security constraints. The main objective of this paper is to propose an algorithm to eliminate a vast majority of linear security constraints in the high-dimensional mixed-integer SCUC problem in order to arrive at an equivalent reduced-order SCUC problem. To this end, we develop a parallel and computationally cheap algorithm for finding a minimal subset of security constraints whose satisfaction guarantees the satisfaction of all security constraints. The proposed algorithm does not depend on the unknown unit commitment parameters and allows the load forecasts to be imprecise. More specifically, a low-order model of the SCUC problem is found based on the topology of the power system, given lower and upper bounds on nodal power injections (to accommodate uncertainties in loads and generation productions), and the normal and emergency line ratings. This algorithm is tested on several power systems with as many as 5500 buses, for which each set of security constraints with millions of conditions is reduced to a minimal subset with only a few hundred conditions.