Validation of a clustering algorithm for voltage stability analysis on the Bosnian electric power system
The validation of a power system clustering algorithm proposed by Glavic et al. on a real-life power system, the Bosnian electric power system, is presented. Voltage collapse is associated with a stress condition of the power system. Theoretical investigations of this complex phenomenon have resulted in many different analyses and solutions. It has been confirmed that the voltage instability problems start locally at the weakest nodes in a reduced area of the network, and the problem spreads to neighbouring nodes, resulting in a cascading phenomenon. Identification of reduced areas in a power system, with coherent behaviour of node voltages, may reduce the computational burden for voltage collapse analysis. A network clustering is proposed to reduce the computational burden associated with the analysis of voltage instability in large power networks. The algorithm is based on a method which was initially developed for coherency based system decomposition into study and external areas based on the linearised dynamic model of power system (satisfactory results have been obtained using the 30-bus New England test system). The algorithm is disturbance-independent, computationally fast and avoids computational burden because it requires only a limited number of basic arithmetic operations on the reduced Jacobian matrix of the load flow model.