Caracterização e modelagem de oscilações de potência para testes em relés de distância

Abstract

Systemic disturbances that affect the load-generation balance may cause rotor angle acceleration or delay of the synchronous generators in relation to the synchronous frequency, giving rise to power swings. The impact of this phenomenon on the transmission line terminals consists essentially of fluctuations in voltage and current values, which may cause improper actuation of the distance function (ANSI 21) in protective relays. In order to prevent this situation, the phenomenon detection and decision-making is done through specific functions designed for power swing blocking (ANSI 68) and out-of-step tripping (ANSI 78). In order to analyze the occurrence of the phenomenon and enable the execution of tests of algorithms of functions 68-78, we performed the characterization of the main quantities that are affected during power swings, and the analytical and numerical equation of two models designed for voltage and current signals generation. The characterization included implementation in MATLAB® and validation in the Alternative Transients Program (ATP) of instantaneous and phasor quantities current, voltage, active power, reactive power, impedance and impedance velocity during oscillations. Power swing signals generation in non-commercial software is generally limited as time, impedance velocity, and trajectory parameters may not be available. The two models (test systems) equated in this work allow simulation of stable, unstable and in special, editable trajectory power swings, which can be parameterized by the user. The reading of the generated signals and the subsequent decision making (blocking, unblocking or tripping) is performed by the protection functions 68-78, implemented through the algorithms “Concentric Characteristics Method” and “Swing Center Voltage Method”, which are compared. The results were satisfactory in the simulated cases, since the instantaneous signals generated were able to perform the parameterized impedance path and both algorithms could correctly identify the phenomenon.