A Multiobjective Minimax Regret Robust VAr Planning Model

Abstract

This paper proposes a risk-based mixed integer quadratically-constrained programming model for the long-term VAr planning problem. Risk aversion is included in the proposed model by means of regret-based optimization to quantify the load shedding risk because of a reactive power deficit. The expected operation and expansion costs of new installed reactive power sources and load shedding risk are jointly minimized. Uncertainty in the active and reactive load demands has been included in the model. An epsilon-constraint approach is used to characterize the optimal efficient frontier. Also, discrete tap settings of tap-changing transformers are modeled as a set of mixed integer linear equations which are embedded into an ac optimal convex power flow. Computational results are obtained from a realistic South and South-East Brazilian power system to illustrate the proposed methodology. Finally, conclusions are duly drawn.