WORST-CASE IDENTIFICATION OF REACTIVE POWER MARGIN AND LOCAL WEAKNESSOF POWER-SYSTEMS

This paper has the main objectives of evaluating the worst-case VAR ma rgin of power systems and identifying the most vulnerable busbars. One possible method of achieving these objectives is to progressively inc rease system-wide reactive power (VAR) demands on power systems and to perform load flow after each VAR increase. The process is continued u p to the point where load flow diverges. This method is inefficient an d subjective, and would most likely fail to reach critical stability d ue to numerical problems. A more sophisticated method is to directly l ocate critical stability by solving an optimization problem. By evalua ting the system VAR margin, traditional optimization approaches usuall y pre-specifies a disturbance scenario, which distributes the VAR incr eases for stressing the power system. However, different disturbance s cenarios will stress the power system towards different critical point s, which will lead to different VAR margins. To estimate the system's capability to withstand VAR disturbance, the worst disturbance scenari o should be identified. Traditional optimization approaches did not us ually lead to the worst case. Worst-case identification of disturbance scenario is treated in this paper as a separate optimization problem with the VAR disturbance scenario taken as the decision variables. Apa rt from providing the worst-case VAR margin and the associated disturb ance scenario, the proposed method also highlights local weakness of t he study power system and relative effectiveness of control measures. The paper presents a systematic method of worst-case identification by incorporating genetic algorithm (GA) and nonlinear programming techni ques in two levels. In order to achieve an accurate and reliable estim ation, the method performs feasibility checks during optimization on V AR disturbance scenarios, generator reactive limits, and voltage const raints at regulated busbars. (C) 1998 Elsevier Science S.A.