Consideração de técnicas de otimização aplicadas ao problema de colapso de tensão

Abstract

This thesis studies three topics related to voltage collapse: Load margin increase, load shedding and sensitivity analysis. For each study, optimization techniques are employed. For load margin increase, a new strategy to avoid voltage collapse is proposed. With the help of interior points-based optimization algorithms, some control actions are proposed in order to minimize the voltage collapse symptoms. Even though an operating point close to voltage collapse is associated with high system loss, the literature shows that minimizing the loss may not enlarge the system margin. Rather than that, this work proposes to reduce the system loss and maximize the voltage level in a certain area of interest, yielding a better load margin with no risk of over excitation. The second part of the work is related to load shedding analysis. The literature shows that optimization techniques provide the best results for this purpose. On the other hand, such methods are time consuming. As a consequence of that, this thesis works on three different methodologies for load shedding analysis. The first one is based on the direct method, widely spread in the literature for load margin calculation, but not usually applied for load shedding studies. The second approach is based on interior points method. For this, a Lagrangian function is created using only the objective function and the active and reactive power equations. Finally, the problem of topological changes is addressed. When it comes to topological changes, the problem of building new transmission lines, construction of new plants or capacitor shunt installation are very expensive. It drives one to be careful about the planning actions recommended. For this, these actions need to be exhaustively studied. With this purpose, this thesis presents a methodology based on optimization theory. Such a methodology employs the Lagrange multipliers, obtained by a interior points algorithm, as an index which identifies effectively the best control actions to be taken in order to achieve a goal.